1<html>
2<head>
3<title>pcre2test specification</title>
4</head>
5<body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6<h1>pcre2test man page</h1>
7<p>
8Return to the <a href="index.html">PCRE2 index page</a>.
9</p>
10<p>
11This page is part of the PCRE2 HTML documentation. It was generated
12automatically from the original man page. If there is any nonsense in it,
13please consult the man page, in case the conversion went wrong.
14<br>
15<ul>
16<li><a name="TOC1" href="#SEC1">SYNOPSIS</a>
17<li><a name="TOC2" href="#SEC2">PCRE2's 8-BIT, 16-BIT AND 32-BIT LIBRARIES</a>
18<li><a name="TOC3" href="#SEC3">INPUT ENCODING</a>
19<li><a name="TOC4" href="#SEC4">COMMAND LINE OPTIONS</a>
20<li><a name="TOC5" href="#SEC5">DESCRIPTION</a>
21<li><a name="TOC6" href="#SEC6">COMMAND LINES</a>
22<li><a name="TOC7" href="#SEC7">MODIFIER SYNTAX</a>
23<li><a name="TOC8" href="#SEC8">PATTERN SYNTAX</a>
24<li><a name="TOC9" href="#SEC9">SUBJECT LINE SYNTAX</a>
25<li><a name="TOC10" href="#SEC10">PATTERN MODIFIERS</a>
26<li><a name="TOC11" href="#SEC11">SUBJECT MODIFIERS</a>
27<li><a name="TOC12" href="#SEC12">THE ALTERNATIVE MATCHING FUNCTION</a>
28<li><a name="TOC13" href="#SEC13">DEFAULT OUTPUT FROM pcre2test</a>
29<li><a name="TOC14" href="#SEC14">OUTPUT FROM THE ALTERNATIVE MATCHING FUNCTION</a>
30<li><a name="TOC15" href="#SEC15">RESTARTING AFTER A PARTIAL MATCH</a>
31<li><a name="TOC16" href="#SEC16">CALLOUTS</a>
32<li><a name="TOC17" href="#SEC17">NON-PRINTING CHARACTERS</a>
33<li><a name="TOC18" href="#SEC18">SAVING AND RESTORING COMPILED PATTERNS</a>
34<li><a name="TOC19" href="#SEC19">SEE ALSO</a>
35<li><a name="TOC20" href="#SEC20">AUTHOR</a>
36<li><a name="TOC21" href="#SEC21">REVISION</a>
37</ul>
38<br><a name="SEC1" href="#TOC1">SYNOPSIS</a><br>
39<P>
40<b>pcre2test [options] [input file [output file]]</b>
41<br>
42<br>
43<b>pcre2test</b> is a test program for the PCRE2 regular expression libraries,
44but it can also be used for experimenting with regular expressions. This
45document describes the features of the test program; for details of the regular
46expressions themselves, see the
47<a href="pcre2pattern.html"><b>pcre2pattern</b></a>
48documentation. For details of the PCRE2 library function calls and their
49options, see the
50<a href="pcre2api.html"><b>pcre2api</b></a>
51documentation.
52</P>
53<P>
54The input for <b>pcre2test</b> is a sequence of regular expression patterns and
55subject strings to be matched. There are also command lines for setting
56defaults and controlling some special actions. The output shows the result of
57each match attempt. Modifiers on external or internal command lines, the
58patterns, and the subject lines specify PCRE2 function options, control how the
59subject is processed, and what output is produced.
60</P>
61<P>
62As the original fairly simple PCRE library evolved, it acquired many different
63features, and as a result, the original <b>pcretest</b> program ended up with a
64lot of options in a messy, arcane syntax for testing all the features. The
65move to the new PCRE2 API provided an opportunity to re-implement the test
66program as <b>pcre2test</b>, with a cleaner modifier syntax. Nevertheless, there
67are still many obscure modifiers, some of which are specifically designed for
68use in conjunction with the test script and data files that are distributed as
69part of PCRE2. All the modifiers are documented here, some without much
70justification, but many of them are unlikely to be of use except when testing
71the libraries.
72</P>
73<br><a name="SEC2" href="#TOC1">PCRE2's 8-BIT, 16-BIT AND 32-BIT LIBRARIES</a><br>
74<P>
75Different versions of the PCRE2 library can be built to support character
76strings that are encoded in 8-bit, 16-bit, or 32-bit code units. One, two, or
77all three of these libraries may be simultaneously installed. The
78<b>pcre2test</b> program can be used to test all the libraries. However, its own
79input and output are always in 8-bit format. When testing the 16-bit or 32-bit
80libraries, patterns and subject strings are converted to 16-bit or 32-bit
81format before being passed to the library functions. Results are converted back
82to 8-bit code units for output.
83</P>
84<P>
85In the rest of this document, the names of library functions and structures
86are given in generic form, for example, <b>pcre_compile()</b>. The actual
87names used in the libraries have a suffix _8, _16, or _32, as appropriate.
88<a name="inputencoding"></a></P>
89<br><a name="SEC3" href="#TOC1">INPUT ENCODING</a><br>
90<P>
91Input to <b>pcre2test</b> is processed line by line, either by calling the C
92library's <b>fgets()</b> function, or via the <b>libreadline</b> library. In some
93Windows environments character 26 (hex 1A) causes an immediate end of file, and
94no further data is read, so this character should be avoided unless you really
95want that action.
96</P>
97<P>
98The input is processed using using C's string functions, so must not
99contain binary zeros, even though in Unix-like environments, <b>fgets()</b>
100treats any bytes other than newline as data characters. An error is generated
101if a binary zero is encountered. By default subject lines are processed for
102backslash escapes, which makes it possible to include any data value in strings
103that are passed to the library for matching. For patterns, there is a facility
104for specifying some or all of the 8-bit input characters as hexadecimal pairs,
105which makes it possible to include binary zeros.
106</P>
107<br><b>
108Input for the 16-bit and 32-bit libraries
109</b><br>
110<P>
111When testing the 16-bit or 32-bit libraries, there is a need to be able to
112generate character code points greater than 255 in the strings that are passed
113to the library. For subject lines, backslash escapes can be used. In addition,
114when the <b>utf</b> modifier (see
115<a href="#optionmodifiers">"Setting compilation options"</a>
116below) is set, the pattern and any following subject lines are interpreted as
117UTF-8 strings and translated to UTF-16 or UTF-32 as appropriate.
118</P>
119<P>
120For non-UTF testing of wide characters, the <b>utf8_input</b> modifier can be
121used. This is mutually exclusive with <b>utf</b>, and is allowed only in 16-bit
122or 32-bit mode. It causes the pattern and following subject lines to be treated
123as UTF-8 according to the original definition (RFC 2279), which allows for
124character values up to 0x7fffffff. Each character is placed in one 16-bit or
12532-bit code unit (in the 16-bit case, values greater than 0xffff cause an error
126to occur).
127</P>
128<P>
129UTF-8 (in its original definition) is not capable of encoding values greater
130than 0x7fffffff, but such values can be handled by the 32-bit library. When
131testing this library in non-UTF mode with <b>utf8_input</b> set, if any
132character is preceded by the byte 0xff (which is an invalid byte in UTF-8)
1330x80000000 is added to the character's value. This is the only way of passing
134such code points in a pattern string. For subject strings, using an escape
135sequence is preferable.
136</P>
137<br><a name="SEC4" href="#TOC1">COMMAND LINE OPTIONS</a><br>
138<P>
139<b>-8</b>
140If the 8-bit library has been built, this option causes it to be used (this is
141the default). If the 8-bit library has not been built, this option causes an
142error.
143</P>
144<P>
145<b>-16</b>
146If the 16-bit library has been built, this option causes it to be used. If only
147the 16-bit library has been built, this is the default. If the 16-bit library
148has not been built, this option causes an error.
149</P>
150<P>
151<b>-32</b>
152If the 32-bit library has been built, this option causes it to be used. If only
153the 32-bit library has been built, this is the default. If the 32-bit library
154has not been built, this option causes an error.
155</P>
156<P>
157<b>-ac</b>
158Behave as if each pattern has the <b>auto_callout</b> modifier, that is, insert
159automatic callouts into every pattern that is compiled.
160</P>
161<P>
162<b>-AC</b>
163As for <b>-ac</b>, but in addition behave as if each subject line has the
164<b>callout_extra</b> modifier, that is, show additional information from
165callouts.
166</P>
167<P>
168<b>-b</b>
169Behave as if each pattern has the <b>fullbincode</b> modifier; the full
170internal binary form of the pattern is output after compilation.
171</P>
172<P>
173<b>-C</b>
174Output the version number of the PCRE2 library, and all available information
175about the optional features that are included, and then exit with zero exit
176code. All other options are ignored. If both -C and -LM are present, whichever
177is first is recognized.
178</P>
179<P>
180<b>-C</b> <i>option</i>
181Output information about a specific build-time option, then exit. This
182functionality is intended for use in scripts such as <b>RunTest</b>. The
183following options output the value and set the exit code as indicated:
184<pre>
185  ebcdic-nl  the code for LF (= NL) in an EBCDIC environment:
186               0x15 or 0x25
187               0 if used in an ASCII environment
188               exit code is always 0
189  linksize   the configured internal link size (2, 3, or 4)
190               exit code is set to the link size
191  newline    the default newline setting:
192               CR, LF, CRLF, ANYCRLF, ANY, or NUL
193               exit code is always 0
194  bsr        the default setting for what \R matches:
195               ANYCRLF or ANY
196               exit code is always 0
197</pre>
198The following options output 1 for true or 0 for false, and set the exit code
199to the same value:
200<pre>
201  backslash-C  \C is supported (not locked out)
202  ebcdic       compiled for an EBCDIC environment
203  jit          just-in-time support is available
204  pcre2-16     the 16-bit library was built
205  pcre2-32     the 32-bit library was built
206  pcre2-8      the 8-bit library was built
207  unicode      Unicode support is available
208</pre>
209If an unknown option is given, an error message is output; the exit code is 0.
210</P>
211<P>
212<b>-d</b>
213Behave as if each pattern has the <b>debug</b> modifier; the internal
214form and information about the compiled pattern is output after compilation;
215<b>-d</b> is equivalent to <b>-b -i</b>.
216</P>
217<P>
218<b>-dfa</b>
219Behave as if each subject line has the <b>dfa</b> modifier; matching is done
220using the <b>pcre2_dfa_match()</b> function instead of the default
221<b>pcre2_match()</b>.
222</P>
223<P>
224<b>-error</b> <i>number[,number,...]</i>
225Call <b>pcre2_get_error_message()</b> for each of the error numbers in the
226comma-separated list, display the resulting messages on the standard output,
227then exit with zero exit code. The numbers may be positive or negative. This is
228a convenience facility for PCRE2 maintainers.
229</P>
230<P>
231<b>-help</b>
232Output a brief summary these options and then exit.
233</P>
234<P>
235<b>-i</b>
236Behave as if each pattern has the <b>info</b> modifier; information about the
237compiled pattern is given after compilation.
238</P>
239<P>
240<b>-jit</b>
241Behave as if each pattern line has the <b>jit</b> modifier; after successful
242compilation, each pattern is passed to the just-in-time compiler, if available.
243</P>
244<P>
245<b>-jitfast</b>
246Behave as if each pattern line has the <b>jitfast</b> modifier; after
247successful compilation, each pattern is passed to the just-in-time compiler, if
248available, and each subject line is passed directly to the JIT matcher via its
249"fast path".
250</P>
251<P>
252<b>-jitverify</b>
253Behave as if each pattern line has the <b>jitverify</b> modifier; after
254successful compilation, each pattern is passed to the just-in-time compiler, if
255available, and the use of JIT for matching is verified.
256</P>
257<P>
258<b>-LM</b>
259List modifiers: write a list of available pattern and subject modifiers to the
260standard output, then exit with zero exit code. All other options are ignored.
261If both -C and -LM are present, whichever is first is recognized.
262</P>
263<P>
264<b>-pattern</b> <i>modifier-list</i>
265Behave as if each pattern line contains the given modifiers.
266</P>
267<P>
268<b>-q</b>
269Do not output the version number of <b>pcre2test</b> at the start of execution.
270</P>
271<P>
272<b>-S</b> <i>size</i>
273On Unix-like systems, set the size of the run-time stack to <i>size</i>
274mebibytes (units of 1024*1024 bytes).
275</P>
276<P>
277<b>-subject</b> <i>modifier-list</i>
278Behave as if each subject line contains the given modifiers.
279</P>
280<P>
281<b>-t</b>
282Run each compile and match many times with a timer, and output the resulting
283times per compile or match. When JIT is used, separate times are given for the
284initial compile and the JIT compile. You can control the number of iterations
285that are used for timing by following <b>-t</b> with a number (as a separate
286item on the command line). For example, "-t 1000" iterates 1000 times. The
287default is to iterate 500,000 times.
288</P>
289<P>
290<b>-tm</b>
291This is like <b>-t</b> except that it times only the matching phase, not the
292compile phase.
293</P>
294<P>
295<b>-T</b> <b>-TM</b>
296These behave like <b>-t</b> and <b>-tm</b>, but in addition, at the end of a run,
297the total times for all compiles and matches are output.
298</P>
299<P>
300<b>-version</b>
301Output the PCRE2 version number and then exit.
302</P>
303<br><a name="SEC5" href="#TOC1">DESCRIPTION</a><br>
304<P>
305If <b>pcre2test</b> is given two filename arguments, it reads from the first and
306writes to the second. If the first name is "-", input is taken from the
307standard input. If <b>pcre2test</b> is given only one argument, it reads from
308that file and writes to stdout. Otherwise, it reads from stdin and writes to
309stdout.
310</P>
311<P>
312When <b>pcre2test</b> is built, a configuration option can specify that it
313should be linked with the <b>libreadline</b> or <b>libedit</b> library. When this
314is done, if the input is from a terminal, it is read using the <b>readline()</b>
315function. This provides line-editing and history facilities. The output from
316the <b>-help</b> option states whether or not <b>readline()</b> will be used.
317</P>
318<P>
319The program handles any number of tests, each of which consists of a set of
320input lines. Each set starts with a regular expression pattern, followed by any
321number of subject lines to be matched against that pattern. In between sets of
322test data, command lines that begin with # may appear. This file format, with
323some restrictions, can also be processed by the <b>perltest.sh</b> script that
324is distributed with PCRE2 as a means of checking that the behaviour of PCRE2
325and Perl is the same. For a specification of <b>perltest.sh</b>, see the
326comments near its beginning. See also the #perltest command below.
327</P>
328<P>
329When the input is a terminal, <b>pcre2test</b> prompts for each line of input,
330using "re&#62;" to prompt for regular expression patterns, and "data&#62;" to prompt
331for subject lines. Command lines starting with # can be entered only in
332response to the "re&#62;" prompt.
333</P>
334<P>
335Each subject line is matched separately and independently. If you want to do
336multi-line matches, you have to use the \n escape sequence (or \r or \r\n,
337etc., depending on the newline setting) in a single line of input to encode the
338newline sequences. There is no limit on the length of subject lines; the input
339buffer is automatically extended if it is too small. There are replication
340features that makes it possible to generate long repetitive pattern or subject
341lines without having to supply them explicitly.
342</P>
343<P>
344An empty line or the end of the file signals the end of the subject lines for a
345test, at which point a new pattern or command line is expected if there is
346still input to be read.
347</P>
348<br><a name="SEC6" href="#TOC1">COMMAND LINES</a><br>
349<P>
350In between sets of test data, a line that begins with # is interpreted as a
351command line. If the first character is followed by white space or an
352exclamation mark, the line is treated as a comment, and ignored. Otherwise, the
353following commands are recognized:
354<pre>
355  #forbid_utf
356</pre>
357Subsequent patterns automatically have the PCRE2_NEVER_UTF and PCRE2_NEVER_UCP
358options set, which locks out the use of the PCRE2_UTF and PCRE2_UCP options and
359the use of (*UTF) and (*UCP) at the start of patterns. This command also forces
360an error if a subsequent pattern contains any occurrences of \P, \p, or \X,
361which are still supported when PCRE2_UTF is not set, but which require Unicode
362property support to be included in the library.
363</P>
364<P>
365This is a trigger guard that is used in test files to ensure that UTF or
366Unicode property tests are not accidentally added to files that are used when
367Unicode support is not included in the library. Setting PCRE2_NEVER_UTF and
368PCRE2_NEVER_UCP as a default can also be obtained by the use of <b>#pattern</b>;
369the difference is that <b>#forbid_utf</b> cannot be unset, and the automatic
370options are not displayed in pattern information, to avoid cluttering up test
371output.
372<pre>
373  #load &#60;filename&#62;
374</pre>
375This command is used to load a set of precompiled patterns from a file, as
376described in the section entitled "Saving and restoring compiled patterns"
377<a href="#saverestore">below.</a>
378<pre>
379  #loadtables &#60;filename&#62;
380</pre>
381This command is used to load a set of binary character tables that can be
382accessed by the tables=3 qualifier. Such tables can be created by the
383<b>pcre2_dftables</b> program with the -b option.
384<pre>
385  #newline_default [&#60;newline-list&#62;]
386</pre>
387When PCRE2 is built, a default newline convention can be specified. This
388determines which characters and/or character pairs are recognized as indicating
389a newline in a pattern or subject string. The default can be overridden when a
390pattern is compiled. The standard test files contain tests of various newline
391conventions, but the majority of the tests expect a single linefeed to be
392recognized as a newline by default. Without special action the tests would fail
393when PCRE2 is compiled with either CR or CRLF as the default newline.
394</P>
395<P>
396The #newline_default command specifies a list of newline types that are
397acceptable as the default. The types must be one of CR, LF, CRLF, ANYCRLF,
398ANY, or NUL (in upper or lower case), for example:
399<pre>
400  #newline_default LF Any anyCRLF
401</pre>
402If the default newline is in the list, this command has no effect. Otherwise,
403except when testing the POSIX API, a <b>newline</b> modifier that specifies the
404first newline convention in the list (LF in the above example) is added to any
405pattern that does not already have a <b>newline</b> modifier. If the newline
406list is empty, the feature is turned off. This command is present in a number
407of the standard test input files.
408</P>
409<P>
410When the POSIX API is being tested there is no way to override the default
411newline convention, though it is possible to set the newline convention from
412within the pattern. A warning is given if the <b>posix</b> or <b>posix_nosub</b>
413modifier is used when <b>#newline_default</b> would set a default for the
414non-POSIX API.
415<pre>
416  #pattern &#60;modifier-list&#62;
417</pre>
418This command sets a default modifier list that applies to all subsequent
419patterns. Modifiers on a pattern can change these settings.
420<pre>
421  #perltest
422</pre>
423This line is used in test files that can also be processed by <b>perltest.sh</b>
424to confirm that Perl gives the same results as PCRE2. Subsequent tests are
425checked for the use of <b>pcre2test</b> features that are incompatible with the
426<b>perltest.sh</b> script.
427</P>
428<P>
429Patterns must use '/' as their delimiter, and only certain modifiers are
430supported. Comment lines, #pattern commands, and #subject commands that set or
431unset "mark" are recognized and acted on. The #perltest, #forbid_utf, and
432#newline_default commands, which are needed in the relevant pcre2test files,
433are silently ignored. All other command lines are ignored, but give a warning
434message. The <b>#perltest</b> command helps detect tests that are accidentally
435put in the wrong file or use the wrong delimiter. For more details of the
436<b>perltest.sh</b> script see the comments it contains.
437<pre>
438  #pop [&#60;modifiers&#62;]
439  #popcopy [&#60;modifiers&#62;]
440</pre>
441These commands are used to manipulate the stack of compiled patterns, as
442described in the section entitled "Saving and restoring compiled patterns"
443<a href="#saverestore">below.</a>
444<pre>
445  #save &#60;filename&#62;
446</pre>
447This command is used to save a set of compiled patterns to a file, as described
448in the section entitled "Saving and restoring compiled patterns"
449<a href="#saverestore">below.</a>
450<pre>
451  #subject &#60;modifier-list&#62;
452</pre>
453This command sets a default modifier list that applies to all subsequent
454subject lines. Modifiers on a subject line can change these settings.
455</P>
456<br><a name="SEC7" href="#TOC1">MODIFIER SYNTAX</a><br>
457<P>
458Modifier lists are used with both pattern and subject lines. Items in a list
459are separated by commas followed by optional white space. Trailing whitespace
460in a modifier list is ignored. Some modifiers may be given for both patterns
461and subject lines, whereas others are valid only for one or the other. Each
462modifier has a long name, for example "anchored", and some of them must be
463followed by an equals sign and a value, for example, "offset=12". Values cannot
464contain comma characters, but may contain spaces. Modifiers that do not take
465values may be preceded by a minus sign to turn off a previous setting.
466</P>
467<P>
468A few of the more common modifiers can also be specified as single letters, for
469example "i" for "caseless". In documentation, following the Perl convention,
470these are written with a slash ("the /i modifier") for clarity. Abbreviated
471modifiers must all be concatenated in the first item of a modifier list. If the
472first item is not recognized as a long modifier name, it is interpreted as a
473sequence of these abbreviations. For example:
474<pre>
475  /abc/ig,newline=cr,jit=3
476</pre>
477This is a pattern line whose modifier list starts with two one-letter modifiers
478(/i and /g). The lower-case abbreviated modifiers are the same as used in Perl.
479</P>
480<br><a name="SEC8" href="#TOC1">PATTERN SYNTAX</a><br>
481<P>
482A pattern line must start with one of the following characters (common symbols,
483excluding pattern meta-characters):
484<pre>
485  / ! " ' ` - = _ : ; , % & @ ~
486</pre>
487This is interpreted as the pattern's delimiter. A regular expression may be
488continued over several input lines, in which case the newline characters are
489included within it. It is possible to include the delimiter within the pattern
490by escaping it with a backslash, for example
491<pre>
492  /abc\/def/
493</pre>
494If you do this, the escape and the delimiter form part of the pattern, but
495since the delimiters are all non-alphanumeric, this does not affect its
496interpretation. If the terminating delimiter is immediately followed by a
497backslash, for example,
498<pre>
499  /abc/\
500</pre>
501then a backslash is added to the end of the pattern. This is done to provide a
502way of testing the error condition that arises if a pattern finishes with a
503backslash, because
504<pre>
505  /abc\/
506</pre>
507is interpreted as the first line of a pattern that starts with "abc/", causing
508pcre2test to read the next line as a continuation of the regular expression.
509</P>
510<P>
511A pattern can be followed by a modifier list (details below).
512</P>
513<br><a name="SEC9" href="#TOC1">SUBJECT LINE SYNTAX</a><br>
514<P>
515Before each subject line is passed to <b>pcre2_match()</b> or
516<b>pcre2_dfa_match()</b>, leading and trailing white space is removed, and the
517line is scanned for backslash escapes, unless the <b>subject_literal</b>
518modifier was set for the pattern. The following provide a means of encoding
519non-printing characters in a visible way:
520<pre>
521  \a         alarm (BEL, \x07)
522  \b         backspace (\x08)
523  \e         escape (\x27)
524  \f         form feed (\x0c)
525  \n         newline (\x0a)
526  \r         carriage return (\x0d)
527  \t         tab (\x09)
528  \v         vertical tab (\x0b)
529  \nnn       octal character (up to 3 octal digits); always
530               a byte unless &#62; 255 in UTF-8 or 16-bit or 32-bit mode
531  \o{dd...}  octal character (any number of octal digits}
532  \xhh       hexadecimal byte (up to 2 hex digits)
533  \x{hh...}  hexadecimal character (any number of hex digits)
534</pre>
535The use of \x{hh...} is not dependent on the use of the <b>utf</b> modifier on
536the pattern. It is recognized always. There may be any number of hexadecimal
537digits inside the braces; invalid values provoke error messages.
538</P>
539<P>
540Note that \xhh specifies one byte rather than one character in UTF-8 mode;
541this makes it possible to construct invalid UTF-8 sequences for testing
542purposes. On the other hand, \x{hh} is interpreted as a UTF-8 character in
543UTF-8 mode, generating more than one byte if the value is greater than 127.
544When testing the 8-bit library not in UTF-8 mode, \x{hh} generates one byte
545for values less than 256, and causes an error for greater values.
546</P>
547<P>
548In UTF-16 mode, all 4-digit \x{hhhh} values are accepted. This makes it
549possible to construct invalid UTF-16 sequences for testing purposes.
550</P>
551<P>
552In UTF-32 mode, all 4- to 8-digit \x{...} values are accepted. This makes it
553possible to construct invalid UTF-32 sequences for testing purposes.
554</P>
555<P>
556There is a special backslash sequence that specifies replication of one or more
557characters:
558<pre>
559  \[&#60;characters&#62;]{&#60;count&#62;}
560</pre>
561This makes it possible to test long strings without having to provide them as
562part of the file. For example:
563<pre>
564  \[abc]{4}
565</pre>
566is converted to "abcabcabcabc". This feature does not support nesting. To
567include a closing square bracket in the characters, code it as \x5D.
568</P>
569<P>
570A backslash followed by an equals sign marks the end of the subject string and
571the start of a modifier list. For example:
572<pre>
573  abc\=notbol,notempty
574</pre>
575If the subject string is empty and \= is followed by whitespace, the line is
576treated as a comment line, and is not used for matching. For example:
577<pre>
578  \= This is a comment.
579  abc\= This is an invalid modifier list.
580</pre>
581A backslash followed by any other non-alphanumeric character just escapes that
582character. A backslash followed by anything else causes an error. However, if
583the very last character in the line is a backslash (and there is no modifier
584list), it is ignored. This gives a way of passing an empty line as data, since
585a real empty line terminates the data input.
586</P>
587<P>
588If the <b>subject_literal</b> modifier is set for a pattern, all subject lines
589that follow are treated as literals, with no special treatment of backslashes.
590No replication is possible, and any subject modifiers must be set as defaults
591by a <b>#subject</b> command.
592</P>
593<br><a name="SEC10" href="#TOC1">PATTERN MODIFIERS</a><br>
594<P>
595There are several types of modifier that can appear in pattern lines. Except
596where noted below, they may also be used in <b>#pattern</b> commands. A
597pattern's modifier list can add to or override default modifiers that were set
598by a previous <b>#pattern</b> command.
599<a name="optionmodifiers"></a></P>
600<br><b>
601Setting compilation options
602</b><br>
603<P>
604The following modifiers set options for <b>pcre2_compile()</b>. Most of them set
605bits in the options argument of that function, but those whose names start with
606PCRE2_EXTRA are additional options that are set in the compile context. For the
607main options, there are some single-letter abbreviations that are the same as
608Perl options. There is special handling for /x: if a second x is present,
609PCRE2_EXTENDED is converted into PCRE2_EXTENDED_MORE as in Perl. A third
610appearance adds PCRE2_EXTENDED as well, though this makes no difference to the
611way <b>pcre2_compile()</b> behaves. See
612<a href="pcre2api.html"><b>pcre2api</b></a>
613for a description of the effects of these options.
614<pre>
615      allow_empty_class         set PCRE2_ALLOW_EMPTY_CLASS
616      allow_surrogate_escapes   set PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES
617      alt_bsux                  set PCRE2_ALT_BSUX
618      alt_circumflex            set PCRE2_ALT_CIRCUMFLEX
619      alt_verbnames             set PCRE2_ALT_VERBNAMES
620      anchored                  set PCRE2_ANCHORED
621      auto_callout              set PCRE2_AUTO_CALLOUT
622      bad_escape_is_literal     set PCRE2_EXTRA_BAD_ESCAPE_IS_LITERAL
623  /i  caseless                  set PCRE2_CASELESS
624      dollar_endonly            set PCRE2_DOLLAR_ENDONLY
625  /s  dotall                    set PCRE2_DOTALL
626      dupnames                  set PCRE2_DUPNAMES
627      endanchored               set PCRE2_ENDANCHORED
628      escaped_cr_is_lf          set PCRE2_EXTRA_ESCAPED_CR_IS_LF
629  /x  extended                  set PCRE2_EXTENDED
630  /xx extended_more             set PCRE2_EXTENDED_MORE
631      extra_alt_bsux            set PCRE2_EXTRA_ALT_BSUX
632      firstline                 set PCRE2_FIRSTLINE
633      literal                   set PCRE2_LITERAL
634      match_line                set PCRE2_EXTRA_MATCH_LINE
635      match_invalid_utf         set PCRE2_MATCH_INVALID_UTF
636      match_unset_backref       set PCRE2_MATCH_UNSET_BACKREF
637      match_word                set PCRE2_EXTRA_MATCH_WORD
638  /m  multiline                 set PCRE2_MULTILINE
639      never_backslash_c         set PCRE2_NEVER_BACKSLASH_C
640      never_ucp                 set PCRE2_NEVER_UCP
641      never_utf                 set PCRE2_NEVER_UTF
642  /n  no_auto_capture           set PCRE2_NO_AUTO_CAPTURE
643      no_auto_possess           set PCRE2_NO_AUTO_POSSESS
644      no_dotstar_anchor         set PCRE2_NO_DOTSTAR_ANCHOR
645      no_start_optimize         set PCRE2_NO_START_OPTIMIZE
646      no_utf_check              set PCRE2_NO_UTF_CHECK
647      ucp                       set PCRE2_UCP
648      ungreedy                  set PCRE2_UNGREEDY
649      use_offset_limit          set PCRE2_USE_OFFSET_LIMIT
650      utf                       set PCRE2_UTF
651</pre>
652As well as turning on the PCRE2_UTF option, the <b>utf</b> modifier causes all
653non-printing characters in output strings to be printed using the \x{hh...}
654notation. Otherwise, those less than 0x100 are output in hex without the curly
655brackets. Setting <b>utf</b> in 16-bit or 32-bit mode also causes pattern and
656subject strings to be translated to UTF-16 or UTF-32, respectively, before
657being passed to library functions.
658<a name="controlmodifiers"></a></P>
659<br><b>
660Setting compilation controls
661</b><br>
662<P>
663The following modifiers affect the compilation process or request information
664about the pattern. There are single-letter abbreviations for some that are
665heavily used in the test files.
666<pre>
667      bsr=[anycrlf|unicode]     specify \R handling
668  /B  bincode                   show binary code without lengths
669      callout_info              show callout information
670      convert=&#60;options&#62;         request foreign pattern conversion
671      convert_glob_escape=c     set glob escape character
672      convert_glob_separator=c  set glob separator character
673      convert_length            set convert buffer length
674      debug                     same as info,fullbincode
675      framesize                 show matching frame size
676      fullbincode               show binary code with lengths
677  /I  info                      show info about compiled pattern
678      hex                       unquoted characters are hexadecimal
679      jit[=&#60;number&#62;]            use JIT
680      jitfast                   use JIT fast path
681      jitverify                 verify JIT use
682      locale=&#60;name&#62;             use this locale
683      max_pattern_length=&#60;n&#62;    set the maximum pattern length
684      memory                    show memory used
685      newline=&#60;type&#62;            set newline type
686      null_context              compile with a NULL context
687      parens_nest_limit=&#60;n&#62;     set maximum parentheses depth
688      posix                     use the POSIX API
689      posix_nosub               use the POSIX API with REG_NOSUB
690      push                      push compiled pattern onto the stack
691      pushcopy                  push a copy onto the stack
692      stackguard=&#60;number&#62;       test the stackguard feature
693      subject_literal           treat all subject lines as literal
694      tables=[0|1|2|3]          select internal tables
695      use_length                do not zero-terminate the pattern
696      utf8_input                treat input as UTF-8
697</pre>
698The effects of these modifiers are described in the following sections.
699</P>
700<br><b>
701Newline and \R handling
702</b><br>
703<P>
704The <b>bsr</b> modifier specifies what \R in a pattern should match. If it is
705set to "anycrlf", \R matches CR, LF, or CRLF only. If it is set to "unicode",
706\R matches any Unicode newline sequence. The default can be specified when
707PCRE2 is built; if it is not, the default is set to Unicode.
708</P>
709<P>
710The <b>newline</b> modifier specifies which characters are to be interpreted as
711newlines, both in the pattern and in subject lines. The type must be one of CR,
712LF, CRLF, ANYCRLF, ANY, or NUL (in upper or lower case).
713</P>
714<br><b>
715Information about a pattern
716</b><br>
717<P>
718The <b>debug</b> modifier is a shorthand for <b>info,fullbincode</b>, requesting
719all available information.
720</P>
721<P>
722The <b>bincode</b> modifier causes a representation of the compiled code to be
723output after compilation. This information does not contain length and offset
724values, which ensures that the same output is generated for different internal
725link sizes and different code unit widths. By using <b>bincode</b>, the same
726regression tests can be used in different environments.
727</P>
728<P>
729The <b>fullbincode</b> modifier, by contrast, <i>does</i> include length and
730offset values. This is used in a few special tests that run only for specific
731code unit widths and link sizes, and is also useful for one-off tests.
732</P>
733<P>
734The <b>info</b> modifier requests information about the compiled pattern
735(whether it is anchored, has a fixed first character, and so on). The
736information is obtained from the <b>pcre2_pattern_info()</b> function. Here are
737some typical examples:
738<pre>
739    re&#62; /(?i)(^a|^b)/m,info
740  Capture group count = 1
741  Compile options: multiline
742  Overall options: caseless multiline
743  First code unit at start or follows newline
744  Subject length lower bound = 1
745
746    re&#62; /(?i)abc/info
747  Capture group count = 0
748  Compile options: &#60;none&#62;
749  Overall options: caseless
750  First code unit = 'a' (caseless)
751  Last code unit = 'c' (caseless)
752  Subject length lower bound = 3
753</pre>
754"Compile options" are those specified by modifiers; "overall options" have
755added options that are taken or deduced from the pattern. If both sets of
756options are the same, just a single "options" line is output; if there are no
757options, the line is omitted. "First code unit" is where any match must start;
758if there is more than one they are listed as "starting code units". "Last code
759unit" is the last literal code unit that must be present in any match. This is
760not necessarily the last character. These lines are omitted if no starting or
761ending code units are recorded. The subject length line is omitted when
762<b>no_start_optimize</b> is set because the minimum length is not calculated
763when it can never be used.
764</P>
765<P>
766The <b>framesize</b> modifier shows the size, in bytes, of the storage frames
767used by <b>pcre2_match()</b> for handling backtracking. The size depends on the
768number of capturing parentheses in the pattern.
769</P>
770<P>
771The <b>callout_info</b> modifier requests information about all the callouts in
772the pattern. A list of them is output at the end of any other information that
773is requested. For each callout, either its number or string is given, followed
774by the item that follows it in the pattern.
775</P>
776<br><b>
777Passing a NULL context
778</b><br>
779<P>
780Normally, <b>pcre2test</b> passes a context block to <b>pcre2_compile()</b>. If
781the <b>null_context</b> modifier is set, however, NULL is passed. This is for
782testing that <b>pcre2_compile()</b> behaves correctly in this case (it uses
783default values).
784</P>
785<br><b>
786Specifying pattern characters in hexadecimal
787</b><br>
788<P>
789The <b>hex</b> modifier specifies that the characters of the pattern, except for
790substrings enclosed in single or double quotes, are to be interpreted as pairs
791of hexadecimal digits. This feature is provided as a way of creating patterns
792that contain binary zeros and other non-printing characters. White space is
793permitted between pairs of digits. For example, this pattern contains three
794characters:
795<pre>
796  /ab 32 59/hex
797</pre>
798Parts of such a pattern are taken literally if quoted. This pattern contains
799nine characters, only two of which are specified in hexadecimal:
800<pre>
801  /ab "literal" 32/hex
802</pre>
803Either single or double quotes may be used. There is no way of including
804the delimiter within a substring. The <b>hex</b> and <b>expand</b> modifiers are
805mutually exclusive.
806</P>
807<br><b>
808Specifying the pattern's length
809</b><br>
810<P>
811By default, patterns are passed to the compiling functions as zero-terminated
812strings but can be passed by length instead of being zero-terminated. The
813<b>use_length</b> modifier causes this to happen. Using a length happens
814automatically (whether or not <b>use_length</b> is set) when <b>hex</b> is set,
815because patterns specified in hexadecimal may contain binary zeros.
816</P>
817<P>
818If <b>hex</b> or <b>use_length</b> is used with the POSIX wrapper API (see
819<a href="#posixwrapper">"Using the POSIX wrapper API"</a>
820below), the REG_PEND extension is used to pass the pattern's length.
821</P>
822<br><b>
823Specifying wide characters in 16-bit and 32-bit modes
824</b><br>
825<P>
826In 16-bit and 32-bit modes, all input is automatically treated as UTF-8 and
827translated to UTF-16 or UTF-32 when the <b>utf</b> modifier is set. For testing
828the 16-bit and 32-bit libraries in non-UTF mode, the <b>utf8_input</b> modifier
829can be used. It is mutually exclusive with <b>utf</b>. Input lines are
830interpreted as UTF-8 as a means of specifying wide characters. More details are
831given in
832<a href="#inputencoding">"Input encoding"</a>
833above.
834</P>
835<br><b>
836Generating long repetitive patterns
837</b><br>
838<P>
839Some tests use long patterns that are very repetitive. Instead of creating a
840very long input line for such a pattern, you can use a special repetition
841feature, similar to the one described for subject lines above. If the
842<b>expand</b> modifier is present on a pattern, parts of the pattern that have
843the form
844<pre>
845  \[&#60;characters&#62;]{&#60;count&#62;}
846</pre>
847are expanded before the pattern is passed to <b>pcre2_compile()</b>. For
848example, \[AB]{6000} is expanded to "ABAB..." 6000 times. This construction
849cannot be nested. An initial "\[" sequence is recognized only if "]{" followed
850by decimal digits and "}" is found later in the pattern. If not, the characters
851remain in the pattern unaltered. The <b>expand</b> and <b>hex</b> modifiers are
852mutually exclusive.
853</P>
854<P>
855If part of an expanded pattern looks like an expansion, but is really part of
856the actual pattern, unwanted expansion can be avoided by giving two values in
857the quantifier. For example, \[AB]{6000,6000} is not recognized as an
858expansion item.
859</P>
860<P>
861If the <b>info</b> modifier is set on an expanded pattern, the result of the
862expansion is included in the information that is output.
863</P>
864<br><b>
865JIT compilation
866</b><br>
867<P>
868Just-in-time (JIT) compiling is a heavyweight optimization that can greatly
869speed up pattern matching. See the
870<a href="pcre2jit.html"><b>pcre2jit</b></a>
871documentation for details. JIT compiling happens, optionally, after a pattern
872has been successfully compiled into an internal form. The JIT compiler converts
873this to optimized machine code. It needs to know whether the match-time options
874PCRE2_PARTIAL_HARD and PCRE2_PARTIAL_SOFT are going to be used, because
875different code is generated for the different cases. See the <b>partial</b>
876modifier in "Subject Modifiers"
877<a href="#subjectmodifiers">below</a>
878for details of how these options are specified for each match attempt.
879</P>
880<P>
881JIT compilation is requested by the <b>jit</b> pattern modifier, which may
882optionally be followed by an equals sign and a number in the range 0 to 7.
883The three bits that make up the number specify which of the three JIT operating
884modes are to be compiled:
885<pre>
886  1  compile JIT code for non-partial matching
887  2  compile JIT code for soft partial matching
888  4  compile JIT code for hard partial matching
889</pre>
890The possible values for the <b>jit</b> modifier are therefore:
891<pre>
892  0  disable JIT
893  1  normal matching only
894  2  soft partial matching only
895  3  normal and soft partial matching
896  4  hard partial matching only
897  6  soft and hard partial matching only
898  7  all three modes
899</pre>
900If no number is given, 7 is assumed. The phrase "partial matching" means a call
901to <b>pcre2_match()</b> with either the PCRE2_PARTIAL_SOFT or the
902PCRE2_PARTIAL_HARD option set. Note that such a call may return a complete
903match; the options enable the possibility of a partial match, but do not
904require it. Note also that if you request JIT compilation only for partial
905matching (for example, jit=2) but do not set the <b>partial</b> modifier on a
906subject line, that match will not use JIT code because none was compiled for
907non-partial matching.
908</P>
909<P>
910If JIT compilation is successful, the compiled JIT code will automatically be
911used when an appropriate type of match is run, except when incompatible
912run-time options are specified. For more details, see the
913<a href="pcre2jit.html"><b>pcre2jit</b></a>
914documentation. See also the <b>jitstack</b> modifier below for a way of
915setting the size of the JIT stack.
916</P>
917<P>
918If the <b>jitfast</b> modifier is specified, matching is done using the JIT
919"fast path" interface, <b>pcre2_jit_match()</b>, which skips some of the sanity
920checks that are done by <b>pcre2_match()</b>, and of course does not work when
921JIT is not supported. If <b>jitfast</b> is specified without <b>jit</b>, jit=7 is
922assumed.
923</P>
924<P>
925If the <b>jitverify</b> modifier is specified, information about the compiled
926pattern shows whether JIT compilation was or was not successful. If
927<b>jitverify</b> is specified without <b>jit</b>, jit=7 is assumed. If JIT
928compilation is successful when <b>jitverify</b> is set, the text "(JIT)" is
929added to the first output line after a match or non match when JIT-compiled
930code was actually used in the match.
931</P>
932<br><b>
933Setting a locale
934</b><br>
935<P>
936The <b>locale</b> modifier must specify the name of a locale, for example:
937<pre>
938  /pattern/locale=fr_FR
939</pre>
940The given locale is set, <b>pcre2_maketables()</b> is called to build a set of
941character tables for the locale, and this is then passed to
942<b>pcre2_compile()</b> when compiling the regular expression. The same tables
943are used when matching the following subject lines. The <b>locale</b> modifier
944applies only to the pattern on which it appears, but can be given in a
945<b>#pattern</b> command if a default is needed. Setting a locale and alternate
946character tables are mutually exclusive.
947</P>
948<br><b>
949Showing pattern memory
950</b><br>
951<P>
952The <b>memory</b> modifier causes the size in bytes of the memory used to hold
953the compiled pattern to be output. This does not include the size of the
954<b>pcre2_code</b> block; it is just the actual compiled data. If the pattern is
955subsequently passed to the JIT compiler, the size of the JIT compiled code is
956also output. Here is an example:
957<pre>
958    re&#62; /a(b)c/jit,memory
959  Memory allocation (code space): 21
960  Memory allocation (JIT code): 1910
961
962</PRE>
963</P>
964<br><b>
965Limiting nested parentheses
966</b><br>
967<P>
968The <b>parens_nest_limit</b> modifier sets a limit on the depth of nested
969parentheses in a pattern. Breaching the limit causes a compilation error.
970The default for the library is set when PCRE2 is built, but <b>pcre2test</b>
971sets its own default of 220, which is required for running the standard test
972suite.
973</P>
974<br><b>
975Limiting the pattern length
976</b><br>
977<P>
978The <b>max_pattern_length</b> modifier sets a limit, in code units, to the
979length of pattern that <b>pcre2_compile()</b> will accept. Breaching the limit
980causes a compilation error. The default is the largest number a PCRE2_SIZE
981variable can hold (essentially unlimited).
982<a name="posixwrapper"></a></P>
983<br><b>
984Using the POSIX wrapper API
985</b><br>
986<P>
987The <b>posix</b> and <b>posix_nosub</b> modifiers cause <b>pcre2test</b> to call
988PCRE2 via the POSIX wrapper API rather than its native API. When
989<b>posix_nosub</b> is used, the POSIX option REG_NOSUB is passed to
990<b>regcomp()</b>. The POSIX wrapper supports only the 8-bit library. Note that
991it does not imply POSIX matching semantics; for more detail see the
992<a href="pcre2posix.html"><b>pcre2posix</b></a>
993documentation. The following pattern modifiers set options for the
994<b>regcomp()</b> function:
995<pre>
996  caseless           REG_ICASE
997  multiline          REG_NEWLINE
998  dotall             REG_DOTALL     )
999  ungreedy           REG_UNGREEDY   ) These options are not part of
1000  ucp                REG_UCP        )   the POSIX standard
1001  utf                REG_UTF8       )
1002</pre>
1003The <b>regerror_buffsize</b> modifier specifies a size for the error buffer that
1004is passed to <b>regerror()</b> in the event of a compilation error. For example:
1005<pre>
1006  /abc/posix,regerror_buffsize=20
1007</pre>
1008This provides a means of testing the behaviour of <b>regerror()</b> when the
1009buffer is too small for the error message. If this modifier has not been set, a
1010large buffer is used.
1011</P>
1012<P>
1013The <b>aftertext</b> and <b>allaftertext</b> subject modifiers work as described
1014below. All other modifiers are either ignored, with a warning message, or cause
1015an error.
1016</P>
1017<P>
1018The pattern is passed to <b>regcomp()</b> as a zero-terminated string by
1019default, but if the <b>use_length</b> or <b>hex</b> modifiers are set, the
1020REG_PEND extension is used to pass it by length.
1021</P>
1022<br><b>
1023Testing the stack guard feature
1024</b><br>
1025<P>
1026The <b>stackguard</b> modifier is used to test the use of
1027<b>pcre2_set_compile_recursion_guard()</b>, a function that is provided to
1028enable stack availability to be checked during compilation (see the
1029<a href="pcre2api.html"><b>pcre2api</b></a>
1030documentation for details). If the number specified by the modifier is greater
1031than zero, <b>pcre2_set_compile_recursion_guard()</b> is called to set up
1032callback from <b>pcre2_compile()</b> to a local function. The argument it
1033receives is the current nesting parenthesis depth; if this is greater than the
1034value given by the modifier, non-zero is returned, causing the compilation to
1035be aborted.
1036</P>
1037<br><b>
1038Using alternative character tables
1039</b><br>
1040<P>
1041The value specified for the <b>tables</b> modifier must be one of the digits 0,
10421, 2, or 3. It causes a specific set of built-in character tables to be passed
1043to <b>pcre2_compile()</b>. This is used in the PCRE2 tests to check behaviour
1044with different character tables. The digit specifies the tables as follows:
1045<pre>
1046  0   do not pass any special character tables
1047  1   the default ASCII tables, as distributed in
1048        pcre2_chartables.c.dist
1049  2   a set of tables defining ISO 8859 characters
1050  3   a set of tables loaded by the #loadtables command
1051</pre>
1052In tables 2, some characters whose codes are greater than 128 are identified as
1053letters, digits, spaces, etc. Tables 3 can be used only after a
1054<b>#loadtables</b> command has loaded them from a binary file. Setting alternate
1055character tables and a locale are mutually exclusive.
1056</P>
1057<br><b>
1058Setting certain match controls
1059</b><br>
1060<P>
1061The following modifiers are really subject modifiers, and are described under
1062"Subject Modifiers" below. However, they may be included in a pattern's
1063modifier list, in which case they are applied to every subject line that is
1064processed with that pattern. These modifiers do not affect the compilation
1065process.
1066<pre>
1067      aftertext                   show text after match
1068      allaftertext                show text after captures
1069      allcaptures                 show all captures
1070      allvector                   show the entire ovector
1071      allusedtext                 show all consulted text
1072      altglobal                   alternative global matching
1073  /g  global                      global matching
1074      jitstack=&#60;n&#62;                set size of JIT stack
1075      mark                        show mark values
1076      replace=&#60;string&#62;            specify a replacement string
1077      startchar                   show starting character when relevant
1078      substitute_callout          use substitution callouts
1079      substitute_extended         use PCRE2_SUBSTITUTE_EXTENDED
1080      substitute_literal          use PCRE2_SUBSTITUTE_LITERAL
1081      substitute_matched          use PCRE2_SUBSTITUTE_MATCHED
1082      substitute_overflow_length  use PCRE2_SUBSTITUTE_OVERFLOW_LENGTH
1083      substitute_replacement_only use PCRE2_SUBSTITUTE_REPLACEMENT_ONLY
1084      substitute_skip=&#60;n&#62;         skip substitution &#60;n&#62;
1085      substitute_stop=&#60;n&#62;         skip substitution &#60;n&#62; and following
1086      substitute_unknown_unset    use PCRE2_SUBSTITUTE_UNKNOWN_UNSET
1087      substitute_unset_empty      use PCRE2_SUBSTITUTE_UNSET_EMPTY
1088</pre>
1089These modifiers may not appear in a <b>#pattern</b> command. If you want them as
1090defaults, set them in a <b>#subject</b> command.
1091</P>
1092<br><b>
1093Specifying literal subject lines
1094</b><br>
1095<P>
1096If the <b>subject_literal</b> modifier is present on a pattern, all the subject
1097lines that it matches are taken as literal strings, with no interpretation of
1098backslashes. It is not possible to set subject modifiers on such lines, but any
1099that are set as defaults by a <b>#subject</b> command are recognized.
1100</P>
1101<br><b>
1102Saving a compiled pattern
1103</b><br>
1104<P>
1105When a pattern with the <b>push</b> modifier is successfully compiled, it is
1106pushed onto a stack of compiled patterns, and <b>pcre2test</b> expects the next
1107line to contain a new pattern (or a command) instead of a subject line. This
1108facility is used when saving compiled patterns to a file, as described in the
1109section entitled "Saving and restoring compiled patterns"
1110<a href="#saverestore">below.</a>
1111If <b>pushcopy</b> is used instead of <b>push</b>, a copy of the compiled
1112pattern is stacked, leaving the original as current, ready to match the
1113following input lines. This provides a way of testing the
1114<b>pcre2_code_copy()</b> function.
1115The <b>push</b> and <b>pushcopy </b> modifiers are incompatible with compilation
1116modifiers such as <b>global</b> that act at match time. Any that are specified
1117are ignored (for the stacked copy), with a warning message, except for
1118<b>replace</b>, which causes an error. Note that <b>jitverify</b>, which is
1119allowed, does not carry through to any subsequent matching that uses a stacked
1120pattern.
1121</P>
1122<br><b>
1123Testing foreign pattern conversion
1124</b><br>
1125<P>
1126The experimental foreign pattern conversion functions in PCRE2 can be tested by
1127setting the <b>convert</b> modifier. Its argument is a colon-separated list of
1128options, which set the equivalent option for the <b>pcre2_pattern_convert()</b>
1129function:
1130<pre>
1131  glob                    PCRE2_CONVERT_GLOB
1132  glob_no_starstar        PCRE2_CONVERT_GLOB_NO_STARSTAR
1133  glob_no_wild_separator  PCRE2_CONVERT_GLOB_NO_WILD_SEPARATOR
1134  posix_basic             PCRE2_CONVERT_POSIX_BASIC
1135  posix_extended          PCRE2_CONVERT_POSIX_EXTENDED
1136  unset                   Unset all options
1137</pre>
1138The "unset" value is useful for turning off a default that has been set by a
1139<b>#pattern</b> command. When one of these options is set, the input pattern is
1140passed to <b>pcre2_pattern_convert()</b>. If the conversion is successful, the
1141result is reflected in the output and then passed to <b>pcre2_compile()</b>. The
1142normal <b>utf</b> and <b>no_utf_check</b> options, if set, cause the
1143PCRE2_CONVERT_UTF and PCRE2_CONVERT_NO_UTF_CHECK options to be passed to
1144<b>pcre2_pattern_convert()</b>.
1145</P>
1146<P>
1147By default, the conversion function is allowed to allocate a buffer for its
1148output. However, if the <b>convert_length</b> modifier is set to a value greater
1149than zero, <b>pcre2test</b> passes a buffer of the given length. This makes it
1150possible to test the length check.
1151</P>
1152<P>
1153The <b>convert_glob_escape</b> and <b>convert_glob_separator</b> modifiers can be
1154used to specify the escape and separator characters for glob processing,
1155overriding the defaults, which are operating-system dependent.
1156<a name="subjectmodifiers"></a></P>
1157<br><a name="SEC11" href="#TOC1">SUBJECT MODIFIERS</a><br>
1158<P>
1159The modifiers that can appear in subject lines and the <b>#subject</b>
1160command are of two types.
1161</P>
1162<br><b>
1163Setting match options
1164</b><br>
1165<P>
1166The following modifiers set options for <b>pcre2_match()</b> or
1167<b>pcre2_dfa_match()</b>. See
1168<a href="pcreapi.html"><b>pcreapi</b></a>
1169for a description of their effects.
1170<pre>
1171      anchored                  set PCRE2_ANCHORED
1172      endanchored               set PCRE2_ENDANCHORED
1173      dfa_restart               set PCRE2_DFA_RESTART
1174      dfa_shortest              set PCRE2_DFA_SHORTEST
1175      no_jit                    set PCRE2_NO_JIT
1176      no_utf_check              set PCRE2_NO_UTF_CHECK
1177      notbol                    set PCRE2_NOTBOL
1178      notempty                  set PCRE2_NOTEMPTY
1179      notempty_atstart          set PCRE2_NOTEMPTY_ATSTART
1180      noteol                    set PCRE2_NOTEOL
1181      partial_hard (or ph)      set PCRE2_PARTIAL_HARD
1182      partial_soft (or ps)      set PCRE2_PARTIAL_SOFT
1183</pre>
1184The partial matching modifiers are provided with abbreviations because they
1185appear frequently in tests.
1186</P>
1187<P>
1188If the <b>posix</b> or <b>posix_nosub</b> modifier was present on the pattern,
1189causing the POSIX wrapper API to be used, the only option-setting modifiers
1190that have any effect are <b>notbol</b>, <b>notempty</b>, and <b>noteol</b>,
1191causing REG_NOTBOL, REG_NOTEMPTY, and REG_NOTEOL, respectively, to be passed to
1192<b>regexec()</b>. The other modifiers are ignored, with a warning message.
1193</P>
1194<P>
1195There is one additional modifier that can be used with the POSIX wrapper. It is
1196ignored (with a warning) if used for non-POSIX matching.
1197<pre>
1198      posix_startend=&#60;n&#62;[:&#60;m&#62;]
1199</pre>
1200This causes the subject string to be passed to <b>regexec()</b> using the
1201REG_STARTEND option, which uses offsets to specify which part of the string is
1202searched. If only one number is given, the end offset is passed as the end of
1203the subject string. For more detail of REG_STARTEND, see the
1204<a href="pcre2posix.html"><b>pcre2posix</b></a>
1205documentation. If the subject string contains binary zeros (coded as escapes
1206such as \x{00} because <b>pcre2test</b> does not support actual binary zeros in
1207its input), you must use <b>posix_startend</b> to specify its length.
1208</P>
1209<br><b>
1210Setting match controls
1211</b><br>
1212<P>
1213The following modifiers affect the matching process or request additional
1214information. Some of them may also be specified on a pattern line (see above),
1215in which case they apply to every subject line that is matched against that
1216pattern.
1217<pre>
1218      aftertext                  show text after match
1219      allaftertext               show text after captures
1220      allcaptures                show all captures
1221      allvector                  show the entire ovector
1222      allusedtext                show all consulted text (non-JIT only)
1223      altglobal                  alternative global matching
1224      callout_capture            show captures at callout time
1225      callout_data=&#60;n&#62;           set a value to pass via callouts
1226      callout_error=&#60;n&#62;[:&#60;m&#62;]    control callout error
1227      callout_extra              show extra callout information
1228      callout_fail=&#60;n&#62;[:&#60;m&#62;]     control callout failure
1229      callout_no_where           do not show position of a callout
1230      callout_none               do not supply a callout function
1231      copy=&#60;number or name&#62;      copy captured substring
1232      depth_limit=&#60;n&#62;            set a depth limit
1233      dfa                        use <b>pcre2_dfa_match()</b>
1234      find_limits                find match and depth limits
1235      get=&#60;number or name&#62;       extract captured substring
1236      getall                     extract all captured substrings
1237  /g  global                     global matching
1238      heap_limit=&#60;n&#62;             set a limit on heap memory (Kbytes)
1239      jitstack=&#60;n&#62;               set size of JIT stack
1240      mark                       show mark values
1241      match_limit=&#60;n&#62;            set a match limit
1242      memory                     show heap memory usage
1243      null_context               match with a NULL context
1244      offset=&#60;n&#62;                 set starting offset
1245      offset_limit=&#60;n&#62;           set offset limit
1246      ovector=&#60;n&#62;                set size of output vector
1247      recursion_limit=&#60;n&#62;        obsolete synonym for depth_limit
1248      replace=&#60;string&#62;           specify a replacement string
1249      startchar                  show startchar when relevant
1250      startoffset=&#60;n&#62;            same as offset=&#60;n&#62;
1251      substitute_callout         use substitution callouts
1252      substitute_extedded        use PCRE2_SUBSTITUTE_EXTENDED
1253      substitute_literal         use PCRE2_SUBSTITUTE_LITERAL
1254      substitute_matched         use PCRE2_SUBSTITUTE_MATCHED
1255      substitute_overflow_length use PCRE2_SUBSTITUTE_OVERFLOW_LENGTH
1256      substitute_replacement_only use PCRE2_SUBSTITUTE_REPLACEMENT_ONLY
1257      substitute_skip=&#60;n&#62;        skip substitution number n
1258      substitute_stop=&#60;n&#62;        skip substitution number n and greater
1259      substitute_unknown_unset   use PCRE2_SUBSTITUTE_UNKNOWN_UNSET
1260      substitute_unset_empty     use PCRE2_SUBSTITUTE_UNSET_EMPTY
1261      zero_terminate             pass the subject as zero-terminated
1262</pre>
1263The effects of these modifiers are described in the following sections. When
1264matching via the POSIX wrapper API, the <b>aftertext</b>, <b>allaftertext</b>,
1265and <b>ovector</b> subject modifiers work as described below. All other
1266modifiers are either ignored, with a warning message, or cause an error.
1267</P>
1268<br><b>
1269Showing more text
1270</b><br>
1271<P>
1272The <b>aftertext</b> modifier requests that as well as outputting the part of
1273the subject string that matched the entire pattern, <b>pcre2test</b> should in
1274addition output the remainder of the subject string. This is useful for tests
1275where the subject contains multiple copies of the same substring. The
1276<b>allaftertext</b> modifier requests the same action for captured substrings as
1277well as the main matched substring. In each case the remainder is output on the
1278following line with a plus character following the capture number.
1279</P>
1280<P>
1281The <b>allusedtext</b> modifier requests that all the text that was consulted
1282during a successful pattern match by the interpreter should be shown, for both
1283full and partial matches. This feature is not supported for JIT matching, and
1284if requested with JIT it is ignored (with a warning message). Setting this
1285modifier affects the output if there is a lookbehind at the start of a match,
1286or, for a complete match, a lookahead at the end, or if \K is used in the
1287pattern. Characters that precede or follow the start and end of the actual
1288match are indicated in the output by '&#60;' or '&#62;' characters underneath them.
1289Here is an example:
1290<pre>
1291    re&#62; /(?&#60;=pqr)abc(?=xyz)/
1292  data&#62; 123pqrabcxyz456\=allusedtext
1293   0: pqrabcxyz
1294      &#60;&#60;&#60;   &#62;&#62;&#62;
1295  data&#62; 123pqrabcxy\=ph,allusedtext
1296  Partial match: pqrabcxy
1297                 &#60;&#60;&#60;
1298</pre>
1299The first, complete match shows that the matched string is "abc", with the
1300preceding and following strings "pqr" and "xyz" having been consulted during
1301the match (when processing the assertions). The partial match can indicate only
1302the preceding string.
1303</P>
1304<P>
1305The <b>startchar</b> modifier requests that the starting character for the match
1306be indicated, if it is different to the start of the matched string. The only
1307time when this occurs is when \K has been processed as part of the match. In
1308this situation, the output for the matched string is displayed from the
1309starting character instead of from the match point, with circumflex characters
1310under the earlier characters. For example:
1311<pre>
1312    re&#62; /abc\Kxyz/
1313  data&#62; abcxyz\=startchar
1314   0: abcxyz
1315      ^^^
1316</pre>
1317Unlike <b>allusedtext</b>, the <b>startchar</b> modifier can be used with JIT.
1318However, these two modifiers are mutually exclusive.
1319</P>
1320<br><b>
1321Showing the value of all capture groups
1322</b><br>
1323<P>
1324The <b>allcaptures</b> modifier requests that the values of all potential
1325captured parentheses be output after a match. By default, only those up to the
1326highest one actually used in the match are output (corresponding to the return
1327code from <b>pcre2_match()</b>). Groups that did not take part in the match
1328are output as "&#60;unset&#62;". This modifier is not relevant for DFA matching (which
1329does no capturing) and does not apply when <b>replace</b> is specified; it is
1330ignored, with a warning message, if present.
1331</P>
1332<br><b>
1333Showing the entire ovector, for all outcomes
1334</b><br>
1335<P>
1336The <b>allvector</b> modifier requests that the entire ovector be shown,
1337whatever the outcome of the match. Compare <b>allcaptures</b>, which shows only
1338up to the maximum number of capture groups for the pattern, and then only for a
1339successful complete non-DFA match. This modifier, which acts after any match
1340result, and also for DFA matching, provides a means of checking that there are
1341no unexpected modifications to ovector fields. Before each match attempt, the
1342ovector is filled with a special value, and if this is found in both elements
1343of a capturing pair, "&#60;unchanged&#62;" is output. After a successful match, this
1344applies to all groups after the maximum capture group for the pattern. In other
1345cases it applies to the entire ovector. After a partial match, the first two
1346elements are the only ones that should be set. After a DFA match, the amount of
1347ovector that is used depends on the number of matches that were found.
1348</P>
1349<br><b>
1350Testing pattern callouts
1351</b><br>
1352<P>
1353A callout function is supplied when <b>pcre2test</b> calls the library matching
1354functions, unless <b>callout_none</b> is specified. Its behaviour can be
1355controlled by various modifiers listed above whose names begin with
1356<b>callout_</b>. Details are given in the section entitled "Callouts"
1357<a href="#callouts">below.</a>
1358Testing callouts from <b>pcre2_substitute()</b> is decribed separately in
1359"Testing the substitution function"
1360<a href="#substitution">below.</a>
1361</P>
1362<br><b>
1363Finding all matches in a string
1364</b><br>
1365<P>
1366Searching for all possible matches within a subject can be requested by the
1367<b>global</b> or <b>altglobal</b> modifier. After finding a match, the matching
1368function is called again to search the remainder of the subject. The difference
1369between <b>global</b> and <b>altglobal</b> is that the former uses the
1370<i>start_offset</i> argument to <b>pcre2_match()</b> or <b>pcre2_dfa_match()</b>
1371to start searching at a new point within the entire string (which is what Perl
1372does), whereas the latter passes over a shortened subject. This makes a
1373difference to the matching process if the pattern begins with a lookbehind
1374assertion (including \b or \B).
1375</P>
1376<P>
1377If an empty string is matched, the next match is done with the
1378PCRE2_NOTEMPTY_ATSTART and PCRE2_ANCHORED flags set, in order to search for
1379another, non-empty, match at the same point in the subject. If this match
1380fails, the start offset is advanced, and the normal match is retried. This
1381imitates the way Perl handles such cases when using the <b>/g</b> modifier or
1382the <b>split()</b> function. Normally, the start offset is advanced by one
1383character, but if the newline convention recognizes CRLF as a newline, and the
1384current character is CR followed by LF, an advance of two characters occurs.
1385</P>
1386<br><b>
1387Testing substring extraction functions
1388</b><br>
1389<P>
1390The <b>copy</b> and <b>get</b> modifiers can be used to test the
1391<b>pcre2_substring_copy_xxx()</b> and <b>pcre2_substring_get_xxx()</b> functions.
1392They can be given more than once, and each can specify a capture group name or
1393number, for example:
1394<pre>
1395   abcd\=copy=1,copy=3,get=G1
1396</pre>
1397If the <b>#subject</b> command is used to set default copy and/or get lists,
1398these can be unset by specifying a negative number to cancel all numbered
1399groups and an empty name to cancel all named groups.
1400</P>
1401<P>
1402The <b>getall</b> modifier tests <b>pcre2_substring_list_get()</b>, which
1403extracts all captured substrings.
1404</P>
1405<P>
1406If the subject line is successfully matched, the substrings extracted by the
1407convenience functions are output with C, G, or L after the string number
1408instead of a colon. This is in addition to the normal full list. The string
1409length (that is, the return from the extraction function) is given in
1410parentheses after each substring, followed by the name when the extraction was
1411by name.
1412<a name="substitution"></a></P>
1413<br><b>
1414Testing the substitution function
1415</b><br>
1416<P>
1417If the <b>replace</b> modifier is set, the <b>pcre2_substitute()</b> function is
1418called instead of one of the matching functions (or after one call of
1419<b>pcre2_match()</b> in the case of PCRE2_SUBSTITUTE_MATCHED). Note that
1420replacement strings cannot contain commas, because a comma signifies the end of
1421a modifier. This is not thought to be an issue in a test program.
1422</P>
1423<P>
1424Unlike subject strings, <b>pcre2test</b> does not process replacement strings
1425for escape sequences. In UTF mode, a replacement string is checked to see if it
1426is a valid UTF-8 string. If so, it is correctly converted to a UTF string of
1427the appropriate code unit width. If it is not a valid UTF-8 string, the
1428individual code units are copied directly. This provides a means of passing an
1429invalid UTF-8 string for testing purposes.
1430</P>
1431<P>
1432The following modifiers set options (in additional to the normal match options)
1433for <b>pcre2_substitute()</b>:
1434<pre>
1435  global                      PCRE2_SUBSTITUTE_GLOBAL
1436  substitute_extended         PCRE2_SUBSTITUTE_EXTENDED
1437  substitute_literal          PCRE2_SUBSTITUTE_LITERAL
1438  substitute_matched          PCRE2_SUBSTITUTE_MATCHED
1439  substitute_overflow_length  PCRE2_SUBSTITUTE_OVERFLOW_LENGTH
1440  substitute_replacement_only PCRE2_SUBSTITUTE_REPLACEMENT_ONLY
1441  substitute_unknown_unset    PCRE2_SUBSTITUTE_UNKNOWN_UNSET
1442  substitute_unset_empty      PCRE2_SUBSTITUTE_UNSET_EMPTY
1443</pre>
1444See the
1445<a href="pcre2api.html"><b>pcre2api</b></a>
1446documentation for details of these options.
1447</P>
1448<P>
1449After a successful substitution, the modified string is output, preceded by the
1450number of replacements. This may be zero if there were no matches. Here is a
1451simple example of a substitution test:
1452<pre>
1453  /abc/replace=xxx
1454      =abc=abc=
1455   1: =xxx=abc=
1456      =abc=abc=\=global
1457   2: =xxx=xxx=
1458</pre>
1459Subject and replacement strings should be kept relatively short (fewer than 256
1460characters) for substitution tests, as fixed-size buffers are used. To make it
1461easy to test for buffer overflow, if the replacement string starts with a
1462number in square brackets, that number is passed to <b>pcre2_substitute()</b> as
1463the size of the output buffer, with the replacement string starting at the next
1464character. Here is an example that tests the edge case:
1465<pre>
1466  /abc/
1467      123abc123\=replace=[10]XYZ
1468   1: 123XYZ123
1469      123abc123\=replace=[9]XYZ
1470  Failed: error -47: no more memory
1471</pre>
1472The default action of <b>pcre2_substitute()</b> is to return
1473PCRE2_ERROR_NOMEMORY when the output buffer is too small. However, if the
1474PCRE2_SUBSTITUTE_OVERFLOW_LENGTH option is set (by using the
1475<b>substitute_overflow_length</b> modifier), <b>pcre2_substitute()</b> continues
1476to go through the motions of matching and substituting (but not doing any
1477callouts), in order to compute the size of buffer that is required. When this
1478happens, <b>pcre2test</b> shows the required buffer length (which includes space
1479for the trailing zero) as part of the error message. For example:
1480<pre>
1481  /abc/substitute_overflow_length
1482      123abc123\=replace=[9]XYZ
1483  Failed: error -47: no more memory: 10 code units are needed
1484</pre>
1485A replacement string is ignored with POSIX and DFA matching. Specifying partial
1486matching provokes an error return ("bad option value") from
1487<b>pcre2_substitute()</b>.
1488</P>
1489<br><b>
1490Testing substitute callouts
1491</b><br>
1492<P>
1493If the <b>substitute_callout</b> modifier is set, a substitution callout
1494function is set up. The <b>null_context</b> modifier must not be set, because
1495the address of the callout function is passed in a match context. When the
1496callout function is called (after each substitution), details of the the input
1497and output strings are output. For example:
1498<pre>
1499  /abc/g,replace=&#60;$0&#62;,substitute_callout
1500      abcdefabcpqr
1501   1(1) Old 0 3 "abc" New 0 5 "&#60;abc&#62;"
1502   2(1) Old 6 9 "abc" New 8 13 "&#60;abc&#62;"
1503   2: &#60;abc&#62;def&#60;abc&#62;pqr
1504</pre>
1505The first number on each callout line is the count of matches. The
1506parenthesized number is the number of pairs that are set in the ovector (that
1507is, one more than the number of capturing groups that were set). Then are
1508listed the offsets of the old substring, its contents, and the same for the
1509replacement.
1510</P>
1511<P>
1512By default, the substitution callout function returns zero, which accepts the
1513replacement and causes matching to continue if /g was used. Two further
1514modifiers can be used to test other return values. If <b>substitute_skip</b> is
1515set to a value greater than zero the callout function returns +1 for the match
1516of that number, and similarly <b>substitute_stop</b> returns -1. These cause the
1517replacement to be rejected, and -1 causes no further matching to take place. If
1518either of them are set, <b>substitute_callout</b> is assumed. For example:
1519<pre>
1520  /abc/g,replace=&#60;$0&#62;,substitute_skip=1
1521      abcdefabcpqr
1522   1(1) Old 0 3 "abc" New 0 5 "&#60;abc&#62; SKIPPED"
1523   2(1) Old 6 9 "abc" New 6 11 "&#60;abc&#62;"
1524   2: abcdef&#60;abc&#62;pqr
1525      abcdefabcpqr\=substitute_stop=1
1526   1(1) Old 0 3 "abc" New 0 5 "&#60;abc&#62; STOPPED"
1527   1: abcdefabcpqr
1528</pre>
1529If both are set for the same number, stop takes precedence. Only a single skip
1530or stop is supported, which is sufficient for testing that the feature works.
1531</P>
1532<br><b>
1533Setting the JIT stack size
1534</b><br>
1535<P>
1536The <b>jitstack</b> modifier provides a way of setting the maximum stack size
1537that is used by the just-in-time optimization code. It is ignored if JIT
1538optimization is not being used. The value is a number of kibibytes (units of
15391024 bytes). Setting zero reverts to the default of 32KiB. Providing a stack
1540that is larger than the default is necessary only for very complicated
1541patterns. If <b>jitstack</b> is set non-zero on a subject line it overrides any
1542value that was set on the pattern.
1543</P>
1544<br><b>
1545Setting heap, match, and depth limits
1546</b><br>
1547<P>
1548The <b>heap_limit</b>, <b>match_limit</b>, and <b>depth_limit</b> modifiers set
1549the appropriate limits in the match context. These values are ignored when the
1550<b>find_limits</b> modifier is specified.
1551</P>
1552<br><b>
1553Finding minimum limits
1554</b><br>
1555<P>
1556If the <b>find_limits</b> modifier is present on a subject line, <b>pcre2test</b>
1557calls the relevant matching function several times, setting different values in
1558the match context via <b>pcre2_set_heap_limit()</b>,
1559<b>pcre2_set_match_limit()</b>, or <b>pcre2_set_depth_limit()</b> until it finds
1560the minimum values for each parameter that allows the match to complete without
1561error. If JIT is being used, only the match limit is relevant.
1562</P>
1563<P>
1564When using this modifier, the pattern should not contain any limit settings
1565such as (*LIMIT_MATCH=...) within it. If such a setting is present and is
1566lower than the minimum matching value, the minimum value cannot be found
1567because <b>pcre2_set_match_limit()</b> etc. are only able to reduce the value of
1568an in-pattern limit; they cannot increase it.
1569</P>
1570<P>
1571For non-DFA matching, the minimum <i>depth_limit</i> number is a measure of how
1572much nested backtracking happens (that is, how deeply the pattern's tree is
1573searched). In the case of DFA matching, <i>depth_limit</i> controls the depth of
1574recursive calls of the internal function that is used for handling pattern
1575recursion, lookaround assertions, and atomic groups.
1576</P>
1577<P>
1578For non-DFA matching, the <i>match_limit</i> number is a measure of the amount
1579of backtracking that takes place, and learning the minimum value can be
1580instructive. For most simple matches, the number is quite small, but for
1581patterns with very large numbers of matching possibilities, it can become large
1582very quickly with increasing length of subject string. In the case of DFA
1583matching, <i>match_limit</i> controls the total number of calls, both recursive
1584and non-recursive, to the internal matching function, thus controlling the
1585overall amount of computing resource that is used.
1586</P>
1587<P>
1588For both kinds of matching, the <i>heap_limit</i> number, which is in kibibytes
1589(units of 1024 bytes), limits the amount of heap memory used for matching. A
1590value of zero disables the use of any heap memory; many simple pattern matches
1591can be done without using the heap, so zero is not an unreasonable setting.
1592</P>
1593<br><b>
1594Showing MARK names
1595</b><br>
1596<P>
1597The <b>mark</b> modifier causes the names from backtracking control verbs that
1598are returned from calls to <b>pcre2_match()</b> to be displayed. If a mark is
1599returned for a match, non-match, or partial match, <b>pcre2test</b> shows it.
1600For a match, it is on a line by itself, tagged with "MK:". Otherwise, it
1601is added to the non-match message.
1602</P>
1603<br><b>
1604Showing memory usage
1605</b><br>
1606<P>
1607The <b>memory</b> modifier causes <b>pcre2test</b> to log the sizes of all heap
1608memory allocation and freeing calls that occur during a call to
1609<b>pcre2_match()</b> or <b>pcre2_dfa_match()</b>. These occur only when a match
1610requires a bigger vector than the default for remembering backtracking points
1611(<b>pcre2_match()</b>) or for internal workspace (<b>pcre2_dfa_match()</b>). In
1612many cases there will be no heap memory used and therefore no additional
1613output. No heap memory is allocated during matching with JIT, so in that case
1614the <b>memory</b> modifier never has any effect. For this modifier to work, the
1615<b>null_context</b> modifier must not be set on both the pattern and the
1616subject, though it can be set on one or the other.
1617</P>
1618<br><b>
1619Setting a starting offset
1620</b><br>
1621<P>
1622The <b>offset</b> modifier sets an offset in the subject string at which
1623matching starts. Its value is a number of code units, not characters.
1624</P>
1625<br><b>
1626Setting an offset limit
1627</b><br>
1628<P>
1629The <b>offset_limit</b> modifier sets a limit for unanchored matches. If a match
1630cannot be found starting at or before this offset in the subject, a "no match"
1631return is given. The data value is a number of code units, not characters. When
1632this modifier is used, the <b>use_offset_limit</b> modifier must have been set
1633for the pattern; if not, an error is generated.
1634</P>
1635<br><b>
1636Setting the size of the output vector
1637</b><br>
1638<P>
1639The <b>ovector</b> modifier applies only to the subject line in which it
1640appears, though of course it can also be used to set a default in a
1641<b>#subject</b> command. It specifies the number of pairs of offsets that are
1642available for storing matching information. The default is 15.
1643</P>
1644<P>
1645A value of zero is useful when testing the POSIX API because it causes
1646<b>regexec()</b> to be called with a NULL capture vector. When not testing the
1647POSIX API, a value of zero is used to cause
1648<b>pcre2_match_data_create_from_pattern()</b> to be called, in order to create a
1649match block of exactly the right size for the pattern. (It is not possible to
1650create a match block with a zero-length ovector; there is always at least one
1651pair of offsets.)
1652</P>
1653<br><b>
1654Passing the subject as zero-terminated
1655</b><br>
1656<P>
1657By default, the subject string is passed to a native API matching function with
1658its correct length. In order to test the facility for passing a zero-terminated
1659string, the <b>zero_terminate</b> modifier is provided. It causes the length to
1660be passed as PCRE2_ZERO_TERMINATED. When matching via the POSIX interface,
1661this modifier is ignored, with a warning.
1662</P>
1663<P>
1664When testing <b>pcre2_substitute()</b>, this modifier also has the effect of
1665passing the replacement string as zero-terminated.
1666</P>
1667<br><b>
1668Passing a NULL context
1669</b><br>
1670<P>
1671Normally, <b>pcre2test</b> passes a context block to <b>pcre2_match()</b>,
1672<b>pcre2_dfa_match()</b>, <b>pcre2_jit_match()</b> or <b>pcre2_substitute()</b>.
1673If the <b>null_context</b> modifier is set, however, NULL is passed. This is for
1674testing that the matching and substitution functions behave correctly in this
1675case (they use default values). This modifier cannot be used with the
1676<b>find_limits</b> or <b>substitute_callout</b> modifiers.
1677</P>
1678<br><a name="SEC12" href="#TOC1">THE ALTERNATIVE MATCHING FUNCTION</a><br>
1679<P>
1680By default, <b>pcre2test</b> uses the standard PCRE2 matching function,
1681<b>pcre2_match()</b> to match each subject line. PCRE2 also supports an
1682alternative matching function, <b>pcre2_dfa_match()</b>, which operates in a
1683different way, and has some restrictions. The differences between the two
1684functions are described in the
1685<a href="pcre2matching.html"><b>pcre2matching</b></a>
1686documentation.
1687</P>
1688<P>
1689If the <b>dfa</b> modifier is set, the alternative matching function is used.
1690This function finds all possible matches at a given point in the subject. If,
1691however, the <b>dfa_shortest</b> modifier is set, processing stops after the
1692first match is found. This is always the shortest possible match.
1693</P>
1694<br><a name="SEC13" href="#TOC1">DEFAULT OUTPUT FROM pcre2test</a><br>
1695<P>
1696This section describes the output when the normal matching function,
1697<b>pcre2_match()</b>, is being used.
1698</P>
1699<P>
1700When a match succeeds, <b>pcre2test</b> outputs the list of captured substrings,
1701starting with number 0 for the string that matched the whole pattern.
1702Otherwise, it outputs "No match" when the return is PCRE2_ERROR_NOMATCH, or
1703"Partial match:" followed by the partially matching substring when the
1704return is PCRE2_ERROR_PARTIAL. (Note that this is the
1705entire substring that was inspected during the partial match; it may include
1706characters before the actual match start if a lookbehind assertion, \K, \b,
1707or \B was involved.)
1708</P>
1709<P>
1710For any other return, <b>pcre2test</b> outputs the PCRE2 negative error number
1711and a short descriptive phrase. If the error is a failed UTF string check, the
1712code unit offset of the start of the failing character is also output. Here is
1713an example of an interactive <b>pcre2test</b> run.
1714<pre>
1715  $ pcre2test
1716  PCRE2 version 10.22 2016-07-29
1717
1718    re&#62; /^abc(\d+)/
1719  data&#62; abc123
1720   0: abc123
1721   1: 123
1722  data&#62; xyz
1723  No match
1724</pre>
1725Unset capturing substrings that are not followed by one that is set are not
1726shown by <b>pcre2test</b> unless the <b>allcaptures</b> modifier is specified. In
1727the following example, there are two capturing substrings, but when the first
1728data line is matched, the second, unset substring is not shown. An "internal"
1729unset substring is shown as "&#60;unset&#62;", as for the second data line.
1730<pre>
1731    re&#62; /(a)|(b)/
1732  data&#62; a
1733   0: a
1734   1: a
1735  data&#62; b
1736   0: b
1737   1: &#60;unset&#62;
1738   2: b
1739</pre>
1740If the strings contain any non-printing characters, they are output as \xhh
1741escapes if the value is less than 256 and UTF mode is not set. Otherwise they
1742are output as \x{hh...} escapes. See below for the definition of non-printing
1743characters. If the <b>aftertext</b> modifier is set, the output for substring
17440 is followed by the the rest of the subject string, identified by "0+" like
1745this:
1746<pre>
1747    re&#62; /cat/aftertext
1748  data&#62; cataract
1749   0: cat
1750   0+ aract
1751</pre>
1752If global matching is requested, the results of successive matching attempts
1753are output in sequence, like this:
1754<pre>
1755    re&#62; /\Bi(\w\w)/g
1756  data&#62; Mississippi
1757   0: iss
1758   1: ss
1759   0: iss
1760   1: ss
1761   0: ipp
1762   1: pp
1763</pre>
1764"No match" is output only if the first match attempt fails. Here is an example
1765of a failure message (the offset 4 that is specified by the <b>offset</b>
1766modifier is past the end of the subject string):
1767<pre>
1768    re&#62; /xyz/
1769  data&#62; xyz\=offset=4
1770  Error -24 (bad offset value)
1771</PRE>
1772</P>
1773<P>
1774Note that whereas patterns can be continued over several lines (a plain "&#62;"
1775prompt is used for continuations), subject lines may not. However newlines can
1776be included in a subject by means of the \n escape (or \r, \r\n, etc.,
1777depending on the newline sequence setting).
1778</P>
1779<br><a name="SEC14" href="#TOC1">OUTPUT FROM THE ALTERNATIVE MATCHING FUNCTION</a><br>
1780<P>
1781When the alternative matching function, <b>pcre2_dfa_match()</b>, is used, the
1782output consists of a list of all the matches that start at the first point in
1783the subject where there is at least one match. For example:
1784<pre>
1785    re&#62; /(tang|tangerine|tan)/
1786  data&#62; yellow tangerine\=dfa
1787   0: tangerine
1788   1: tang
1789   2: tan
1790</pre>
1791Using the normal matching function on this data finds only "tang". The
1792longest matching string is always given first (and numbered zero). After a
1793PCRE2_ERROR_PARTIAL return, the output is "Partial match:", followed by the
1794partially matching substring. Note that this is the entire substring that was
1795inspected during the partial match; it may include characters before the actual
1796match start if a lookbehind assertion, \b, or \B was involved. (\K is not
1797supported for DFA matching.)
1798</P>
1799<P>
1800If global matching is requested, the search for further matches resumes
1801at the end of the longest match. For example:
1802<pre>
1803    re&#62; /(tang|tangerine|tan)/g
1804  data&#62; yellow tangerine and tangy sultana\=dfa
1805   0: tangerine
1806   1: tang
1807   2: tan
1808   0: tang
1809   1: tan
1810   0: tan
1811</pre>
1812The alternative matching function does not support substring capture, so the
1813modifiers that are concerned with captured substrings are not relevant.
1814</P>
1815<br><a name="SEC15" href="#TOC1">RESTARTING AFTER A PARTIAL MATCH</a><br>
1816<P>
1817When the alternative matching function has given the PCRE2_ERROR_PARTIAL
1818return, indicating that the subject partially matched the pattern, you can
1819restart the match with additional subject data by means of the
1820<b>dfa_restart</b> modifier. For example:
1821<pre>
1822    re&#62; /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
1823  data&#62; 23ja\=ps,dfa
1824  Partial match: 23ja
1825  data&#62; n05\=dfa,dfa_restart
1826   0: n05
1827</pre>
1828For further information about partial matching, see the
1829<a href="pcre2partial.html"><b>pcre2partial</b></a>
1830documentation.
1831<a name="callouts"></a></P>
1832<br><a name="SEC16" href="#TOC1">CALLOUTS</a><br>
1833<P>
1834If the pattern contains any callout requests, <b>pcre2test</b>'s callout
1835function is called during matching unless <b>callout_none</b> is specified. This
1836works with both matching functions, and with JIT, though there are some
1837differences in behaviour. The output for callouts with numerical arguments and
1838those with string arguments is slightly different.
1839</P>
1840<br><b>
1841Callouts with numerical arguments
1842</b><br>
1843<P>
1844By default, the callout function displays the callout number, the start and
1845current positions in the subject text at the callout time, and the next pattern
1846item to be tested. For example:
1847<pre>
1848  ---&#62;pqrabcdef
1849    0    ^  ^     \d
1850</pre>
1851This output indicates that callout number 0 occurred for a match attempt
1852starting at the fourth character of the subject string, when the pointer was at
1853the seventh character, and when the next pattern item was \d. Just
1854one circumflex is output if the start and current positions are the same, or if
1855the current position precedes the start position, which can happen if the
1856callout is in a lookbehind assertion.
1857</P>
1858<P>
1859Callouts numbered 255 are assumed to be automatic callouts, inserted as a
1860result of the <b>auto_callout</b> pattern modifier. In this case, instead of
1861showing the callout number, the offset in the pattern, preceded by a plus, is
1862output. For example:
1863<pre>
1864    re&#62; /\d?[A-E]\*/auto_callout
1865  data&#62; E*
1866  ---&#62;E*
1867   +0 ^      \d?
1868   +3 ^      [A-E]
1869   +8 ^^     \*
1870  +10 ^ ^
1871   0: E*
1872</pre>
1873If a pattern contains (*MARK) items, an additional line is output whenever
1874a change of latest mark is passed to the callout function. For example:
1875<pre>
1876    re&#62; /a(*MARK:X)bc/auto_callout
1877  data&#62; abc
1878  ---&#62;abc
1879   +0 ^       a
1880   +1 ^^      (*MARK:X)
1881  +10 ^^      b
1882  Latest Mark: X
1883  +11 ^ ^     c
1884  +12 ^  ^
1885   0: abc
1886</pre>
1887The mark changes between matching "a" and "b", but stays the same for the rest
1888of the match, so nothing more is output. If, as a result of backtracking, the
1889mark reverts to being unset, the text "&#60;unset&#62;" is output.
1890</P>
1891<br><b>
1892Callouts with string arguments
1893</b><br>
1894<P>
1895The output for a callout with a string argument is similar, except that instead
1896of outputting a callout number before the position indicators, the callout
1897string and its offset in the pattern string are output before the reflection of
1898the subject string, and the subject string is reflected for each callout. For
1899example:
1900<pre>
1901    re&#62; /^ab(?C'first')cd(?C"second")ef/
1902  data&#62; abcdefg
1903  Callout (7): 'first'
1904  ---&#62;abcdefg
1905      ^ ^         c
1906  Callout (20): "second"
1907  ---&#62;abcdefg
1908      ^   ^       e
1909   0: abcdef
1910
1911</PRE>
1912</P>
1913<br><b>
1914Callout modifiers
1915</b><br>
1916<P>
1917The callout function in <b>pcre2test</b> returns zero (carry on matching) by
1918default, but you can use a <b>callout_fail</b> modifier in a subject line to
1919change this and other parameters of the callout (see below).
1920</P>
1921<P>
1922If the <b>callout_capture</b> modifier is set, the current captured groups are
1923output when a callout occurs. This is useful only for non-DFA matching, as
1924<b>pcre2_dfa_match()</b> does not support capturing, so no captures are ever
1925shown.
1926</P>
1927<P>
1928The normal callout output, showing the callout number or pattern offset (as
1929described above) is suppressed if the <b>callout_no_where</b> modifier is set.
1930</P>
1931<P>
1932When using the interpretive matching function <b>pcre2_match()</b> without JIT,
1933setting the <b>callout_extra</b> modifier causes additional output from
1934<b>pcre2test</b>'s callout function to be generated. For the first callout in a
1935match attempt at a new starting position in the subject, "New match attempt" is
1936output. If there has been a backtrack since the last callout (or start of
1937matching if this is the first callout), "Backtrack" is output, followed by "No
1938other matching paths" if the backtrack ended the previous match attempt. For
1939example:
1940<pre>
1941   re&#62; /(a+)b/auto_callout,no_start_optimize,no_auto_possess
1942  data&#62; aac\=callout_extra
1943  New match attempt
1944  ---&#62;aac
1945   +0 ^       (
1946   +1 ^       a+
1947   +3 ^ ^     )
1948   +4 ^ ^     b
1949  Backtrack
1950  ---&#62;aac
1951   +3 ^^      )
1952   +4 ^^      b
1953  Backtrack
1954  No other matching paths
1955  New match attempt
1956  ---&#62;aac
1957   +0  ^      (
1958   +1  ^      a+
1959   +3  ^^     )
1960   +4  ^^     b
1961  Backtrack
1962  No other matching paths
1963  New match attempt
1964  ---&#62;aac
1965   +0   ^     (
1966   +1   ^     a+
1967  Backtrack
1968  No other matching paths
1969  New match attempt
1970  ---&#62;aac
1971   +0    ^    (
1972   +1    ^    a+
1973  No match
1974</pre>
1975Notice that various optimizations must be turned off if you want all possible
1976matching paths to be scanned. If <b>no_start_optimize</b> is not used, there is
1977an immediate "no match", without any callouts, because the starting
1978optimization fails to find "b" in the subject, which it knows must be present
1979for any match. If <b>no_auto_possess</b> is not used, the "a+" item is turned
1980into "a++", which reduces the number of backtracks.
1981</P>
1982<P>
1983The <b>callout_extra</b> modifier has no effect if used with the DFA matching
1984function, or with JIT.
1985</P>
1986<br><b>
1987Return values from callouts
1988</b><br>
1989<P>
1990The default return from the callout function is zero, which allows matching to
1991continue. The <b>callout_fail</b> modifier can be given one or two numbers. If
1992there is only one number, 1 is returned instead of 0 (causing matching to
1993backtrack) when a callout of that number is reached. If two numbers (&#60;n&#62;:&#60;m&#62;)
1994are given, 1 is returned when callout &#60;n&#62; is reached and there have been at
1995least &#60;m&#62; callouts. The <b>callout_error</b> modifier is similar, except that
1996PCRE2_ERROR_CALLOUT is returned, causing the entire matching process to be
1997aborted. If both these modifiers are set for the same callout number,
1998<b>callout_error</b> takes precedence. Note that callouts with string arguments
1999are always given the number zero.
2000</P>
2001<P>
2002The <b>callout_data</b> modifier can be given an unsigned or a negative number.
2003This is set as the "user data" that is passed to the matching function, and
2004passed back when the callout function is invoked. Any value other than zero is
2005used as a return from <b>pcre2test</b>'s callout function.
2006</P>
2007<P>
2008Inserting callouts can be helpful when using <b>pcre2test</b> to check
2009complicated regular expressions. For further information about callouts, see
2010the
2011<a href="pcre2callout.html"><b>pcre2callout</b></a>
2012documentation.
2013</P>
2014<br><a name="SEC17" href="#TOC1">NON-PRINTING CHARACTERS</a><br>
2015<P>
2016When <b>pcre2test</b> is outputting text in the compiled version of a pattern,
2017bytes other than 32-126 are always treated as non-printing characters and are
2018therefore shown as hex escapes.
2019</P>
2020<P>
2021When <b>pcre2test</b> is outputting text that is a matched part of a subject
2022string, it behaves in the same way, unless a different locale has been set for
2023the pattern (using the <b>locale</b> modifier). In this case, the
2024<b>isprint()</b> function is used to distinguish printing and non-printing
2025characters.
2026<a name="saverestore"></a></P>
2027<br><a name="SEC18" href="#TOC1">SAVING AND RESTORING COMPILED PATTERNS</a><br>
2028<P>
2029It is possible to save compiled patterns on disc or elsewhere, and reload them
2030later, subject to a number of restrictions. JIT data cannot be saved. The host
2031on which the patterns are reloaded must be running the same version of PCRE2,
2032with the same code unit width, and must also have the same endianness, pointer
2033width and PCRE2_SIZE type. Before compiled patterns can be saved they must be
2034serialized, that is, converted to a stream of bytes. A single byte stream may
2035contain any number of compiled patterns, but they must all use the same
2036character tables. A single copy of the tables is included in the byte stream
2037(its size is 1088 bytes).
2038</P>
2039<P>
2040The functions whose names begin with <b>pcre2_serialize_</b> are used
2041for serializing and de-serializing. They are described in the
2042<a href="pcre2serialize.html"><b>pcre2serialize</b></a>
2043documentation. In this section we describe the features of <b>pcre2test</b> that
2044can be used to test these functions.
2045</P>
2046<P>
2047Note that "serialization" in PCRE2 does not convert compiled patterns to an
2048abstract format like Java or .NET. It just makes a reloadable byte code stream.
2049Hence the restrictions on reloading mentioned above.
2050</P>
2051<P>
2052In <b>pcre2test</b>, when a pattern with <b>push</b> modifier is successfully
2053compiled, it is pushed onto a stack of compiled patterns, and <b>pcre2test</b>
2054expects the next line to contain a new pattern (or command) instead of a
2055subject line. By contrast, the <b>pushcopy</b> modifier causes a copy of the
2056compiled pattern to be stacked, leaving the original available for immediate
2057matching. By using <b>push</b> and/or <b>pushcopy</b>, a number of patterns can
2058be compiled and retained. These modifiers are incompatible with <b>posix</b>,
2059and control modifiers that act at match time are ignored (with a message) for
2060the stacked patterns. The <b>jitverify</b> modifier applies only at compile
2061time.
2062</P>
2063<P>
2064The command
2065<pre>
2066  #save &#60;filename&#62;
2067</pre>
2068causes all the stacked patterns to be serialized and the result written to the
2069named file. Afterwards, all the stacked patterns are freed. The command
2070<pre>
2071  #load &#60;filename&#62;
2072</pre>
2073reads the data in the file, and then arranges for it to be de-serialized, with
2074the resulting compiled patterns added to the pattern stack. The pattern on the
2075top of the stack can be retrieved by the #pop command, which must be followed
2076by lines of subjects that are to be matched with the pattern, terminated as
2077usual by an empty line or end of file. This command may be followed by a
2078modifier list containing only
2079<a href="#controlmodifiers">control modifiers</a>
2080that act after a pattern has been compiled. In particular, <b>hex</b>,
2081<b>posix</b>, <b>posix_nosub</b>, <b>push</b>, and <b>pushcopy</b> are not allowed,
2082nor are any
2083<a href="#optionmodifiers">option-setting modifiers.</a>
2084The JIT modifiers are, however permitted. Here is an example that saves and
2085reloads two patterns.
2086<pre>
2087  /abc/push
2088  /xyz/push
2089  #save tempfile
2090  #load tempfile
2091  #pop info
2092  xyz
2093
2094  #pop jit,bincode
2095  abc
2096</pre>
2097If <b>jitverify</b> is used with #pop, it does not automatically imply
2098<b>jit</b>, which is different behaviour from when it is used on a pattern.
2099</P>
2100<P>
2101The #popcopy command is analagous to the <b>pushcopy</b> modifier in that it
2102makes current a copy of the topmost stack pattern, leaving the original still
2103on the stack.
2104</P>
2105<br><a name="SEC19" href="#TOC1">SEE ALSO</a><br>
2106<P>
2107<b>pcre2</b>(3), <b>pcre2api</b>(3), <b>pcre2callout</b>(3),
2108<b>pcre2jit</b>, <b>pcre2matching</b>(3), <b>pcre2partial</b>(d),
2109<b>pcre2pattern</b>(3), <b>pcre2serialize</b>(3).
2110</P>
2111<br><a name="SEC20" href="#TOC1">AUTHOR</a><br>
2112<P>
2113Philip Hazel
2114<br>
2115University Computing Service
2116<br>
2117Cambridge, England.
2118<br>
2119</P>
2120<br><a name="SEC21" href="#TOC1">REVISION</a><br>
2121<P>
2122Last updated: 14 September 2020
2123<br>
2124Copyright &copy; 1997-2020 University of Cambridge.
2125<br>
2126<p>
2127Return to the <a href="index.html">PCRE2 index page</a>.
2128</p>
2129