1libpng-manual.txt - A description on how to use and modify libpng
2
3 libpng version 1.6.22beta03 - February 19, 2016
4 Updated and distributed by Glenn Randers-Pehrson
5 <glennrp at users.sourceforge.net>
6 Copyright (c) 1998-2016 Glenn Randers-Pehrson
7
8 This document is released under the libpng license.
9 For conditions of distribution and use, see the disclaimer
10 and license in png.h
11
12 Based on:
13
14 libpng versions 0.97, January 1998, through 1.6.22beta03 - February 19, 2016
15 Updated and distributed by Glenn Randers-Pehrson
16 Copyright (c) 1998-2016 Glenn Randers-Pehrson
17
18 libpng 1.0 beta 6 - version 0.96 - May 28, 1997
19 Updated and distributed by Andreas Dilger
20 Copyright (c) 1996, 1997 Andreas Dilger
21
22 libpng 1.0 beta 2 - version 0.88 - January 26, 1996
23 For conditions of distribution and use, see copyright
24 notice in png.h. Copyright (c) 1995, 1996 Guy Eric
25 Schalnat, Group 42, Inc.
26
27 Updated/rewritten per request in the libpng FAQ
28 Copyright (c) 1995, 1996 Frank J. T. Wojcik
29 December 18, 1995 & January 20, 1996
30
31 TABLE OF CONTENTS
32
33    I. Introduction
34   II. Structures
35  III. Reading
36   IV. Writing
37    V. Simplified API
38   VI. Modifying/Customizing libpng
39  VII. MNG support
40 VIII. Changes to Libpng from version 0.88
41   IX. Changes to Libpng from version 1.0.x to 1.2.x
42    X. Changes to Libpng from version 1.0.x/1.2.x to 1.4.x
43   XI. Changes to Libpng from version 1.4.x to 1.5.x
44  XII. Changes to Libpng from version 1.5.x to 1.6.x
45 XIII. Detecting libpng
46  XIV. Source code repository
47   XV. Coding style
48  XVI. Y2K Compliance in libpng
49
50I. Introduction
51
52This file describes how to use and modify the PNG reference library
53(known as libpng) for your own use.  In addition to this
54file, example.c is a good starting point for using the library, as
55it is heavily commented and should include everything most people
56will need.  We assume that libpng is already installed; see the
57INSTALL file for instructions on how to configure and install libpng.
58
59For examples of libpng usage, see the files "example.c", "pngtest.c",
60and the files in the "contrib" directory, all of which are included in
61the libpng distribution.
62
63Libpng was written as a companion to the PNG specification, as a way
64of reducing the amount of time and effort it takes to support the PNG
65file format in application programs.
66
67The PNG specification (second edition), November 2003, is available as
68a W3C Recommendation and as an ISO Standard (ISO/IEC 15948:2004 (E)) at
69<http://www.w3.org/TR/2003/REC-PNG-20031110/
70The W3C and ISO documents have identical technical content.
71
72The PNG-1.2 specification is available at
73<http://png-mng.sourceforge.net/pub/png/spec/1.2/>.
74It is technically equivalent
75to the PNG specification (second edition) but has some additional material.
76
77The PNG-1.0 specification is available as RFC 2083
78<http://png-mng.sourceforge.net/pub/png/spec/1.0/> and as a
79W3C Recommendation <http://www.w3.org/TR/REC-png-961001>.
80
81Some additional chunks are described in the special-purpose public chunks
82documents at <http://www.libpng.org/pub/png/spec/register/>
83
84Other information
85about PNG, and the latest version of libpng, can be found at the PNG home
86page, <http://www.libpng.org/pub/png/>.
87
88Most users will not have to modify the library significantly; advanced
89users may want to modify it more.  All attempts were made to make it as
90complete as possible, while keeping the code easy to understand.
91Currently, this library only supports C.  Support for other languages
92is being considered.
93
94Libpng has been designed to handle multiple sessions at one time,
95to be easily modifiable, to be portable to the vast majority of
96machines (ANSI, K&R, 16-, 32-, and 64-bit) available, and to be easy
97to use.  The ultimate goal of libpng is to promote the acceptance of
98the PNG file format in whatever way possible.  While there is still
99work to be done (see the TODO file), libpng should cover the
100majority of the needs of its users.
101
102Libpng uses zlib for its compression and decompression of PNG files.
103Further information about zlib, and the latest version of zlib, can
104be found at the zlib home page, <http://zlib.net/>.
105The zlib compression utility is a general purpose utility that is
106useful for more than PNG files, and can be used without libpng.
107See the documentation delivered with zlib for more details.
108You can usually find the source files for the zlib utility wherever you
109find the libpng source files.
110
111Libpng is thread safe, provided the threads are using different
112instances of the structures.  Each thread should have its own
113png_struct and png_info instances, and thus its own image.
114Libpng does not protect itself against two threads using the
115same instance of a structure.
116
117II. Structures
118
119There are two main structures that are important to libpng, png_struct
120and png_info.  Both are internal structures that are no longer exposed
121in the libpng interface (as of libpng 1.5.0).
122
123The png_info structure is designed to provide information about the
124PNG file.  At one time, the fields of png_info were intended to be
125directly accessible to the user.  However, this tended to cause problems
126with applications using dynamically loaded libraries, and as a result
127a set of interface functions for png_info (the png_get_*() and png_set_*()
128functions) was developed, and direct access to the png_info fields was
129deprecated..
130
131The png_struct structure is the object used by the library to decode a
132single image.  As of 1.5.0 this structure is also not exposed.
133
134Almost all libpng APIs require a pointer to a png_struct as the first argument.
135Many (in particular the png_set and png_get APIs) also require a pointer
136to png_info as the second argument.  Some application visible macros
137defined in png.h designed for basic data access (reading and writing
138integers in the PNG format) don't take a png_info pointer, but it's almost
139always safe to assume that a (png_struct*) has to be passed to call an API
140function.
141
142You can have more than one png_info structure associated with an image,
143as illustrated in pngtest.c, one for information valid prior to the
144IDAT chunks and another (called "end_info" below) for things after them.
145
146The png.h header file is an invaluable reference for programming with libpng.
147And while I'm on the topic, make sure you include the libpng header file:
148
149#include <png.h>
150
151and also (as of libpng-1.5.0) the zlib header file, if you need it:
152
153#include <zlib.h>
154
155Types
156
157The png.h header file defines a number of integral types used by the
158APIs.  Most of these are fairly obvious; for example types corresponding
159to integers of particular sizes and types for passing color values.
160
161One exception is how non-integral numbers are handled.  For application
162convenience most APIs that take such numbers have C (double) arguments;
163however, internally PNG, and libpng, use 32 bit signed integers and encode
164the value by multiplying by 100,000.  As of libpng 1.5.0 a convenience
165macro PNG_FP_1 is defined in png.h along with a type (png_fixed_point)
166which is simply (png_int_32).
167
168All APIs that take (double) arguments also have a matching API that
169takes the corresponding fixed point integer arguments.  The fixed point
170API has the same name as the floating point one with "_fixed" appended.
171The actual range of values permitted in the APIs is frequently less than
172the full range of (png_fixed_point) (-21474 to +21474).  When APIs require
173a non-negative argument the type is recorded as png_uint_32 above.  Consult
174the header file and the text below for more information.
175
176Special care must be take with sCAL chunk handling because the chunk itself
177uses non-integral values encoded as strings containing decimal floating point
178numbers.  See the comments in the header file.
179
180Configuration
181
182The main header file function declarations are frequently protected by C
183preprocessing directives of the form:
184
185    #ifdef PNG_feature_SUPPORTED
186    declare-function
187    #endif
188    ...
189    #ifdef PNG_feature_SUPPORTED
190    use-function
191    #endif
192
193The library can be built without support for these APIs, although a
194standard build will have all implemented APIs.  Application programs
195should check the feature macros before using an API for maximum
196portability.  From libpng 1.5.0 the feature macros set during the build
197of libpng are recorded in the header file "pnglibconf.h" and this file
198is always included by png.h.
199
200If you don't need to change the library configuration from the default, skip to
201the next section ("Reading").
202
203Notice that some of the makefiles in the 'scripts' directory and (in 1.5.0) all
204of the build project files in the 'projects' directory simply copy
205scripts/pnglibconf.h.prebuilt to pnglibconf.h.  This means that these build
206systems do not permit easy auto-configuration of the library - they only
207support the default configuration.
208
209The easiest way to make minor changes to the libpng configuration when
210auto-configuration is supported is to add definitions to the command line
211using (typically) CPPFLAGS.  For example:
212
213CPPFLAGS=-DPNG_NO_FLOATING_ARITHMETIC
214
215will change the internal libpng math implementation for gamma correction and
216other arithmetic calculations to fixed point, avoiding the need for fast
217floating point support.  The result can be seen in the generated pnglibconf.h -
218make sure it contains the changed feature macro setting.
219
220If you need to make more extensive configuration changes - more than one or two
221feature macro settings - you can either add -DPNG_USER_CONFIG to the build
222command line and put a list of feature macro settings in pngusr.h or you can set
223DFA_XTRA (a makefile variable) to a file containing the same information in the
224form of 'option' settings.
225
226A. Changing pnglibconf.h
227
228A variety of methods exist to build libpng.  Not all of these support
229reconfiguration of pnglibconf.h.  To reconfigure pnglibconf.h it must either be
230rebuilt from scripts/pnglibconf.dfa using awk or it must be edited by hand.
231
232Hand editing is achieved by copying scripts/pnglibconf.h.prebuilt to
233pnglibconf.h and changing the lines defining the supported features, paying
234very close attention to the 'option' information in scripts/pnglibconf.dfa
235that describes those features and their requirements.  This is easy to get
236wrong.
237
238B. Configuration using DFA_XTRA
239
240Rebuilding from pnglibconf.dfa is easy if a functioning 'awk', or a later
241variant such as 'nawk' or 'gawk', is available.  The configure build will
242automatically find an appropriate awk and build pnglibconf.h.
243The scripts/pnglibconf.mak file contains a set of make rules for doing the
244same thing if configure is not used, and many of the makefiles in the scripts
245directory use this approach.
246
247When rebuilding simply write a new file containing changed options and set
248DFA_XTRA to the name of this file.  This causes the build to append the new file
249to the end of scripts/pnglibconf.dfa.  The pngusr.dfa file should contain lines
250of the following forms:
251
252everything = off
253
254This turns all optional features off.  Include it at the start of pngusr.dfa to
255make it easier to build a minimal configuration.  You will need to turn at least
256some features on afterward to enable either reading or writing code, or both.
257
258option feature on
259option feature off
260
261Enable or disable a single feature.  This will automatically enable other
262features required by a feature that is turned on or disable other features that
263require a feature which is turned off.  Conflicting settings will cause an error
264message to be emitted by awk.
265
266setting feature default value
267
268Changes the default value of setting 'feature' to 'value'.  There are a small
269number of settings listed at the top of pnglibconf.h, they are documented in the
270source code.  Most of these values have performance implications for the library
271but most of them have no visible effect on the API.  Some can also be overridden
272from the API.
273
274This method of building a customized pnglibconf.h is illustrated in
275contrib/pngminim/*.  See the "$(PNGCONF):" target in the makefile and
276pngusr.dfa in these directories.
277
278C. Configuration using PNG_USER_CONFIG
279
280If -DPNG_USER_CONFIG is added to the CPPFLAGS when pnglibconf.h is built,
281the file pngusr.h will automatically be included before the options in
282scripts/pnglibconf.dfa are processed.  Your pngusr.h file should contain only
283macro definitions turning features on or off or setting settings.
284
285Apart from the global setting "everything = off" all the options listed above
286can be set using macros in pngusr.h:
287
288#define PNG_feature_SUPPORTED
289
290is equivalent to:
291
292option feature on
293
294#define PNG_NO_feature
295
296is equivalent to:
297
298option feature off
299
300#define PNG_feature value
301
302is equivalent to:
303
304setting feature default value
305
306Notice that in both cases, pngusr.dfa and pngusr.h, the contents of the
307pngusr file you supply override the contents of scripts/pnglibconf.dfa
308
309If confusing or incomprehensible behavior results it is possible to
310examine the intermediate file pnglibconf.dfn to find the full set of
311dependency information for each setting and option.  Simply locate the
312feature in the file and read the C comments that precede it.
313
314This method is also illustrated in the contrib/pngminim/* makefiles and
315pngusr.h.
316
317III. Reading
318
319We'll now walk you through the possible functions to call when reading
320in a PNG file sequentially, briefly explaining the syntax and purpose
321of each one.  See example.c and png.h for more detail.  While
322progressive reading is covered in the next section, you will still
323need some of the functions discussed in this section to read a PNG
324file.
325
326Setup
327
328You will want to do the I/O initialization(*) before you get into libpng,
329so if it doesn't work, you don't have much to undo.  Of course, you
330will also want to insure that you are, in fact, dealing with a PNG
331file.  Libpng provides a simple check to see if a file is a PNG file.
332To use it, pass in the first 1 to 8 bytes of the file to the function
333png_sig_cmp(), and it will return 0 (false) if the bytes match the
334corresponding bytes of the PNG signature, or nonzero (true) otherwise.
335Of course, the more bytes you pass in, the greater the accuracy of the
336prediction.
337
338If you are intending to keep the file pointer open for use in libpng,
339you must ensure you don't read more than 8 bytes from the beginning
340of the file, and you also have to make a call to png_set_sig_bytes()
341with the number of bytes you read from the beginning.  Libpng will
342then only check the bytes (if any) that your program didn't read.
343
344(*): If you are not using the standard I/O functions, you will need
345to replace them with custom functions.  See the discussion under
346Customizing libpng.
347
348    FILE *fp = fopen(file_name, "rb");
349    if (!fp)
350    {
351       return (ERROR);
352    }
353
354    if (fread(header, 1, number, fp) != number)
355    {
356       return (ERROR);
357    }
358
359    is_png = !png_sig_cmp(header, 0, number);
360    if (!is_png)
361    {
362       return (NOT_PNG);
363    }
364
365Next, png_struct and png_info need to be allocated and initialized.  In
366order to ensure that the size of these structures is correct even with a
367dynamically linked libpng, there are functions to initialize and
368allocate the structures.  We also pass the library version, optional
369pointers to error handling functions, and a pointer to a data struct for
370use by the error functions, if necessary (the pointer and functions can
371be NULL if the default error handlers are to be used).  See the section
372on Changes to Libpng below regarding the old initialization functions.
373The structure allocation functions quietly return NULL if they fail to
374create the structure, so your application should check for that.
375
376    png_structp png_ptr = png_create_read_struct
377        (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
378        user_error_fn, user_warning_fn);
379
380    if (!png_ptr)
381       return (ERROR);
382
383    png_infop info_ptr = png_create_info_struct(png_ptr);
384
385    if (!info_ptr)
386    {
387       png_destroy_read_struct(&png_ptr,
388           (png_infopp)NULL, (png_infopp)NULL);
389       return (ERROR);
390    }
391
392If you want to use your own memory allocation routines,
393use a libpng that was built with PNG_USER_MEM_SUPPORTED defined, and use
394png_create_read_struct_2() instead of png_create_read_struct():
395
396    png_structp png_ptr = png_create_read_struct_2
397        (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
398        user_error_fn, user_warning_fn, (png_voidp)
399        user_mem_ptr, user_malloc_fn, user_free_fn);
400
401The error handling routines passed to png_create_read_struct()
402and the memory alloc/free routines passed to png_create_struct_2()
403are only necessary if you are not using the libpng supplied error
404handling and memory alloc/free functions.
405
406When libpng encounters an error, it expects to longjmp back
407to your routine.  Therefore, you will need to call setjmp and pass
408your png_jmpbuf(png_ptr).  If you read the file from different
409routines, you will need to update the longjmp buffer every time you enter
410a new routine that will call a png_*() function.
411
412See your documentation of setjmp/longjmp for your compiler for more
413information on setjmp/longjmp.  See the discussion on libpng error
414handling in the Customizing Libpng section below for more information
415on the libpng error handling.  If an error occurs, and libpng longjmp's
416back to your setjmp, you will want to call png_destroy_read_struct() to
417free any memory.
418
419    if (setjmp(png_jmpbuf(png_ptr)))
420    {
421       png_destroy_read_struct(&png_ptr, &info_ptr,
422           &end_info);
423       fclose(fp);
424       return (ERROR);
425    }
426
427Pass (png_infopp)NULL instead of &end_info if you didn't create
428an end_info structure.
429
430If you would rather avoid the complexity of setjmp/longjmp issues,
431you can compile libpng with PNG_NO_SETJMP, in which case
432errors will result in a call to PNG_ABORT() which defaults to abort().
433
434You can #define PNG_ABORT() to a function that does something
435more useful than abort(), as long as your function does not
436return.
437
438Now you need to set up the input code.  The default for libpng is to
439use the C function fread().  If you use this, you will need to pass a
440valid FILE * in the function png_init_io().  Be sure that the file is
441opened in binary mode.  If you wish to handle reading data in another
442way, you need not call the png_init_io() function, but you must then
443implement the libpng I/O methods discussed in the Customizing Libpng
444section below.
445
446    png_init_io(png_ptr, fp);
447
448If you had previously opened the file and read any of the signature from
449the beginning in order to see if this was a PNG file, you need to let
450libpng know that there are some bytes missing from the start of the file.
451
452    png_set_sig_bytes(png_ptr, number);
453
454You can change the zlib compression buffer size to be used while
455reading compressed data with
456
457    png_set_compression_buffer_size(png_ptr, buffer_size);
458
459where the default size is 8192 bytes.  Note that the buffer size
460is changed immediately and the buffer is reallocated immediately,
461instead of setting a flag to be acted upon later.
462
463If you want CRC errors to be handled in a different manner than
464the default, use
465
466    png_set_crc_action(png_ptr, crit_action, ancil_action);
467
468The values for png_set_crc_action() say how libpng is to handle CRC errors in
469ancillary and critical chunks, and whether to use the data contained
470therein.  Note that it is impossible to "discard" data in a critical
471chunk.
472
473Choices for (int) crit_action are
474   PNG_CRC_DEFAULT      0  error/quit
475   PNG_CRC_ERROR_QUIT   1  error/quit
476   PNG_CRC_WARN_USE     3  warn/use data
477   PNG_CRC_QUIET_USE    4  quiet/use data
478   PNG_CRC_NO_CHANGE    5  use the current value
479
480Choices for (int) ancil_action are
481   PNG_CRC_DEFAULT      0  error/quit
482   PNG_CRC_ERROR_QUIT   1  error/quit
483   PNG_CRC_WARN_DISCARD 2  warn/discard data
484   PNG_CRC_WARN_USE     3  warn/use data
485   PNG_CRC_QUIET_USE    4  quiet/use data
486   PNG_CRC_NO_CHANGE    5  use the current value
487
488Setting up callback code
489
490You can set up a callback function to handle any unknown chunks in the
491input stream. You must supply the function
492
493    read_chunk_callback(png_structp png_ptr,
494         png_unknown_chunkp chunk);
495    {
496       /* The unknown chunk structure contains your
497          chunk data, along with similar data for any other
498          unknown chunks: */
499
500           png_byte name[5];
501           png_byte *data;
502           png_size_t size;
503
504       /* Note that libpng has already taken care of
505          the CRC handling */
506
507       /* put your code here.  Search for your chunk in the
508          unknown chunk structure, process it, and return one
509          of the following: */
510
511       return (-n); /* chunk had an error */
512       return (0); /* did not recognize */
513       return (n); /* success */
514    }
515
516(You can give your function another name that you like instead of
517"read_chunk_callback")
518
519To inform libpng about your function, use
520
521    png_set_read_user_chunk_fn(png_ptr, user_chunk_ptr,
522        read_chunk_callback);
523
524This names not only the callback function, but also a user pointer that
525you can retrieve with
526
527    png_get_user_chunk_ptr(png_ptr);
528
529If you call the png_set_read_user_chunk_fn() function, then all unknown
530chunks which the callback does not handle will be saved when read.  You can
531cause them to be discarded by returning '1' ("handled") instead of '0'.  This
532behavior will change in libpng 1.7 and the default handling set by the
533png_set_keep_unknown_chunks() function, described below, will be used when the
534callback returns 0.  If you want the existing behavior you should set the global
535default to PNG_HANDLE_CHUNK_IF_SAFE now; this is compatible with all current
536versions of libpng and with 1.7.  Libpng 1.6 issues a warning if you keep the
537default, or PNG_HANDLE_CHUNK_NEVER, and the callback returns 0.
538
539At this point, you can set up a callback function that will be
540called after each row has been read, which you can use to control
541a progress meter or the like.  It's demonstrated in pngtest.c.
542You must supply a function
543
544    void read_row_callback(png_structp png_ptr,
545       png_uint_32 row, int pass);
546    {
547      /* put your code here */
548    }
549
550(You can give it another name that you like instead of "read_row_callback")
551
552To inform libpng about your function, use
553
554    png_set_read_status_fn(png_ptr, read_row_callback);
555
556When this function is called the row has already been completely processed and
557the 'row' and 'pass' refer to the next row to be handled.  For the
558non-interlaced case the row that was just handled is simply one less than the
559passed in row number, and pass will always be 0.  For the interlaced case the
560same applies unless the row value is 0, in which case the row just handled was
561the last one from one of the preceding passes.  Because interlacing may skip a
562pass you cannot be sure that the preceding pass is just 'pass-1', if you really
563need to know what the last pass is record (row,pass) from the callback and use
564the last recorded value each time.
565
566As with the user transform you can find the output row using the
567PNG_ROW_FROM_PASS_ROW macro.
568
569Unknown-chunk handling
570
571Now you get to set the way the library processes unknown chunks in the
572input PNG stream. Both known and unknown chunks will be read.  Normal
573behavior is that known chunks will be parsed into information in
574various info_ptr members while unknown chunks will be discarded. This
575behavior can be wasteful if your application will never use some known
576chunk types. To change this, you can call:
577
578    png_set_keep_unknown_chunks(png_ptr, keep,
579        chunk_list, num_chunks);
580
581    keep       - 0: default unknown chunk handling
582                 1: ignore; do not keep
583                 2: keep only if safe-to-copy
584                 3: keep even if unsafe-to-copy
585
586               You can use these definitions:
587                 PNG_HANDLE_CHUNK_AS_DEFAULT   0
588                 PNG_HANDLE_CHUNK_NEVER        1
589                 PNG_HANDLE_CHUNK_IF_SAFE      2
590                 PNG_HANDLE_CHUNK_ALWAYS       3
591
592    chunk_list - list of chunks affected (a byte string,
593                 five bytes per chunk, NULL or '\0' if
594                 num_chunks is positive; ignored if
595                 numchunks <= 0).
596
597    num_chunks - number of chunks affected; if 0, all
598                 unknown chunks are affected.  If positive,
599                 only the chunks in the list are affected,
600                 and if negative all unknown chunks and
601                 all known chunks except for the IHDR,
602                 PLTE, tRNS, IDAT, and IEND chunks are
603                 affected.
604
605Unknown chunks declared in this way will be saved as raw data onto a
606list of png_unknown_chunk structures.  If a chunk that is normally
607known to libpng is named in the list, it will be handled as unknown,
608according to the "keep" directive.  If a chunk is named in successive
609instances of png_set_keep_unknown_chunks(), the final instance will
610take precedence.  The IHDR and IEND chunks should not be named in
611chunk_list; if they are, libpng will process them normally anyway.
612If you know that your application will never make use of some particular
613chunks, use PNG_HANDLE_CHUNK_NEVER (or 1) as demonstrated below.
614
615Here is an example of the usage of png_set_keep_unknown_chunks(),
616where the private "vpAg" chunk will later be processed by a user chunk
617callback function:
618
619    png_byte vpAg[5]={118, 112,  65, 103, (png_byte) '\0'};
620
621    #if defined(PNG_UNKNOWN_CHUNKS_SUPPORTED)
622      png_byte unused_chunks[]=
623      {
624        104,  73,  83,  84, (png_byte) '\0',   /* hIST */
625        105,  84,  88, 116, (png_byte) '\0',   /* iTXt */
626        112,  67,  65,  76, (png_byte) '\0',   /* pCAL */
627        115,  67,  65,  76, (png_byte) '\0',   /* sCAL */
628        115,  80,  76,  84, (png_byte) '\0',   /* sPLT */
629        116,  73,  77,  69, (png_byte) '\0',   /* tIME */
630      };
631    #endif
632
633    ...
634
635    #if defined(PNG_UNKNOWN_CHUNKS_SUPPORTED)
636      /* ignore all unknown chunks
637       * (use global setting "2" for libpng16 and earlier):
638       */
639      png_set_keep_unknown_chunks(read_ptr, 2, NULL, 0);
640
641      /* except for vpAg: */
642      png_set_keep_unknown_chunks(read_ptr, 2, vpAg, 1);
643
644      /* also ignore unused known chunks: */
645      png_set_keep_unknown_chunks(read_ptr, 1, unused_chunks,
646         (int)(sizeof unused_chunks)/5);
647    #endif
648
649User limits
650
651The PNG specification allows the width and height of an image to be as
652large as 2^31-1 (0x7fffffff), or about 2.147 billion rows and columns.
653For safety, libpng imposes a default limit of 1 million rows and columns.
654Larger images will be rejected immediately with a png_error() call. If
655you wish to change these limits, you can use
656
657   png_set_user_limits(png_ptr, width_max, height_max);
658
659to set your own limits (libpng may reject some very wide images
660anyway because of potential buffer overflow conditions).
661
662You should put this statement after you create the PNG structure and
663before calling png_read_info(), png_read_png(), or png_process_data().
664
665When writing a PNG datastream, put this statement before calling
666png_write_info() or png_write_png().
667
668If you need to retrieve the limits that are being applied, use
669
670   width_max = png_get_user_width_max(png_ptr);
671   height_max = png_get_user_height_max(png_ptr);
672
673The PNG specification sets no limit on the number of ancillary chunks
674allowed in a PNG datastream.  By default, libpng imposes a limit of
675a total of 1000 sPLT, tEXt, iTXt, zTXt, and unknown chunks to be stored.
676If you have set up both info_ptr and end_info_ptr, the limit applies
677separately to each.  You can change the limit on the total number of such
678chunks that will be stored, with
679
680   png_set_chunk_cache_max(png_ptr, user_chunk_cache_max);
681
682where 0x7fffffffL means unlimited.  You can retrieve this limit with
683
684   chunk_cache_max = png_get_chunk_cache_max(png_ptr);
685
686Libpng imposes a limit of 8 Megabytes (8,000,000 bytes) on the amount of
687memory that a compressed chunk other than IDAT can occupy, when decompressed.
688You can change this limit with
689
690   png_set_chunk_malloc_max(png_ptr, user_chunk_malloc_max);
691
692and you can retrieve the limit with
693
694   chunk_malloc_max = png_get_chunk_malloc_max(png_ptr);
695
696Any chunks that would cause either of these limits to be exceeded will
697be ignored.
698
699Information about your system
700
701If you intend to display the PNG or to incorporate it in other image data you
702need to tell libpng information about your display or drawing surface so that
703libpng can convert the values in the image to match the display.
704
705From libpng-1.5.4 this information can be set before reading the PNG file
706header.  In earlier versions png_set_gamma() existed but behaved incorrectly if
707called before the PNG file header had been read and png_set_alpha_mode() did not
708exist.
709
710If you need to support versions prior to libpng-1.5.4 test the version number
711as illustrated below using "PNG_LIBPNG_VER >= 10504" and follow the procedures
712described in the appropriate manual page.
713
714You give libpng the encoding expected by your system expressed as a 'gamma'
715value.  You can also specify a default encoding for the PNG file in
716case the required information is missing from the file.  By default libpng
717assumes that the PNG data matches your system, to keep this default call:
718
719   png_set_gamma(png_ptr, screen_gamma, output_gamma);
720
721or you can use the fixed point equivalent:
722
723   png_set_gamma_fixed(png_ptr, PNG_FP_1*screen_gamma,
724      PNG_FP_1*output_gamma);
725
726If you don't know the gamma for your system it is probably 2.2 - a good
727approximation to the IEC standard for display systems (sRGB).  If images are
728too contrasty or washed out you got the value wrong - check your system
729documentation!
730
731Many systems permit the system gamma to be changed via a lookup table in the
732display driver, a few systems, including older Macs, change the response by
733default.  As of 1.5.4 three special values are available to handle common
734situations:
735
736   PNG_DEFAULT_sRGB: Indicates that the system conforms to the
737                     IEC 61966-2-1 standard.  This matches almost
738                     all systems.
739   PNG_GAMMA_MAC_18: Indicates that the system is an older
740                     (pre Mac OS 10.6) Apple Macintosh system with
741                     the default settings.
742   PNG_GAMMA_LINEAR: Just the fixed point value for 1.0 - indicates
743                     that the system expects data with no gamma
744                     encoding.
745
746You would use the linear (unencoded) value if you need to process the pixel
747values further because this avoids the need to decode and re-encode each
748component value whenever arithmetic is performed.  A lot of graphics software
749uses linear values for this reason, often with higher precision component values
750to preserve overall accuracy.
751
752
753The output_gamma value expresses how to decode the output values, not how
754they are encoded.  The values used correspond to the normal numbers used to
755describe the overall gamma of a computer display system; for example 2.2 for
756an sRGB conformant system.  The values are scaled by 100000 in the _fixed
757version of the API (so 220000 for sRGB.)
758
759The inverse of the value is always used to provide a default for the PNG file
760encoding if it has no gAMA chunk and if png_set_gamma() has not been called
761to override the PNG gamma information.
762
763When the ALPHA_OPTIMIZED mode is selected the output gamma is used to encode
764opaque pixels however pixels with lower alpha values are not encoded,
765regardless of the output gamma setting.
766
767When the standard Porter Duff handling is requested with mode 1 the output
768encoding is set to be linear and the output_gamma value is only relevant
769as a default for input data that has no gamma information.  The linear output
770encoding will be overridden if png_set_gamma() is called - the results may be
771highly unexpected!
772
773The following numbers are derived from the sRGB standard and the research
774behind it.  sRGB is defined to be approximated by a PNG gAMA chunk value of
7750.45455 (1/2.2) for PNG.  The value implicitly includes any viewing
776correction required to take account of any differences in the color
777environment of the original scene and the intended display environment; the
778value expresses how to *decode* the image for display, not how the original
779data was *encoded*.
780
781sRGB provides a peg for the PNG standard by defining a viewing environment.
782sRGB itself, and earlier TV standards, actually use a more complex transform
783(a linear portion then a gamma 2.4 power law) than PNG can express.  (PNG is
784limited to simple power laws.)  By saying that an image for direct display on
785an sRGB conformant system should be stored with a gAMA chunk value of 45455
786(11.3.3.2 and 11.3.3.5 of the ISO PNG specification) the PNG specification
787makes it possible to derive values for other display systems and
788environments.
789
790The Mac value is deduced from the sRGB based on an assumption that the actual
791extra viewing correction used in early Mac display systems was implemented as
792a power 1.45 lookup table.
793
794Any system where a programmable lookup table is used or where the behavior of
795the final display device characteristics can be changed requires system
796specific code to obtain the current characteristic.  However this can be
797difficult and most PNG gamma correction only requires an approximate value.
798
799By default, if png_set_alpha_mode() is not called, libpng assumes that all
800values are unencoded, linear, values and that the output device also has a
801linear characteristic.  This is only very rarely correct - it is invariably
802better to call png_set_alpha_mode() with PNG_DEFAULT_sRGB than rely on the
803default if you don't know what the right answer is!
804
805The special value PNG_GAMMA_MAC_18 indicates an older Mac system (pre Mac OS
80610.6) which used a correction table to implement a somewhat lower gamma on an
807otherwise sRGB system.
808
809Both these values are reserved (not simple gamma values) in order to allow
810more precise correction internally in the future.
811
812NOTE: the values can be passed to either the fixed or floating
813point APIs, but the floating point API will also accept floating point
814values.
815
816The second thing you may need to tell libpng about is how your system handles
817alpha channel information.  Some, but not all, PNG files contain an alpha
818channel.  To display these files correctly you need to compose the data onto a
819suitable background, as described in the PNG specification.
820
821Libpng only supports composing onto a single color (using png_set_background;
822see below).  Otherwise you must do the composition yourself and, in this case,
823you may need to call png_set_alpha_mode:
824
825   #if PNG_LIBPNG_VER >= 10504
826      png_set_alpha_mode(png_ptr, mode, screen_gamma);
827   #else
828      png_set_gamma(png_ptr, screen_gamma, 1.0/screen_gamma);
829   #endif
830
831The screen_gamma value is the same as the argument to png_set_gamma; however,
832how it affects the output depends on the mode.  png_set_alpha_mode() sets the
833file gamma default to 1/screen_gamma, so normally you don't need to call
834png_set_gamma.  If you need different defaults call png_set_gamma() before
835png_set_alpha_mode() - if you call it after it will override the settings made
836by png_set_alpha_mode().
837
838The mode is as follows:
839
840    PNG_ALPHA_PNG: The data is encoded according to the PNG
841specification.  Red, green and blue, or gray, components are
842gamma encoded color values and are not premultiplied by the
843alpha value.  The alpha value is a linear measure of the
844contribution of the pixel to the corresponding final output pixel.
845
846You should normally use this format if you intend to perform
847color correction on the color values; most, maybe all, color
848correction software has no handling for the alpha channel and,
849anyway, the math to handle pre-multiplied component values is
850unnecessarily complex.
851
852Before you do any arithmetic on the component values you need
853to remove the gamma encoding and multiply out the alpha
854channel.  See the PNG specification for more detail.  It is
855important to note that when an image with an alpha channel is
856scaled, linear encoded, pre-multiplied component values must
857be used!
858
859The remaining modes assume you don't need to do any further color correction or
860that if you do, your color correction software knows all about alpha (it
861probably doesn't!).  They 'associate' the alpha with the color information by
862storing color channel values that have been scaled by the alpha.  The
863advantage is that the color channels can be resampled (the image can be
864scaled) in this form.  The disadvantage is that normal practice is to store
865linear, not (gamma) encoded, values and this requires 16-bit channels for
866still images rather than the 8-bit channels that are just about sufficient if
867gamma encoding is used.  In addition all non-transparent pixel values,
868including completely opaque ones, must be gamma encoded to produce the final
869image.  These are the 'STANDARD', 'ASSOCIATED' or 'PREMULTIPLIED' modes
870described below (the latter being the two common names for associated alpha
871color channels). Note that PNG files always contain non-associated color
872channels; png_set_alpha_mode() with one of the modes causes the decoder to
873convert the pixels to an associated form before returning them to your
874application.
875
876Since it is not necessary to perform arithmetic on opaque color values so
877long as they are not to be resampled and are in the final color space it is
878possible to optimize the handling of alpha by storing the opaque pixels in
879the PNG format (adjusted for the output color space) while storing partially
880opaque pixels in the standard, linear, format.  The accuracy required for
881standard alpha composition is relatively low, because the pixels are
882isolated, therefore typically the accuracy loss in storing 8-bit linear
883values is acceptable.  (This is not true if the alpha channel is used to
884simulate transparency over large areas - use 16 bits or the PNG mode in
885this case!)  This is the 'OPTIMIZED' mode.  For this mode a pixel is
886treated as opaque only if the alpha value is equal to the maximum value.
887
888    PNG_ALPHA_STANDARD:  The data libpng produces is encoded in the
889standard way assumed by most correctly written graphics software.
890The gamma encoding will be removed by libpng and the
891linear component values will be pre-multiplied by the
892alpha channel.
893
894With this format the final image must be re-encoded to
895match the display gamma before the image is displayed.
896If your system doesn't do that, yet still seems to
897perform arithmetic on the pixels without decoding them,
898it is broken - check out the modes below.
899
900With PNG_ALPHA_STANDARD libpng always produces linear
901component values, whatever screen_gamma you supply.  The
902screen_gamma value is, however, used as a default for
903the file gamma if the PNG file has no gamma information.
904
905If you call png_set_gamma() after png_set_alpha_mode() you
906will override the linear encoding.  Instead the
907pre-multiplied pixel values will be gamma encoded but
908the alpha channel will still be linear.  This may
909actually match the requirements of some broken software,
910but it is unlikely.
911
912While linear 8-bit data is often used it has
913insufficient precision for any image with a reasonable
914dynamic range.  To avoid problems, and if your software
915supports it, use png_set_expand_16() to force all
916components to 16 bits.
917
918    PNG_ALPHA_OPTIMIZED: This mode is the same as PNG_ALPHA_STANDARD
919except that completely opaque pixels are gamma encoded according to
920the screen_gamma value.  Pixels with alpha less than 1.0
921will still have linear components.
922
923Use this format if you have control over your
924compositing software and so don't do other arithmetic
925(such as scaling) on the data you get from libpng.  Your
926compositing software can simply copy opaque pixels to
927the output but still has linear values for the
928non-opaque pixels.
929
930In normal compositing, where the alpha channel encodes
931partial pixel coverage (as opposed to broad area
932translucency), the inaccuracies of the 8-bit
933representation of non-opaque pixels are irrelevant.
934
935You can also try this format if your software is broken;
936it might look better.
937
938    PNG_ALPHA_BROKEN: This is PNG_ALPHA_STANDARD; however, all component
939values, including the alpha channel are gamma encoded.  This is
940broken because, in practice, no implementation that uses this choice
941correctly undoes the encoding before handling alpha composition.  Use this
942choice only if other serious errors in the software or hardware you use
943mandate it.  In most cases of broken software or hardware the bug in the
944final display manifests as a subtle halo around composited parts of the
945image.  You may not even perceive this as a halo; the composited part of
946the image may simply appear separate from the background, as though it had
947been cut out of paper and pasted on afterward.
948
949If you don't have to deal with bugs in software or hardware, or if you can fix
950them, there are three recommended ways of using png_set_alpha_mode():
951
952   png_set_alpha_mode(png_ptr, PNG_ALPHA_PNG,
953       screen_gamma);
954
955You can do color correction on the result (libpng does not currently
956support color correction internally).  When you handle the alpha channel
957you need to undo the gamma encoding and multiply out the alpha.
958
959   png_set_alpha_mode(png_ptr, PNG_ALPHA_STANDARD,
960       screen_gamma);
961   png_set_expand_16(png_ptr);
962
963If you are using the high level interface, don't call png_set_expand_16();
964instead pass PNG_TRANSFORM_EXPAND_16 to the interface.
965
966With this mode you can't do color correction, but you can do arithmetic,
967including composition and scaling, on the data without further processing.
968
969   png_set_alpha_mode(png_ptr, PNG_ALPHA_OPTIMIZED,
970       screen_gamma);
971
972You can avoid the expansion to 16-bit components with this mode, but you
973lose the ability to scale the image or perform other linear arithmetic.
974All you can do is compose the result onto a matching output.  Since this
975mode is libpng-specific you also need to write your own composition
976software.
977
978The following are examples of calls to png_set_alpha_mode to achieve the
979required overall gamma correction and, where necessary, alpha
980premultiplication.
981
982    png_set_alpha_mode(pp, PNG_ALPHA_PNG, PNG_DEFAULT_sRGB);
983
984This is the default libpng handling of the alpha channel - it is not
985pre-multiplied into the color components.  In addition the call states
986that the output is for a sRGB system and causes all PNG files without gAMA
987chunks to be assumed to be encoded using sRGB.
988
989    png_set_alpha_mode(pp, PNG_ALPHA_PNG, PNG_GAMMA_MAC);
990
991In this case the output is assumed to be something like an sRGB conformant
992display preceeded by a power-law lookup table of power 1.45.  This is how
993early Mac systems behaved.
994
995    png_set_alpha_mode(pp, PNG_ALPHA_STANDARD, PNG_GAMMA_LINEAR);
996
997This is the classic Jim Blinn approach and will work in academic
998environments where everything is done by the book.  It has the shortcoming
999of assuming that input PNG data with no gamma information is linear - this
1000is unlikely to be correct unless the PNG files where generated locally.
1001Most of the time the output precision will be so low as to show
1002significant banding in dark areas of the image.
1003
1004    png_set_expand_16(pp);
1005    png_set_alpha_mode(pp, PNG_ALPHA_STANDARD, PNG_DEFAULT_sRGB);
1006
1007This is a somewhat more realistic Jim Blinn inspired approach.  PNG files
1008are assumed to have the sRGB encoding if not marked with a gamma value and
1009the output is always 16 bits per component.  This permits accurate scaling
1010and processing of the data.  If you know that your input PNG files were
1011generated locally you might need to replace PNG_DEFAULT_sRGB with the
1012correct value for your system.
1013
1014    png_set_alpha_mode(pp, PNG_ALPHA_OPTIMIZED, PNG_DEFAULT_sRGB);
1015
1016If you just need to composite the PNG image onto an existing background
1017and if you control the code that does this you can use the optimization
1018setting.  In this case you just copy completely opaque pixels to the
1019output.  For pixels that are not completely transparent (you just skip
1020those) you do the composition math using png_composite or png_composite_16
1021below then encode the resultant 8-bit or 16-bit values to match the output
1022encoding.
1023
1024    Other cases
1025
1026If neither the PNG nor the standard linear encoding work for you because
1027of the software or hardware you use then you have a big problem.  The PNG
1028case will probably result in halos around the image.  The linear encoding
1029will probably result in a washed out, too bright, image (it's actually too
1030contrasty.)  Try the ALPHA_OPTIMIZED mode above - this will probably
1031substantially reduce the halos.  Alternatively try:
1032
1033    png_set_alpha_mode(pp, PNG_ALPHA_BROKEN, PNG_DEFAULT_sRGB);
1034
1035This option will also reduce the halos, but there will be slight dark
1036halos round the opaque parts of the image where the background is light.
1037In the OPTIMIZED mode the halos will be light halos where the background
1038is dark.  Take your pick - the halos are unavoidable unless you can get
1039your hardware/software fixed!  (The OPTIMIZED approach is slightly
1040faster.)
1041
1042When the default gamma of PNG files doesn't match the output gamma.
1043If you have PNG files with no gamma information png_set_alpha_mode allows
1044you to provide a default gamma, but it also sets the ouput gamma to the
1045matching value.  If you know your PNG files have a gamma that doesn't
1046match the output you can take advantage of the fact that
1047png_set_alpha_mode always sets the output gamma but only sets the PNG
1048default if it is not already set:
1049
1050    png_set_alpha_mode(pp, PNG_ALPHA_PNG, PNG_DEFAULT_sRGB);
1051    png_set_alpha_mode(pp, PNG_ALPHA_PNG, PNG_GAMMA_MAC);
1052
1053The first call sets both the default and the output gamma values, the
1054second call overrides the output gamma without changing the default.  This
1055is easier than achieving the same effect with png_set_gamma.  You must use
1056PNG_ALPHA_PNG for the first call - internal checking in png_set_alpha will
1057fire if more than one call to png_set_alpha_mode and png_set_background is
1058made in the same read operation, however multiple calls with PNG_ALPHA_PNG
1059are ignored.
1060
1061If you don't need, or can't handle, the alpha channel you can call
1062png_set_background() to remove it by compositing against a fixed color.  Don't
1063call png_set_strip_alpha() to do this - it will leave spurious pixel values in
1064transparent parts of this image.
1065
1066   png_set_background(png_ptr, &background_color,
1067       PNG_BACKGROUND_GAMMA_SCREEN, 0, 1);
1068
1069The background_color is an RGB or grayscale value according to the data format
1070libpng will produce for you.  Because you don't yet know the format of the PNG
1071file, if you call png_set_background at this point you must arrange for the
1072format produced by libpng to always have 8-bit or 16-bit components and then
1073store the color as an 8-bit or 16-bit color as appropriate.  The color contains
1074separate gray and RGB component values, so you can let libpng produce gray or
1075RGB output according to the input format, but low bit depth grayscale images
1076must always be converted to at least 8-bit format.  (Even though low bit depth
1077grayscale images can't have an alpha channel they can have a transparent
1078color!)
1079
1080You set the transforms you need later, either as flags to the high level
1081interface or libpng API calls for the low level interface.  For reference the
1082settings and API calls required are:
1083
10848-bit values:
1085   PNG_TRANSFORM_SCALE_16 | PNG_EXPAND
1086   png_set_expand(png_ptr); png_set_scale_16(png_ptr);
1087
1088   If you must get exactly the same inaccurate results
1089   produced by default in versions prior to libpng-1.5.4,
1090   use PNG_TRANSFORM_STRIP_16 and png_set_strip_16(png_ptr)
1091   instead.
1092
109316-bit values:
1094   PNG_TRANSFORM_EXPAND_16
1095   png_set_expand_16(png_ptr);
1096
1097In either case palette image data will be expanded to RGB.  If you just want
1098color data you can add PNG_TRANSFORM_GRAY_TO_RGB or png_set_gray_to_rgb(png_ptr)
1099to the list.
1100
1101Calling png_set_background before the PNG file header is read will not work
1102prior to libpng-1.5.4.  Because the failure may result in unexpected warnings or
1103errors it is therefore much safer to call png_set_background after the head has
1104been read.  Unfortunately this means that prior to libpng-1.5.4 it cannot be
1105used with the high level interface.
1106
1107The high-level read interface
1108
1109At this point there are two ways to proceed; through the high-level
1110read interface, or through a sequence of low-level read operations.
1111You can use the high-level interface if (a) you are willing to read
1112the entire image into memory, and (b) the input transformations
1113you want to do are limited to the following set:
1114
1115    PNG_TRANSFORM_IDENTITY      No transformation
1116    PNG_TRANSFORM_SCALE_16      Strip 16-bit samples to
1117                                8-bit accurately
1118    PNG_TRANSFORM_STRIP_16      Chop 16-bit samples to
1119                                8-bit less accurately
1120    PNG_TRANSFORM_STRIP_ALPHA   Discard the alpha channel
1121    PNG_TRANSFORM_PACKING       Expand 1, 2 and 4-bit
1122                                samples to bytes
1123    PNG_TRANSFORM_PACKSWAP      Change order of packed
1124                                pixels to LSB first
1125    PNG_TRANSFORM_EXPAND        Perform set_expand()
1126    PNG_TRANSFORM_INVERT_MONO   Invert monochrome images
1127    PNG_TRANSFORM_SHIFT         Normalize pixels to the
1128                                sBIT depth
1129    PNG_TRANSFORM_BGR           Flip RGB to BGR, RGBA
1130                                to BGRA
1131    PNG_TRANSFORM_SWAP_ALPHA    Flip RGBA to ARGB or GA
1132                                to AG
1133    PNG_TRANSFORM_INVERT_ALPHA  Change alpha from opacity
1134                                to transparency
1135    PNG_TRANSFORM_SWAP_ENDIAN   Byte-swap 16-bit samples
1136    PNG_TRANSFORM_GRAY_TO_RGB   Expand grayscale samples
1137                                to RGB (or GA to RGBA)
1138    PNG_TRANSFORM_EXPAND_16     Expand samples to 16 bits
1139
1140(This excludes setting a background color, doing gamma transformation,
1141quantizing, and setting filler.)  If this is the case, simply do this:
1142
1143    png_read_png(png_ptr, info_ptr, png_transforms, NULL)
1144
1145where png_transforms is an integer containing the bitwise OR of some
1146set of transformation flags.  This call is equivalent to png_read_info(),
1147followed the set of transformations indicated by the transform mask,
1148then png_read_image(), and finally png_read_end().
1149
1150(The final parameter of this call is not yet used.  Someday it might point
1151to transformation parameters required by some future input transform.)
1152
1153You must use png_transforms and not call any png_set_transform() functions
1154when you use png_read_png().
1155
1156After you have called png_read_png(), you can retrieve the image data
1157with
1158
1159   row_pointers = png_get_rows(png_ptr, info_ptr);
1160
1161where row_pointers is an array of pointers to the pixel data for each row:
1162
1163   png_bytep row_pointers[height];
1164
1165If you know your image size and pixel size ahead of time, you can allocate
1166row_pointers prior to calling png_read_png() with
1167
1168   if (height > PNG_UINT_32_MAX/(sizeof (png_byte)))
1169      png_error (png_ptr,
1170          "Image is too tall to process in memory");
1171
1172   if (width > PNG_UINT_32_MAX/pixel_size)
1173      png_error (png_ptr,
1174          "Image is too wide to process in memory");
1175
1176   row_pointers = png_malloc(png_ptr,
1177       height*(sizeof (png_bytep)));
1178
1179   for (int i=0; i<height, i++)
1180      row_pointers[i]=NULL;  /* security precaution */
1181
1182   for (int i=0; i<height, i++)
1183      row_pointers[i]=png_malloc(png_ptr,
1184          width*pixel_size);
1185
1186   png_set_rows(png_ptr, info_ptr, &row_pointers);
1187
1188Alternatively you could allocate your image in one big block and define
1189row_pointers[i] to point into the proper places in your block.
1190
1191If you use png_set_rows(), the application is responsible for freeing
1192row_pointers (and row_pointers[i], if they were separately allocated).
1193
1194If you don't allocate row_pointers ahead of time, png_read_png() will
1195do it, and it'll be free'ed by libpng when you call png_destroy_*().
1196
1197The low-level read interface
1198
1199If you are going the low-level route, you are now ready to read all
1200the file information up to the actual image data.  You do this with a
1201call to png_read_info().
1202
1203    png_read_info(png_ptr, info_ptr);
1204
1205This will process all chunks up to but not including the image data.
1206
1207This also copies some of the data from the PNG file into the decode structure
1208for use in later transformations.  Important information copied in is:
1209
12101) The PNG file gamma from the gAMA chunk.  This overwrites the default value
1211provided by an earlier call to png_set_gamma or png_set_alpha_mode.
1212
12132) Prior to libpng-1.5.4 the background color from a bKGd chunk.  This
1214damages the information provided by an earlier call to png_set_background
1215resulting in unexpected behavior.  Libpng-1.5.4 no longer does this.
1216
12173) The number of significant bits in each component value.  Libpng uses this to
1218optimize gamma handling by reducing the internal lookup table sizes.
1219
12204) The transparent color information from a tRNS chunk.  This can be modified by
1221a later call to png_set_tRNS.
1222
1223Querying the info structure
1224
1225Functions are used to get the information from the info_ptr once it
1226has been read.  Note that these fields may not be completely filled
1227in until png_read_end() has read the chunk data following the image.
1228
1229    png_get_IHDR(png_ptr, info_ptr, &width, &height,
1230       &bit_depth, &color_type, &interlace_type,
1231       &compression_type, &filter_method);
1232
1233    width          - holds the width of the image
1234                     in pixels (up to 2^31).
1235
1236    height         - holds the height of the image
1237                     in pixels (up to 2^31).
1238
1239    bit_depth      - holds the bit depth of one of the
1240                     image channels.  (valid values are
1241                     1, 2, 4, 8, 16 and depend also on
1242                     the color_type.  See also
1243                     significant bits (sBIT) below).
1244
1245    color_type     - describes which color/alpha channels
1246                         are present.
1247                     PNG_COLOR_TYPE_GRAY
1248                        (bit depths 1, 2, 4, 8, 16)
1249                     PNG_COLOR_TYPE_GRAY_ALPHA
1250                        (bit depths 8, 16)
1251                     PNG_COLOR_TYPE_PALETTE
1252                        (bit depths 1, 2, 4, 8)
1253                     PNG_COLOR_TYPE_RGB
1254                        (bit_depths 8, 16)
1255                     PNG_COLOR_TYPE_RGB_ALPHA
1256                        (bit_depths 8, 16)
1257
1258                     PNG_COLOR_MASK_PALETTE
1259                     PNG_COLOR_MASK_COLOR
1260                     PNG_COLOR_MASK_ALPHA
1261
1262    interlace_type - (PNG_INTERLACE_NONE or
1263                     PNG_INTERLACE_ADAM7)
1264
1265    compression_type - (must be PNG_COMPRESSION_TYPE_BASE
1266                     for PNG 1.0)
1267
1268    filter_method  - (must be PNG_FILTER_TYPE_BASE
1269                     for PNG 1.0, and can also be
1270                     PNG_INTRAPIXEL_DIFFERENCING if
1271                     the PNG datastream is embedded in
1272                     a MNG-1.0 datastream)
1273
1274    Any of width, height, color_type, bit_depth,
1275    interlace_type, compression_type, or filter_method can
1276    be NULL if you are not interested in their values.
1277
1278    Note that png_get_IHDR() returns 32-bit data into
1279    the application's width and height variables.
1280    This is an unsafe situation if these are not png_uint_32
1281    variables.  In such situations, the
1282    png_get_image_width() and png_get_image_height()
1283    functions described below are safer.
1284
1285    width            = png_get_image_width(png_ptr,
1286                         info_ptr);
1287
1288    height           = png_get_image_height(png_ptr,
1289                         info_ptr);
1290
1291    bit_depth        = png_get_bit_depth(png_ptr,
1292                         info_ptr);
1293
1294    color_type       = png_get_color_type(png_ptr,
1295                         info_ptr);
1296
1297    interlace_type   = png_get_interlace_type(png_ptr,
1298                         info_ptr);
1299
1300    compression_type = png_get_compression_type(png_ptr,
1301                         info_ptr);
1302
1303    filter_method    = png_get_filter_type(png_ptr,
1304                         info_ptr);
1305
1306    channels = png_get_channels(png_ptr, info_ptr);
1307
1308    channels       - number of channels of info for the
1309                     color type (valid values are 1 (GRAY,
1310                     PALETTE), 2 (GRAY_ALPHA), 3 (RGB),
1311                     4 (RGB_ALPHA or RGB + filler byte))
1312
1313    rowbytes = png_get_rowbytes(png_ptr, info_ptr);
1314
1315    rowbytes       - number of bytes needed to hold a row
1316
1317    signature = png_get_signature(png_ptr, info_ptr);
1318
1319    signature      - holds the signature read from the
1320                     file (if any).  The data is kept in
1321                     the same offset it would be if the
1322                     whole signature were read (i.e. if an
1323                     application had already read in 4
1324                     bytes of signature before starting
1325                     libpng, the remaining 4 bytes would
1326                     be in signature[4] through signature[7]
1327                     (see png_set_sig_bytes())).
1328
1329These are also important, but their validity depends on whether the chunk
1330has been read.  The png_get_valid(png_ptr, info_ptr, PNG_INFO_<chunk>) and
1331png_get_<chunk>(png_ptr, info_ptr, ...) functions return non-zero if the
1332data has been read, or zero if it is missing.  The parameters to the
1333png_get_<chunk> are set directly if they are simple data types, or a
1334pointer into the info_ptr is returned for any complex types.
1335
1336The colorspace data from gAMA, cHRM, sRGB, iCCP, and sBIT chunks
1337is simply returned to give the application information about how the
1338image was encoded.  Libpng itself only does transformations using the file
1339gamma when combining semitransparent pixels with the background color, and,
1340since libpng-1.6.0, when converting between 8-bit sRGB and 16-bit linear pixels
1341within the simplified API.  Libpng also uses the file gamma when converting
1342RGB to gray, beginning with libpng-1.0.5, if the application calls
1343png_set_rgb_to_gray()).
1344
1345    png_get_PLTE(png_ptr, info_ptr, &palette,
1346                     &num_palette);
1347
1348    palette        - the palette for the file
1349                     (array of png_color)
1350
1351    num_palette    - number of entries in the palette
1352
1353    png_get_gAMA(png_ptr, info_ptr, &file_gamma);
1354    png_get_gAMA_fixed(png_ptr, info_ptr, &int_file_gamma);
1355
1356    file_gamma     - the gamma at which the file is
1357                     written (PNG_INFO_gAMA)
1358
1359    int_file_gamma - 100,000 times the gamma at which the
1360                     file is written
1361
1362    png_get_cHRM(png_ptr, info_ptr,  &white_x, &white_y, &red_x,
1363                     &red_y, &green_x, &green_y, &blue_x, &blue_y)
1364    png_get_cHRM_XYZ(png_ptr, info_ptr, &red_X, &red_Y, &red_Z,
1365                     &green_X, &green_Y, &green_Z, &blue_X, &blue_Y,
1366                     &blue_Z)
1367    png_get_cHRM_fixed(png_ptr, info_ptr, &int_white_x,
1368                     &int_white_y, &int_red_x, &int_red_y,
1369                     &int_green_x, &int_green_y, &int_blue_x,
1370                     &int_blue_y)
1371    png_get_cHRM_XYZ_fixed(png_ptr, info_ptr, &int_red_X, &int_red_Y,
1372                     &int_red_Z, &int_green_X, &int_green_Y,
1373                     &int_green_Z, &int_blue_X, &int_blue_Y,
1374                     &int_blue_Z)
1375
1376    {white,red,green,blue}_{x,y}
1377                     A color space encoding specified using the
1378                     chromaticities of the end points and the
1379                     white point. (PNG_INFO_cHRM)
1380
1381    {red,green,blue}_{X,Y,Z}
1382                     A color space encoding specified using the
1383                     encoding end points - the CIE tristimulus
1384                     specification of the intended color of the red,
1385                     green and blue channels in the PNG RGB data.
1386                     The white point is simply the sum of the three
1387                     end points. (PNG_INFO_cHRM)
1388
1389    png_get_sRGB(png_ptr, info_ptr, &srgb_intent);
1390
1391    srgb_intent -    the rendering intent (PNG_INFO_sRGB)
1392                     The presence of the sRGB chunk
1393                     means that the pixel data is in the
1394                     sRGB color space.  This chunk also
1395                     implies specific values of gAMA and
1396                     cHRM.
1397
1398    png_get_iCCP(png_ptr, info_ptr, &name,
1399       &compression_type, &profile, &proflen);
1400
1401    name             - The profile name.
1402
1403    compression_type - The compression type; always
1404                       PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
1405                       You may give NULL to this argument to
1406                       ignore it.
1407
1408    profile          - International Color Consortium color
1409                       profile data. May contain NULs.
1410
1411    proflen          - length of profile data in bytes.
1412
1413    png_get_sBIT(png_ptr, info_ptr, &sig_bit);
1414
1415    sig_bit        - the number of significant bits for
1416                     (PNG_INFO_sBIT) each of the gray,
1417                     red, green, and blue channels,
1418                     whichever are appropriate for the
1419                     given color type (png_color_16)
1420
1421    png_get_tRNS(png_ptr, info_ptr, &trans_alpha,
1422                     &num_trans, &trans_color);
1423
1424    trans_alpha    - array of alpha (transparency)
1425                     entries for palette (PNG_INFO_tRNS)
1426
1427    num_trans      - number of transparent entries
1428                     (PNG_INFO_tRNS)
1429
1430    trans_color    - graylevel or color sample values of
1431                     the single transparent color for
1432                     non-paletted images (PNG_INFO_tRNS)
1433
1434    png_get_hIST(png_ptr, info_ptr, &hist);
1435                     (PNG_INFO_hIST)
1436
1437    hist           - histogram of palette (array of
1438                     png_uint_16)
1439
1440    png_get_tIME(png_ptr, info_ptr, &mod_time);
1441
1442    mod_time       - time image was last modified
1443                    (PNG_VALID_tIME)
1444
1445    png_get_bKGD(png_ptr, info_ptr, &background);
1446
1447    background     - background color (of type
1448                     png_color_16p) (PNG_VALID_bKGD)
1449                     valid 16-bit red, green and blue
1450                     values, regardless of color_type
1451
1452    num_comments   = png_get_text(png_ptr, info_ptr,
1453                     &text_ptr, &num_text);
1454
1455    num_comments   - number of comments
1456
1457    text_ptr       - array of png_text holding image
1458                     comments
1459
1460    text_ptr[i].compression - type of compression used
1461                 on "text" PNG_TEXT_COMPRESSION_NONE
1462                           PNG_TEXT_COMPRESSION_zTXt
1463                           PNG_ITXT_COMPRESSION_NONE
1464                           PNG_ITXT_COMPRESSION_zTXt
1465
1466    text_ptr[i].key   - keyword for comment.  Must contain
1467                         1-79 characters.
1468
1469    text_ptr[i].text  - text comments for current
1470                         keyword.  Can be empty.
1471
1472    text_ptr[i].text_length - length of text string,
1473                 after decompression, 0 for iTXt
1474
1475    text_ptr[i].itxt_length - length of itxt string,
1476                 after decompression, 0 for tEXt/zTXt
1477
1478    text_ptr[i].lang  - language of comment (empty
1479                         string for unknown).
1480
1481    text_ptr[i].lang_key  - keyword in UTF-8
1482                         (empty string for unknown).
1483
1484    Note that the itxt_length, lang, and lang_key
1485    members of the text_ptr structure only exist when the
1486    library is built with iTXt chunk support.  Prior to
1487    libpng-1.4.0 the library was built by default without
1488    iTXt support. Also note that when iTXt is supported,
1489    they contain NULL pointers when the "compression"
1490    field contains PNG_TEXT_COMPRESSION_NONE or
1491    PNG_TEXT_COMPRESSION_zTXt.
1492
1493    num_text       - number of comments (same as
1494                     num_comments; you can put NULL here
1495                     to avoid the duplication)
1496
1497    Note while png_set_text() will accept text, language,
1498    and translated keywords that can be NULL pointers, the
1499    structure returned by png_get_text will always contain
1500    regular zero-terminated C strings.  They might be
1501    empty strings but they will never be NULL pointers.
1502
1503    num_spalettes = png_get_sPLT(png_ptr, info_ptr,
1504       &palette_ptr);
1505
1506    num_spalettes  - number of sPLT chunks read.
1507
1508    palette_ptr    - array of palette structures holding
1509                     contents of one or more sPLT chunks
1510                     read.
1511
1512    png_get_oFFs(png_ptr, info_ptr, &offset_x, &offset_y,
1513       &unit_type);
1514
1515    offset_x       - positive offset from the left edge
1516                     of the screen (can be negative)
1517
1518    offset_y       - positive offset from the top edge
1519                     of the screen (can be negative)
1520
1521    unit_type      - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER
1522
1523    png_get_pHYs(png_ptr, info_ptr, &res_x, &res_y,
1524       &unit_type);
1525
1526    res_x          - pixels/unit physical resolution in
1527                     x direction
1528
1529    res_y          - pixels/unit physical resolution in
1530                     x direction
1531
1532    unit_type      - PNG_RESOLUTION_UNKNOWN,
1533                     PNG_RESOLUTION_METER
1534
1535    png_get_sCAL(png_ptr, info_ptr, &unit, &width,
1536       &height)
1537
1538    unit        - physical scale units (an integer)
1539
1540    width       - width of a pixel in physical scale units
1541
1542    height      - height of a pixel in physical scale units
1543                 (width and height are doubles)
1544
1545    png_get_sCAL_s(png_ptr, info_ptr, &unit, &width,
1546       &height)
1547
1548    unit        - physical scale units (an integer)
1549
1550    width       - width of a pixel in physical scale units
1551                  (expressed as a string)
1552
1553    height      - height of a pixel in physical scale units
1554                 (width and height are strings like "2.54")
1555
1556    num_unknown_chunks = png_get_unknown_chunks(png_ptr,
1557       info_ptr, &unknowns)
1558
1559    unknowns          - array of png_unknown_chunk
1560                        structures holding unknown chunks
1561
1562    unknowns[i].name  - name of unknown chunk
1563
1564    unknowns[i].data  - data of unknown chunk
1565
1566    unknowns[i].size  - size of unknown chunk's data
1567
1568    unknowns[i].location - position of chunk in file
1569
1570    The value of "i" corresponds to the order in which the
1571    chunks were read from the PNG file or inserted with the
1572    png_set_unknown_chunks() function.
1573
1574    The value of "location" is a bitwise "or" of
1575
1576         PNG_HAVE_IHDR  (0x01)
1577         PNG_HAVE_PLTE  (0x02)
1578         PNG_AFTER_IDAT (0x08)
1579
1580The data from the pHYs chunk can be retrieved in several convenient
1581forms:
1582
1583    res_x = png_get_x_pixels_per_meter(png_ptr,
1584       info_ptr)
1585
1586    res_y = png_get_y_pixels_per_meter(png_ptr,
1587       info_ptr)
1588
1589    res_x_and_y = png_get_pixels_per_meter(png_ptr,
1590       info_ptr)
1591
1592    res_x = png_get_x_pixels_per_inch(png_ptr,
1593       info_ptr)
1594
1595    res_y = png_get_y_pixels_per_inch(png_ptr,
1596       info_ptr)
1597
1598    res_x_and_y = png_get_pixels_per_inch(png_ptr,
1599       info_ptr)
1600
1601    aspect_ratio = png_get_pixel_aspect_ratio(png_ptr,
1602       info_ptr)
1603
1604    Each of these returns 0 [signifying "unknown"] if
1605       the data is not present or if res_x is 0;
1606       res_x_and_y is 0 if res_x != res_y
1607
1608    Note that because of the way the resolutions are
1609       stored internally, the inch conversions won't
1610       come out to exactly even number.  For example,
1611       72 dpi is stored as 0.28346 pixels/meter, and
1612       when this is retrieved it is 71.9988 dpi, so
1613       be sure to round the returned value appropriately
1614       if you want to display a reasonable-looking result.
1615
1616The data from the oFFs chunk can be retrieved in several convenient
1617forms:
1618
1619    x_offset = png_get_x_offset_microns(png_ptr, info_ptr);
1620
1621    y_offset = png_get_y_offset_microns(png_ptr, info_ptr);
1622
1623    x_offset = png_get_x_offset_inches(png_ptr, info_ptr);
1624
1625    y_offset = png_get_y_offset_inches(png_ptr, info_ptr);
1626
1627    Each of these returns 0 [signifying "unknown" if both
1628       x and y are 0] if the data is not present or if the
1629       chunk is present but the unit is the pixel.  The
1630       remark about inexact inch conversions applies here
1631       as well, because a value in inches can't always be
1632       converted to microns and back without some loss
1633       of precision.
1634
1635For more information, see the
1636PNG specification for chunk contents.  Be careful with trusting
1637rowbytes, as some of the transformations could increase the space
1638needed to hold a row (expand, filler, gray_to_rgb, etc.).
1639See png_read_update_info(), below.
1640
1641A quick word about text_ptr and num_text.  PNG stores comments in
1642keyword/text pairs, one pair per chunk, with no limit on the number
1643of text chunks, and a 2^31 byte limit on their size.  While there are
1644suggested keywords, there is no requirement to restrict the use to these
1645strings.  It is strongly suggested that keywords and text be sensible
1646to humans (that's the point), so don't use abbreviations.  Non-printing
1647symbols are not allowed.  See the PNG specification for more details.
1648There is also no requirement to have text after the keyword.
1649
1650Keywords should be limited to 79 Latin-1 characters without leading or
1651trailing spaces, but non-consecutive spaces are allowed within the
1652keyword.  It is possible to have the same keyword any number of times.
1653The text_ptr is an array of png_text structures, each holding a
1654pointer to a language string, a pointer to a keyword and a pointer to
1655a text string.  The text string, language code, and translated
1656keyword may be empty or NULL pointers.  The keyword/text
1657pairs are put into the array in the order that they are received.
1658However, some or all of the text chunks may be after the image, so, to
1659make sure you have read all the text chunks, don't mess with these
1660until after you read the stuff after the image.  This will be
1661mentioned again below in the discussion that goes with png_read_end().
1662
1663Input transformations
1664
1665After you've read the header information, you can set up the library
1666to handle any special transformations of the image data.  The various
1667ways to transform the data will be described in the order that they
1668should occur.  This is important, as some of these change the color
1669type and/or bit depth of the data, and some others only work on
1670certain color types and bit depths.
1671
1672Transformations you request are ignored if they don't have any meaning for a
1673particular input data format.  However some transformations can have an effect
1674as a result of a previous transformation.  If you specify a contradictory set of
1675transformations, for example both adding and removing the alpha channel, you
1676cannot predict the final result.
1677
1678The color used for the transparency values should be supplied in the same
1679format/depth as the current image data.  It is stored in the same format/depth
1680as the image data in a tRNS chunk, so this is what libpng expects for this data.
1681
1682The color used for the background value depends on the need_expand argument as
1683described below.
1684
1685Data will be decoded into the supplied row buffers packed into bytes
1686unless the library has been told to transform it into another format.
1687For example, 4 bit/pixel paletted or grayscale data will be returned
16882 pixels/byte with the leftmost pixel in the high-order bits of the byte,
1689unless png_set_packing() is called.  8-bit RGB data will be stored
1690in RGB RGB RGB format unless png_set_filler() or png_set_add_alpha()
1691is called to insert filler bytes, either before or after each RGB triplet.
1692
169316-bit RGB data will be returned RRGGBB RRGGBB, with the most significant
1694byte of the color value first, unless png_set_scale_16() is called to
1695transform it to regular RGB RGB triplets, or png_set_filler() or
1696png_set_add alpha() is called to insert two filler bytes, either before
1697or after each RRGGBB triplet.  Similarly, 8-bit or 16-bit grayscale data can
1698be modified with png_set_filler(), png_set_add_alpha(), png_set_strip_16(),
1699or png_set_scale_16().
1700
1701The following code transforms grayscale images of less than 8 to 8 bits,
1702changes paletted images to RGB, and adds a full alpha channel if there is
1703transparency information in a tRNS chunk.  This is most useful on
1704grayscale images with bit depths of 2 or 4 or if there is a multiple-image
1705viewing application that wishes to treat all images in the same way.
1706
1707    if (color_type == PNG_COLOR_TYPE_PALETTE)
1708        png_set_palette_to_rgb(png_ptr);
1709
1710    if (png_get_valid(png_ptr, info_ptr,
1711        PNG_INFO_tRNS)) png_set_tRNS_to_alpha(png_ptr);
1712
1713    if (color_type == PNG_COLOR_TYPE_GRAY &&
1714        bit_depth < 8) png_set_expand_gray_1_2_4_to_8(png_ptr);
1715
1716The first two functions are actually aliases for png_set_expand(), added
1717in libpng version 1.0.4, with the function names expanded to improve code
1718readability.  In some future version they may actually do different
1719things.
1720
1721As of libpng version 1.2.9, png_set_expand_gray_1_2_4_to_8() was
1722added.  It expands the sample depth without changing tRNS to alpha.
1723
1724As of libpng version 1.5.2, png_set_expand_16() was added.  It behaves as
1725png_set_expand(); however, the resultant channels have 16 bits rather than 8.
1726Use this when the output color or gray channels are made linear to avoid fairly
1727severe accuracy loss.
1728
1729   if (bit_depth < 16)
1730      png_set_expand_16(png_ptr);
1731
1732PNG can have files with 16 bits per channel.  If you only can handle
17338 bits per channel, this will strip the pixels down to 8-bit.
1734
1735    if (bit_depth == 16)
1736#if PNG_LIBPNG_VER >= 10504
1737       png_set_scale_16(png_ptr);
1738#else
1739       png_set_strip_16(png_ptr);
1740#endif
1741
1742(The more accurate "png_set_scale_16()" API became available in libpng version
17431.5.4).
1744
1745If you need to process the alpha channel on the image separately from the image
1746data (for example if you convert it to a bitmap mask) it is possible to have
1747libpng strip the channel leaving just RGB or gray data:
1748
1749    if (color_type & PNG_COLOR_MASK_ALPHA)
1750       png_set_strip_alpha(png_ptr);
1751
1752If you strip the alpha channel you need to find some other way of dealing with
1753the information.  If, instead, you want to convert the image to an opaque
1754version with no alpha channel use png_set_background; see below.
1755
1756As of libpng version 1.5.2, almost all useful expansions are supported, the
1757major ommissions are conversion of grayscale to indexed images (which can be
1758done trivially in the application) and conversion of indexed to grayscale (which
1759can be done by a trivial manipulation of the palette.)
1760
1761In the following table, the 01 means grayscale with depth<8, 31 means
1762indexed with depth<8, other numerals represent the color type, "T" means
1763the tRNS chunk is present, A means an alpha channel is present, and O
1764means tRNS or alpha is present but all pixels in the image are opaque.
1765
1766  FROM  01  31   0  0T  0O   2  2T  2O   3  3T  3O  4A  4O  6A  6O
1767   TO
1768   01    -  [G]  -   -   -   -   -   -   -   -   -   -   -   -   -
1769   31   [Q]  Q  [Q] [Q] [Q]  Q   Q   Q   Q   Q   Q  [Q] [Q]  Q   Q
1770    0    1   G   +   .   .   G   G   G   G   G   G   B   B  GB  GB
1771   0T    lt  Gt  t   +   .   Gt  G   G   Gt  G   G   Bt  Bt GBt GBt
1772   0O    lt  Gt  t   .   +   Gt  Gt  G   Gt  Gt  G   Bt  Bt GBt GBt
1773    2    C   P   C   C   C   +   .   .   C   -   -  CB  CB   B   B
1774   2T    Ct  -   Ct  C   C   t   +   t   -   -   -  CBt CBt  Bt  Bt
1775   2O    Ct  -   Ct  C   C   t   t   +   -   -   -  CBt CBt  Bt  Bt
1776    3   [Q]  p  [Q] [Q] [Q]  Q   Q   Q   +   .   .  [Q] [Q]  Q   Q
1777   3T   [Qt] p  [Qt][Q] [Q]  Qt  Qt  Qt  t   +   t  [Qt][Qt] Qt  Qt
1778   3O   [Qt] p  [Qt][Q] [Q]  Qt  Qt  Qt  t   t   +  [Qt][Qt] Qt  Qt
1779   4A    lA  G   A   T   T   GA  GT  GT  GA  GT  GT  +   BA  G  GBA
1780   4O    lA GBA  A   T   T   GA  GT  GT  GA  GT  GT  BA  +  GBA  G
1781   6A    CA  PA  CA  C   C   A   T  tT   PA  P   P   C  CBA  +   BA
1782   6O    CA PBA  CA  C   C   A  tT   T   PA  P   P  CBA  C   BA  +
1783
1784Within the matrix,
1785     "+" identifies entries where 'from' and 'to' are the same.
1786     "-" means the transformation is not supported.
1787     "." means nothing is necessary (a tRNS chunk can just be ignored).
1788     "t" means the transformation is obtained by png_set_tRNS.
1789     "A" means the transformation is obtained by png_set_add_alpha().
1790     "X" means the transformation is obtained by png_set_expand().
1791     "1" means the transformation is obtained by
1792         png_set_expand_gray_1_2_4_to_8() (and by png_set_expand()
1793         if there is no transparency in the original or the final
1794         format).
1795     "C" means the transformation is obtained by png_set_gray_to_rgb().
1796     "G" means the transformation is obtained by png_set_rgb_to_gray().
1797     "P" means the transformation is obtained by
1798         png_set_expand_palette_to_rgb().
1799     "p" means the transformation is obtained by png_set_packing().
1800     "Q" means the transformation is obtained by png_set_quantize().
1801     "T" means the transformation is obtained by
1802         png_set_tRNS_to_alpha().
1803     "B" means the transformation is obtained by
1804         png_set_background(), or png_strip_alpha().
1805
1806When an entry has multiple transforms listed all are required to cause the
1807right overall transformation.  When two transforms are separated by a comma
1808either will do the job.  When transforms are enclosed in [] the transform should
1809do the job but this is currently unimplemented - a different format will result
1810if the suggested transformations are used.
1811
1812In PNG files, the alpha channel in an image
1813is the level of opacity.  If you need the alpha channel in an image to
1814be the level of transparency instead of opacity, you can invert the
1815alpha channel (or the tRNS chunk data) after it's read, so that 0 is
1816fully opaque and 255 (in 8-bit or paletted images) or 65535 (in 16-bit
1817images) is fully transparent, with
1818
1819    png_set_invert_alpha(png_ptr);
1820
1821PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as
1822they can, resulting in, for example, 8 pixels per byte for 1 bit
1823files.  This code expands to 1 pixel per byte without changing the
1824values of the pixels:
1825
1826    if (bit_depth < 8)
1827       png_set_packing(png_ptr);
1828
1829PNG files have possible bit depths of 1, 2, 4, 8, and 16.  All pixels
1830stored in a PNG image have been "scaled" or "shifted" up to the next
1831higher possible bit depth (e.g. from 5 bits/sample in the range [0,31]
1832to 8 bits/sample in the range [0, 255]).  However, it is also possible
1833to convert the PNG pixel data back to the original bit depth of the
1834image.  This call reduces the pixels back down to the original bit depth:
1835
1836    png_color_8p sig_bit;
1837
1838    if (png_get_sBIT(png_ptr, info_ptr, &sig_bit))
1839       png_set_shift(png_ptr, sig_bit);
1840
1841PNG files store 3-color pixels in red, green, blue order.  This code
1842changes the storage of the pixels to blue, green, red:
1843
1844    if (color_type == PNG_COLOR_TYPE_RGB ||
1845        color_type == PNG_COLOR_TYPE_RGB_ALPHA)
1846       png_set_bgr(png_ptr);
1847
1848PNG files store RGB pixels packed into 3 or 6 bytes. This code expands them
1849into 4 or 8 bytes for windowing systems that need them in this format:
1850
1851    if (color_type == PNG_COLOR_TYPE_RGB)
1852       png_set_filler(png_ptr, filler, PNG_FILLER_BEFORE);
1853
1854where "filler" is the 8-bit or 16-bit number to fill with, and the location
1855is either PNG_FILLER_BEFORE or PNG_FILLER_AFTER, depending upon whether
1856you want the filler before the RGB or after. When filling an 8-bit pixel,
1857the least significant 8 bits of the number are used, if a 16-bit number is
1858supplied.  This transformation does not affect images that already have full
1859alpha channels.  To add an opaque alpha channel, use filler=0xffff and
1860PNG_FILLER_AFTER which will generate RGBA pixels.
1861
1862Note that png_set_filler() does not change the color type.  If you want
1863to do that, you can add a true alpha channel with
1864
1865    if (color_type == PNG_COLOR_TYPE_RGB ||
1866       color_type == PNG_COLOR_TYPE_GRAY)
1867       png_set_add_alpha(png_ptr, filler, PNG_FILLER_AFTER);
1868
1869where "filler" contains the alpha value to assign to each pixel.
1870The png_set_add_alpha() function was added in libpng-1.2.7.
1871
1872If you are reading an image with an alpha channel, and you need the
1873data as ARGB instead of the normal PNG format RGBA:
1874
1875    if (color_type == PNG_COLOR_TYPE_RGB_ALPHA)
1876       png_set_swap_alpha(png_ptr);
1877
1878For some uses, you may want a grayscale image to be represented as
1879RGB.  This code will do that conversion:
1880
1881    if (color_type == PNG_COLOR_TYPE_GRAY ||
1882        color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
1883       png_set_gray_to_rgb(png_ptr);
1884
1885Conversely, you can convert an RGB or RGBA image to grayscale or grayscale
1886with alpha.
1887
1888    if (color_type == PNG_COLOR_TYPE_RGB ||
1889        color_type == PNG_COLOR_TYPE_RGB_ALPHA)
1890       png_set_rgb_to_gray(png_ptr, error_action,
1891          double red_weight, double green_weight);
1892
1893    error_action = 1: silently do the conversion
1894
1895    error_action = 2: issue a warning if the original
1896                      image has any pixel where
1897                      red != green or red != blue
1898
1899    error_action = 3: issue an error and abort the
1900                      conversion if the original
1901                      image has any pixel where
1902                      red != green or red != blue
1903
1904    red_weight:       weight of red component
1905
1906    green_weight:     weight of green component
1907                      If either weight is negative, default
1908                      weights are used.
1909
1910In the corresponding fixed point API the red_weight and green_weight values are
1911simply scaled by 100,000:
1912
1913    png_set_rgb_to_gray(png_ptr, error_action,
1914       png_fixed_point red_weight,
1915       png_fixed_point green_weight);
1916
1917If you have set error_action = 1 or 2, you can
1918later check whether the image really was gray, after processing
1919the image rows, with the png_get_rgb_to_gray_status(png_ptr) function.
1920It will return a png_byte that is zero if the image was gray or
19211 if there were any non-gray pixels.  Background and sBIT data
1922will be silently converted to grayscale, using the green channel
1923data for sBIT, regardless of the error_action setting.
1924
1925The default values come from the PNG file cHRM chunk if present; otherwise, the
1926defaults correspond to the ITU-R recommendation 709, and also the sRGB color
1927space, as recommended in the Charles Poynton's Colour FAQ,
1928Copyright (c) 2006-11-28 Charles Poynton, in section 9:
1929
1930<http://www.poynton.com/notes/colour_and_gamma/ColorFAQ.html#RTFToC9>
1931
1932    Y = 0.2126 * R + 0.7152 * G + 0.0722 * B
1933
1934Previous versions of this document, 1998 through 2002, recommended a slightly
1935different formula:
1936
1937    Y = 0.212671 * R + 0.715160 * G + 0.072169 * B
1938
1939Libpng uses an integer approximation:
1940
1941    Y = (6968 * R + 23434 * G + 2366 * B)/32768
1942
1943The calculation is done in a linear colorspace, if the image gamma
1944can be determined.
1945
1946The png_set_background() function has been described already; it tells libpng to
1947composite images with alpha or simple transparency against the supplied
1948background color.  For compatibility with versions of libpng earlier than
1949libpng-1.5.4 it is recommended that you call the function after reading the file
1950header, even if you don't want to use the color in a bKGD chunk, if one exists.
1951
1952If the PNG file contains a bKGD chunk (PNG_INFO_bKGD valid),
1953you may use this color, or supply another color more suitable for
1954the current display (e.g., the background color from a web page).  You
1955need to tell libpng how the color is represented, both the format of the
1956component values in the color (the number of bits) and the gamma encoding of the
1957color.  The function takes two arguments, background_gamma_mode and need_expand
1958to convey this information; however, only two combinations are likely to be
1959useful:
1960
1961    png_color_16 my_background;
1962    png_color_16p image_background;
1963
1964    if (png_get_bKGD(png_ptr, info_ptr, &image_background))
1965       png_set_background(png_ptr, image_background,
1966           PNG_BACKGROUND_GAMMA_FILE, 1/*needs to be expanded*/, 1);
1967    else
1968       png_set_background(png_ptr, &my_background,
1969           PNG_BACKGROUND_GAMMA_SCREEN, 0/*do not expand*/, 1);
1970
1971The second call was described above - my_background is in the format of the
1972final, display, output produced by libpng.  Because you now know the format of
1973the PNG it is possible to avoid the need to choose either 8-bit or 16-bit
1974output and to retain palette images (the palette colors will be modified
1975appropriately and the tRNS chunk removed.)  However, if you are doing this,
1976take great care not to ask for transformations without checking first that
1977they apply!
1978
1979In the first call the background color has the original bit depth and color type
1980of the PNG file.  So, for palette images the color is supplied as a palette
1981index and for low bit greyscale images the color is a reduced bit value in
1982image_background->gray.
1983
1984If you didn't call png_set_gamma() before reading the file header, for example
1985if you need your code to remain compatible with older versions of libpng prior
1986to libpng-1.5.4, this is the place to call it.
1987
1988Do not call it if you called png_set_alpha_mode(); doing so will damage the
1989settings put in place by png_set_alpha_mode().  (If png_set_alpha_mode() is
1990supported then you can certainly do png_set_gamma() before reading the PNG
1991header.)
1992
1993This API unconditionally sets the screen and file gamma values, so it will
1994override the value in the PNG file unless it is called before the PNG file
1995reading starts.  For this reason you must always call it with the PNG file
1996value when you call it in this position:
1997
1998   if (png_get_gAMA(png_ptr, info_ptr, &file_gamma))
1999      png_set_gamma(png_ptr, screen_gamma, file_gamma);
2000
2001   else
2002      png_set_gamma(png_ptr, screen_gamma, 0.45455);
2003
2004If you need to reduce an RGB file to a paletted file, or if a paletted
2005file has more entries than will fit on your screen, png_set_quantize()
2006will do that.  Note that this is a simple match quantization that merely
2007finds the closest color available.  This should work fairly well with
2008optimized palettes, but fairly badly with linear color cubes.  If you
2009pass a palette that is larger than maximum_colors, the file will
2010reduce the number of colors in the palette so it will fit into
2011maximum_colors.  If there is a histogram, libpng will use it to make
2012more intelligent choices when reducing the palette.  If there is no
2013histogram, it may not do as good a job.
2014
2015   if (color_type & PNG_COLOR_MASK_COLOR)
2016   {
2017      if (png_get_valid(png_ptr, info_ptr,
2018          PNG_INFO_PLTE))
2019      {
2020         png_uint_16p histogram = NULL;
2021
2022         png_get_hIST(png_ptr, info_ptr,
2023             &histogram);
2024         png_set_quantize(png_ptr, palette, num_palette,
2025            max_screen_colors, histogram, 1);
2026      }
2027
2028      else
2029      {
2030         png_color std_color_cube[MAX_SCREEN_COLORS] =
2031            { ... colors ... };
2032
2033         png_set_quantize(png_ptr, std_color_cube,
2034            MAX_SCREEN_COLORS, MAX_SCREEN_COLORS,
2035            NULL,0);
2036      }
2037   }
2038
2039PNG files describe monochrome as black being zero and white being one.
2040The following code will reverse this (make black be one and white be
2041zero):
2042
2043   if (bit_depth == 1 && color_type == PNG_COLOR_TYPE_GRAY)
2044      png_set_invert_mono(png_ptr);
2045
2046This function can also be used to invert grayscale and gray-alpha images:
2047
2048   if (color_type == PNG_COLOR_TYPE_GRAY ||
2049       color_type == PNG_COLOR_TYPE_GRAY_ALPHA)
2050      png_set_invert_mono(png_ptr);
2051
2052PNG files store 16-bit pixels in network byte order (big-endian,
2053ie. most significant bits first).  This code changes the storage to the
2054other way (little-endian, i.e. least significant bits first, the
2055way PCs store them):
2056
2057    if (bit_depth == 16)
2058       png_set_swap(png_ptr);
2059
2060If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you
2061need to change the order the pixels are packed into bytes, you can use:
2062
2063    if (bit_depth < 8)
2064       png_set_packswap(png_ptr);
2065
2066Finally, you can write your own transformation function if none of
2067the existing ones meets your needs.  This is done by setting a callback
2068with
2069
2070    png_set_read_user_transform_fn(png_ptr,
2071        read_transform_fn);
2072
2073You must supply the function
2074
2075    void read_transform_fn(png_structp png_ptr, png_row_infop
2076        row_info, png_bytep data)
2077
2078See pngtest.c for a working example.  Your function will be called
2079after all of the other transformations have been processed.  Take care with
2080interlaced images if you do the interlace yourself - the width of the row is the
2081width in 'row_info', not the overall image width.
2082
2083If supported, libpng provides two information routines that you can use to find
2084where you are in processing the image:
2085
2086   png_get_current_pass_number(png_structp png_ptr);
2087   png_get_current_row_number(png_structp png_ptr);
2088
2089Don't try using these outside a transform callback - firstly they are only
2090supported if user transforms are supported, secondly they may well return
2091unexpected results unless the row is actually being processed at the moment they
2092are called.
2093
2094With interlaced
2095images the value returned is the row in the input sub-image image.  Use
2096PNG_ROW_FROM_PASS_ROW(row, pass) and PNG_COL_FROM_PASS_COL(col, pass) to
2097find the output pixel (x,y) given an interlaced sub-image pixel (row,col,pass).
2098
2099The discussion of interlace handling above contains more information on how to
2100use these values.
2101
2102You can also set up a pointer to a user structure for use by your
2103callback function, and you can inform libpng that your transform
2104function will change the number of channels or bit depth with the
2105function
2106
2107    png_set_user_transform_info(png_ptr, user_ptr,
2108        user_depth, user_channels);
2109
2110The user's application, not libpng, is responsible for allocating and
2111freeing any memory required for the user structure.
2112
2113You can retrieve the pointer via the function
2114png_get_user_transform_ptr().  For example:
2115
2116    voidp read_user_transform_ptr =
2117        png_get_user_transform_ptr(png_ptr);
2118
2119The last thing to handle is interlacing; this is covered in detail below,
2120but you must call the function here if you want libpng to handle expansion
2121of the interlaced image.
2122
2123    number_of_passes = png_set_interlace_handling(png_ptr);
2124
2125After setting the transformations, libpng can update your png_info
2126structure to reflect any transformations you've requested with this
2127call.
2128
2129    png_read_update_info(png_ptr, info_ptr);
2130
2131This is most useful to update the info structure's rowbytes
2132field so you can use it to allocate your image memory.  This function
2133will also update your palette with the correct screen_gamma and
2134background if these have been given with the calls above.  You may
2135only call png_read_update_info() once with a particular info_ptr.
2136
2137After you call png_read_update_info(), you can allocate any
2138memory you need to hold the image.  The row data is simply
2139raw byte data for all forms of images.  As the actual allocation
2140varies among applications, no example will be given.  If you
2141are allocating one large chunk, you will need to build an
2142array of pointers to each row, as it will be needed for some
2143of the functions below.
2144
2145Remember: Before you call png_read_update_info(), the png_get_*()
2146functions return the values corresponding to the original PNG image.
2147After you call png_read_update_info the values refer to the image
2148that libpng will output.  Consequently you must call all the png_set_
2149functions before you call png_read_update_info().  This is particularly
2150important for png_set_interlace_handling() - if you are going to call
2151png_read_update_info() you must call png_set_interlace_handling() before
2152it unless you want to receive interlaced output.
2153
2154Reading image data
2155
2156After you've allocated memory, you can read the image data.
2157The simplest way to do this is in one function call.  If you are
2158allocating enough memory to hold the whole image, you can just
2159call png_read_image() and libpng will read in all the image data
2160and put it in the memory area supplied.  You will need to pass in
2161an array of pointers to each row.
2162
2163This function automatically handles interlacing, so you don't
2164need to call png_set_interlace_handling() (unless you call
2165png_read_update_info()) or call this function multiple times, or any
2166of that other stuff necessary with png_read_rows().
2167
2168   png_read_image(png_ptr, row_pointers);
2169
2170where row_pointers is:
2171
2172   png_bytep row_pointers[height];
2173
2174You can point to void or char or whatever you use for pixels.
2175
2176If you don't want to read in the whole image at once, you can
2177use png_read_rows() instead.  If there is no interlacing (check
2178interlace_type == PNG_INTERLACE_NONE), this is simple:
2179
2180    png_read_rows(png_ptr, row_pointers, NULL,
2181        number_of_rows);
2182
2183where row_pointers is the same as in the png_read_image() call.
2184
2185If you are doing this just one row at a time, you can do this with
2186a single row_pointer instead of an array of row_pointers:
2187
2188    png_bytep row_pointer = row;
2189    png_read_row(png_ptr, row_pointer, NULL);
2190
2191If the file is interlaced (interlace_type != 0 in the IHDR chunk), things
2192get somewhat harder.  The only current (PNG Specification version 1.2)
2193interlacing type for PNG is (interlace_type == PNG_INTERLACE_ADAM7);
2194a somewhat complicated 2D interlace scheme, known as Adam7, that
2195breaks down an image into seven smaller images of varying size, based
2196on an 8x8 grid.  This number is defined (from libpng 1.5) as
2197PNG_INTERLACE_ADAM7_PASSES in png.h
2198
2199libpng can fill out those images or it can give them to you "as is".
2200It is almost always better to have libpng handle the interlacing for you.
2201If you want the images filled out, there are two ways to do that.  The one
2202mentioned in the PNG specification is to expand each pixel to cover
2203those pixels that have not been read yet (the "rectangle" method).
2204This results in a blocky image for the first pass, which gradually
2205smooths out as more pixels are read.  The other method is the "sparkle"
2206method, where pixels are drawn only in their final locations, with the
2207rest of the image remaining whatever colors they were initialized to
2208before the start of the read.  The first method usually looks better,
2209but tends to be slower, as there are more pixels to put in the rows.
2210
2211If, as is likely, you want libpng to expand the images, call this before
2212calling png_start_read_image() or png_read_update_info():
2213
2214    if (interlace_type == PNG_INTERLACE_ADAM7)
2215       number_of_passes
2216           = png_set_interlace_handling(png_ptr);
2217
2218This will return the number of passes needed.  Currently, this is seven,
2219but may change if another interlace type is added.  This function can be
2220called even if the file is not interlaced, where it will return one pass.
2221You then need to read the whole image 'number_of_passes' times.  Each time
2222will distribute the pixels from the current pass to the correct place in
2223the output image, so you need to supply the same rows to png_read_rows in
2224each pass.
2225
2226If you are not going to display the image after each pass, but are
2227going to wait until the entire image is read in, use the sparkle
2228effect.  This effect is faster and the end result of either method
2229is exactly the same.  If you are planning on displaying the image
2230after each pass, the "rectangle" effect is generally considered the
2231better looking one.
2232
2233If you only want the "sparkle" effect, just call png_read_rows() as
2234normal, with the third parameter NULL.  Make sure you make pass over
2235the image number_of_passes times, and you don't change the data in the
2236rows between calls.  You can change the locations of the data, just
2237not the data.  Each pass only writes the pixels appropriate for that
2238pass, and assumes the data from previous passes is still valid.
2239
2240    png_read_rows(png_ptr, row_pointers, NULL,
2241        number_of_rows);
2242
2243If you only want the first effect (the rectangles), do the same as
2244before except pass the row buffer in the third parameter, and leave
2245the second parameter NULL.
2246
2247    png_read_rows(png_ptr, NULL, row_pointers,
2248        number_of_rows);
2249
2250If you don't want libpng to handle the interlacing details, just call
2251png_read_rows() PNG_INTERLACE_ADAM7_PASSES times to read in all the images.
2252Each of the images is a valid image by itself; however, you will almost
2253certainly need to distribute the pixels from each sub-image to the
2254correct place.  This is where everything gets very tricky.
2255
2256If you want to retrieve the separate images you must pass the correct
2257number of rows to each successive call of png_read_rows().  The calculation
2258gets pretty complicated for small images, where some sub-images may
2259not even exist because either their width or height ends up zero.
2260libpng provides two macros to help you in 1.5 and later versions:
2261
2262   png_uint_32 width = PNG_PASS_COLS(image_width, pass_number);
2263   png_uint_32 height = PNG_PASS_ROWS(image_height, pass_number);
2264
2265Respectively these tell you the width and height of the sub-image
2266corresponding to the numbered pass.  'pass' is in in the range 0 to 6 -
2267this can be confusing because the specification refers to the same passes
2268as 1 to 7!  Be careful, you must check both the width and height before
2269calling png_read_rows() and not call it for that pass if either is zero.
2270
2271You can, of course, read each sub-image row by row.  If you want to
2272produce optimal code to make a pixel-by-pixel transformation of an
2273interlaced image this is the best approach; read each row of each pass,
2274transform it, and write it out to a new interlaced image.
2275
2276If you want to de-interlace the image yourself libpng provides further
2277macros to help that tell you where to place the pixels in the output image.
2278Because the interlacing scheme is rectangular - sub-image pixels are always
2279arranged on a rectangular grid - all you need to know for each pass is the
2280starting column and row in the output image of the first pixel plus the
2281spacing between each pixel.  As of libpng 1.5 there are four macros to
2282retrieve this information:
2283
2284   png_uint_32 x = PNG_PASS_START_COL(pass);
2285   png_uint_32 y = PNG_PASS_START_ROW(pass);
2286   png_uint_32 xStep = 1U << PNG_PASS_COL_SHIFT(pass);
2287   png_uint_32 yStep = 1U << PNG_PASS_ROW_SHIFT(pass);
2288
2289These allow you to write the obvious loop:
2290
2291   png_uint_32 input_y = 0;
2292   png_uint_32 output_y = PNG_PASS_START_ROW(pass);
2293
2294   while (output_y < output_image_height)
2295   {
2296      png_uint_32 input_x = 0;
2297      png_uint_32 output_x = PNG_PASS_START_COL(pass);
2298
2299      while (output_x < output_image_width)
2300      {
2301         image[output_y][output_x] =
2302             subimage[pass][input_y][input_x++];
2303
2304         output_x += xStep;
2305      }
2306
2307      ++input_y;
2308      output_y += yStep;
2309   }
2310
2311Notice that the steps between successive output rows and columns are
2312returned as shifts.  This is possible because the pixels in the subimages
2313are always a power of 2 apart - 1, 2, 4 or 8 pixels - in the original
2314image.  In practice you may need to directly calculate the output coordinate
2315given an input coordinate.  libpng provides two further macros for this
2316purpose:
2317
2318   png_uint_32 output_x = PNG_COL_FROM_PASS_COL(input_x, pass);
2319   png_uint_32 output_y = PNG_ROW_FROM_PASS_ROW(input_y, pass);
2320
2321Finally a pair of macros are provided to tell you if a particular image
2322row or column appears in a given pass:
2323
2324   int col_in_pass = PNG_COL_IN_INTERLACE_PASS(output_x, pass);
2325   int row_in_pass = PNG_ROW_IN_INTERLACE_PASS(output_y, pass);
2326
2327Bear in mind that you will probably also need to check the width and height
2328of the pass in addition to the above to be sure the pass even exists!
2329
2330With any luck you are convinced by now that you don't want to do your own
2331interlace handling.  In reality normally the only good reason for doing this
2332is if you are processing PNG files on a pixel-by-pixel basis and don't want
2333to load the whole file into memory when it is interlaced.
2334
2335libpng includes a test program, pngvalid, that illustrates reading and
2336writing of interlaced images.  If you can't get interlacing to work in your
2337code and don't want to leave it to libpng (the recommended approach), see
2338how pngvalid.c does it.
2339
2340Finishing a sequential read
2341
2342After you are finished reading the image through the
2343low-level interface, you can finish reading the file.
2344
2345If you want to use a different crc action for handling CRC errors in
2346chunks after the image data, you can call png_set_crc_action()
2347again at this point.
2348
2349If you are interested in comments or time, which may be stored either
2350before or after the image data, you should pass the separate png_info
2351struct if you want to keep the comments from before and after the image
2352separate.
2353
2354    png_infop end_info = png_create_info_struct(png_ptr);
2355
2356    if (!end_info)
2357    {
2358       png_destroy_read_struct(&png_ptr, &info_ptr,
2359           (png_infopp)NULL);
2360       return (ERROR);
2361    }
2362
2363   png_read_end(png_ptr, end_info);
2364
2365If you are not interested, you should still call png_read_end()
2366but you can pass NULL, avoiding the need to create an end_info structure.
2367If you do this, libpng will not process any chunks after IDAT other than
2368skipping over them and perhaps (depending on whether you have called
2369png_set_crc_action) checking their CRCs while looking for the IEND chunk.
2370
2371   png_read_end(png_ptr, (png_infop)NULL);
2372
2373If you don't call png_read_end(), then your file pointer will be
2374left pointing to the first chunk after the last IDAT, which is probably
2375not what you want if you expect to read something beyond the end of
2376the PNG datastream.
2377
2378When you are done, you can free all memory allocated by libpng like this:
2379
2380   png_destroy_read_struct(&png_ptr, &info_ptr,
2381       &end_info);
2382
2383or, if you didn't create an end_info structure,
2384
2385   png_destroy_read_struct(&png_ptr, &info_ptr,
2386       (png_infopp)NULL);
2387
2388It is also possible to individually free the info_ptr members that
2389point to libpng-allocated storage with the following function:
2390
2391    png_free_data(png_ptr, info_ptr, mask, seq)
2392
2393    mask - identifies data to be freed, a mask
2394           containing the bitwise OR of one or
2395           more of
2396             PNG_FREE_PLTE, PNG_FREE_TRNS,
2397             PNG_FREE_HIST, PNG_FREE_ICCP,
2398             PNG_FREE_PCAL, PNG_FREE_ROWS,
2399             PNG_FREE_SCAL, PNG_FREE_SPLT,
2400             PNG_FREE_TEXT, PNG_FREE_UNKN,
2401           or simply PNG_FREE_ALL
2402
2403    seq  - sequence number of item to be freed
2404           (-1 for all items)
2405
2406This function may be safely called when the relevant storage has
2407already been freed, or has not yet been allocated, or was allocated
2408by the user and not by libpng,  and will in those cases do nothing.
2409The "seq" parameter is ignored if only one item of the selected data
2410type, such as PLTE, is allowed.  If "seq" is not -1, and multiple items
2411are allowed for the data type identified in the mask, such as text or
2412sPLT, only the n'th item in the structure is freed, where n is "seq".
2413
2414The default behavior is only to free data that was allocated internally
2415by libpng.  This can be changed, so that libpng will not free the data,
2416or so that it will free data that was allocated by the user with png_malloc()
2417or png_calloc() and passed in via a png_set_*() function, with
2418
2419    png_data_freer(png_ptr, info_ptr, freer, mask)
2420
2421    freer  - one of
2422               PNG_DESTROY_WILL_FREE_DATA
2423               PNG_SET_WILL_FREE_DATA
2424               PNG_USER_WILL_FREE_DATA
2425
2426    mask   - which data elements are affected
2427             same choices as in png_free_data()
2428
2429This function only affects data that has already been allocated.
2430You can call this function after reading the PNG data but before calling
2431any png_set_*() functions, to control whether the user or the png_set_*()
2432function is responsible for freeing any existing data that might be present,
2433and again after the png_set_*() functions to control whether the user
2434or png_destroy_*() is supposed to free the data.  When the user assumes
2435responsibility for libpng-allocated data, the application must use
2436png_free() to free it, and when the user transfers responsibility to libpng
2437for data that the user has allocated, the user must have used png_malloc()
2438or png_calloc() to allocate it.
2439
2440If you allocated your row_pointers in a single block, as suggested above in
2441the description of the high level read interface, you must not transfer
2442responsibility for freeing it to the png_set_rows or png_read_destroy function,
2443because they would also try to free the individual row_pointers[i].
2444
2445If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword
2446separately, do not transfer responsibility for freeing text_ptr to libpng,
2447because when libpng fills a png_text structure it combines these members with
2448the key member, and png_free_data() will free only text_ptr.key.  Similarly,
2449if you transfer responsibility for free'ing text_ptr from libpng to your
2450application, your application must not separately free those members.
2451
2452The png_free_data() function will turn off the "valid" flag for anything
2453it frees.  If you need to turn the flag off for a chunk that was freed by
2454your application instead of by libpng, you can use
2455
2456    png_set_invalid(png_ptr, info_ptr, mask);
2457
2458    mask - identifies the chunks to be made invalid,
2459           containing the bitwise OR of one or
2460           more of
2461             PNG_INFO_gAMA, PNG_INFO_sBIT,
2462             PNG_INFO_cHRM, PNG_INFO_PLTE,
2463             PNG_INFO_tRNS, PNG_INFO_bKGD,
2464             PNG_INFO_hIST, PNG_INFO_pHYs,
2465             PNG_INFO_oFFs, PNG_INFO_tIME,
2466             PNG_INFO_pCAL, PNG_INFO_sRGB,
2467             PNG_INFO_iCCP, PNG_INFO_sPLT,
2468             PNG_INFO_sCAL, PNG_INFO_IDAT
2469
2470For a more compact example of reading a PNG image, see the file example.c.
2471
2472Reading PNG files progressively
2473
2474The progressive reader is slightly different from the non-progressive
2475reader.  Instead of calling png_read_info(), png_read_rows(), and
2476png_read_end(), you make one call to png_process_data(), which calls
2477callbacks when it has the info, a row, or the end of the image.  You
2478set up these callbacks with png_set_progressive_read_fn().  You don't
2479have to worry about the input/output functions of libpng, as you are
2480giving the library the data directly in png_process_data().  I will
2481assume that you have read the section on reading PNG files above,
2482so I will only highlight the differences (although I will show
2483all of the code).
2484
2485png_structp png_ptr;
2486png_infop info_ptr;
2487
2488 /*  An example code fragment of how you would
2489     initialize the progressive reader in your
2490     application. */
2491 int
2492 initialize_png_reader()
2493 {
2494    png_ptr = png_create_read_struct
2495        (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
2496         user_error_fn, user_warning_fn);
2497
2498    if (!png_ptr)
2499        return (ERROR);
2500
2501    info_ptr = png_create_info_struct(png_ptr);
2502
2503    if (!info_ptr)
2504    {
2505       png_destroy_read_struct(&png_ptr,
2506          (png_infopp)NULL, (png_infopp)NULL);
2507       return (ERROR);
2508    }
2509
2510    if (setjmp(png_jmpbuf(png_ptr)))
2511    {
2512       png_destroy_read_struct(&png_ptr, &info_ptr,
2513          (png_infopp)NULL);
2514       return (ERROR);
2515    }
2516
2517    /* This one's new.  You can provide functions
2518       to be called when the header info is valid,
2519       when each row is completed, and when the image
2520       is finished.  If you aren't using all functions,
2521       you can specify NULL parameters.  Even when all
2522       three functions are NULL, you need to call
2523       png_set_progressive_read_fn().  You can use
2524       any struct as the user_ptr (cast to a void pointer
2525       for the function call), and retrieve the pointer
2526       from inside the callbacks using the function
2527
2528          png_get_progressive_ptr(png_ptr);
2529
2530       which will return a void pointer, which you have
2531       to cast appropriately.
2532     */
2533    png_set_progressive_read_fn(png_ptr, (void *)user_ptr,
2534        info_callback, row_callback, end_callback);
2535
2536    return 0;
2537 }
2538
2539 /* A code fragment that you call as you receive blocks
2540   of data */
2541 int
2542 process_data(png_bytep buffer, png_uint_32 length)
2543 {
2544    if (setjmp(png_jmpbuf(png_ptr)))
2545    {
2546       png_destroy_read_struct(&png_ptr, &info_ptr,
2547           (png_infopp)NULL);
2548       return (ERROR);
2549    }
2550
2551    /* This one's new also.  Simply give it a chunk
2552       of data from the file stream (in order, of
2553       course).  On machines with segmented memory
2554       models machines, don't give it any more than
2555       64K.  The library seems to run fine with sizes
2556       of 4K. Although you can give it much less if
2557       necessary (I assume you can give it chunks of
2558       1 byte, I haven't tried less than 256 bytes
2559       yet).  When this function returns, you may
2560       want to display any rows that were generated
2561       in the row callback if you don't already do
2562       so there.
2563     */
2564    png_process_data(png_ptr, info_ptr, buffer, length);
2565
2566    /* At this point you can call png_process_data_skip if
2567       you want to handle data the library will skip yourself;
2568       it simply returns the number of bytes to skip (and stops
2569       libpng skipping that number of bytes on the next
2570       png_process_data call).
2571    return 0;
2572 }
2573
2574 /* This function is called (as set by
2575    png_set_progressive_read_fn() above) when enough data
2576    has been supplied so all of the header has been
2577    read.
2578 */
2579 void
2580 info_callback(png_structp png_ptr, png_infop info)
2581 {
2582    /* Do any setup here, including setting any of
2583       the transformations mentioned in the Reading
2584       PNG files section.  For now, you _must_ call
2585       either png_start_read_image() or
2586       png_read_update_info() after all the
2587       transformations are set (even if you don't set
2588       any).  You may start getting rows before
2589       png_process_data() returns, so this is your
2590       last chance to prepare for that.
2591
2592       This is where you turn on interlace handling,
2593       assuming you don't want to do it yourself.
2594
2595       If you need to you can stop the processing of
2596       your original input data at this point by calling
2597       png_process_data_pause.  This returns the number
2598       of unprocessed bytes from the last png_process_data
2599       call - it is up to you to ensure that the next call
2600       sees these bytes again.  If you don't want to bother
2601       with this you can get libpng to cache the unread
2602       bytes by setting the 'save' parameter (see png.h) but
2603       then libpng will have to copy the data internally.
2604     */
2605 }
2606
2607 /* This function is called when each row of image
2608    data is complete */
2609 void
2610 row_callback(png_structp png_ptr, png_bytep new_row,
2611    png_uint_32 row_num, int pass)
2612 {
2613    /* If the image is interlaced, and you turned
2614       on the interlace handler, this function will
2615       be called for every row in every pass.  Some
2616       of these rows will not be changed from the
2617       previous pass.  When the row is not changed,
2618       the new_row variable will be NULL.  The rows
2619       and passes are called in order, so you don't
2620       really need the row_num and pass, but I'm
2621       supplying them because it may make your life
2622       easier.
2623
2624       If you did not turn on interlace handling then
2625       the callback is called for each row of each
2626       sub-image when the image is interlaced.  In this
2627       case 'row_num' is the row in the sub-image, not
2628       the row in the output image as it is in all other
2629       cases.
2630
2631       For the non-NULL rows of interlaced images when
2632       you have switched on libpng interlace handling,
2633       you must call png_progressive_combine_row()
2634       passing in the row and the old row.  You can
2635       call this function for NULL rows (it will just
2636       return) and for non-interlaced images (it just
2637       does the memcpy for you) if it will make the
2638       code easier.  Thus, you can just do this for
2639       all cases if you switch on interlace handling;
2640     */
2641
2642        png_progressive_combine_row(png_ptr, old_row,
2643          new_row);
2644
2645    /* where old_row is what was displayed
2646       previously for the row.  Note that the first
2647       pass (pass == 0, really) will completely cover
2648       the old row, so the rows do not have to be
2649       initialized.  After the first pass (and only
2650       for interlaced images), you will have to pass
2651       the current row, and the function will combine
2652       the old row and the new row.
2653
2654       You can also call png_process_data_pause in this
2655       callback - see above.
2656    */
2657 }
2658
2659 void
2660 end_callback(png_structp png_ptr, png_infop info)
2661 {
2662    /* This function is called after the whole image
2663       has been read, including any chunks after the
2664       image (up to and including the IEND).  You
2665       will usually have the same info chunk as you
2666       had in the header, although some data may have
2667       been added to the comments and time fields.
2668
2669       Most people won't do much here, perhaps setting
2670       a flag that marks the image as finished.
2671     */
2672 }
2673
2674
2675
2676IV. Writing
2677
2678Much of this is very similar to reading.  However, everything of
2679importance is repeated here, so you won't have to constantly look
2680back up in the reading section to understand writing.
2681
2682Setup
2683
2684You will want to do the I/O initialization before you get into libpng,
2685so if it doesn't work, you don't have anything to undo. If you are not
2686using the standard I/O functions, you will need to replace them with
2687custom writing functions.  See the discussion under Customizing libpng.
2688
2689    FILE *fp = fopen(file_name, "wb");
2690
2691    if (!fp)
2692       return (ERROR);
2693
2694Next, png_struct and png_info need to be allocated and initialized.
2695As these can be both relatively large, you may not want to store these
2696on the stack, unless you have stack space to spare.  Of course, you
2697will want to check if they return NULL.  If you are also reading,
2698you won't want to name your read structure and your write structure
2699both "png_ptr"; you can call them anything you like, such as
2700"read_ptr" and "write_ptr".  Look at pngtest.c, for example.
2701
2702    png_structp png_ptr = png_create_write_struct
2703       (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
2704        user_error_fn, user_warning_fn);
2705
2706    if (!png_ptr)
2707       return (ERROR);
2708
2709    png_infop info_ptr = png_create_info_struct(png_ptr);
2710    if (!info_ptr)
2711    {
2712       png_destroy_write_struct(&png_ptr,
2713           (png_infopp)NULL);
2714       return (ERROR);
2715    }
2716
2717If you want to use your own memory allocation routines,
2718define PNG_USER_MEM_SUPPORTED and use
2719png_create_write_struct_2() instead of png_create_write_struct():
2720
2721    png_structp png_ptr = png_create_write_struct_2
2722       (PNG_LIBPNG_VER_STRING, (png_voidp)user_error_ptr,
2723        user_error_fn, user_warning_fn, (png_voidp)
2724        user_mem_ptr, user_malloc_fn, user_free_fn);
2725
2726After you have these structures, you will need to set up the
2727error handling.  When libpng encounters an error, it expects to
2728longjmp() back to your routine.  Therefore, you will need to call
2729setjmp() and pass the png_jmpbuf(png_ptr).  If you
2730write the file from different routines, you will need to update
2731the png_jmpbuf(png_ptr) every time you enter a new routine that will
2732call a png_*() function.  See your documentation of setjmp/longjmp
2733for your compiler for more information on setjmp/longjmp.  See
2734the discussion on libpng error handling in the Customizing Libpng
2735section below for more information on the libpng error handling.
2736
2737    if (setjmp(png_jmpbuf(png_ptr)))
2738    {
2739    png_destroy_write_struct(&png_ptr, &info_ptr);
2740       fclose(fp);
2741       return (ERROR);
2742    }
2743    ...
2744    return;
2745
2746If you would rather avoid the complexity of setjmp/longjmp issues,
2747you can compile libpng with PNG_NO_SETJMP, in which case
2748errors will result in a call to PNG_ABORT() which defaults to abort().
2749
2750You can #define PNG_ABORT() to a function that does something
2751more useful than abort(), as long as your function does not
2752return.
2753
2754Checking for invalid palette index on write was added at libpng
27551.5.10.  If a pixel contains an invalid (out-of-range) index libpng issues
2756a benign error.  This is enabled by default because this condition is an
2757error according to the PNG specification, Clause 11.3.2, but the error can
2758be ignored in each png_ptr with
2759
2760   png_set_check_for_invalid_index(png_ptr, 0);
2761
2762If the error is ignored, or if png_benign_error() treats it as a warning,
2763any invalid pixels are written as-is by the encoder, resulting in an
2764invalid PNG datastream as output.  In this case the application is
2765responsible for ensuring that the pixel indexes are in range when it writes
2766a PLTE chunk with fewer entries than the bit depth would allow.
2767
2768Now you need to set up the output code.  The default for libpng is to
2769use the C function fwrite().  If you use this, you will need to pass a
2770valid FILE * in the function png_init_io().  Be sure that the file is
2771opened in binary mode.  Again, if you wish to handle writing data in
2772another way, see the discussion on libpng I/O handling in the Customizing
2773Libpng section below.
2774
2775    png_init_io(png_ptr, fp);
2776
2777If you are embedding your PNG into a datastream such as MNG, and don't
2778want libpng to write the 8-byte signature, or if you have already
2779written the signature in your application, use
2780
2781    png_set_sig_bytes(png_ptr, 8);
2782
2783to inform libpng that it should not write a signature.
2784
2785Write callbacks
2786
2787At this point, you can set up a callback function that will be
2788called after each row has been written, which you can use to control
2789a progress meter or the like.  It's demonstrated in pngtest.c.
2790You must supply a function
2791
2792    void write_row_callback(png_structp png_ptr, png_uint_32 row,
2793       int pass);
2794    {
2795      /* put your code here */
2796    }
2797
2798(You can give it another name that you like instead of "write_row_callback")
2799
2800To inform libpng about your function, use
2801
2802    png_set_write_status_fn(png_ptr, write_row_callback);
2803
2804When this function is called the row has already been completely processed and
2805it has also been written out.  The 'row' and 'pass' refer to the next row to be
2806handled.  For the
2807non-interlaced case the row that was just handled is simply one less than the
2808passed in row number, and pass will always be 0.  For the interlaced case the
2809same applies unless the row value is 0, in which case the row just handled was
2810the last one from one of the preceding passes.  Because interlacing may skip a
2811pass you cannot be sure that the preceding pass is just 'pass-1', if you really
2812need to know what the last pass is record (row,pass) from the callback and use
2813the last recorded value each time.
2814
2815As with the user transform you can find the output row using the
2816PNG_ROW_FROM_PASS_ROW macro.
2817
2818You now have the option of modifying how the compression library will
2819run.  The following functions are mainly for testing, but may be useful
2820in some cases, like if you need to write PNG files extremely fast and
2821are willing to give up some compression, or if you want to get the
2822maximum possible compression at the expense of slower writing.  If you
2823have no special needs in this area, let the library do what it wants by
2824not calling this function at all, as it has been tuned to deliver a good
2825speed/compression ratio. The second parameter to png_set_filter() is
2826the filter method, for which the only valid values are 0 (as of the
2827July 1999 PNG specification, version 1.2) or 64 (if you are writing
2828a PNG datastream that is to be embedded in a MNG datastream).  The third
2829parameter is a flag that indicates which filter type(s) are to be tested
2830for each scanline.  See the PNG specification for details on the specific
2831filter types.
2832
2833
2834    /* turn on or off filtering, and/or choose
2835       specific filters.  You can use either a single
2836       PNG_FILTER_VALUE_NAME or the bitwise OR of one
2837       or more PNG_FILTER_NAME masks.
2838     */
2839    png_set_filter(png_ptr, 0,
2840       PNG_FILTER_NONE  | PNG_FILTER_VALUE_NONE |
2841       PNG_FILTER_SUB   | PNG_FILTER_VALUE_SUB  |
2842       PNG_FILTER_UP    | PNG_FILTER_VALUE_UP   |
2843       PNG_FILTER_AVG   | PNG_FILTER_VALUE_AVG  |
2844       PNG_FILTER_PAETH | PNG_FILTER_VALUE_PAETH|
2845       PNG_ALL_FILTERS);
2846
2847If an application wants to start and stop using particular filters during
2848compression, it should start out with all of the filters (to ensure that
2849the previous row of pixels will be stored in case it's needed later),
2850and then add and remove them after the start of compression.
2851
2852If you are writing a PNG datastream that is to be embedded in a MNG
2853datastream, the second parameter can be either 0 or 64.
2854
2855The png_set_compression_*() functions interface to the zlib compression
2856library, and should mostly be ignored unless you really know what you are
2857doing.  The only generally useful call is png_set_compression_level()
2858which changes how much time zlib spends on trying to compress the image
2859data.  See the Compression Library (zlib.h and algorithm.txt, distributed
2860with zlib) for details on the compression levels.
2861
2862    #include zlib.h
2863
2864    /* Set the zlib compression level */
2865    png_set_compression_level(png_ptr,
2866        Z_BEST_COMPRESSION);
2867
2868    /* Set other zlib parameters for compressing IDAT */
2869    png_set_compression_mem_level(png_ptr, 8);
2870    png_set_compression_strategy(png_ptr,
2871        Z_DEFAULT_STRATEGY);
2872    png_set_compression_window_bits(png_ptr, 15);
2873    png_set_compression_method(png_ptr, 8);
2874    png_set_compression_buffer_size(png_ptr, 8192)
2875
2876    /* Set zlib parameters for text compression
2877     * If you don't call these, the parameters
2878     * fall back on those defined for IDAT chunks
2879     */
2880    png_set_text_compression_mem_level(png_ptr, 8);
2881    png_set_text_compression_strategy(png_ptr,
2882        Z_DEFAULT_STRATEGY);
2883    png_set_text_compression_window_bits(png_ptr, 15);
2884    png_set_text_compression_method(png_ptr, 8);
2885
2886Setting the contents of info for output
2887
2888You now need to fill in the png_info structure with all the data you
2889wish to write before the actual image.  Note that the only thing you
2890are allowed to write after the image is the text chunks and the time
2891chunk (as of PNG Specification 1.2, anyway).  See png_write_end() and
2892the latest PNG specification for more information on that.  If you
2893wish to write them before the image, fill them in now, and flag that
2894data as being valid.  If you want to wait until after the data, don't
2895fill them until png_write_end().  For all the fields in png_info and
2896their data types, see png.h.  For explanations of what the fields
2897contain, see the PNG specification.
2898
2899Some of the more important parts of the png_info are:
2900
2901    png_set_IHDR(png_ptr, info_ptr, width, height,
2902       bit_depth, color_type, interlace_type,
2903       compression_type, filter_method)
2904
2905    width          - holds the width of the image
2906                     in pixels (up to 2^31).
2907
2908    height         - holds the height of the image
2909                     in pixels (up to 2^31).
2910
2911    bit_depth      - holds the bit depth of one of the
2912                     image channels.
2913                     (valid values are 1, 2, 4, 8, 16
2914                     and depend also on the
2915                     color_type.  See also significant
2916                     bits (sBIT) below).
2917
2918    color_type     - describes which color/alpha
2919                     channels are present.
2920                     PNG_COLOR_TYPE_GRAY
2921                        (bit depths 1, 2, 4, 8, 16)
2922                     PNG_COLOR_TYPE_GRAY_ALPHA
2923                        (bit depths 8, 16)
2924                     PNG_COLOR_TYPE_PALETTE
2925                        (bit depths 1, 2, 4, 8)
2926                     PNG_COLOR_TYPE_RGB
2927                        (bit_depths 8, 16)
2928                     PNG_COLOR_TYPE_RGB_ALPHA
2929                        (bit_depths 8, 16)
2930
2931                     PNG_COLOR_MASK_PALETTE
2932                     PNG_COLOR_MASK_COLOR
2933                     PNG_COLOR_MASK_ALPHA
2934
2935    interlace_type - PNG_INTERLACE_NONE or
2936                     PNG_INTERLACE_ADAM7
2937
2938    compression_type - (must be
2939                     PNG_COMPRESSION_TYPE_DEFAULT)
2940
2941    filter_method  - (must be PNG_FILTER_TYPE_DEFAULT
2942                     or, if you are writing a PNG to
2943                     be embedded in a MNG datastream,
2944                     can also be
2945                     PNG_INTRAPIXEL_DIFFERENCING)
2946
2947If you call png_set_IHDR(), the call must appear before any of the
2948other png_set_*() functions, because they might require access to some of
2949the IHDR settings.  The remaining png_set_*() functions can be called
2950in any order.
2951
2952If you wish, you can reset the compression_type, interlace_type, or
2953filter_method later by calling png_set_IHDR() again; if you do this, the
2954width, height, bit_depth, and color_type must be the same in each call.
2955
2956    png_set_PLTE(png_ptr, info_ptr, palette,
2957       num_palette);
2958
2959    palette        - the palette for the file
2960                     (array of png_color)
2961    num_palette    - number of entries in the palette
2962
2963
2964    png_set_gAMA(png_ptr, info_ptr, file_gamma);
2965    png_set_gAMA_fixed(png_ptr, info_ptr, int_file_gamma);
2966
2967    file_gamma     - the gamma at which the image was
2968                     created (PNG_INFO_gAMA)
2969
2970    int_file_gamma - 100,000 times the gamma at which
2971                     the image was created
2972
2973    png_set_cHRM(png_ptr, info_ptr,  white_x, white_y, red_x, red_y,
2974                     green_x, green_y, blue_x, blue_y)
2975    png_set_cHRM_XYZ(png_ptr, info_ptr, red_X, red_Y, red_Z, green_X,
2976                     green_Y, green_Z, blue_X, blue_Y, blue_Z)
2977    png_set_cHRM_fixed(png_ptr, info_ptr, int_white_x, int_white_y,
2978                     int_red_x, int_red_y, int_green_x, int_green_y,
2979                     int_blue_x, int_blue_y)
2980    png_set_cHRM_XYZ_fixed(png_ptr, info_ptr, int_red_X, int_red_Y,
2981                     int_red_Z, int_green_X, int_green_Y, int_green_Z,
2982                     int_blue_X, int_blue_Y, int_blue_Z)
2983
2984    {white,red,green,blue}_{x,y}
2985                     A color space encoding specified using the chromaticities
2986                     of the end points and the white point.
2987
2988    {red,green,blue}_{X,Y,Z}
2989                     A color space encoding specified using the encoding end
2990                     points - the CIE tristimulus specification of the intended
2991                     color of the red, green and blue channels in the PNG RGB
2992                     data.  The white point is simply the sum of the three end
2993                     points.
2994
2995    png_set_sRGB(png_ptr, info_ptr, srgb_intent);
2996
2997    srgb_intent    - the rendering intent
2998                     (PNG_INFO_sRGB) The presence of
2999                     the sRGB chunk means that the pixel
3000                     data is in the sRGB color space.
3001                     This chunk also implies specific
3002                     values of gAMA and cHRM.  Rendering
3003                     intent is the CSS-1 property that
3004                     has been defined by the International
3005                     Color Consortium
3006                     (http://www.color.org).
3007                     It can be one of
3008                     PNG_sRGB_INTENT_SATURATION,
3009                     PNG_sRGB_INTENT_PERCEPTUAL,
3010                     PNG_sRGB_INTENT_ABSOLUTE, or
3011                     PNG_sRGB_INTENT_RELATIVE.
3012
3013
3014    png_set_sRGB_gAMA_and_cHRM(png_ptr, info_ptr,
3015       srgb_intent);
3016
3017    srgb_intent    - the rendering intent
3018                     (PNG_INFO_sRGB) The presence of the
3019                     sRGB chunk means that the pixel
3020                     data is in the sRGB color space.
3021                     This function also causes gAMA and
3022                     cHRM chunks with the specific values
3023                     that are consistent with sRGB to be
3024                     written.
3025
3026    png_set_iCCP(png_ptr, info_ptr, name, compression_type,
3027                       profile, proflen);
3028
3029    name             - The profile name.
3030
3031    compression_type - The compression type; always
3032                       PNG_COMPRESSION_TYPE_BASE for PNG 1.0.
3033                       You may give NULL to this argument to
3034                       ignore it.
3035
3036    profile          - International Color Consortium color
3037                       profile data. May contain NULs.
3038
3039    proflen          - length of profile data in bytes.
3040
3041    png_set_sBIT(png_ptr, info_ptr, sig_bit);
3042
3043    sig_bit        - the number of significant bits for
3044                     (PNG_INFO_sBIT) each of the gray, red,
3045                     green, and blue channels, whichever are
3046                     appropriate for the given color type
3047                     (png_color_16)
3048
3049    png_set_tRNS(png_ptr, info_ptr, trans_alpha,
3050       num_trans, trans_color);
3051
3052    trans_alpha    - array of alpha (transparency)
3053                     entries for palette (PNG_INFO_tRNS)
3054
3055    num_trans      - number of transparent entries
3056                     (PNG_INFO_tRNS)
3057
3058    trans_color    - graylevel or color sample values
3059                     (in order red, green, blue) of the
3060                     single transparent color for
3061                     non-paletted images (PNG_INFO_tRNS)
3062
3063    png_set_hIST(png_ptr, info_ptr, hist);
3064
3065    hist           - histogram of palette (array of
3066                     png_uint_16) (PNG_INFO_hIST)
3067
3068    png_set_tIME(png_ptr, info_ptr, mod_time);
3069
3070    mod_time       - time image was last modified
3071                     (PNG_VALID_tIME)
3072
3073    png_set_bKGD(png_ptr, info_ptr, background);
3074
3075    background     - background color (of type
3076                     png_color_16p) (PNG_VALID_bKGD)
3077
3078    png_set_text(png_ptr, info_ptr, text_ptr, num_text);
3079
3080    text_ptr       - array of png_text holding image
3081                     comments
3082
3083    text_ptr[i].compression - type of compression used
3084                 on "text" PNG_TEXT_COMPRESSION_NONE
3085                           PNG_TEXT_COMPRESSION_zTXt
3086                           PNG_ITXT_COMPRESSION_NONE
3087                           PNG_ITXT_COMPRESSION_zTXt
3088    text_ptr[i].key   - keyword for comment.  Must contain
3089                 1-79 characters.
3090    text_ptr[i].text  - text comments for current
3091                         keyword.  Can be NULL or empty.
3092    text_ptr[i].text_length - length of text string,
3093                 after decompression, 0 for iTXt
3094    text_ptr[i].itxt_length - length of itxt string,
3095                 after decompression, 0 for tEXt/zTXt
3096    text_ptr[i].lang  - language of comment (NULL or
3097                         empty for unknown).
3098    text_ptr[i].translated_keyword  - keyword in UTF-8 (NULL
3099                         or empty for unknown).
3100
3101    Note that the itxt_length, lang, and lang_key
3102    members of the text_ptr structure only exist when the
3103    library is built with iTXt chunk support.  Prior to
3104    libpng-1.4.0 the library was built by default without
3105    iTXt support. Also note that when iTXt is supported,
3106    they contain NULL pointers when the "compression"
3107    field contains PNG_TEXT_COMPRESSION_NONE or
3108    PNG_TEXT_COMPRESSION_zTXt.
3109
3110    num_text       - number of comments
3111
3112    png_set_sPLT(png_ptr, info_ptr, &palette_ptr,
3113       num_spalettes);
3114
3115    palette_ptr    - array of png_sPLT_struct structures
3116                     to be added to the list of palettes
3117                     in the info structure.
3118    num_spalettes  - number of palette structures to be
3119                     added.
3120
3121    png_set_oFFs(png_ptr, info_ptr, offset_x, offset_y,
3122        unit_type);
3123
3124    offset_x  - positive offset from the left
3125                     edge of the screen
3126
3127    offset_y  - positive offset from the top
3128                     edge of the screen
3129
3130    unit_type - PNG_OFFSET_PIXEL, PNG_OFFSET_MICROMETER
3131
3132    png_set_pHYs(png_ptr, info_ptr, res_x, res_y,
3133        unit_type);
3134
3135    res_x       - pixels/unit physical resolution
3136                  in x direction
3137
3138    res_y       - pixels/unit physical resolution
3139                  in y direction
3140
3141    unit_type   - PNG_RESOLUTION_UNKNOWN,
3142                  PNG_RESOLUTION_METER
3143
3144    png_set_sCAL(png_ptr, info_ptr, unit, width, height)
3145
3146    unit        - physical scale units (an integer)
3147
3148    width       - width of a pixel in physical scale units
3149
3150    height      - height of a pixel in physical scale units
3151                  (width and height are doubles)
3152
3153    png_set_sCAL_s(png_ptr, info_ptr, unit, width, height)
3154
3155    unit        - physical scale units (an integer)
3156
3157    width       - width of a pixel in physical scale units
3158                  expressed as a string
3159
3160    height      - height of a pixel in physical scale units
3161                 (width and height are strings like "2.54")
3162
3163    png_set_unknown_chunks(png_ptr, info_ptr, &unknowns,
3164       num_unknowns)
3165
3166    unknowns          - array of png_unknown_chunk
3167                        structures holding unknown chunks
3168    unknowns[i].name  - name of unknown chunk
3169    unknowns[i].data  - data of unknown chunk
3170    unknowns[i].size  - size of unknown chunk's data
3171    unknowns[i].location - position to write chunk in file
3172                           0: do not write chunk
3173                           PNG_HAVE_IHDR: before PLTE
3174                           PNG_HAVE_PLTE: before IDAT
3175                           PNG_AFTER_IDAT: after IDAT
3176
3177The "location" member is set automatically according to
3178what part of the output file has already been written.
3179You can change its value after calling png_set_unknown_chunks()
3180as demonstrated in pngtest.c.  Within each of the "locations",
3181the chunks are sequenced according to their position in the
3182structure (that is, the value of "i", which is the order in which
3183the chunk was either read from the input file or defined with
3184png_set_unknown_chunks).
3185
3186A quick word about text and num_text.  text is an array of png_text
3187structures.  num_text is the number of valid structures in the array.
3188Each png_text structure holds a language code, a keyword, a text value,
3189and a compression type.
3190
3191The compression types have the same valid numbers as the compression
3192types of the image data.  Currently, the only valid number is zero.
3193However, you can store text either compressed or uncompressed, unlike
3194images, which always have to be compressed.  So if you don't want the
3195text compressed, set the compression type to PNG_TEXT_COMPRESSION_NONE.
3196Because tEXt and zTXt chunks don't have a language field, if you
3197specify PNG_TEXT_COMPRESSION_NONE or PNG_TEXT_COMPRESSION_zTXt
3198any language code or translated keyword will not be written out.
3199
3200Until text gets around a few hundred bytes, it is not worth compressing it.
3201After the text has been written out to the file, the compression type
3202is set to PNG_TEXT_COMPRESSION_NONE_WR or PNG_TEXT_COMPRESSION_zTXt_WR,
3203so that it isn't written out again at the end (in case you are calling
3204png_write_end() with the same struct).
3205
3206The keywords that are given in the PNG Specification are:
3207
3208    Title            Short (one line) title or
3209                     caption for image
3210
3211    Author           Name of image's creator
3212
3213    Description      Description of image (possibly long)
3214
3215    Copyright        Copyright notice
3216
3217    Creation Time    Time of original image creation
3218                     (usually RFC 1123 format, see below)
3219
3220    Software         Software used to create the image
3221
3222    Disclaimer       Legal disclaimer
3223
3224    Warning          Warning of nature of content
3225
3226    Source           Device used to create the image
3227
3228    Comment          Miscellaneous comment; conversion
3229                     from other image format
3230
3231The keyword-text pairs work like this.  Keywords should be short
3232simple descriptions of what the comment is about.  Some typical
3233keywords are found in the PNG specification, as is some recommendations
3234on keywords.  You can repeat keywords in a file.  You can even write
3235some text before the image and some after.  For example, you may want
3236to put a description of the image before the image, but leave the
3237disclaimer until after, so viewers working over modem connections
3238don't have to wait for the disclaimer to go over the modem before
3239they start seeing the image.  Finally, keywords should be full
3240words, not abbreviations.  Keywords and text are in the ISO 8859-1
3241(Latin-1) character set (a superset of regular ASCII) and can not
3242contain NUL characters, and should not contain control or other
3243unprintable characters.  To make the comments widely readable, stick
3244with basic ASCII, and avoid machine specific character set extensions
3245like the IBM-PC character set.  The keyword must be present, but
3246you can leave off the text string on non-compressed pairs.
3247Compressed pairs must have a text string, as only the text string
3248is compressed anyway, so the compression would be meaningless.
3249
3250PNG supports modification time via the png_time structure.  Two
3251conversion routines are provided, png_convert_from_time_t() for
3252time_t and png_convert_from_struct_tm() for struct tm.  The
3253time_t routine uses gmtime().  You don't have to use either of
3254these, but if you wish to fill in the png_time structure directly,
3255you should provide the time in universal time (GMT) if possible
3256instead of your local time.  Note that the year number is the full
3257year (e.g. 1998, rather than 98 - PNG is year 2000 compliant!), and
3258that months start with 1.
3259
3260If you want to store the time of the original image creation, you should
3261use a plain tEXt chunk with the "Creation Time" keyword.  This is
3262necessary because the "creation time" of a PNG image is somewhat vague,
3263depending on whether you mean the PNG file, the time the image was
3264created in a non-PNG format, a still photo from which the image was
3265scanned, or possibly the subject matter itself.  In order to facilitate
3266machine-readable dates, it is recommended that the "Creation Time"
3267tEXt chunk use RFC 1123 format dates (e.g. "22 May 1997 18:07:10 GMT"),
3268although this isn't a requirement.  Unlike the tIME chunk, the
3269"Creation Time" tEXt chunk is not expected to be automatically changed
3270by the software.  To facilitate the use of RFC 1123 dates, a function
3271png_convert_to_rfc1123_buffer(buffer, png_timep) is provided to
3272convert from PNG time to an RFC 1123 format string.  The caller must provide
3273a writeable buffer of at least 29 bytes.
3274
3275Writing unknown chunks
3276
3277You can use the png_set_unknown_chunks function to queue up private chunks
3278for writing.  You give it a chunk name, location, raw data, and a size.  You
3279also must use png_set_keep_unknown_chunks() to ensure that libpng will
3280handle them.  That's all there is to it.  The chunks will be written by the
3281next following png_write_info_before_PLTE, png_write_info, or png_write_end
3282function, depending upon the specified location.  Any chunks previously
3283read into the info structure's unknown-chunk list will also be written out
3284in a sequence that satisfies the PNG specification's ordering rules.
3285
3286Here is an example of writing two private chunks, prVt and miNE:
3287
3288    #ifdef PNG_WRITE_UNKNOWN_CHUNKS_SUPPORTED
3289    /* Set unknown chunk data */
3290    png_unknown_chunk unk_chunk[2];
3291    strcpy((char *) unk_chunk[0].name, "prVt";
3292    unk_chunk[0].data = (unsigned char *) "PRIVATE DATA";
3293    unk_chunk[0].size = strlen(unk_chunk[0].data)+1;
3294    unk_chunk[0].location = PNG_HAVE_IHDR;
3295    strcpy((char *) unk_chunk[1].name, "miNE";
3296    unk_chunk[1].data = (unsigned char *) "MY CHUNK DATA";
3297    unk_chunk[1].size = strlen(unk_chunk[0].data)+1;
3298    unk_chunk[1].location = PNG_AFTER_IDAT;
3299    png_set_unknown_chunks(write_ptr, write_info_ptr,
3300        unk_chunk, 2);
3301    /* Needed because miNE is not safe-to-copy */
3302    png_set_keep_unknown_chunks(png, PNG_HANDLE_CHUNK_ALWAYS,
3303       (png_bytep) "miNE", 1);
3304    # if PNG_LIBPNG_VER < 10600
3305      /* Deal with unknown chunk location bug in 1.5.x and earlier */
3306      png_set_unknown_chunk_location(png, info, 0, PNG_HAVE_IHDR);
3307      png_set_unknown_chunk_location(png, info, 1, PNG_AFTER_IDAT);
3308    # endif
3309    # if PNG_LIBPNG_VER < 10500
3310      /* PNG_AFTER_IDAT writes two copies of the chunk prior to libpng-1.5.0,
3311       * one before IDAT and another after IDAT, so don't use it; only use
3312       * PNG_HAVE_IHDR location.  This call resets the location previously
3313       * set by assignment and png_set_unknown_chunk_location() for chunk 1.
3314       */
3315      png_set_unknown_chunk_location(png, info, 1, PNG_HAVE_IHDR);
3316    # endif
3317    #endif
3318
3319The high-level write interface
3320
3321At this point there are two ways to proceed; through the high-level
3322write interface, or through a sequence of low-level write operations.
3323You can use the high-level interface if your image data is present
3324in the info structure.  All defined output
3325transformations are permitted, enabled by the following masks.
3326
3327    PNG_TRANSFORM_IDENTITY      No transformation
3328    PNG_TRANSFORM_PACKING       Pack 1, 2 and 4-bit samples
3329    PNG_TRANSFORM_PACKSWAP      Change order of packed
3330                                pixels to LSB first
3331    PNG_TRANSFORM_INVERT_MONO   Invert monochrome images
3332    PNG_TRANSFORM_SHIFT         Normalize pixels to the
3333                                sBIT depth
3334    PNG_TRANSFORM_BGR           Flip RGB to BGR, RGBA
3335                                to BGRA
3336    PNG_TRANSFORM_SWAP_ALPHA    Flip RGBA to ARGB or GA
3337                                to AG
3338    PNG_TRANSFORM_INVERT_ALPHA  Change alpha from opacity
3339                                to transparency
3340    PNG_TRANSFORM_SWAP_ENDIAN   Byte-swap 16-bit samples
3341    PNG_TRANSFORM_STRIP_FILLER        Strip out filler
3342                                      bytes (deprecated).
3343    PNG_TRANSFORM_STRIP_FILLER_BEFORE Strip out leading
3344                                      filler bytes
3345    PNG_TRANSFORM_STRIP_FILLER_AFTER  Strip out trailing
3346                                      filler bytes
3347
3348If you have valid image data in the info structure (you can use
3349png_set_rows() to put image data in the info structure), simply do this:
3350
3351    png_write_png(png_ptr, info_ptr, png_transforms, NULL)
3352
3353where png_transforms is an integer containing the bitwise OR of some set of
3354transformation flags.  This call is equivalent to png_write_info(),
3355followed the set of transformations indicated by the transform mask,
3356then png_write_image(), and finally png_write_end().
3357
3358(The final parameter of this call is not yet used.  Someday it might point
3359to transformation parameters required by some future output transform.)
3360
3361You must use png_transforms and not call any png_set_transform() functions
3362when you use png_write_png().
3363
3364The low-level write interface
3365
3366If you are going the low-level route instead, you are now ready to
3367write all the file information up to the actual image data.  You do
3368this with a call to png_write_info().
3369
3370    png_write_info(png_ptr, info_ptr);
3371
3372Note that there is one transformation you may need to do before
3373png_write_info().  In PNG files, the alpha channel in an image is the
3374level of opacity.  If your data is supplied as a level of transparency,
3375you can invert the alpha channel before you write it, so that 0 is
3376fully transparent and 255 (in 8-bit or paletted images) or 65535
3377(in 16-bit images) is fully opaque, with
3378
3379    png_set_invert_alpha(png_ptr);
3380
3381This must appear before png_write_info() instead of later with the
3382other transformations because in the case of paletted images the tRNS
3383chunk data has to be inverted before the tRNS chunk is written.  If
3384your image is not a paletted image, the tRNS data (which in such cases
3385represents a single color to be rendered as transparent) won't need to
3386be changed, and you can safely do this transformation after your
3387png_write_info() call.
3388
3389If you need to write a private chunk that you want to appear before
3390the PLTE chunk when PLTE is present, you can write the PNG info in
3391two steps, and insert code to write your own chunk between them:
3392
3393    png_write_info_before_PLTE(png_ptr, info_ptr);
3394    png_set_unknown_chunks(png_ptr, info_ptr, ...);
3395    png_write_info(png_ptr, info_ptr);
3396
3397After you've written the file information, you can set up the library
3398to handle any special transformations of the image data.  The various
3399ways to transform the data will be described in the order that they
3400should occur.  This is important, as some of these change the color
3401type and/or bit depth of the data, and some others only work on
3402certain color types and bit depths.  Even though each transformation
3403checks to see if it has data that it can do something with, you should
3404make sure to only enable a transformation if it will be valid for the
3405data.  For example, don't swap red and blue on grayscale data.
3406
3407PNG files store RGB pixels packed into 3 or 6 bytes.  This code tells
3408the library to strip input data that has 4 or 8 bytes per pixel down
3409to 3 or 6 bytes (or strip 2 or 4-byte grayscale+filler data to 1 or 2
3410bytes per pixel).
3411
3412    png_set_filler(png_ptr, 0, PNG_FILLER_BEFORE);
3413
3414where the 0 is unused, and the location is either PNG_FILLER_BEFORE or
3415PNG_FILLER_AFTER, depending upon whether the filler byte in the pixel
3416is stored XRGB or RGBX.
3417
3418PNG files pack pixels of bit depths 1, 2, and 4 into bytes as small as
3419they can, resulting in, for example, 8 pixels per byte for 1 bit files.
3420If the data is supplied at 1 pixel per byte, use this code, which will
3421correctly pack the pixels into a single byte:
3422
3423    png_set_packing(png_ptr);
3424
3425PNG files reduce possible bit depths to 1, 2, 4, 8, and 16.  If your
3426data is of another bit depth, you can write an sBIT chunk into the
3427file so that decoders can recover the original data if desired.
3428
3429    /* Set the true bit depth of the image data */
3430    if (color_type & PNG_COLOR_MASK_COLOR)
3431    {
3432       sig_bit.red = true_bit_depth;
3433       sig_bit.green = true_bit_depth;
3434       sig_bit.blue = true_bit_depth;
3435    }
3436
3437    else
3438    {
3439       sig_bit.gray = true_bit_depth;
3440    }
3441
3442    if (color_type & PNG_COLOR_MASK_ALPHA)
3443    {
3444       sig_bit.alpha = true_bit_depth;
3445    }
3446
3447    png_set_sBIT(png_ptr, info_ptr, &sig_bit);
3448
3449If the data is stored in the row buffer in a bit depth other than
3450one supported by PNG (e.g. 3 bit data in the range 0-7 for a 4-bit PNG),
3451this will scale the values to appear to be the correct bit depth as
3452is required by PNG.
3453
3454    png_set_shift(png_ptr, &sig_bit);
3455
3456PNG files store 16-bit pixels in network byte order (big-endian,
3457ie. most significant bits first).  This code would be used if they are
3458supplied the other way (little-endian, i.e. least significant bits
3459first, the way PCs store them):
3460
3461    if (bit_depth > 8)
3462       png_set_swap(png_ptr);
3463
3464If you are using packed-pixel images (1, 2, or 4 bits/pixel), and you
3465need to change the order the pixels are packed into bytes, you can use:
3466
3467    if (bit_depth < 8)
3468       png_set_packswap(png_ptr);
3469
3470PNG files store 3 color pixels in red, green, blue order.  This code
3471would be used if they are supplied as blue, green, red:
3472
3473    png_set_bgr(png_ptr);
3474
3475PNG files describe monochrome as black being zero and white being
3476one. This code would be used if the pixels are supplied with this reversed
3477(black being one and white being zero):
3478
3479    png_set_invert_mono(png_ptr);
3480
3481Finally, you can write your own transformation function if none of
3482the existing ones meets your needs.  This is done by setting a callback
3483with
3484
3485    png_set_write_user_transform_fn(png_ptr,
3486       write_transform_fn);
3487
3488You must supply the function
3489
3490    void write_transform_fn(png_structp png_ptr, png_row_infop
3491       row_info, png_bytep data)
3492
3493See pngtest.c for a working example.  Your function will be called
3494before any of the other transformations are processed.  If supported
3495libpng also supplies an information routine that may be called from
3496your callback:
3497
3498   png_get_current_row_number(png_ptr);
3499   png_get_current_pass_number(png_ptr);
3500
3501This returns the current row passed to the transform.  With interlaced
3502images the value returned is the row in the input sub-image image.  Use
3503PNG_ROW_FROM_PASS_ROW(row, pass) and PNG_COL_FROM_PASS_COL(col, pass) to
3504find the output pixel (x,y) given an interlaced sub-image pixel (row,col,pass).
3505
3506The discussion of interlace handling above contains more information on how to
3507use these values.
3508
3509You can also set up a pointer to a user structure for use by your
3510callback function.
3511
3512    png_set_user_transform_info(png_ptr, user_ptr, 0, 0);
3513
3514The user_channels and user_depth parameters of this function are ignored
3515when writing; you can set them to zero as shown.
3516
3517You can retrieve the pointer via the function png_get_user_transform_ptr().
3518For example:
3519
3520    voidp write_user_transform_ptr =
3521       png_get_user_transform_ptr(png_ptr);
3522
3523It is possible to have libpng flush any pending output, either manually,
3524or automatically after a certain number of lines have been written.  To
3525flush the output stream a single time call:
3526
3527    png_write_flush(png_ptr);
3528
3529and to have libpng flush the output stream periodically after a certain
3530number of scanlines have been written, call:
3531
3532    png_set_flush(png_ptr, nrows);
3533
3534Note that the distance between rows is from the last time png_write_flush()
3535was called, or the first row of the image if it has never been called.
3536So if you write 50 lines, and then png_set_flush 25, it will flush the
3537output on the next scanline, and every 25 lines thereafter, unless
3538png_write_flush() is called before 25 more lines have been written.
3539If nrows is too small (less than about 10 lines for a 640 pixel wide
3540RGB image) the image compression may decrease noticeably (although this
3541may be acceptable for real-time applications).  Infrequent flushing will
3542only degrade the compression performance by a few percent over images
3543that do not use flushing.
3544
3545Writing the image data
3546
3547That's it for the transformations.  Now you can write the image data.
3548The simplest way to do this is in one function call.  If you have the
3549whole image in memory, you can just call png_write_image() and libpng
3550will write the image.  You will need to pass in an array of pointers to
3551each row.  This function automatically handles interlacing, so you don't
3552need to call png_set_interlace_handling() or call this function multiple
3553times, or any of that other stuff necessary with png_write_rows().
3554
3555    png_write_image(png_ptr, row_pointers);
3556
3557where row_pointers is:
3558
3559    png_byte *row_pointers[height];
3560
3561You can point to void or char or whatever you use for pixels.
3562
3563If you don't want to write the whole image at once, you can
3564use png_write_rows() instead.  If the file is not interlaced,
3565this is simple:
3566
3567    png_write_rows(png_ptr, row_pointers,
3568       number_of_rows);
3569
3570row_pointers is the same as in the png_write_image() call.
3571
3572If you are just writing one row at a time, you can do this with
3573a single row_pointer instead of an array of row_pointers:
3574
3575    png_bytep row_pointer = row;
3576
3577    png_write_row(png_ptr, row_pointer);
3578
3579When the file is interlaced, things can get a good deal more complicated.
3580The only currently (as of the PNG Specification version 1.2, dated July
35811999) defined interlacing scheme for PNG files is the "Adam7" interlace
3582scheme, that breaks down an image into seven smaller images of varying
3583size.  libpng will build these images for you, or you can do them
3584yourself.  If you want to build them yourself, see the PNG specification
3585for details of which pixels to write when.
3586
3587If you don't want libpng to handle the interlacing details, just
3588use png_set_interlace_handling() and call png_write_rows() the
3589correct number of times to write all the sub-images
3590(png_set_interlace_handling() returns the number of sub-images.)
3591
3592If you want libpng to build the sub-images, call this before you start
3593writing any rows:
3594
3595    number_of_passes = png_set_interlace_handling(png_ptr);
3596
3597This will return the number of passes needed.  Currently, this is seven,
3598but may change if another interlace type is added.
3599
3600Then write the complete image number_of_passes times.
3601
3602    png_write_rows(png_ptr, row_pointers, number_of_rows);
3603
3604Think carefully before you write an interlaced image.  Typically code that
3605reads such images reads all the image data into memory, uncompressed, before
3606doing any processing.  Only code that can display an image on the fly can
3607take advantage of the interlacing and even then the image has to be exactly
3608the correct size for the output device, because scaling an image requires
3609adjacent pixels and these are not available until all the passes have been
3610read.
3611
3612If you do write an interlaced image you will hardly ever need to handle
3613the interlacing yourself.  Call png_set_interlace_handling() and use the
3614approach described above.
3615
3616The only time it is conceivable that you will really need to write an
3617interlaced image pass-by-pass is when you have read one pass by pass and
3618made some pixel-by-pixel transformation to it, as described in the read
3619code above.  In this case use the PNG_PASS_ROWS and PNG_PASS_COLS macros
3620to determine the size of each sub-image in turn and simply write the rows
3621you obtained from the read code.
3622
3623Finishing a sequential write
3624
3625After you are finished writing the image, you should finish writing
3626the file.  If you are interested in writing comments or time, you should
3627pass an appropriately filled png_info pointer.  If you are not interested,
3628you can pass NULL.
3629
3630    png_write_end(png_ptr, info_ptr);
3631
3632When you are done, you can free all memory used by libpng like this:
3633
3634    png_destroy_write_struct(&png_ptr, &info_ptr);
3635
3636It is also possible to individually free the info_ptr members that
3637point to libpng-allocated storage with the following function:
3638
3639    png_free_data(png_ptr, info_ptr, mask, seq)
3640
3641    mask  - identifies data to be freed, a mask
3642            containing the bitwise OR of one or
3643            more of
3644              PNG_FREE_PLTE, PNG_FREE_TRNS,
3645              PNG_FREE_HIST, PNG_FREE_ICCP,
3646              PNG_FREE_PCAL, PNG_FREE_ROWS,
3647              PNG_FREE_SCAL, PNG_FREE_SPLT,
3648              PNG_FREE_TEXT, PNG_FREE_UNKN,
3649            or simply PNG_FREE_ALL
3650
3651    seq   - sequence number of item to be freed
3652            (-1 for all items)
3653
3654This function may be safely called when the relevant storage has
3655already been freed, or has not yet been allocated, or was allocated
3656by the user  and not by libpng,  and will in those cases do nothing.
3657The "seq" parameter is ignored if only one item of the selected data
3658type, such as PLTE, is allowed.  If "seq" is not -1, and multiple items
3659are allowed for the data type identified in the mask, such as text or
3660sPLT, only the n'th item in the structure is freed, where n is "seq".
3661
3662If you allocated data such as a palette that you passed in to libpng
3663with png_set_*, you must not free it until just before the call to
3664png_destroy_write_struct().
3665
3666The default behavior is only to free data that was allocated internally
3667by libpng.  This can be changed, so that libpng will not free the data,
3668or so that it will free data that was allocated by the user with png_malloc()
3669or png_calloc() and passed in via a png_set_*() function, with
3670
3671    png_data_freer(png_ptr, info_ptr, freer, mask)
3672
3673    freer  - one of
3674               PNG_DESTROY_WILL_FREE_DATA
3675               PNG_SET_WILL_FREE_DATA
3676               PNG_USER_WILL_FREE_DATA
3677
3678    mask   - which data elements are affected
3679             same choices as in png_free_data()
3680
3681For example, to transfer responsibility for some data from a read structure
3682to a write structure, you could use
3683
3684    png_data_freer(read_ptr, read_info_ptr,
3685       PNG_USER_WILL_FREE_DATA,
3686       PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)
3687
3688    png_data_freer(write_ptr, write_info_ptr,
3689       PNG_DESTROY_WILL_FREE_DATA,
3690       PNG_FREE_PLTE|PNG_FREE_tRNS|PNG_FREE_hIST)
3691
3692thereby briefly reassigning responsibility for freeing to the user but
3693immediately afterwards reassigning it once more to the write_destroy
3694function.  Having done this, it would then be safe to destroy the read
3695structure and continue to use the PLTE, tRNS, and hIST data in the write
3696structure.
3697
3698This function only affects data that has already been allocated.
3699You can call this function before calling after the png_set_*() functions
3700to control whether the user or png_destroy_*() is supposed to free the data.
3701When the user assumes responsibility for libpng-allocated data, the
3702application must use
3703png_free() to free it, and when the user transfers responsibility to libpng
3704for data that the user has allocated, the user must have used png_malloc()
3705or png_calloc() to allocate it.
3706
3707If you allocated text_ptr.text, text_ptr.lang, and text_ptr.translated_keyword
3708separately, do not transfer responsibility for freeing text_ptr to libpng,
3709because when libpng fills a png_text structure it combines these members with
3710the key member, and png_free_data() will free only text_ptr.key.  Similarly,
3711if you transfer responsibility for free'ing text_ptr from libpng to your
3712application, your application must not separately free those members.
3713For a more compact example of writing a PNG image, see the file example.c.
3714
3715V. Simplified API
3716
3717The simplified API, which became available in libpng-1.6.0, hides the details
3718of both libpng and the PNG file format itself.
3719It allows PNG files to be read into a very limited number of
3720in-memory bitmap formats or to be written from the same formats.  If these
3721formats do not accommodate your needs then you can, and should, use the more
3722sophisticated APIs above - these support a wide variety of in-memory formats
3723and a wide variety of sophisticated transformations to those formats as well
3724as a wide variety of APIs to manipulate ancilliary information.
3725
3726To read a PNG file using the simplified API:
3727
3728  1) Declare a 'png_image' structure (see below) on the stack, set the
3729     version field to PNG_IMAGE_VERSION and the 'opaque' pointer to NULL
3730     (this is REQUIRED, your program may crash if you don't do it.)
3731
3732  2) Call the appropriate png_image_begin_read... function.
3733
3734  3) Set the png_image 'format' member to the required sample format.
3735
3736  4) Allocate a buffer for the image and, if required, the color-map.
3737
3738  5) Call png_image_finish_read to read the image and, if required, the
3739     color-map into your buffers.
3740
3741There are no restrictions on the format of the PNG input itself; all valid
3742color types, bit depths, and interlace methods are acceptable, and the
3743input image is transformed as necessary to the requested in-memory format
3744during the png_image_finish_read() step.  The only caveat is that if you
3745request a color-mapped image from a PNG that is full-color or makes
3746complex use of an alpha channel the transformation is extremely lossy and the
3747result may look terrible.
3748
3749To write a PNG file using the simplified API:
3750
3751  1) Declare a 'png_image' structure on the stack and memset()
3752     it to all zero.
3753
3754  2) Initialize the members of the structure that describe the
3755     image, setting the 'format' member to the format of the
3756     image samples.
3757
3758  3) Call the appropriate png_image_write... function with a
3759     pointer to the image and, if necessary, the color-map to write
3760     the PNG data.
3761
3762png_image is a structure that describes the in-memory format of an image
3763when it is being read or defines the in-memory format of an image that you
3764need to write.  The "png_image" structure contains the following members:
3765
3766   png_controlp opaque  Initialize to NULL, free with png_image_free
3767   png_uint_32  version Set to PNG_IMAGE_VERSION
3768   png_uint_32  width   Image width in pixels (columns)
3769   png_uint_32  height  Image height in pixels (rows)
3770   png_uint_32  format  Image format as defined below
3771   png_uint_32  flags   A bit mask containing informational flags
3772   png_uint_32  colormap_entries; Number of entries in the color-map
3773   png_uint_32  warning_or_error;
3774   char         message[64];
3775
3776In the event of an error or warning the "warning_or_error"
3777field will be set to a non-zero value and the 'message' field will contain
3778a '\0' terminated string with the libpng error or warning message.  If both
3779warnings and an error were encountered, only the error is recorded.  If there
3780are multiple warnings, only the first one is recorded.
3781
3782The upper 30 bits of the "warning_or_error" value are reserved; the low two
3783bits contain a two bit code such that a value more than 1 indicates a failure
3784in the API just called:
3785
3786   0 - no warning or error
3787   1 - warning
3788   2 - error
3789   3 - error preceded by warning
3790
3791The pixels (samples) of the image have one to four channels whose components
3792have original values in the range 0 to 1.0:
3793
3794  1: A single gray or luminance channel (G).
3795  2: A gray/luminance channel and an alpha channel (GA).
3796  3: Three red, green, blue color channels (RGB).
3797  4: Three color channels and an alpha channel (RGBA).
3798
3799The channels are encoded in one of two ways:
3800
3801  a) As a small integer, value 0..255, contained in a single byte.  For the
3802alpha channel the original value is simply value/255.  For the color or
3803luminance channels the value is encoded according to the sRGB specification
3804and matches the 8-bit format expected by typical display devices.
3805
3806The color/gray channels are not scaled (pre-multiplied) by the alpha
3807channel and are suitable for passing to color management software.
3808
3809  b) As a value in the range 0..65535, contained in a 2-byte integer, in
3810the native byte order of the platform on which the application is running.
3811All channels can be converted to the original value by dividing by 65535; all
3812channels are linear.  Color channels use the RGB encoding (RGB end-points) of
3813the sRGB specification.  This encoding is identified by the
3814PNG_FORMAT_FLAG_LINEAR flag below.
3815
3816When the simplified API needs to convert between sRGB and linear colorspaces,
3817the actual sRGB transfer curve defined in the sRGB specification (see the
3818article at http://en.wikipedia.org/wiki/SRGB) is used, not the gamma=1/2.2
3819approximation used elsewhere in libpng.
3820
3821When an alpha channel is present it is expected to denote pixel coverage
3822of the color or luminance channels and is returned as an associated alpha
3823channel: the color/gray channels are scaled (pre-multiplied) by the alpha
3824value.
3825
3826The samples are either contained directly in the image data, between 1 and 8
3827bytes per pixel according to the encoding, or are held in a color-map indexed
3828by bytes in the image data.  In the case of a color-map the color-map entries
3829are individual samples, encoded as above, and the image data has one byte per
3830pixel to select the relevant sample from the color-map.
3831
3832PNG_FORMAT_*
3833
3834The #defines to be used in png_image::format.  Each #define identifies a
3835particular layout of channel data and, if present, alpha values.  There are
3836separate defines for each of the two component encodings.
3837
3838A format is built up using single bit flag values.  All combinations are
3839valid.  Formats can be built up from the flag values or you can use one of
3840the predefined values below.  When testing formats always use the FORMAT_FLAG
3841macros to test for individual features - future versions of the library may
3842add new flags.
3843
3844When reading or writing color-mapped images the format should be set to the
3845format of the entries in the color-map then png_image_{read,write}_colormap
3846called to read or write the color-map and set the format correctly for the
3847image data.  Do not set the PNG_FORMAT_FLAG_COLORMAP bit directly!
3848
3849NOTE: libpng can be built with particular features disabled. If you see
3850compiler errors because the definition of one of the following flags has been
3851compiled out it is because libpng does not have the required support.  It is
3852possible, however, for the libpng configuration to enable the format on just
3853read or just write; in that case you may see an error at run time.
3854You can guard against this by checking for the definition of the
3855appropriate "_SUPPORTED" macro, one of:
3856
3857   PNG_SIMPLIFIED_{READ,WRITE}_{BGR,AFIRST}_SUPPORTED
3858
3859   PNG_FORMAT_FLAG_ALPHA    format with an alpha channel
3860   PNG_FORMAT_FLAG_COLOR    color format: otherwise grayscale
3861   PNG_FORMAT_FLAG_LINEAR   2-byte channels else 1-byte
3862   PNG_FORMAT_FLAG_COLORMAP image data is color-mapped
3863   PNG_FORMAT_FLAG_BGR      BGR colors, else order is RGB
3864   PNG_FORMAT_FLAG_AFIRST   alpha channel comes first
3865
3866Supported formats are as follows.  Future versions of libpng may support more
3867formats; for compatibility with older versions simply check if the format
3868macro is defined using #ifdef.  These defines describe the in-memory layout
3869of the components of the pixels of the image.
3870
3871First the single byte (sRGB) formats:
3872
3873   PNG_FORMAT_GRAY
3874   PNG_FORMAT_GA
3875   PNG_FORMAT_AG
3876   PNG_FORMAT_RGB
3877   PNG_FORMAT_BGR
3878   PNG_FORMAT_RGBA
3879   PNG_FORMAT_ARGB
3880   PNG_FORMAT_BGRA
3881   PNG_FORMAT_ABGR
3882
3883Then the linear 2-byte formats.  When naming these "Y" is used to
3884indicate a luminance (gray) channel.  The component order within the pixel
3885is always the same - there is no provision for swapping the order of the
3886components in the linear format.  The components are 16-bit integers in
3887the native byte order for your platform, and there is no provision for
3888swapping the bytes to a different endian condition.
3889
3890   PNG_FORMAT_LINEAR_Y
3891   PNG_FORMAT_LINEAR_Y_ALPHA
3892   PNG_FORMAT_LINEAR_RGB
3893   PNG_FORMAT_LINEAR_RGB_ALPHA
3894
3895With color-mapped formats the image data is one byte for each pixel. The byte
3896is an index into the color-map which is formatted as above.  To obtain a
3897color-mapped format it is sufficient just to add the PNG_FOMAT_FLAG_COLORMAP
3898to one of the above definitions, or you can use one of the definitions below.
3899
3900   PNG_FORMAT_RGB_COLORMAP
3901   PNG_FORMAT_BGR_COLORMAP
3902   PNG_FORMAT_RGBA_COLORMAP
3903   PNG_FORMAT_ARGB_COLORMAP
3904   PNG_FORMAT_BGRA_COLORMAP
3905   PNG_FORMAT_ABGR_COLORMAP
3906
3907PNG_IMAGE macros
3908
3909These are convenience macros to derive information from a png_image
3910structure.  The PNG_IMAGE_SAMPLE_ macros return values appropriate to the
3911actual image sample values - either the entries in the color-map or the
3912pixels in the image.  The PNG_IMAGE_PIXEL_ macros return corresponding values
3913for the pixels and will always return 1 for color-mapped formats.  The
3914remaining macros return information about the rows in the image and the
3915complete image.
3916
3917NOTE: All the macros that take a png_image::format parameter are compile time
3918constants if the format parameter is, itself, a constant.  Therefore these
3919macros can be used in array declarations and case labels where required.
3920Similarly the macros are also pre-processor constants (sizeof is not used) so
3921they can be used in #if tests.
3922
3923  PNG_IMAGE_SAMPLE_CHANNELS(fmt)
3924    Returns the total number of channels in a given format: 1..4
3925
3926  PNG_IMAGE_SAMPLE_COMPONENT_SIZE(fmt)
3927    Returns the size in bytes of a single component of a pixel or color-map
3928    entry (as appropriate) in the image: 1 or 2.
3929
3930  PNG_IMAGE_SAMPLE_SIZE(fmt)
3931    This is the size of the sample data for one sample.  If the image is
3932    color-mapped it is the size of one color-map entry (and image pixels are
3933    one byte in size), otherwise it is the size of one image pixel.
3934
3935  PNG_IMAGE_MAXIMUM_COLORMAP_COMPONENTS(fmt)
3936    The maximum size of the color-map required by the format expressed in a
3937    count of components.  This can be used to compile-time allocate a
3938    color-map:
3939
3940    png_uint_16 colormap[PNG_IMAGE_MAXIMUM_COLORMAP_COMPONENTS(linear_fmt)];
3941
3942    png_byte colormap[PNG_IMAGE_MAXIMUM_COLORMAP_COMPONENTS(sRGB_fmt)];
3943
3944    Alternatively use the PNG_IMAGE_COLORMAP_SIZE macro below to use the
3945    information from one of the png_image_begin_read_ APIs and dynamically
3946    allocate the required memory.
3947
3948  PNG_IMAGE_COLORMAP_SIZE(fmt)
3949   The size of the color-map required by the format; this is the size of the
3950   color-map buffer passed to the png_image_{read,write}_colormap APIs. It is
3951   a fixed number determined by the format so can easily be allocated on the
3952   stack if necessary.
3953
3954Corresponding information about the pixels
3955
3956  PNG_IMAGE_PIXEL_CHANNELS(fmt)
3957   The number of separate channels (components) in a pixel; 1 for a
3958   color-mapped image.
3959
3960  PNG_IMAGE_PIXEL_COMPONENT_SIZE(fmt)\
3961   The size, in bytes, of each component in a pixel; 1 for a color-mapped
3962   image.
3963
3964  PNG_IMAGE_PIXEL_SIZE(fmt)
3965   The size, in bytes, of a complete pixel; 1 for a color-mapped image.
3966
3967Information about the whole row, or whole image
3968
3969  PNG_IMAGE_ROW_STRIDE(image)
3970   Returns the total number of components in a single row of the image; this
3971   is the minimum 'row stride', the minimum count of components between each
3972   row.  For a color-mapped image this is the minimum number of bytes in a
3973   row.
3974
3975   If you need the stride measured in bytes, row_stride_bytes is
3976   PNG_IMAGE_ROW_STRIDE(image) * PNG_IMAGE_PIXEL_COMPONENT_SIZE(fmt)
3977   plus any padding bytes that your application might need, for example
3978   to start the next row on a 4-byte boundary.
3979
3980  PNG_IMAGE_BUFFER_SIZE(image, row_stride)
3981   Return the size, in bytes, of an image buffer given a png_image and a row
3982   stride - the number of components to leave space for in each row.
3983
3984  PNG_IMAGE_SIZE(image)
3985   Return the size, in bytes, of the image in memory given just a png_image;
3986   the row stride is the minimum stride required for the image.
3987
3988  PNG_IMAGE_COLORMAP_SIZE(image)
3989   Return the size, in bytes, of the color-map of this image.  If the image
3990   format is not a color-map format this will return a size sufficient for
3991   256 entries in the given format; check PNG_FORMAT_FLAG_COLORMAP if
3992   you don't want to allocate a color-map in this case.
3993
3994PNG_IMAGE_FLAG_*
3995
3996Flags containing additional information about the image are held in
3997the 'flags' field of png_image.
3998
3999  PNG_IMAGE_FLAG_COLORSPACE_NOT_sRGB == 0x01
4000    This indicates the the RGB values of the in-memory bitmap do not
4001    correspond to the red, green and blue end-points defined by sRGB.
4002
4003  PNG_IMAGE_FLAG_FAST == 0x02
4004   On write emphasise speed over compression; the resultant PNG file will be
4005   larger but will be produced significantly faster, particular for large
4006   images.  Do not use this option for images which will be distributed, only
4007   used it when producing intermediate files that will be read back in
4008   repeatedly.  For a typical 24-bit image the option will double the read
4009   speed at the cost of increasing the image size by 25%, however for many
4010   more compressible images the PNG file can be 10 times larger with only a
4011   slight speed gain.
4012
4013  PNG_IMAGE_FLAG_16BIT_sRGB == 0x04
4014    On read if the image is a 16-bit per component image and there is no gAMA
4015    or sRGB chunk assume that the components are sRGB encoded.  Notice that
4016    images output by the simplified API always have gamma information; setting
4017    this flag only affects the interpretation of 16-bit images from an
4018    external source.  It is recommended that the application expose this flag
4019    to the user; the user can normally easily recognize the difference between
4020    linear and sRGB encoding.  This flag has no effect on write - the data
4021    passed to the write APIs must have the correct encoding (as defined
4022    above.)
4023
4024    If the flag is not set (the default) input 16-bit per component data is
4025    assumed to be linear.
4026
4027    NOTE: the flag can only be set after the png_image_begin_read_ call,
4028    because that call initializes the 'flags' field.
4029
4030READ APIs
4031
4032   The png_image passed to the read APIs must have been initialized by setting
4033   the png_controlp field 'opaque' to NULL (or, better, memset the whole thing.)
4034
4035   int png_image_begin_read_from_file( png_imagep image,
4036     const char *file_name)
4037
4038     The named file is opened for read and the image header
4039     is filled in from the PNG header in the file.
4040
4041   int png_image_begin_read_from_stdio (png_imagep image,
4042     FILE* file)
4043
4044      The PNG header is read from the stdio FILE object.
4045
4046   int png_image_begin_read_from_memory(png_imagep image,
4047      png_const_voidp memory, png_size_t size)
4048
4049      The PNG header is read from the given memory buffer.
4050
4051   int png_image_finish_read(png_imagep image,
4052      png_colorp background, void *buffer,
4053      png_int_32 row_stride, void *colormap));
4054
4055      Finish reading the image into the supplied buffer and
4056      clean up the png_image structure.
4057
4058      row_stride is the step, in png_byte or png_uint_16 units
4059      as appropriate, between adjacent rows.  A positive stride
4060      indicates that the top-most row is first in the buffer -
4061      the normal top-down arrangement.  A negative stride
4062      indicates that the bottom-most row is first in the buffer.
4063
4064      background need only be supplied if an alpha channel must
4065      be removed from a png_byte format and the removal is to be
4066      done by compositing on a solid color; otherwise it may be
4067      NULL and any composition will be done directly onto the
4068      buffer.  The value is an sRGB color to use for the
4069      background, for grayscale output the green channel is used.
4070
4071      For linear output removing the alpha channel is always done
4072      by compositing on black.
4073
4074   void png_image_free(png_imagep image)
4075
4076      Free any data allocated by libpng in image->opaque,
4077      setting the pointer to NULL.  May be called at any time
4078      after the structure is initialized.
4079
4080When the simplified API needs to convert between sRGB and linear colorspaces,
4081the actual sRGB transfer curve defined in the sRGB specification (see the
4082article at http://en.wikipedia.org/wiki/SRGB) is used, not the gamma=1/2.2
4083approximation used elsewhere in libpng.
4084
4085WRITE APIS
4086
4087For write you must initialize a png_image structure to describe the image to
4088be written:
4089
4090   version: must be set to PNG_IMAGE_VERSION
4091   opaque: must be initialized to NULL
4092   width: image width in pixels
4093   height: image height in rows
4094   format: the format of the data you wish to write
4095   flags: set to 0 unless one of the defined flags applies; set
4096      PNG_IMAGE_FLAG_COLORSPACE_NOT_sRGB for color format images
4097      where the RGB values do not correspond to the colors in sRGB.
4098   colormap_entries: set to the number of entries in the color-map (0 to 256)
4099
4100   int png_image_write_to_file, (png_imagep image,
4101      const char *file, int convert_to_8bit, const void *buffer,
4102      png_int_32 row_stride, const void *colormap));
4103
4104      Write the image to the named file.
4105
4106   int png_image_write_to_memory (png_imagep image, void *memory,
4107      png_alloc_size_t * PNG_RESTRICT memory_bytes,
4108      int convert_to_8_bit, const void *buffer, ptrdiff_t row_stride,
4109      const void *colormap));
4110
4111      Write the image to memory.
4112
4113   int png_image_write_to_stdio(png_imagep image, FILE *file,
4114      int convert_to_8_bit, const void *buffer,
4115      png_int_32 row_stride, const void *colormap)
4116
4117      Write the image to the given (FILE*).
4118
4119With all write APIs if image is in one of the linear formats with
4120(png_uint_16) data then setting convert_to_8_bit will cause the output to be
4121a (png_byte) PNG gamma encoded according to the sRGB specification, otherwise
4122a 16-bit linear encoded PNG file is written.
4123
4124With all APIs row_stride is handled as in the read APIs - it is the spacing
4125from one row to the next in component sized units (float) and if negative
4126indicates a bottom-up row layout in the buffer.  If you pass zero, libpng will
4127calculate the row_stride for you from the width and number of channels.
4128
4129Note that the write API does not support interlacing, sub-8-bit pixels,
4130indexed (paletted) images, or most ancillary chunks.
4131
4132VI. Modifying/Customizing libpng
4133
4134There are two issues here.  The first is changing how libpng does
4135standard things like memory allocation, input/output, and error handling.
4136The second deals with more complicated things like adding new chunks,
4137adding new transformations, and generally changing how libpng works.
4138Both of those are compile-time issues; that is, they are generally
4139determined at the time the code is written, and there is rarely a need
4140to provide the user with a means of changing them.
4141
4142Memory allocation, input/output, and error handling
4143
4144All of the memory allocation, input/output, and error handling in libpng
4145goes through callbacks that are user-settable.  The default routines are
4146in pngmem.c, pngrio.c, pngwio.c, and pngerror.c, respectively.  To change
4147these functions, call the appropriate png_set_*_fn() function.
4148
4149Memory allocation is done through the functions png_malloc(), png_calloc(),
4150and png_free().  The png_malloc() and png_free() functions currently just
4151call the standard C functions and png_calloc() calls png_malloc() and then
4152clears the newly allocated memory to zero; note that png_calloc(png_ptr, size)
4153is not the same as the calloc(number, size) function provided by stdlib.h.
4154There is limited support for certain systems with segmented memory
4155architectures and the types of pointers declared by png.h match this; you
4156will have to use appropriate pointers in your application.  If you prefer
4157to use a different method of allocating and freeing data, you can use
4158png_create_read_struct_2() or png_create_write_struct_2() to register your
4159own functions as described above.  These functions also provide a void
4160pointer that can be retrieved via
4161
4162    mem_ptr=png_get_mem_ptr(png_ptr);
4163
4164Your replacement memory functions must have prototypes as follows:
4165
4166    png_voidp malloc_fn(png_structp png_ptr,
4167       png_alloc_size_t size);
4168
4169    void free_fn(png_structp png_ptr, png_voidp ptr);
4170
4171Your malloc_fn() must return NULL in case of failure.  The png_malloc()
4172function will normally call png_error() if it receives a NULL from the
4173system memory allocator or from your replacement malloc_fn().
4174
4175Your free_fn() will never be called with a NULL ptr, since libpng's
4176png_free() checks for NULL before calling free_fn().
4177
4178Input/Output in libpng is done through png_read() and png_write(),
4179which currently just call fread() and fwrite().  The FILE * is stored in
4180png_struct and is initialized via png_init_io().  If you wish to change
4181the method of I/O, the library supplies callbacks that you can set
4182through the function png_set_read_fn() and png_set_write_fn() at run
4183time, instead of calling the png_init_io() function.  These functions
4184also provide a void pointer that can be retrieved via the function
4185png_get_io_ptr().  For example:
4186
4187    png_set_read_fn(png_structp read_ptr,
4188        voidp read_io_ptr, png_rw_ptr read_data_fn)
4189
4190    png_set_write_fn(png_structp write_ptr,
4191        voidp write_io_ptr, png_rw_ptr write_data_fn,
4192        png_flush_ptr output_flush_fn);
4193
4194    voidp read_io_ptr = png_get_io_ptr(read_ptr);
4195    voidp write_io_ptr = png_get_io_ptr(write_ptr);
4196
4197The replacement I/O functions must have prototypes as follows:
4198
4199    void user_read_data(png_structp png_ptr,
4200        png_bytep data, png_size_t length);
4201
4202    void user_write_data(png_structp png_ptr,
4203        png_bytep data, png_size_t length);
4204
4205    void user_flush_data(png_structp png_ptr);
4206
4207The user_read_data() function is responsible for detecting and
4208handling end-of-data errors.
4209
4210Supplying NULL for the read, write, or flush functions sets them back
4211to using the default C stream functions, which expect the io_ptr to
4212point to a standard *FILE structure.  It is probably a mistake
4213to use NULL for one of write_data_fn and output_flush_fn but not both
4214of them, unless you have built libpng with PNG_NO_WRITE_FLUSH defined.
4215It is an error to read from a write stream, and vice versa.
4216
4217Error handling in libpng is done through png_error() and png_warning().
4218Errors handled through png_error() are fatal, meaning that png_error()
4219should never return to its caller.  Currently, this is handled via
4220setjmp() and longjmp() (unless you have compiled libpng with
4221PNG_NO_SETJMP, in which case it is handled via PNG_ABORT()),
4222but you could change this to do things like exit() if you should wish,
4223as long as your function does not return.
4224
4225On non-fatal errors, png_warning() is called
4226to print a warning message, and then control returns to the calling code.
4227By default png_error() and png_warning() print a message on stderr via
4228fprintf() unless the library is compiled with PNG_NO_CONSOLE_IO defined
4229(because you don't want the messages) or PNG_NO_STDIO defined (because
4230fprintf() isn't available).  If you wish to change the behavior of the error
4231functions, you will need to set up your own message callbacks.  These
4232functions are normally supplied at the time that the png_struct is created.
4233It is also possible to redirect errors and warnings to your own replacement
4234functions after png_create_*_struct() has been called by calling:
4235
4236    png_set_error_fn(png_structp png_ptr,
4237        png_voidp error_ptr, png_error_ptr error_fn,
4238        png_error_ptr warning_fn);
4239
4240    png_voidp error_ptr = png_get_error_ptr(png_ptr);
4241
4242If NULL is supplied for either error_fn or warning_fn, then the libpng
4243default function will be used, calling fprintf() and/or longjmp() if a
4244problem is encountered.  The replacement error functions should have
4245parameters as follows:
4246
4247    void user_error_fn(png_structp png_ptr,
4248        png_const_charp error_msg);
4249
4250    void user_warning_fn(png_structp png_ptr,
4251        png_const_charp warning_msg);
4252
4253The motivation behind using setjmp() and longjmp() is the C++ throw and
4254catch exception handling methods.  This makes the code much easier to write,
4255as there is no need to check every return code of every function call.
4256However, there are some uncertainties about the status of local variables
4257after a longjmp, so the user may want to be careful about doing anything
4258after setjmp returns non-zero besides returning itself.  Consult your
4259compiler documentation for more details.  For an alternative approach, you
4260may wish to use the "cexcept" facility (see http://cexcept.sourceforge.net),
4261which is illustrated in pngvalid.c and in contrib/visupng.
4262
4263Beginning in libpng-1.4.0, the png_set_benign_errors() API became available.
4264You can use this to handle certain errors (normally handled as errors)
4265as warnings.
4266
4267    png_set_benign_errors (png_ptr, int allowed);
4268
4269    allowed: 0: treat png_benign_error() as an error.
4270             1: treat png_benign_error() as a warning.
4271
4272As of libpng-1.6.0, the default condition is to treat benign errors as
4273warnings while reading and as errors while writing.
4274
4275Custom chunks
4276
4277If you need to read or write custom chunks, you may need to get deeper
4278into the libpng code.  The library now has mechanisms for storing
4279and writing chunks of unknown type; you can even declare callbacks
4280for custom chunks.  However, this may not be good enough if the
4281library code itself needs to know about interactions between your
4282chunk and existing `intrinsic' chunks.
4283
4284If you need to write a new intrinsic chunk, first read the PNG
4285specification. Acquire a first level of understanding of how it works.
4286Pay particular attention to the sections that describe chunk names,
4287and look at how other chunks were designed, so you can do things
4288similarly.  Second, check out the sections of libpng that read and
4289write chunks.  Try to find a chunk that is similar to yours and use
4290it as a template.  More details can be found in the comments inside
4291the code.  It is best to handle private or unknown chunks in a generic method,
4292via callback functions, instead of by modifying libpng functions. This
4293is illustrated in pngtest.c, which uses a callback function to handle a
4294private "vpAg" chunk and the new "sTER" chunk, which are both unknown to
4295libpng.
4296
4297If you wish to write your own transformation for the data, look through
4298the part of the code that does the transformations, and check out some of
4299the simpler ones to get an idea of how they work.  Try to find a similar
4300transformation to the one you want to add and copy off of it.  More details
4301can be found in the comments inside the code itself.
4302
4303Configuring for gui/windowing platforms:
4304
4305You will need to write new error and warning functions that use the GUI
4306interface, as described previously, and set them to be the error and
4307warning functions at the time that png_create_*_struct() is called,
4308in order to have them available during the structure initialization.
4309They can be changed later via png_set_error_fn().  On some compilers,
4310you may also have to change the memory allocators (png_malloc, etc.).
4311
4312Configuring zlib:
4313
4314There are special functions to configure the compression.  Perhaps the
4315most useful one changes the compression level, which currently uses
4316input compression values in the range 0 - 9.  The library normally
4317uses the default compression level (Z_DEFAULT_COMPRESSION = 6).  Tests
4318have shown that for a large majority of images, compression values in
4319the range 3-6 compress nearly as well as higher levels, and do so much
4320faster.  For online applications it may be desirable to have maximum speed
4321(Z_BEST_SPEED = 1).  With versions of zlib after v0.99, you can also
4322specify no compression (Z_NO_COMPRESSION = 0), but this would create
4323files larger than just storing the raw bitmap.  You can specify the
4324compression level by calling:
4325
4326    #include zlib.h
4327    png_set_compression_level(png_ptr, level);
4328
4329Another useful one is to reduce the memory level used by the library.
4330The memory level defaults to 8, but it can be lowered if you are
4331short on memory (running DOS, for example, where you only have 640K).
4332Note that the memory level does have an effect on compression; among
4333other things, lower levels will result in sections of incompressible
4334data being emitted in smaller stored blocks, with a correspondingly
4335larger relative overhead of up to 15% in the worst case.
4336
4337    #include zlib.h
4338    png_set_compression_mem_level(png_ptr, level);
4339
4340The other functions are for configuring zlib.  They are not recommended
4341for normal use and may result in writing an invalid PNG file.  See
4342zlib.h for more information on what these mean.
4343
4344    #include zlib.h
4345    png_set_compression_strategy(png_ptr,
4346        strategy);
4347
4348    png_set_compression_window_bits(png_ptr,
4349        window_bits);
4350
4351    png_set_compression_method(png_ptr, method);
4352
4353This controls the size of the IDAT chunks (default 8192):
4354
4355    png_set_compression_buffer_size(png_ptr, size);
4356
4357As of libpng version 1.5.4, additional APIs became
4358available to set these separately for non-IDAT
4359compressed chunks such as zTXt, iTXt, and iCCP:
4360
4361    #include zlib.h
4362    #if PNG_LIBPNG_VER >= 10504
4363    png_set_text_compression_level(png_ptr, level);
4364
4365    png_set_text_compression_mem_level(png_ptr, level);
4366
4367    png_set_text_compression_strategy(png_ptr,
4368        strategy);
4369
4370    png_set_text_compression_window_bits(png_ptr,
4371        window_bits);
4372
4373    png_set_text_compression_method(png_ptr, method);
4374    #endif
4375
4376Controlling row filtering
4377
4378If you want to control whether libpng uses filtering or not, which
4379filters are used, and how it goes about picking row filters, you
4380can call one of these functions.  The selection and configuration
4381of row filters can have a significant impact on the size and
4382encoding speed and a somewhat lesser impact on the decoding speed
4383of an image.  Filtering is enabled by default for RGB and grayscale
4384images (with and without alpha), but not for paletted images nor
4385for any images with bit depths less than 8 bits/pixel.
4386
4387The 'method' parameter sets the main filtering method, which is
4388currently only '0' in the PNG 1.2 specification.  The 'filters'
4389parameter sets which filter(s), if any, should be used for each
4390scanline.  Possible values are PNG_ALL_FILTERS and PNG_NO_FILTERS
4391to turn filtering on and off, respectively.
4392
4393Individual filter types are PNG_FILTER_NONE, PNG_FILTER_SUB,
4394PNG_FILTER_UP, PNG_FILTER_AVG, PNG_FILTER_PAETH, which can be bitwise
4395ORed together with '|' to specify one or more filters to use.
4396These filters are described in more detail in the PNG specification.
4397If you intend to change the filter type during the course of writing
4398the image, you should start with flags set for all of the filters
4399you intend to use so that libpng can initialize its internal
4400structures appropriately for all of the filter types.  (Note that this
4401means the first row must always be adaptively filtered, because libpng
4402currently does not allocate the filter buffers until png_write_row()
4403is called for the first time.)
4404
4405    filters = PNG_FILTER_NONE | PNG_FILTER_SUB
4406              PNG_FILTER_UP | PNG_FILTER_AVG |
4407              PNG_FILTER_PAETH | PNG_ALL_FILTERS;
4408
4409    png_set_filter(png_ptr, PNG_FILTER_TYPE_BASE,
4410       filters);
4411              The second parameter can also be
4412              PNG_INTRAPIXEL_DIFFERENCING if you are
4413              writing a PNG to be embedded in a MNG
4414              datastream.  This parameter must be the
4415              same as the value of filter_method used
4416              in png_set_IHDR().
4417
4418Requesting debug printout
4419
4420The macro definition PNG_DEBUG can be used to request debugging
4421printout.  Set it to an integer value in the range 0 to 3.  Higher
4422numbers result in increasing amounts of debugging information.  The
4423information is printed to the "stderr" file, unless another file
4424name is specified in the PNG_DEBUG_FILE macro definition.
4425
4426When PNG_DEBUG > 0, the following functions (macros) become available:
4427
4428   png_debug(level, message)
4429   png_debug1(level, message, p1)
4430   png_debug2(level, message, p1, p2)
4431
4432in which "level" is compared to PNG_DEBUG to decide whether to print
4433the message, "message" is the formatted string to be printed,
4434and p1 and p2 are parameters that are to be embedded in the string
4435according to printf-style formatting directives.  For example,
4436
4437   png_debug1(2, "foo=%d", foo);
4438
4439is expanded to
4440
4441   if (PNG_DEBUG > 2)
4442      fprintf(PNG_DEBUG_FILE, "foo=%d\n", foo);
4443
4444When PNG_DEBUG is defined but is zero, the macros aren't defined, but you
4445can still use PNG_DEBUG to control your own debugging:
4446
4447   #ifdef PNG_DEBUG
4448       fprintf(stderr, ...
4449   #endif
4450
4451When PNG_DEBUG = 1, the macros are defined, but only png_debug statements
4452having level = 0 will be printed.  There aren't any such statements in
4453this version of libpng, but if you insert some they will be printed.
4454
4455VII.  MNG support
4456
4457The MNG specification (available at http://www.libpng.org/pub/mng) allows
4458certain extensions to PNG for PNG images that are embedded in MNG datastreams.
4459Libpng can support some of these extensions.  To enable them, use the
4460png_permit_mng_features() function:
4461
4462   feature_set = png_permit_mng_features(png_ptr, mask)
4463
4464   mask is a png_uint_32 containing the bitwise OR of the
4465        features you want to enable.  These include
4466        PNG_FLAG_MNG_EMPTY_PLTE
4467        PNG_FLAG_MNG_FILTER_64
4468        PNG_ALL_MNG_FEATURES
4469
4470   feature_set is a png_uint_32 that is the bitwise AND of
4471      your mask with the set of MNG features that is
4472      supported by the version of libpng that you are using.
4473
4474It is an error to use this function when reading or writing a standalone
4475PNG file with the PNG 8-byte signature.  The PNG datastream must be wrapped
4476in a MNG datastream.  As a minimum, it must have the MNG 8-byte signature
4477and the MHDR and MEND chunks.  Libpng does not provide support for these
4478or any other MNG chunks; your application must provide its own support for
4479them.  You may wish to consider using libmng (available at
4480http://www.libmng.com) instead.
4481
4482VIII.  Changes to Libpng from version 0.88
4483
4484It should be noted that versions of libpng later than 0.96 are not
4485distributed by the original libpng author, Guy Schalnat, nor by
4486Andreas Dilger, who had taken over from Guy during 1996 and 1997, and
4487distributed versions 0.89 through 0.96, but rather by another member
4488of the original PNG Group, Glenn Randers-Pehrson.  Guy and Andreas are
4489still alive and well, but they have moved on to other things.
4490
4491The old libpng functions png_read_init(), png_write_init(),
4492png_info_init(), png_read_destroy(), and png_write_destroy() have been
4493moved to PNG_INTERNAL in version 0.95 to discourage their use.  These
4494functions will be removed from libpng version 1.4.0.
4495
4496The preferred method of creating and initializing the libpng structures is
4497via the png_create_read_struct(), png_create_write_struct(), and
4498png_create_info_struct() because they isolate the size of the structures
4499from the application, allow version error checking, and also allow the
4500use of custom error handling routines during the initialization, which
4501the old functions do not.  The functions png_read_destroy() and
4502png_write_destroy() do not actually free the memory that libpng
4503allocated for these structs, but just reset the data structures, so they
4504can be used instead of png_destroy_read_struct() and
4505png_destroy_write_struct() if you feel there is too much system overhead
4506allocating and freeing the png_struct for each image read.
4507
4508Setting the error callbacks via png_set_message_fn() before
4509png_read_init() as was suggested in libpng-0.88 is no longer supported
4510because this caused applications that do not use custom error functions
4511to fail if the png_ptr was not initialized to zero.  It is still possible
4512to set the error callbacks AFTER png_read_init(), or to change them with
4513png_set_error_fn(), which is essentially the same function, but with a new
4514name to force compilation errors with applications that try to use the old
4515method.
4516
4517Support for the sCAL, iCCP, iTXt, and sPLT chunks was added at libpng-1.0.6;
4518however, iTXt support was not enabled by default.
4519
4520Starting with version 1.0.7, you can find out which version of the library
4521you are using at run-time:
4522
4523   png_uint_32 libpng_vn = png_access_version_number();
4524
4525The number libpng_vn is constructed from the major version, minor
4526version with leading zero, and release number with leading zero,
4527(e.g., libpng_vn for version 1.0.7 is 10007).
4528
4529Note that this function does not take a png_ptr, so you can call it
4530before you've created one.
4531
4532You can also check which version of png.h you used when compiling your
4533application:
4534
4535   png_uint_32 application_vn = PNG_LIBPNG_VER;
4536
4537IX.  Changes to Libpng from version 1.0.x to 1.2.x
4538
4539Support for user memory management was enabled by default.  To
4540accomplish this, the functions png_create_read_struct_2(),
4541png_create_write_struct_2(), png_set_mem_fn(), png_get_mem_ptr(),
4542png_malloc_default(), and png_free_default() were added.
4543
4544Support for the iTXt chunk has been enabled by default as of
4545version 1.2.41.
4546
4547Support for certain MNG features was enabled.
4548
4549Support for numbered error messages was added.  However, we never got
4550around to actually numbering the error messages.  The function
4551png_set_strip_error_numbers() was added (Note: the prototype for this
4552function was inadvertently removed from png.h in PNG_NO_ASSEMBLER_CODE
4553builds of libpng-1.2.15.  It was restored in libpng-1.2.36).
4554
4555The png_malloc_warn() function was added at libpng-1.2.3.  This issues
4556a png_warning and returns NULL instead of aborting when it fails to
4557acquire the requested memory allocation.
4558
4559Support for setting user limits on image width and height was enabled
4560by default.  The functions png_set_user_limits(), png_get_user_width_max(),
4561and png_get_user_height_max() were added at libpng-1.2.6.
4562
4563The png_set_add_alpha() function was added at libpng-1.2.7.
4564
4565The function png_set_expand_gray_1_2_4_to_8() was added at libpng-1.2.9.
4566Unlike png_set_gray_1_2_4_to_8(), the new function does not expand the
4567tRNS chunk to alpha. The png_set_gray_1_2_4_to_8() function is
4568deprecated.
4569
4570A number of macro definitions in support of runtime selection of
4571assembler code features (especially Intel MMX code support) were
4572added at libpng-1.2.0:
4573
4574    PNG_ASM_FLAG_MMX_SUPPORT_COMPILED
4575    PNG_ASM_FLAG_MMX_SUPPORT_IN_CPU
4576    PNG_ASM_FLAG_MMX_READ_COMBINE_ROW
4577    PNG_ASM_FLAG_MMX_READ_INTERLACE
4578    PNG_ASM_FLAG_MMX_READ_FILTER_SUB
4579    PNG_ASM_FLAG_MMX_READ_FILTER_UP
4580    PNG_ASM_FLAG_MMX_READ_FILTER_AVG
4581    PNG_ASM_FLAG_MMX_READ_FILTER_PAETH
4582    PNG_ASM_FLAGS_INITIALIZED
4583    PNG_MMX_READ_FLAGS
4584    PNG_MMX_FLAGS
4585    PNG_MMX_WRITE_FLAGS
4586    PNG_MMX_FLAGS
4587
4588We added the following functions in support of runtime
4589selection of assembler code features:
4590
4591    png_get_mmx_flagmask()
4592    png_set_mmx_thresholds()
4593    png_get_asm_flags()
4594    png_get_mmx_bitdepth_threshold()
4595    png_get_mmx_rowbytes_threshold()
4596    png_set_asm_flags()
4597
4598We replaced all of these functions with simple stubs in libpng-1.2.20,
4599when the Intel assembler code was removed due to a licensing issue.
4600
4601These macros are deprecated:
4602
4603    PNG_READ_TRANSFORMS_NOT_SUPPORTED
4604    PNG_PROGRESSIVE_READ_NOT_SUPPORTED
4605    PNG_NO_SEQUENTIAL_READ_SUPPORTED
4606    PNG_WRITE_TRANSFORMS_NOT_SUPPORTED
4607    PNG_READ_ANCILLARY_CHUNKS_NOT_SUPPORTED
4608    PNG_WRITE_ANCILLARY_CHUNKS_NOT_SUPPORTED
4609
4610They have been replaced, respectively, by:
4611
4612    PNG_NO_READ_TRANSFORMS
4613    PNG_NO_PROGRESSIVE_READ
4614    PNG_NO_SEQUENTIAL_READ
4615    PNG_NO_WRITE_TRANSFORMS
4616    PNG_NO_READ_ANCILLARY_CHUNKS
4617    PNG_NO_WRITE_ANCILLARY_CHUNKS
4618
4619PNG_MAX_UINT was replaced with PNG_UINT_31_MAX.  It has been
4620deprecated since libpng-1.0.16 and libpng-1.2.6.
4621
4622The function
4623    png_check_sig(sig, num)
4624was replaced with
4625    !png_sig_cmp(sig, 0, num)
4626It has been deprecated since libpng-0.90.
4627
4628The function
4629    png_set_gray_1_2_4_to_8()
4630which also expands tRNS to alpha was replaced with
4631    png_set_expand_gray_1_2_4_to_8()
4632which does not. It has been deprecated since libpng-1.0.18 and 1.2.9.
4633
4634X.  Changes to Libpng from version 1.0.x/1.2.x to 1.4.x
4635
4636Private libpng prototypes and macro definitions were moved from
4637png.h and pngconf.h into a new pngpriv.h header file.
4638
4639Functions png_set_benign_errors(), png_benign_error(), and
4640png_chunk_benign_error() were added.
4641
4642Support for setting the maximum amount of memory that the application
4643will allocate for reading chunks was added, as a security measure.
4644The functions png_set_chunk_cache_max() and png_get_chunk_cache_max()
4645were added to the library.
4646
4647We implemented support for I/O states by adding png_ptr member io_state
4648and functions png_get_io_chunk_name() and png_get_io_state() in pngget.c
4649
4650We added PNG_TRANSFORM_GRAY_TO_RGB to the available high-level
4651input transforms.
4652
4653Checking for and reporting of errors in the IHDR chunk is more thorough.
4654
4655Support for global arrays was removed, to improve thread safety.
4656
4657Some obsolete/deprecated macros and functions have been removed.
4658
4659Typecasted NULL definitions such as
4660   #define png_voidp_NULL            (png_voidp)NULL
4661were eliminated.  If you used these in your application, just use
4662NULL instead.
4663
4664The png_struct and info_struct members "trans" and "trans_values" were
4665changed to "trans_alpha" and "trans_color", respectively.
4666
4667The obsolete, unused pnggccrd.c and pngvcrd.c files and related makefiles
4668were removed.
4669
4670The PNG_1_0_X and PNG_1_2_X macros were eliminated.
4671
4672The PNG_LEGACY_SUPPORTED macro was eliminated.
4673
4674Many WIN32_WCE #ifdefs were removed.
4675
4676The functions png_read_init(info_ptr), png_write_init(info_ptr),
4677png_info_init(info_ptr), png_read_destroy(), and png_write_destroy()
4678have been removed.  They have been deprecated since libpng-0.95.
4679
4680The png_permit_empty_plte() was removed. It has been deprecated
4681since libpng-1.0.9.  Use png_permit_mng_features() instead.
4682
4683We removed the obsolete stub functions png_get_mmx_flagmask(),
4684png_set_mmx_thresholds(), png_get_asm_flags(),
4685png_get_mmx_bitdepth_threshold(), png_get_mmx_rowbytes_threshold(),
4686png_set_asm_flags(), and png_mmx_supported()
4687
4688We removed the obsolete png_check_sig(), png_memcpy_check(), and
4689png_memset_check() functions.  Instead use !png_sig_cmp(), memcpy(),
4690and memset(), respectively.
4691
4692The function png_set_gray_1_2_4_to_8() was removed. It has been
4693deprecated since libpng-1.0.18 and 1.2.9, when it was replaced with
4694png_set_expand_gray_1_2_4_to_8() because the former function also
4695expanded any tRNS chunk to an alpha channel.
4696
4697Macros for png_get_uint_16, png_get_uint_32, and png_get_int_32
4698were added and are used by default instead of the corresponding
4699functions. Unfortunately,
4700from libpng-1.4.0 until 1.4.4, the png_get_uint_16 macro (but not the
4701function) incorrectly returned a value of type png_uint_32.
4702
4703We changed the prototype for png_malloc() from
4704    png_malloc(png_structp png_ptr, png_uint_32 size)
4705to
4706    png_malloc(png_structp png_ptr, png_alloc_size_t size)
4707
4708This also applies to the prototype for the user replacement malloc_fn().
4709
4710The png_calloc() function was added and is used in place of
4711of "png_malloc(); memset();" except in the case in png_read_png()
4712where the array consists of pointers; in this case a "for" loop is used
4713after the png_malloc() to set the pointers to NULL, to give robust.
4714behavior in case the application runs out of memory part-way through
4715the process.
4716
4717We changed the prototypes of png_get_compression_buffer_size() and
4718png_set_compression_buffer_size() to work with png_size_t instead of
4719png_uint_32.
4720
4721Support for numbered error messages was removed by default, since we
4722never got around to actually numbering the error messages. The function
4723png_set_strip_error_numbers() was removed from the library by default.
4724
4725The png_zalloc() and png_zfree() functions are no longer exported.
4726The png_zalloc() function no longer zeroes out the memory that it
4727allocates.  Applications that called png_zalloc(png_ptr, number, size)
4728can call png_calloc(png_ptr, number*size) instead, and can call
4729png_free() instead of png_zfree().
4730
4731Support for dithering was disabled by default in libpng-1.4.0, because
4732it has not been well tested and doesn't actually "dither".
4733The code was not
4734removed, however, and could be enabled by building libpng with
4735PNG_READ_DITHER_SUPPORTED defined.  In libpng-1.4.2, this support
4736was re-enabled, but the function was renamed png_set_quantize() to
4737reflect more accurately what it actually does.  At the same time,
4738the PNG_DITHER_[RED,GREEN_BLUE]_BITS macros were also renamed to
4739PNG_QUANTIZE_[RED,GREEN,BLUE]_BITS, and PNG_READ_DITHER_SUPPORTED
4740was renamed to PNG_READ_QUANTIZE_SUPPORTED.
4741
4742We removed the trailing '.' from the warning and error messages.
4743
4744XI.  Changes to Libpng from version 1.4.x to 1.5.x
4745
4746From libpng-1.4.0 until 1.4.4, the png_get_uint_16 macro (but not the
4747function) incorrectly returned a value of type png_uint_32.
4748The incorrect macro was removed from libpng-1.4.5.
4749
4750Checking for invalid palette index on write was added at libpng
47511.5.10.  If a pixel contains an invalid (out-of-range) index libpng issues
4752a benign error.  This is enabled by default because this condition is an
4753error according to the PNG specification, Clause 11.3.2, but the error can
4754be ignored in each png_ptr with
4755
4756   png_set_check_for_invalid_index(png_ptr, allowed);
4757
4758      allowed  - one of
4759                 0: disable benign error (accept the
4760                    invalid data without warning).
4761                 1: enable benign error (treat the
4762                    invalid data as an error or a
4763                    warning).
4764
4765If the error is ignored, or if png_benign_error() treats it as a warning,
4766any invalid pixels are decoded as opaque black by the decoder and written
4767as-is by the encoder.
4768
4769Retrieving the maximum palette index found was added at libpng-1.5.15.
4770This statement must appear after png_read_png() or png_read_image() while
4771reading, and after png_write_png() or png_write_image() while writing.
4772
4773   int max_palette = png_get_palette_max(png_ptr, info_ptr);
4774
4775This will return the maximum palette index found in the image, or "-1" if
4776the palette was not checked, or "0" if no palette was found.  Note that this
4777does not account for any palette index used by ancillary chunks such as the
4778bKGD chunk; you must check those separately to determine the maximum
4779palette index actually used.
4780
4781There are no substantial API changes between the non-deprecated parts of
4782the 1.4.5 API and the 1.5.0 API; however, the ability to directly access
4783members of the main libpng control structures, png_struct and png_info,
4784deprecated in earlier versions of libpng, has been completely removed from
4785libpng 1.5, and new private "pngstruct.h", "pnginfo.h", and "pngdebug.h"
4786header files were created.
4787
4788We no longer include zlib.h in png.h.  The include statement has been moved
4789to pngstruct.h, where it is not accessible by applications. Applications that
4790need access to information in zlib.h will need to add the '#include "zlib.h"'
4791directive.  It does not matter whether this is placed prior to or after
4792the '"#include png.h"' directive.
4793
4794The png_sprintf(), png_strcpy(), and png_strncpy() macros are no longer used
4795and were removed.
4796
4797We moved the png_strlen(), png_memcpy(), png_memset(), and png_memcmp()
4798macros into a private header file (pngpriv.h) that is not accessible to
4799applications.
4800
4801In png_get_iCCP, the type of "profile" was changed from png_charpp
4802to png_bytepp, and in png_set_iCCP, from png_charp to png_const_bytep.
4803
4804There are changes of form in png.h, including new and changed macros to
4805declare parts of the API.  Some API functions with arguments that are
4806pointers to data not modified within the function have been corrected to
4807declare these arguments with PNG_CONST.
4808
4809Much of the internal use of C macros to control the library build has also
4810changed and some of this is visible in the exported header files, in
4811particular the use of macros to control data and API elements visible
4812during application compilation may require significant revision to
4813application code.  (It is extremely rare for an application to do this.)
4814
4815Any program that compiled against libpng 1.4 and did not use deprecated
4816features or access internal library structures should compile and work
4817against libpng 1.5, except for the change in the prototype for
4818png_get_iCCP() and png_set_iCCP() API functions mentioned above.
4819
4820libpng 1.5.0 adds PNG_ PASS macros to help in the reading and writing of
4821interlaced images.  The macros return the number of rows and columns in
4822each pass and information that can be used to de-interlace and (if
4823absolutely necessary) interlace an image.
4824
4825libpng 1.5.0 adds an API png_longjmp(png_ptr, value).  This API calls
4826the application-provided png_longjmp_ptr on the internal, but application
4827initialized, longjmp buffer.  It is provided as a convenience to avoid
4828the need to use the png_jmpbuf macro, which had the unnecessary side
4829effect of resetting the internal png_longjmp_ptr value.
4830
4831libpng 1.5.0 includes a complete fixed point API.  By default this is
4832present along with the corresponding floating point API.  In general the
4833fixed point API is faster and smaller than the floating point one because
4834the PNG file format used fixed point, not floating point.  This applies
4835even if the library uses floating point in internal calculations.  A new
4836macro, PNG_FLOATING_ARITHMETIC_SUPPORTED, reveals whether the library
4837uses floating point arithmetic (the default) or fixed point arithmetic
4838internally for performance critical calculations such as gamma correction.
4839In some cases, the gamma calculations may produce slightly different
4840results.  This has changed the results in png_rgb_to_gray and in alpha
4841composition (png_set_background for example). This applies even if the
4842original image was already linear (gamma == 1.0) and, therefore, it is
4843not necessary to linearize the image.  This is because libpng has *not*
4844been changed to optimize that case correctly, yet.
4845
4846Fixed point support for the sCAL chunk comes with an important caveat;
4847the sCAL specification uses a decimal encoding of floating point values
4848and the accuracy of PNG fixed point values is insufficient for
4849representation of these values. Consequently a "string" API
4850(png_get_sCAL_s and png_set_sCAL_s) is the only reliable way of reading
4851arbitrary sCAL chunks in the absence of either the floating point API or
4852internal floating point calculations.  Starting with libpng-1.5.0, both
4853of these functions are present when PNG_sCAL_SUPPORTED is defined.  Prior
4854to libpng-1.5.0, their presence also depended upon PNG_FIXED_POINT_SUPPORTED
4855being defined and PNG_FLOATING_POINT_SUPPORTED not being defined.
4856
4857Applications no longer need to include the optional distribution header
4858file pngusr.h or define the corresponding macros during application
4859build in order to see the correct variant of the libpng API.  From 1.5.0
4860application code can check for the corresponding _SUPPORTED macro:
4861
4862#ifdef PNG_INCH_CONVERSIONS_SUPPORTED
4863   /* code that uses the inch conversion APIs. */
4864#endif
4865
4866This macro will only be defined if the inch conversion functions have been
4867compiled into libpng.  The full set of macros, and whether or not support
4868has been compiled in, are available in the header file pnglibconf.h.
4869This header file is specific to the libpng build.  Notice that prior to
48701.5.0 the _SUPPORTED macros would always have the default definition unless
4871reset by pngusr.h or by explicit settings on the compiler command line.
4872These settings may produce compiler warnings or errors in 1.5.0 because
4873of macro redefinition.
4874
4875Applications can now choose whether to use these macros or to call the
4876corresponding function by defining PNG_USE_READ_MACROS or
4877PNG_NO_USE_READ_MACROS before including png.h.  Notice that this is
4878only supported from 1.5.0; defining PNG_NO_USE_READ_MACROS prior to 1.5.0
4879will lead to a link failure.
4880
4881Prior to libpng-1.5.4, the zlib compressor used the same set of parameters
4882when compressing the IDAT data and textual data such as zTXt and iCCP.
4883In libpng-1.5.4 we reinitialized the zlib stream for each type of data.
4884We added five png_set_text_*() functions for setting the parameters to
4885use with textual data.
4886
4887Prior to libpng-1.5.4, the PNG_READ_16_TO_8_ACCURATE_SCALE_SUPPORTED
4888option was off by default, and slightly inaccurate scaling occurred.
4889This option can no longer be turned off, and the choice of accurate
4890or inaccurate 16-to-8 scaling is by using the new png_set_scale_16_to_8()
4891API for accurate scaling or the old png_set_strip_16_to_8() API for simple
4892chopping.  In libpng-1.5.4, the PNG_READ_16_TO_8_ACCURATE_SCALE_SUPPORTED
4893macro became PNG_READ_SCALE_16_TO_8_SUPPORTED, and the PNG_READ_16_TO_8
4894macro became PNG_READ_STRIP_16_TO_8_SUPPORTED, to enable the two
4895png_set_*_16_to_8() functions separately.
4896
4897Prior to libpng-1.5.4, the png_set_user_limits() function could only be
4898used to reduce the width and height limits from the value of
4899PNG_USER_WIDTH_MAX and PNG_USER_HEIGHT_MAX, although this document said
4900that it could be used to override them.  Now this function will reduce or
4901increase the limits.
4902
4903Starting in libpng-1.5.10, the user limits can be set en masse with the
4904configuration option PNG_SAFE_LIMITS_SUPPORTED.  If this option is enabled,
4905a set of "safe" limits is applied in pngpriv.h.  These can be overridden by
4906application calls to png_set_user_limits(), png_set_user_chunk_cache_max(),
4907and/or png_set_user_malloc_max() that increase or decrease the limits.  Also,
4908in libpng-1.5.10 the default width and height limits were increased
4909from 1,000,000 to 0x7fffffff (i.e., made unlimited).  Therefore, the
4910limits are now
4911                               default      safe
4912   png_user_width_max        0x7fffffff    1,000,000
4913   png_user_height_max       0x7fffffff    1,000,000
4914   png_user_chunk_cache_max  0 (unlimited)   128
4915   png_user_chunk_malloc_max 0 (unlimited) 8,000,000
4916
4917The png_set_option() function (and the "options" member of the png struct) was
4918added to libpng-1.5.15, with option PNG_ARM_NEON.
4919
4920The library now supports a complete fixed point implementation and can
4921thus be used on systems that have no floating point support or very
4922limited or slow support.  Previously gamma correction, an essential part
4923of complete PNG support, required reasonably fast floating point.
4924
4925As part of this the choice of internal implementation has been made
4926independent of the choice of fixed versus floating point APIs and all the
4927missing fixed point APIs have been implemented.
4928
4929The exact mechanism used to control attributes of API functions has
4930changed, as described in the INSTALL file.
4931
4932A new test program, pngvalid, is provided in addition to pngtest.
4933pngvalid validates the arithmetic accuracy of the gamma correction
4934calculations and includes a number of validations of the file format.
4935A subset of the full range of tests is run when "make check" is done
4936(in the 'configure' build.)  pngvalid also allows total allocated memory
4937usage to be evaluated and performs additional memory overwrite validation.
4938
4939Many changes to individual feature macros have been made. The following
4940are the changes most likely to be noticed by library builders who
4941configure libpng:
4942
49431) All feature macros now have consistent naming:
4944
4945#define PNG_NO_feature turns the feature off
4946#define PNG_feature_SUPPORTED turns the feature on
4947
4948pnglibconf.h contains one line for each feature macro which is either:
4949
4950#define PNG_feature_SUPPORTED
4951
4952if the feature is supported or:
4953
4954/*#undef PNG_feature_SUPPORTED*/
4955
4956if it is not.  Library code consistently checks for the 'SUPPORTED' macro.
4957It does not, and libpng applications should not, check for the 'NO' macro
4958which will not normally be defined even if the feature is not supported.
4959The 'NO' macros are only used internally for setting or not setting the
4960corresponding 'SUPPORTED' macros.
4961
4962Compatibility with the old names is provided as follows:
4963
4964PNG_INCH_CONVERSIONS turns on PNG_INCH_CONVERSIONS_SUPPORTED
4965
4966And the following definitions disable the corresponding feature:
4967
4968PNG_SETJMP_NOT_SUPPORTED disables SETJMP
4969PNG_READ_TRANSFORMS_NOT_SUPPORTED disables READ_TRANSFORMS
4970PNG_NO_READ_COMPOSITED_NODIV disables READ_COMPOSITE_NODIV
4971PNG_WRITE_TRANSFORMS_NOT_SUPPORTED disables WRITE_TRANSFORMS
4972PNG_READ_ANCILLARY_CHUNKS_NOT_SUPPORTED disables READ_ANCILLARY_CHUNKS
4973PNG_WRITE_ANCILLARY_CHUNKS_NOT_SUPPORTED disables WRITE_ANCILLARY_CHUNKS
4974
4975Library builders should remove use of the above, inconsistent, names.
4976
49772) Warning and error message formatting was previously conditional on
4978the STDIO feature. The library has been changed to use the
4979CONSOLE_IO feature instead. This means that if CONSOLE_IO is disabled
4980the library no longer uses the printf(3) functions, even though the
4981default read/write implementations use (FILE) style stdio.h functions.
4982
49833) Three feature macros now control the fixed/floating point decisions:
4984
4985PNG_FLOATING_POINT_SUPPORTED enables the floating point APIs
4986
4987PNG_FIXED_POINT_SUPPORTED enables the fixed point APIs; however, in
4988practice these are normally required internally anyway (because the PNG
4989file format is fixed point), therefore in most cases PNG_NO_FIXED_POINT
4990merely stops the function from being exported.
4991
4992PNG_FLOATING_ARITHMETIC_SUPPORTED chooses between the internal floating
4993point implementation or the fixed point one.  Typically the fixed point
4994implementation is larger and slower than the floating point implementation
4995on a system that supports floating point; however, it may be faster on a
4996system which lacks floating point hardware and therefore uses a software
4997emulation.
4998
49994) Added PNG_{READ,WRITE}_INT_FUNCTIONS_SUPPORTED.  This allows the
5000functions to read and write ints to be disabled independently of
5001PNG_USE_READ_MACROS, which allows libpng to be built with the functions
5002even though the default is to use the macros - this allows applications
5003to choose at app buildtime whether or not to use macros (previously
5004impossible because the functions weren't in the default build.)
5005
5006XII.  Changes to Libpng from version 1.5.x to 1.6.x
5007
5008A "simplified API" has been added (see documentation in png.h and a simple
5009example in contrib/examples/pngtopng.c).  The new publicly visible API
5010includes the following:
5011
5012   macros:
5013     PNG_FORMAT_*
5014     PNG_IMAGE_*
5015   structures:
5016     png_control
5017     png_image
5018   read functions
5019     png_image_begin_read_from_file()
5020     png_image_begin_read_from_stdio()
5021     png_image_begin_read_from_memory()
5022     png_image_finish_read()
5023     png_image_free()
5024   write functions
5025     png_image_write_to_file()
5026     png_image_write_to_memory()
5027     png_image_write_to_stdio()
5028
5029Starting with libpng-1.6.0, you can configure libpng to prefix all exported
5030symbols, using the PNG_PREFIX macro.
5031
5032We no longer include string.h in png.h.  The include statement has been moved
5033to pngpriv.h, where it is not accessible by applications.  Applications that
5034need access to information in string.h must add an '#include <string.h>'
5035directive.  It does not matter whether this is placed prior to or after
5036the '#include "png.h"' directive.
5037
5038The following API are now DEPRECATED:
5039   png_info_init_3()
5040   png_convert_to_rfc1123() which has been replaced
5041     with png_convert_to_rfc1123_buffer()
5042   png_malloc_default()
5043   png_free_default()
5044   png_reset_zstream()
5045
5046The following have been removed:
5047   png_get_io_chunk_name(), which has been replaced
5048     with png_get_io_chunk_type().  The new
5049     function returns a 32-bit integer instead of
5050     a string.
5051   The png_sizeof(), png_strlen(), png_memcpy(), png_memcmp(), and
5052     png_memset() macros are no longer used in the libpng sources and
5053     have been removed.  These had already been made invisible to applications
5054     (i.e., defined in the private pngpriv.h header file) since libpng-1.5.0.
5055
5056The signatures of many exported functions were changed, such that
5057   png_structp became png_structrp or png_const_structrp
5058   png_infop became png_inforp or png_const_inforp
5059where "rp" indicates a "restricted pointer".
5060
5061Dropped support for 16-bit platforms. The support for FAR/far types has
5062been eliminated and the definition of png_alloc_size_t is now controlled
5063by a flag so that 'small size_t' systems can select it if necessary.
5064
5065Error detection in some chunks has improved; in particular the iCCP chunk
5066reader now does pretty complete validation of the basic format.  Some bad
5067profiles that were previously accepted are now accepted with a warning or
5068rejected, depending upon the png_set_benign_errors() setting, in particular
5069the very old broken Microsoft/HP 3144-byte sRGB profile.  Starting with
5070libpng-1.6.11, recognizing and checking sRGB profiles can be avoided by
5071means of
5072
5073    #if defined(PNG_SKIP_sRGB_CHECK_PROFILE) && \
5074        defined(PNG_SET_OPTION_SUPPORTED)
5075       png_set_option(png_ptr, PNG_SKIP_sRGB_CHECK_PROFILE,
5076           PNG_OPTION_ON);
5077    #endif
5078
5079It's not a good idea to do this if you are using the "simplified API",
5080which needs to be able to recognize sRGB profiles conveyed via the iCCP
5081chunk.
5082
5083The PNG spec requirement that only grayscale profiles may appear in images
5084with color type 0 or 4 and that even if the image only contains gray pixels,
5085only RGB profiles may appear in images with color type 2, 3, or 6, is now
5086enforced.  The sRGB chunk is allowed to appear in images with any color type
5087and is interpreted by libpng to convey a one-tracer-curve gray profile or a
5088three-tracer-curve RGB profile as appropriate.
5089
5090Libpng 1.5.x erroneously used /MD for Debug DLL builds; if you used the debug
5091builds in your app and you changed your app to use /MD you will need to
5092change it back to /MDd for libpng 1.6.x.
5093
5094Prior to libpng-1.6.0 a warning would be issued if the iTXt chunk contained
5095an empty language field or an empty translated keyword.  Both of these
5096are allowed by the PNG specification, so these warnings are no longer issued.
5097
5098The library now issues an error if the application attempts to set a
5099transform after it calls png_read_update_info() or if it attempts to call
5100both png_read_update_info() and png_start_read_image() or to call either
5101of them more than once.
5102
5103The default condition for benign_errors is now to treat benign errors as
5104warnings while reading and as errors while writing.
5105
5106The library now issues a warning if both background processing and RGB to
5107gray are used when gamma correction happens. As with previous versions of
5108the library the results are numerically very incorrect in this case.
5109
5110There are some minor arithmetic changes in some transforms such as
5111png_set_background(), that might be detected by certain regression tests.
5112
5113Unknown chunk handling has been improved internally, without any API change.
5114This adds more correct option control of the unknown handling, corrects
5115a pre-existing bug where the per-chunk 'keep' setting is ignored, and makes
5116it possible to skip IDAT chunks in the sequential reader.
5117
5118The machine-generated configure files are no longer included in branches
5119libpng16 and later of the GIT repository.  They continue to be included
5120in the tarball releases, however.
5121
5122Libpng-1.6.0 through 1.6.2 used the CMF bytes at the beginning of the IDAT
5123stream to set the size of the sliding window for reading instead of using the
5124default 32-kbyte sliding window size.  It was discovered that there are
5125hundreds of PNG files in the wild that have incorrect CMF bytes that caused
5126zlib to issue the "invalid distance too far back" error and reject the file.
5127Libpng-1.6.3 and later calculate their own safe CMF from the image dimensions,
5128provide a way to revert to the libpng-1.5.x behavior (ignoring the CMF bytes
5129and using a 32-kbyte sliding window), by using
5130
5131    png_set_option(png_ptr, PNG_MAXIMUM_INFLATE_WINDOW,
5132        PNG_OPTION_ON);
5133
5134and provide a tool (contrib/tools/pngfix) for rewriting a PNG file while
5135optimizing the CMF bytes in its IDAT chunk correctly.
5136
5137Libpng-1.6.0 and libpng-1.6.1 wrote uncompressed iTXt chunks with the wrong
5138length, which resulted in PNG files that cannot be read beyond the bad iTXt
5139chunk.  This error was fixed in libpng-1.6.3, and a tool (called
5140contrib/tools/png-fix-itxt) has been added to the libpng distribution.
5141
5142Starting with libpng-1.6.17, the PNG_SAFE_LIMITS macro was eliminated
5143and safe limits are used by default (users who need larger limits
5144can still override them at compile time or run time, as described above).
5145
5146The new limits are
5147                                default   spec limit
5148   png_user_width_max         1,000,000  2,147,483,647
5149   png_user_height_max        1,000,000  2,147,483,647
5150   png_user_chunk_cache_max         128  unlimited
5151   png_user_chunk_malloc_max  8,000,000  unlimited
5152
5153Starting with libpng-1.6.18, a PNG_RELEASE_BUILD macro was added, which allows
5154library builders to control compilation for an installed system (a release build).
5155It can be set for testing debug or beta builds to ensure that they will compile
5156when the build type is switched to RC or STABLE. In essence this overrides the
5157PNG_LIBPNG_BUILD_BASE_TYPE definition which is not directly user controllable.
5158
5159Starting with libpng-1.6.19, attempting to set an over-length PLTE chunk
5160is an error. Previously this requirement of the PNG specification was not
5161enforced, and the palette was always limited to 256 entries. An over-length
5162PLTE chunk found in an input PNG is silently truncated.
5163
5164XIII.  Detecting libpng
5165
5166The png_get_io_ptr() function has been present since libpng-0.88, has never
5167changed, and is unaffected by conditional compilation macros.  It is the
5168best choice for use in configure scripts for detecting the presence of any
5169libpng version since 0.88.  In an autoconf "configure.in" you could use
5170
5171    AC_CHECK_LIB(png, png_get_io_ptr, ...
5172
5173XV. Source code repository
5174
5175Since about February 2009, version 1.2.34, libpng has been under "git" source
5176control.  The git repository was built from old libpng-x.y.z.tar.gz files
5177going back to version 0.70.  You can access the git repository (read only)
5178at
5179
5180    git://git.code.sf.net/p/libpng/code
5181
5182or you can browse it with a web browser by selecting the "code" button at
5183
5184    https://sourceforge.net/projects/libpng
5185
5186Patches can be sent to glennrp at users.sourceforge.net or to
5187png-mng-implement at lists.sourceforge.net or you can upload them to
5188the libpng bug tracker at
5189
5190    http://libpng.sourceforge.net
5191
5192We also accept patches built from the tar or zip distributions, and
5193simple verbal discriptions of bug fixes, reported either to the
5194SourceForge bug tracker, to the png-mng-implement at lists.sf.net
5195mailing list, or directly to glennrp.
5196
5197XV. Coding style
5198
5199Our coding style is similar to the "Allman" style
5200(See http://en.wikipedia.org/wiki/Indent_style#Allman_style), with curly
5201braces on separate lines:
5202
5203    if (condition)
5204    {
5205       action;
5206    }
5207
5208    else if (another condition)
5209    {
5210       another action;
5211    }
5212
5213The braces can be omitted from simple one-line actions:
5214
5215    if (condition)
5216       return (0);
5217
5218We use 3-space indentation, except for continued statements which
5219are usually indented the same as the first line of the statement
5220plus four more spaces.
5221
5222For macro definitions we use 2-space indentation, always leaving the "#"
5223in the first column.
5224
5225    #ifndef PNG_NO_FEATURE
5226    #  ifndef PNG_FEATURE_SUPPORTED
5227    #    define PNG_FEATURE_SUPPORTED
5228    #  endif
5229    #endif
5230
5231Comments appear with the leading "/*" at the same indentation as
5232the statement that follows the comment:
5233
5234    /* Single-line comment */
5235    statement;
5236
5237    /* This is a multiple-line
5238     * comment.
5239     */
5240    statement;
5241
5242Very short comments can be placed after the end of the statement
5243to which they pertain:
5244
5245    statement;    /* comment */
5246
5247We don't use C++ style ("//") comments. We have, however,
5248used them in the past in some now-abandoned MMX assembler
5249code.
5250
5251Functions and their curly braces are not indented, and
5252exported functions are marked with PNGAPI:
5253
5254 /* This is a public function that is visible to
5255  * application programmers. It does thus-and-so.
5256  */
5257 void PNGAPI
5258 png_exported_function(png_ptr, png_info, foo)
5259 {
5260    body;
5261 }
5262
5263The return type and decorations are placed on a separate line
5264ahead of the function name, as illustrated above.
5265
5266The prototypes for all exported functions appear in png.h,
5267above the comment that says
5268
5269    /* Maintainer: Put new public prototypes here ... */
5270
5271We mark all non-exported functions with "/* PRIVATE */"":
5272
5273 void /* PRIVATE */
5274 png_non_exported_function(png_ptr, png_info, foo)
5275 {
5276    body;
5277 }
5278
5279The prototypes for non-exported functions (except for those in
5280pngtest) appear in pngpriv.h above the comment that says
5281
5282  /* Maintainer: Put new private prototypes here ^ */
5283
5284To avoid polluting the global namespace, the names of all exported
5285functions and variables begin with "png_", and all publicly visible C
5286preprocessor macros begin with "PNG".  We request that applications that
5287use libpng *not* begin any of their own symbols with either of these strings.
5288
5289We put a space after the "sizeof" operator and we omit the
5290optional parentheses around its argument when the argument
5291is an expression, not a type name, and we always enclose the
5292sizeof operator, with its argument, in parentheses:
5293
5294  (sizeof (png_uint_32))
5295  (sizeof array)
5296
5297Prior to libpng-1.6.0 we used a "png_sizeof()" macro, formatted as
5298though it were a function.
5299
5300Control keywords if, for, while, and switch are always followed by a space
5301to distinguish them from function calls, which have no trailing space.
5302
5303We put a space after each comma and after each semicolon
5304in "for" statements, and we put spaces before and after each
5305C binary operator and after "for" or "while", and before
5306"?".  We don't put a space between a typecast and the expression
5307being cast, nor do we put one between a function name and the
5308left parenthesis that follows it:
5309
5310    for (i = 2; i > 0; --i)
5311       y[i] = a(x) + (int)b;
5312
5313We prefer #ifdef and #ifndef to #if defined() and #if !defined()
5314when there is only one macro being tested.  We always use parentheses
5315with "defined".
5316
5317We express integer constants that are used as bit masks in hex format,
5318with an even number of lower-case hex digits, and to make them unsigned
5319(e.g., 0x00U, 0xffU, 0x0100U) and long if they are greater than 0x7fff
5320(e.g., 0xffffUL).
5321
5322We prefer to use underscores rather than camelCase in names, except
5323for a few type names that we inherit from zlib.h.
5324
5325We prefer "if (something != 0)" and "if (something == 0)"
5326over "if (something)" and if "(!something)", respectively.
5327
5328We do not use the TAB character for indentation in the C sources.
5329
5330Lines do not exceed 80 characters.
5331
5332Other rules can be inferred by inspecting the libpng source.
5333
5334XVI. Y2K Compliance in libpng
5335
5336Since the PNG Development group is an ad-hoc body, we can't make
5337an official declaration.
5338
5339This is your unofficial assurance that libpng from version 0.71 and
5340upward through 1.6.22beta03 are Y2K compliant.  It is my belief that earlier
5341versions were also Y2K compliant.
5342
5343Libpng only has two year fields.  One is a 2-byte unsigned integer
5344that will hold years up to 65535.  The other, which is deprecated,
5345holds the date in text format, and will hold years up to 9999.
5346
5347The integer is
5348    "png_uint_16 year" in png_time_struct.
5349
5350The string is
5351    "char time_buffer[29]" in png_struct.  This is no longer used
5352in libpng-1.6.x and will be removed from libpng-1.7.0.
5353
5354There are seven time-related functions:
5355
5356    png_convert_to_rfc_1123_buffer() in png.c
5357      (formerly png_convert_to_rfc_1152() in error, and
5358      also formerly png_convert_to_rfc_1123())
5359    png_convert_from_struct_tm() in pngwrite.c, called
5360      in pngwrite.c
5361    png_convert_from_time_t() in pngwrite.c
5362    png_get_tIME() in pngget.c
5363    png_handle_tIME() in pngrutil.c, called in pngread.c
5364    png_set_tIME() in pngset.c
5365    png_write_tIME() in pngwutil.c, called in pngwrite.c
5366
5367All appear to handle dates properly in a Y2K environment.  The
5368png_convert_from_time_t() function calls gmtime() to convert from system
5369clock time, which returns (year - 1900), which we properly convert to
5370the full 4-digit year.  There is a possibility that applications using
5371libpng are not passing 4-digit years into the png_convert_to_rfc_1123()
5372function, or that they are incorrectly passing only a 2-digit year
5373instead of "year - 1900" into the png_convert_from_struct_tm() function,
5374but this is not under our control.  The libpng documentation has always
5375stated that it works with 4-digit years, and the APIs have been
5376documented as such.
5377
5378The tIME chunk itself is also Y2K compliant.  It uses a 2-byte unsigned
5379integer to hold the year, and can hold years as large as 65535.
5380
5381zlib, upon which libpng depends, is also Y2K compliant.  It contains
5382no date-related code.
5383
5384
5385   Glenn Randers-Pehrson
5386   libpng maintainer
5387   PNG Development Group
5388