1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 **********************************************************************
5 * Copyright (C) 1999-2014, International Business Machines
6 * Corporation and others. All Rights Reserved.
7 **********************************************************************
8 *   Date        Name        Description
9 *   11/17/99    aliu        Creation.
10 **********************************************************************
11 */
12 #ifndef TRANSLIT_H
13 #define TRANSLIT_H
14 
15 #include "unicode/utypes.h"
16 
17 /**
18  * \file
19  * \brief C++ API: Tranforms text from one format to another.
20  */
21 
22 #if !UCONFIG_NO_TRANSLITERATION
23 
24 #include "unicode/uobject.h"
25 #include "unicode/unistr.h"
26 #include "unicode/parseerr.h"
27 #include "unicode/utrans.h" // UTransPosition, UTransDirection
28 #include "unicode/strenum.h"
29 
30 U_NAMESPACE_BEGIN
31 
32 class UnicodeFilter;
33 class UnicodeSet;
34 class TransliteratorParser;
35 class NormalizationTransliterator;
36 class TransliteratorIDParser;
37 
38 /**
39  *
40  * <code>Transliterator</code> is an abstract class that
41  * transliterates text from one format to another.  The most common
42  * kind of transliterator is a script, or alphabet, transliterator.
43  * For example, a Russian to Latin transliterator changes Russian text
44  * written in Cyrillic characters to phonetically equivalent Latin
45  * characters.  It does not <em>translate</em> Russian to English!
46  * Transliteration, unlike translation, operates on characters, without
47  * reference to the meanings of words and sentences.
48  *
49  * <p>Although script conversion is its most common use, a
50  * transliterator can actually perform a more general class of tasks.
51  * In fact, <code>Transliterator</code> defines a very general API
52  * which specifies only that a segment of the input text is replaced
53  * by new text.  The particulars of this conversion are determined
54  * entirely by subclasses of <code>Transliterator</code>.
55  *
56  * <p><b>Transliterators are stateless</b>
57  *
58  * <p><code>Transliterator</code> objects are <em>stateless</em>; they
59  * retain no information between calls to
60  * <code>transliterate()</code>.  (However, this does <em>not</em>
61  * mean that threads may share transliterators without synchronizing
62  * them.  Transliterators are not immutable, so they must be
63  * synchronized when shared between threads.)  This might seem to
64  * limit the complexity of the transliteration operation.  In
65  * practice, subclasses perform complex transliterations by delaying
66  * the replacement of text until it is known that no other
67  * replacements are possible.  In other words, although the
68  * <code>Transliterator</code> objects are stateless, the source text
69  * itself embodies all the needed information, and delayed operation
70  * allows arbitrary complexity.
71  *
72  * <p><b>Batch transliteration</b>
73  *
74  * <p>The simplest way to perform transliteration is all at once, on a
75  * string of existing text.  This is referred to as <em>batch</em>
76  * transliteration.  For example, given a string <code>input</code>
77  * and a transliterator <code>t</code>, the call
78  *
79  *     String result = t.transliterate(input);
80  *
81  * will transliterate it and return the result.  Other methods allow
82  * the client to specify a substring to be transliterated and to use
83  * {@link Replaceable } objects instead of strings, in order to
84  * preserve out-of-band information (such as text styles).
85  *
86  * <p><b>Keyboard transliteration</b>
87  *
88  * <p>Somewhat more involved is <em>keyboard</em>, or incremental
89  * transliteration.  This is the transliteration of text that is
90  * arriving from some source (typically the user's keyboard) one
91  * character at a time, or in some other piecemeal fashion.
92  *
93  * <p>In keyboard transliteration, a <code>Replaceable</code> buffer
94  * stores the text.  As text is inserted, as much as possible is
95  * transliterated on the fly.  This means a GUI that displays the
96  * contents of the buffer may show text being modified as each new
97  * character arrives.
98  *
99  * <p>Consider the simple rule-based Transliterator:
100  * <pre>
101  *     th>{theta}
102  *     t>{tau}
103  * </pre>
104  *
105  * When the user types 't', nothing will happen, since the
106  * transliterator is waiting to see if the next character is 'h'.  To
107  * remedy this, we introduce the notion of a cursor, marked by a '|'
108  * in the output string:
109  * <pre>
110  *     t>|{tau}
111  *     {tau}h>{theta}
112  * </pre>
113  *
114  * Now when the user types 't', tau appears, and if the next character
115  * is 'h', the tau changes to a theta.  This is accomplished by
116  * maintaining a cursor position (independent of the insertion point,
117  * and invisible in the GUI) across calls to
118  * <code>transliterate()</code>.  Typically, the cursor will
119  * be coincident with the insertion point, but in a case like the one
120  * above, it will precede the insertion point.
121  *
122  * <p>Keyboard transliteration methods maintain a set of three indices
123  * that are updated with each call to
124  * <code>transliterate()</code>, including the cursor, start,
125  * and limit.  Since these indices are changed by the method, they are
126  * passed in an <code>int[]</code> array. The <code>START</code> index
127  * marks the beginning of the substring that the transliterator will
128  * look at.  It is advanced as text becomes committed (but it is not
129  * the committed index; that's the <code>CURSOR</code>).  The
130  * <code>CURSOR</code> index, described above, marks the point at
131  * which the transliterator last stopped, either because it reached
132  * the end, or because it required more characters to disambiguate
133  * between possible inputs.  The <code>CURSOR</code> can also be
134  * explicitly set by rules in a rule-based Transliterator.
135  * Any characters before the <code>CURSOR</code> index are frozen;
136  * future keyboard transliteration calls within this input sequence
137  * will not change them.  New text is inserted at the
138  * <code>LIMIT</code> index, which marks the end of the substring that
139  * the transliterator looks at.
140  *
141  * <p>Because keyboard transliteration assumes that more characters
142  * are to arrive, it is conservative in its operation.  It only
143  * transliterates when it can do so unambiguously.  Otherwise it waits
144  * for more characters to arrive.  When the client code knows that no
145  * more characters are forthcoming, perhaps because the user has
146  * performed some input termination operation, then it should call
147  * <code>finishTransliteration()</code> to complete any
148  * pending transliterations.
149  *
150  * <p><b>Inverses</b>
151  *
152  * <p>Pairs of transliterators may be inverses of one another.  For
153  * example, if transliterator <b>A</b> transliterates characters by
154  * incrementing their Unicode value (so "abc" -> "def"), and
155  * transliterator <b>B</b> decrements character values, then <b>A</b>
156  * is an inverse of <b>B</b> and vice versa.  If we compose <b>A</b>
157  * with <b>B</b> in a compound transliterator, the result is the
158  * indentity transliterator, that is, a transliterator that does not
159  * change its input text.
160  *
161  * The <code>Transliterator</code> method <code>getInverse()</code>
162  * returns a transliterator's inverse, if one exists, or
163  * <code>null</code> otherwise.  However, the result of
164  * <code>getInverse()</code> usually will <em>not</em> be a true
165  * mathematical inverse.  This is because true inverse transliterators
166  * are difficult to formulate.  For example, consider two
167  * transliterators: <b>AB</b>, which transliterates the character 'A'
168  * to 'B', and <b>BA</b>, which transliterates 'B' to 'A'.  It might
169  * seem that these are exact inverses, since
170  *
171  * \htmlonly<blockquote>\endhtmlonly"A" x <b>AB</b> -> "B"<br>
172  * "B" x <b>BA</b> -> "A"\htmlonly</blockquote>\endhtmlonly
173  *
174  * where 'x' represents transliteration.  However,
175  *
176  * \htmlonly<blockquote>\endhtmlonly"ABCD" x <b>AB</b> -> "BBCD"<br>
177  * "BBCD" x <b>BA</b> -> "AACD"\htmlonly</blockquote>\endhtmlonly
178  *
179  * so <b>AB</b> composed with <b>BA</b> is not the
180  * identity. Nonetheless, <b>BA</b> may be usefully considered to be
181  * <b>AB</b>'s inverse, and it is on this basis that
182  * <b>AB</b><code>.getInverse()</code> could legitimately return
183  * <b>BA</b>.
184  *
185  * <p><b>IDs and display names</b>
186  *
187  * <p>A transliterator is designated by a short identifier string or
188  * <em>ID</em>.  IDs follow the format <em>source-destination</em>,
189  * where <em>source</em> describes the entity being replaced, and
190  * <em>destination</em> describes the entity replacing
191  * <em>source</em>.  The entities may be the names of scripts,
192  * particular sequences of characters, or whatever else it is that the
193  * transliterator converts to or from.  For example, a transliterator
194  * from Russian to Latin might be named "Russian-Latin".  A
195  * transliterator from keyboard escape sequences to Latin-1 characters
196  * might be named "KeyboardEscape-Latin1".  By convention, system
197  * entity names are in English, with the initial letters of words
198  * capitalized; user entity names may follow any format so long as
199  * they do not contain dashes.
200  *
201  * <p>In addition to programmatic IDs, transliterator objects have
202  * display names for presentation in user interfaces, returned by
203  * {@link #getDisplayName }.
204  *
205  * <p><b>Factory methods and registration</b>
206  *
207  * <p>In general, client code should use the factory method
208  * {@link #createInstance } to obtain an instance of a
209  * transliterator given its ID.  Valid IDs may be enumerated using
210  * <code>getAvailableIDs()</code>.  Since transliterators are mutable,
211  * multiple calls to {@link #createInstance } with the same ID will
212  * return distinct objects.
213  *
214  * <p>In addition to the system transliterators registered at startup,
215  * user transliterators may be registered by calling
216  * <code>registerInstance()</code> at run time.  A registered instance
217  * acts a template; future calls to {@link #createInstance } with the ID
218  * of the registered object return clones of that object.  Thus any
219  * object passed to <tt>registerInstance()</tt> must implement
220  * <tt>clone()</tt> propertly.  To register a transliterator subclass
221  * without instantiating it (until it is needed), users may call
222  * {@link #registerFactory }.  In this case, the objects are
223  * instantiated by invoking the zero-argument public constructor of
224  * the class.
225  *
226  * <p><b>Subclassing</b>
227  *
228  * Subclasses must implement the abstract method
229  * <code>handleTransliterate()</code>.  <p>Subclasses should override
230  * the <code>transliterate()</code> method taking a
231  * <code>Replaceable</code> and the <code>transliterate()</code>
232  * method taking a <code>String</code> and <code>StringBuffer</code>
233  * if the performance of these methods can be improved over the
234  * performance obtained by the default implementations in this class.
235  *
236  * <p><b>Rule syntax</b>
237  *
238  * <p>A set of rules determines how to perform translations.
239  * Rules within a rule set are separated by semicolons (';').
240  * To include a literal semicolon, prefix it with a backslash ('\').
241  * Unicode Pattern_White_Space is ignored.
242  * If the first non-blank character on a line is '#',
243  * the entire line is ignored as a comment.
244  *
245  * <p>Each set of rules consists of two groups, one forward, and one
246  * reverse. This is a convention that is not enforced; rules for one
247  * direction may be omitted, with the result that translations in
248  * that direction will not modify the source text. In addition,
249  * bidirectional forward-reverse rules may be specified for
250  * symmetrical transformations.
251  *
252  * <p>Note: Another description of the Transliterator rule syntax is available in
253  * <a href="https://www.unicode.org/reports/tr35/tr35-general.html#Transform_Rules_Syntax">section
254  * Transform Rules Syntax of UTS #35: Unicode LDML</a>.
255  * The rules are shown there using arrow symbols ← and → and ↔.
256  * ICU supports both those and the equivalent ASCII symbols &lt; and &gt; and &lt;&gt;.
257  *
258  * <p>Rule statements take one of the following forms:
259  *
260  * <dl>
261  *     <dt><code>$alefmadda=\\u0622;</code></dt>
262  *     <dd><strong>Variable definition.</strong> The name on the
263  *         left is assigned the text on the right. In this example,
264  *         after this statement, instances of the left hand name,
265  *         &quot;<code>$alefmadda</code>&quot;, will be replaced by
266  *         the Unicode character U+0622. Variable names must begin
267  *         with a letter and consist only of letters, digits, and
268  *         underscores. Case is significant. Duplicate names cause
269  *         an exception to be thrown, that is, variables cannot be
270  *         redefined. The right hand side may contain well-formed
271  *         text of any length, including no text at all (&quot;<code>$empty=;</code>&quot;).
272  *         The right hand side may contain embedded <code>UnicodeSet</code>
273  *         patterns, for example, &quot;<code>$softvowel=[eiyEIY]</code>&quot;.</dd>
274  *     <dt><code>ai&gt;$alefmadda;</code></dt>
275  *     <dd><strong>Forward translation rule.</strong> This rule
276  *         states that the string on the left will be changed to the
277  *         string on the right when performing forward
278  *         transliteration.</dd>
279  *     <dt><code>ai&lt;$alefmadda;</code></dt>
280  *     <dd><strong>Reverse translation rule.</strong> This rule
281  *         states that the string on the right will be changed to
282  *         the string on the left when performing reverse
283  *         transliteration.</dd>
284  * </dl>
285  *
286  * <dl>
287  *     <dt><code>ai&lt;&gt;$alefmadda;</code></dt>
288  *     <dd><strong>Bidirectional translation rule.</strong> This
289  *         rule states that the string on the right will be changed
290  *         to the string on the left when performing forward
291  *         transliteration, and vice versa when performing reverse
292  *         transliteration.</dd>
293  * </dl>
294  *
295  * <p>Translation rules consist of a <em>match pattern</em> and an <em>output
296  * string</em>. The match pattern consists of literal characters,
297  * optionally preceded by context, and optionally followed by
298  * context. Context characters, like literal pattern characters,
299  * must be matched in the text being transliterated. However, unlike
300  * literal pattern characters, they are not replaced by the output
301  * text. For example, the pattern &quot;<code>abc{def}</code>&quot;
302  * indicates the characters &quot;<code>def</code>&quot; must be
303  * preceded by &quot;<code>abc</code>&quot; for a successful match.
304  * If there is a successful match, &quot;<code>def</code>&quot; will
305  * be replaced, but not &quot;<code>abc</code>&quot;. The final '<code>}</code>'
306  * is optional, so &quot;<code>abc{def</code>&quot; is equivalent to
307  * &quot;<code>abc{def}</code>&quot;. Another example is &quot;<code>{123}456</code>&quot;
308  * (or &quot;<code>123}456</code>&quot;) in which the literal
309  * pattern &quot;<code>123</code>&quot; must be followed by &quot;<code>456</code>&quot;.
310  *
311  * <p>The output string of a forward or reverse rule consists of
312  * characters to replace the literal pattern characters. If the
313  * output string contains the character '<code>|</code>', this is
314  * taken to indicate the location of the <em>cursor</em> after
315  * replacement. The cursor is the point in the text at which the
316  * next replacement, if any, will be applied. The cursor is usually
317  * placed within the replacement text; however, it can actually be
318  * placed into the precending or following context by using the
319  * special character '@'. Examples:
320  *
321  * <pre>
322  *     a {foo} z &gt; | @ bar; # foo -&gt; bar, move cursor before a
323  *     {foo} xyz &gt; bar @@|; #&nbsp;foo -&gt; bar, cursor between y and z
324  * </pre>
325  *
326  * <p><b>UnicodeSet</b>
327  *
328  * <p><code>UnicodeSet</code> patterns may appear anywhere that
329  * makes sense. They may appear in variable definitions.
330  * Contrariwise, <code>UnicodeSet</code> patterns may themselves
331  * contain variable references, such as &quot;<code>$a=[a-z];$not_a=[^$a]</code>&quot;,
332  * or &quot;<code>$range=a-z;$ll=[$range]</code>&quot;.
333  *
334  * <p><code>UnicodeSet</code> patterns may also be embedded directly
335  * into rule strings. Thus, the following two rules are equivalent:
336  *
337  * <pre>
338  *     $vowel=[aeiou]; $vowel&gt;'*'; # One way to do this
339  *     [aeiou]&gt;'*'; # Another way
340  * </pre>
341  *
342  * <p>See {@link UnicodeSet} for more documentation and examples.
343  *
344  * <p><b>Segments</b>
345  *
346  * <p>Segments of the input string can be matched and copied to the
347  * output string. This makes certain sets of rules simpler and more
348  * general, and makes reordering possible. For example:
349  *
350  * <pre>
351  *     ([a-z]) &gt; $1 $1; # double lowercase letters
352  *     ([:Lu:]) ([:Ll:]) &gt; $2 $1; # reverse order of Lu-Ll pairs
353  * </pre>
354  *
355  * <p>The segment of the input string to be copied is delimited by
356  * &quot;<code>(</code>&quot; and &quot;<code>)</code>&quot;. Up to
357  * nine segments may be defined. Segments may not overlap. In the
358  * output string, &quot;<code>$1</code>&quot; through &quot;<code>$9</code>&quot;
359  * represent the input string segments, in left-to-right order of
360  * definition.
361  *
362  * <p><b>Anchors</b>
363  *
364  * <p>Patterns can be anchored to the beginning or the end of the text. This is done with the
365  * special characters '<code>^</code>' and '<code>$</code>'. For example:
366  *
367  * <pre>
368  *   ^ a&nbsp;&nbsp; &gt; 'BEG_A'; &nbsp;&nbsp;# match 'a' at start of text
369  *   &nbsp; a&nbsp;&nbsp; &gt; 'A'; # match other instances of 'a'
370  *   &nbsp; z $ &gt; 'END_Z'; &nbsp;&nbsp;# match 'z' at end of text
371  *   &nbsp; z&nbsp;&nbsp; &gt; 'Z';&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; # match other instances of 'z'
372  * </pre>
373  *
374  * <p>It is also possible to match the beginning or the end of the text using a <code>UnicodeSet</code>.
375  * This is done by including a virtual anchor character '<code>$</code>' at the end of the
376  * set pattern. Although this is usually the match chafacter for the end anchor, the set will
377  * match either the beginning or the end of the text, depending on its placement. For
378  * example:
379  *
380  * <pre>
381  *   $x = [a-z$]; &nbsp;&nbsp;# match 'a' through 'z' OR anchor
382  *   $x 1&nbsp;&nbsp;&nbsp; &gt; 2;&nbsp;&nbsp; # match '1' after a-z or at the start
383  *   &nbsp;&nbsp; 3 $x &gt; 4; &nbsp;&nbsp;# match '3' before a-z or at the end
384  * </pre>
385  *
386  * <p><b>Example</b>
387  *
388  * <p>The following example rules illustrate many of the features of
389  * the rule language.
390  *
391  * <table border="0" cellpadding="4">
392  *     <tr>
393  *         <td style="vertical-align: top;">Rule 1.</td>
394  *         <td style="vertical-align: top; write-space: nowrap;"><code>abc{def}&gt;x|y</code></td>
395  *     </tr>
396  *     <tr>
397  *         <td style="vertical-align: top;">Rule 2.</td>
398  *         <td style="vertical-align: top; write-space: nowrap;"><code>xyz&gt;r</code></td>
399  *     </tr>
400  *     <tr>
401  *         <td style="vertical-align: top;">Rule 3.</td>
402  *         <td style="vertical-align: top; write-space: nowrap;"><code>yz&gt;q</code></td>
403  *     </tr>
404  * </table>
405  *
406  * <p>Applying these rules to the string &quot;<code>adefabcdefz</code>&quot;
407  * yields the following results:
408  *
409  * <table border="0" cellpadding="4">
410  *     <tr>
411  *         <td style="vertical-align: top; write-space: nowrap;"><code>|adefabcdefz</code></td>
412  *         <td style="vertical-align: top;">Initial state, no rules match. Advance
413  *         cursor.</td>
414  *     </tr>
415  *     <tr>
416  *         <td style="vertical-align: top; write-space: nowrap;"><code>a|defabcdefz</code></td>
417  *         <td style="vertical-align: top;">Still no match. Rule 1 does not match
418  *         because the preceding context is not present.</td>
419  *     </tr>
420  *     <tr>
421  *         <td style="vertical-align: top; write-space: nowrap;"><code>ad|efabcdefz</code></td>
422  *         <td style="vertical-align: top;">Still no match. Keep advancing until
423  *         there is a match...</td>
424  *     </tr>
425  *     <tr>
426  *         <td style="vertical-align: top; write-space: nowrap;"><code>ade|fabcdefz</code></td>
427  *         <td style="vertical-align: top;">...</td>
428  *     </tr>
429  *     <tr>
430  *         <td style="vertical-align: top; write-space: nowrap;"><code>adef|abcdefz</code></td>
431  *         <td style="vertical-align: top;">...</td>
432  *     </tr>
433  *     <tr>
434  *         <td style="vertical-align: top; write-space: nowrap;"><code>adefa|bcdefz</code></td>
435  *         <td style="vertical-align: top;">...</td>
436  *     </tr>
437  *     <tr>
438  *         <td style="vertical-align: top; write-space: nowrap;"><code>adefab|cdefz</code></td>
439  *         <td style="vertical-align: top;">...</td>
440  *     </tr>
441  *     <tr>
442  *         <td style="vertical-align: top; write-space: nowrap;"><code>adefabc|defz</code></td>
443  *         <td style="vertical-align: top;">Rule 1 matches; replace &quot;<code>def</code>&quot;
444  *         with &quot;<code>xy</code>&quot; and back up the cursor
445  *         to before the '<code>y</code>'.</td>
446  *     </tr>
447  *     <tr>
448  *         <td style="vertical-align: top; write-space: nowrap;"><code>adefabcx|yz</code></td>
449  *         <td style="vertical-align: top;">Although &quot;<code>xyz</code>&quot; is
450  *         present, rule 2 does not match because the cursor is
451  *         before the '<code>y</code>', not before the '<code>x</code>'.
452  *         Rule 3 does match. Replace &quot;<code>yz</code>&quot;
453  *         with &quot;<code>q</code>&quot;.</td>
454  *     </tr>
455  *     <tr>
456  *         <td style="vertical-align: top; write-space: nowrap;"><code>adefabcxq|</code></td>
457  *         <td style="vertical-align: top;">The cursor is at the end;
458  *         transliteration is complete.</td>
459  *     </tr>
460  * </table>
461  *
462  * <p>The order of rules is significant. If multiple rules may match
463  * at some point, the first matching rule is applied.
464  *
465  * <p>Forward and reverse rules may have an empty output string.
466  * Otherwise, an empty left or right hand side of any statement is a
467  * syntax error.
468  *
469  * <p>Single quotes are used to quote any character other than a
470  * digit or letter. To specify a single quote itself, inside or
471  * outside of quotes, use two single quotes in a row. For example,
472  * the rule &quot;<code>'&gt;'&gt;o''clock</code>&quot; changes the
473  * string &quot;<code>&gt;</code>&quot; to the string &quot;<code>o'clock</code>&quot;.
474  *
475  * <p><b>Notes</b>
476  *
477  * <p>While a Transliterator is being built from rules, it checks that
478  * the rules are added in proper order. For example, if the rule
479  * &quot;a&gt;x&quot; is followed by the rule &quot;ab&gt;y&quot;,
480  * then the second rule will throw an exception. The reason is that
481  * the second rule can never be triggered, since the first rule
482  * always matches anything it matches. In other words, the first
483  * rule <em>masks</em> the second rule.
484  *
485  * @author Alan Liu
486  * @stable ICU 2.0
487  */
488 class U_I18N_API Transliterator : public UObject {
489 
490 private:
491 
492     /**
493      * Programmatic name, e.g., "Latin-Arabic".
494      */
495     UnicodeString ID;
496 
497     /**
498      * This transliterator's filter.  Any character for which
499      * <tt>filter.contains()</tt> returns <tt>false</tt> will not be
500      * altered by this transliterator.  If <tt>filter</tt> is
501      * <tt>null</tt> then no filtering is applied.
502      */
503     UnicodeFilter* filter;
504 
505     int32_t maximumContextLength;
506 
507  public:
508 
509     /**
510      * A context integer or pointer for a factory function, passed by
511      * value.
512      * @stable ICU 2.4
513      */
514     union Token {
515         /**
516          * This token, interpreted as a 32-bit integer.
517          * @stable ICU 2.4
518          */
519         int32_t integer;
520         /**
521          * This token, interpreted as a native pointer.
522          * @stable ICU 2.4
523          */
524         void*   pointer;
525     };
526 
527 #ifndef U_HIDE_INTERNAL_API
528     /**
529      * Return a token containing an integer.
530      * @return a token containing an integer.
531      * @internal
532      */
533     inline static Token integerToken(int32_t);
534 
535     /**
536      * Return a token containing a pointer.
537      * @return a token containing a pointer.
538      * @internal
539      */
540     inline static Token pointerToken(void*);
541 #endif  /* U_HIDE_INTERNAL_API */
542 
543     /**
544      * A function that creates and returns a Transliterator.  When
545      * invoked, it will be passed the ID string that is being
546      * instantiated, together with the context pointer that was passed
547      * in when the factory function was first registered.  Many
548      * factory functions will ignore both parameters, however,
549      * functions that are registered to more than one ID may use the
550      * ID or the context parameter to parameterize the transliterator
551      * they create.
552      * @param ID      the string identifier for this transliterator
553      * @param context a context pointer that will be stored and
554      *                later passed to the factory function when an ID matching
555      *                the registration ID is being instantiated with this factory.
556      * @stable ICU 2.4
557      */
558     typedef Transliterator* (U_EXPORT2 *Factory)(const UnicodeString& ID, Token context);
559 
560 protected:
561 
562     /**
563      * Default constructor.
564      * @param ID the string identifier for this transliterator
565      * @param adoptedFilter the filter.  Any character for which
566      * <tt>filter.contains()</tt> returns <tt>false</tt> will not be
567      * altered by this transliterator.  If <tt>filter</tt> is
568      * <tt>null</tt> then no filtering is applied.
569      * @stable ICU 2.4
570      */
571     Transliterator(const UnicodeString& ID, UnicodeFilter* adoptedFilter);
572 
573     /**
574      * Copy constructor.
575      * @stable ICU 2.4
576      */
577     Transliterator(const Transliterator&);
578 
579     /**
580      * Assignment operator.
581      * @stable ICU 2.4
582      */
583     Transliterator& operator=(const Transliterator&);
584 
585     /**
586      * Create a transliterator from a basic ID.  This is an ID
587      * containing only the forward direction source, target, and
588      * variant.
589      * @param id a basic ID of the form S-T or S-T/V.
590      * @param canon canonical ID to assign to the object, or
591      * NULL to leave the ID unchanged
592      * @return a newly created Transliterator or null if the ID is
593      * invalid.
594      * @stable ICU 2.4
595      */
596     static Transliterator* createBasicInstance(const UnicodeString& id,
597                                                const UnicodeString* canon);
598 
599     friend class TransliteratorParser; // for parseID()
600     friend class TransliteratorIDParser; // for createBasicInstance()
601     friend class TransliteratorAlias; // for setID()
602 
603 public:
604 
605     /**
606      * Destructor.
607      * @stable ICU 2.0
608      */
609     virtual ~Transliterator();
610 
611     /**
612      * Implements Cloneable.
613      * All subclasses are encouraged to implement this method if it is
614      * possible and reasonable to do so.  Subclasses that are to be
615      * registered with the system using <tt>registerInstance()</tt>
616      * are required to implement this method.  If a subclass does not
617      * implement clone() properly and is registered with the system
618      * using registerInstance(), then the default clone() implementation
619      * will return null, and calls to createInstance() will fail.
620      *
621      * @return a copy of the object.
622      * @see #registerInstance
623      * @stable ICU 2.0
624      */
625     virtual Transliterator* clone() const;
626 
627     /**
628      * Transliterates a segment of a string, with optional filtering.
629      *
630      * @param text the string to be transliterated
631      * @param start the beginning index, inclusive; <code>0 <= start
632      * <= limit</code>.
633      * @param limit the ending index, exclusive; <code>start <= limit
634      * <= text.length()</code>.
635      * @return The new limit index.  The text previously occupying <code>[start,
636      * limit)</code> has been transliterated, possibly to a string of a different
637      * length, at <code>[start, </code><em>new-limit</em><code>)</code>, where
638      * <em>new-limit</em> is the return value. If the input offsets are out of bounds,
639      * the returned value is -1 and the input string remains unchanged.
640      * @stable ICU 2.0
641      */
642     virtual int32_t transliterate(Replaceable& text,
643                                   int32_t start, int32_t limit) const;
644 
645     /**
646      * Transliterates an entire string in place. Convenience method.
647      * @param text the string to be transliterated
648      * @stable ICU 2.0
649      */
650     virtual void transliterate(Replaceable& text) const;
651 
652     /**
653      * Transliterates the portion of the text buffer that can be
654      * transliterated unambiguosly after new text has been inserted,
655      * typically as a result of a keyboard event.  The new text in
656      * <code>insertion</code> will be inserted into <code>text</code>
657      * at <code>index.limit</code>, advancing
658      * <code>index.limit</code> by <code>insertion.length()</code>.
659      * Then the transliterator will try to transliterate characters of
660      * <code>text</code> between <code>index.cursor</code> and
661      * <code>index.limit</code>.  Characters before
662      * <code>index.cursor</code> will not be changed.
663      *
664      * <p>Upon return, values in <code>index</code> will be updated.
665      * <code>index.start</code> will be advanced to the first
666      * character that future calls to this method will read.
667      * <code>index.cursor</code> and <code>index.limit</code> will
668      * be adjusted to delimit the range of text that future calls to
669      * this method may change.
670      *
671      * <p>Typical usage of this method begins with an initial call
672      * with <code>index.start</code> and <code>index.limit</code>
673      * set to indicate the portion of <code>text</code> to be
674      * transliterated, and <code>index.cursor == index.start</code>.
675      * Thereafter, <code>index</code> can be used without
676      * modification in future calls, provided that all changes to
677      * <code>text</code> are made via this method.
678      *
679      * <p>This method assumes that future calls may be made that will
680      * insert new text into the buffer.  As a result, it only performs
681      * unambiguous transliterations.  After the last call to this
682      * method, there may be untransliterated text that is waiting for
683      * more input to resolve an ambiguity.  In order to perform these
684      * pending transliterations, clients should call {@link
685      * #finishTransliteration } after the last call to this
686      * method has been made.
687      *
688      * @param text the buffer holding transliterated and untransliterated text
689      * @param index an array of three integers.
690      *
691      * <ul><li><code>index.start</code>: the beginning index,
692      * inclusive; <code>0 <= index.start <= index.limit</code>.
693      *
694      * <li><code>index.limit</code>: the ending index, exclusive;
695      * <code>index.start <= index.limit <= text.length()</code>.
696      * <code>insertion</code> is inserted at
697      * <code>index.limit</code>.
698      *
699      * <li><code>index.cursor</code>: the next character to be
700      * considered for transliteration; <code>index.start <=
701      * index.cursor <= index.limit</code>.  Characters before
702      * <code>index.cursor</code> will not be changed by future calls
703      * to this method.</ul>
704      *
705      * @param insertion text to be inserted and possibly
706      * transliterated into the translation buffer at
707      * <code>index.limit</code>.  If <code>null</code> then no text
708      * is inserted.
709      * @param status    Output param to filled in with a success or an error.
710      * @see #handleTransliterate
711      * @exception IllegalArgumentException if <code>index</code>
712      * is invalid
713      * @see UTransPosition
714      * @stable ICU 2.0
715      */
716     virtual void transliterate(Replaceable& text, UTransPosition& index,
717                                const UnicodeString& insertion,
718                                UErrorCode& status) const;
719 
720     /**
721      * Transliterates the portion of the text buffer that can be
722      * transliterated unambiguosly after a new character has been
723      * inserted, typically as a result of a keyboard event.  This is a
724      * convenience method.
725      * @param text the buffer holding transliterated and
726      * untransliterated text
727      * @param index an array of three integers.
728      * @param insertion text to be inserted and possibly
729      * transliterated into the translation buffer at
730      * <code>index.limit</code>.
731      * @param status    Output param to filled in with a success or an error.
732      * @see #transliterate(Replaceable&, UTransPosition&, const UnicodeString&, UErrorCode&) const
733      * @stable ICU 2.0
734      */
735     virtual void transliterate(Replaceable& text, UTransPosition& index,
736                                UChar32 insertion,
737                                UErrorCode& status) const;
738 
739     /**
740      * Transliterates the portion of the text buffer that can be
741      * transliterated unambiguosly.  This is a convenience method; see
742      * {@link
743      * #transliterate(Replaceable&, UTransPosition&, const UnicodeString&, UErrorCode&) const }
744      * for details.
745      * @param text the buffer holding transliterated and
746      * untransliterated text
747      * @param index an array of three integers.
748      * @param status    Output param to filled in with a success or an error.
749      * @see #transliterate(Replaceable&, UTransPosition&, const UnicodeString&, UErrorCode &) const
750      * @stable ICU 2.0
751      */
752     virtual void transliterate(Replaceable& text, UTransPosition& index,
753                                UErrorCode& status) const;
754 
755     /**
756      * Finishes any pending transliterations that were waiting for
757      * more characters.  Clients should call this method as the last
758      * call after a sequence of one or more calls to
759      * <code>transliterate()</code>.
760      * @param text the buffer holding transliterated and
761      * untransliterated text.
762      * @param index the array of indices previously passed to {@link
763      * #transliterate }
764      * @stable ICU 2.0
765      */
766     virtual void finishTransliteration(Replaceable& text,
767                                        UTransPosition& index) const;
768 
769 private:
770 
771     /**
772      * This internal method does incremental transliteration.  If the
773      * 'insertion' is non-null then we append it to 'text' before
774      * proceeding.  This method calls through to the pure virtual
775      * framework method handleTransliterate() to do the actual
776      * work.
777      * @param text the buffer holding transliterated and
778      * untransliterated text
779      * @param index an array of three integers.  See {@link
780      * #transliterate(Replaceable, int[], String)}.
781      * @param insertion text to be inserted and possibly
782      * transliterated into the translation buffer at
783      * <code>index.limit</code>.
784      * @param status    Output param to filled in with a success or an error.
785      */
786     void _transliterate(Replaceable& text,
787                         UTransPosition& index,
788                         const UnicodeString* insertion,
789                         UErrorCode &status) const;
790 
791 protected:
792 
793     /**
794      * Abstract method that concrete subclasses define to implement
795      * their transliteration algorithm.  This method handles both
796      * incremental and non-incremental transliteration.  Let
797      * <code>originalStart</code> refer to the value of
798      * <code>pos.start</code> upon entry.
799      *
800      * <ul>
801      *  <li>If <code>incremental</code> is false, then this method
802      *  should transliterate all characters between
803      *  <code>pos.start</code> and <code>pos.limit</code>. Upon return
804      *  <code>pos.start</code> must == <code> pos.limit</code>.</li>
805      *
806      *  <li>If <code>incremental</code> is true, then this method
807      *  should transliterate all characters between
808      *  <code>pos.start</code> and <code>pos.limit</code> that can be
809      *  unambiguously transliterated, regardless of future insertions
810      *  of text at <code>pos.limit</code>.  Upon return,
811      *  <code>pos.start</code> should be in the range
812      *  [<code>originalStart</code>, <code>pos.limit</code>).
813      *  <code>pos.start</code> should be positioned such that
814      *  characters [<code>originalStart</code>, <code>
815      *  pos.start</code>) will not be changed in the future by this
816      *  transliterator and characters [<code>pos.start</code>,
817      *  <code>pos.limit</code>) are unchanged.</li>
818      * </ul>
819      *
820      * <p>Implementations of this method should also obey the
821      * following invariants:</p>
822      *
823      * <ul>
824      *  <li> <code>pos.limit</code> and <code>pos.contextLimit</code>
825      *  should be updated to reflect changes in length of the text
826      *  between <code>pos.start</code> and <code>pos.limit</code>. The
827      *  difference <code> pos.contextLimit - pos.limit</code> should
828      *  not change.</li>
829      *
830      *  <li><code>pos.contextStart</code> should not change.</li>
831      *
832      *  <li>Upon return, neither <code>pos.start</code> nor
833      *  <code>pos.limit</code> should be less than
834      *  <code>originalStart</code>.</li>
835      *
836      *  <li>Text before <code>originalStart</code> and text after
837      *  <code>pos.limit</code> should not change.</li>
838      *
839      *  <li>Text before <code>pos.contextStart</code> and text after
840      *  <code> pos.contextLimit</code> should be ignored.</li>
841      * </ul>
842      *
843      * <p>Subclasses may safely assume that all characters in
844      * [<code>pos.start</code>, <code>pos.limit</code>) are filtered.
845      * In other words, the filter has already been applied by the time
846      * this method is called.  See
847      * <code>filteredTransliterate()</code>.
848      *
849      * <p>This method is <b>not</b> for public consumption.  Calling
850      * this method directly will transliterate
851      * [<code>pos.start</code>, <code>pos.limit</code>) without
852      * applying the filter. End user code should call <code>
853      * transliterate()</code> instead of this method. Subclass code
854      * and wrapping transliterators should call
855      * <code>filteredTransliterate()</code> instead of this method.<p>
856      *
857      * @param text the buffer holding transliterated and
858      * untransliterated text
859      *
860      * @param pos the indices indicating the start, limit, context
861      * start, and context limit of the text.
862      *
863      * @param incremental if true, assume more text may be inserted at
864      * <code>pos.limit</code> and act accordingly.  Otherwise,
865      * transliterate all text between <code>pos.start</code> and
866      * <code>pos.limit</code> and move <code>pos.start</code> up to
867      * <code>pos.limit</code>.
868      *
869      * @see #transliterate
870      * @stable ICU 2.4
871      */
872     virtual void handleTransliterate(Replaceable& text,
873                                      UTransPosition& pos,
874                                      UBool incremental) const = 0;
875 
876 public:
877     /**
878      * Transliterate a substring of text, as specified by index, taking filters
879      * into account.  This method is for subclasses that need to delegate to
880      * another transliterator.
881      * @param text the text to be transliterated
882      * @param index the position indices
883      * @param incremental if TRUE, then assume more characters may be inserted
884      * at index.limit, and postpone processing to accomodate future incoming
885      * characters
886      * @stable ICU 2.4
887      */
888     virtual void filteredTransliterate(Replaceable& text,
889                                        UTransPosition& index,
890                                        UBool incremental) const;
891 
892 private:
893 
894     /**
895      * Top-level transliteration method, handling filtering, incremental and
896      * non-incremental transliteration, and rollback.  All transliteration
897      * public API methods eventually call this method with a rollback argument
898      * of TRUE.  Other entities may call this method but rollback should be
899      * FALSE.
900      *
901      * <p>If this transliterator has a filter, break up the input text into runs
902      * of unfiltered characters.  Pass each run to
903      * subclass.handleTransliterate().
904      *
905      * <p>In incremental mode, if rollback is TRUE, perform a special
906      * incremental procedure in which several passes are made over the input
907      * text, adding one character at a time, and committing successful
908      * transliterations as they occur.  Unsuccessful transliterations are rolled
909      * back and retried with additional characters to give correct results.
910      *
911      * @param text the text to be transliterated
912      * @param index the position indices
913      * @param incremental if TRUE, then assume more characters may be inserted
914      * at index.limit, and postpone processing to accomodate future incoming
915      * characters
916      * @param rollback if TRUE and if incremental is TRUE, then perform special
917      * incremental processing, as described above, and undo partial
918      * transliterations where necessary.  If incremental is FALSE then this
919      * parameter is ignored.
920      */
921     virtual void filteredTransliterate(Replaceable& text,
922                                        UTransPosition& index,
923                                        UBool incremental,
924                                        UBool rollback) const;
925 
926 public:
927 
928     /**
929      * Returns the length of the longest context required by this transliterator.
930      * This is <em>preceding</em> context.  The default implementation supplied
931      * by <code>Transliterator</code> returns zero; subclasses
932      * that use preceding context should override this method to return the
933      * correct value.  For example, if a transliterator translates "ddd" (where
934      * d is any digit) to "555" when preceded by "(ddd)", then the preceding
935      * context length is 5, the length of "(ddd)".
936      *
937      * @return The maximum number of preceding context characters this
938      * transliterator needs to examine
939      * @stable ICU 2.0
940      */
941     int32_t getMaximumContextLength(void) const;
942 
943 protected:
944 
945     /**
946      * Method for subclasses to use to set the maximum context length.
947      * @param maxContextLength the new value to be set.
948      * @see #getMaximumContextLength
949      * @stable ICU 2.4
950      */
951     void setMaximumContextLength(int32_t maxContextLength);
952 
953 public:
954 
955     /**
956      * Returns a programmatic identifier for this transliterator.
957      * If this identifier is passed to <code>createInstance()</code>, it
958      * will return this object, if it has been registered.
959      * @return a programmatic identifier for this transliterator.
960      * @see #registerInstance
961      * @see #registerFactory
962      * @see #getAvailableIDs
963      * @stable ICU 2.0
964      */
965     virtual const UnicodeString& getID(void) const;
966 
967     /**
968      * Returns a name for this transliterator that is appropriate for
969      * display to the user in the default locale.  See {@link
970      * #getDisplayName } for details.
971      * @param ID     the string identifier for this transliterator
972      * @param result Output param to receive the display name
973      * @return       A reference to 'result'.
974      * @stable ICU 2.0
975      */
976     static UnicodeString& U_EXPORT2 getDisplayName(const UnicodeString& ID,
977                                          UnicodeString& result);
978 
979     /**
980      * Returns a name for this transliterator that is appropriate for
981      * display to the user in the given locale.  This name is taken
982      * from the locale resource data in the standard manner of the
983      * <code>java.text</code> package.
984      *
985      * <p>If no localized names exist in the system resource bundles,
986      * a name is synthesized using a localized
987      * <code>MessageFormat</code> pattern from the resource data.  The
988      * arguments to this pattern are an integer followed by one or two
989      * strings.  The integer is the number of strings, either 1 or 2.
990      * The strings are formed by splitting the ID for this
991      * transliterator at the first '-'.  If there is no '-', then the
992      * entire ID forms the only string.
993      * @param ID       the string identifier for this transliterator
994      * @param inLocale the Locale in which the display name should be
995      *                 localized.
996      * @param result   Output param to receive the display name
997      * @return         A reference to 'result'.
998      * @stable ICU 2.0
999      */
1000     static UnicodeString& U_EXPORT2 getDisplayName(const UnicodeString& ID,
1001                                          const Locale& inLocale,
1002                                          UnicodeString& result);
1003 
1004     /**
1005      * Returns the filter used by this transliterator, or <tt>NULL</tt>
1006      * if this transliterator uses no filter.
1007      * @return the filter used by this transliterator, or <tt>NULL</tt>
1008      *         if this transliterator uses no filter.
1009      * @stable ICU 2.0
1010      */
1011     const UnicodeFilter* getFilter(void) const;
1012 
1013     /**
1014      * Returns the filter used by this transliterator, or <tt>NULL</tt> if this
1015      * transliterator uses no filter.  The caller must eventually delete the
1016      * result.  After this call, this transliterator's filter is set to
1017      * <tt>NULL</tt>.
1018      * @return the filter used by this transliterator, or <tt>NULL</tt> if this
1019      *         transliterator uses no filter.
1020      * @stable ICU 2.4
1021      */
1022     UnicodeFilter* orphanFilter(void);
1023 
1024     /**
1025      * Changes the filter used by this transliterator.  If the filter
1026      * is set to <tt>null</tt> then no filtering will occur.
1027      *
1028      * <p>Callers must take care if a transliterator is in use by
1029      * multiple threads.  The filter should not be changed by one
1030      * thread while another thread may be transliterating.
1031      * @param adoptedFilter the new filter to be adopted.
1032      * @stable ICU 2.0
1033      */
1034     void adoptFilter(UnicodeFilter* adoptedFilter);
1035 
1036     /**
1037      * Returns this transliterator's inverse.  See the class
1038      * documentation for details.  This implementation simply inverts
1039      * the two entities in the ID and attempts to retrieve the
1040      * resulting transliterator.  That is, if <code>getID()</code>
1041      * returns "A-B", then this method will return the result of
1042      * <code>createInstance("B-A")</code>, or <code>null</code> if that
1043      * call fails.
1044      *
1045      * <p>Subclasses with knowledge of their inverse may wish to
1046      * override this method.
1047      *
1048      * @param status Output param to filled in with a success or an error.
1049      * @return a transliterator that is an inverse, not necessarily
1050      * exact, of this transliterator, or <code>null</code> if no such
1051      * transliterator is registered.
1052      * @see #registerInstance
1053      * @stable ICU 2.0
1054      */
1055     Transliterator* createInverse(UErrorCode& status) const;
1056 
1057     /**
1058      * Returns a <code>Transliterator</code> object given its ID.
1059      * The ID must be either a system transliterator ID or a ID registered
1060      * using <code>registerInstance()</code>.
1061      *
1062      * @param ID a valid ID, as enumerated by <code>getAvailableIDs()</code>
1063      * @param dir        either FORWARD or REVERSE.
1064      * @param parseError Struct to recieve information on position
1065      *                   of error if an error is encountered
1066      * @param status     Output param to filled in with a success or an error.
1067      * @return A <code>Transliterator</code> object with the given ID
1068      * @see #registerInstance
1069      * @see #getAvailableIDs
1070      * @see #getID
1071      * @stable ICU 2.0
1072      */
1073     static Transliterator* U_EXPORT2 createInstance(const UnicodeString& ID,
1074                                           UTransDirection dir,
1075                                           UParseError& parseError,
1076                                           UErrorCode& status);
1077 
1078     /**
1079      * Returns a <code>Transliterator</code> object given its ID.
1080      * The ID must be either a system transliterator ID or a ID registered
1081      * using <code>registerInstance()</code>.
1082      * @param ID a valid ID, as enumerated by <code>getAvailableIDs()</code>
1083      * @param dir        either FORWARD or REVERSE.
1084      * @param status     Output param to filled in with a success or an error.
1085      * @return A <code>Transliterator</code> object with the given ID
1086      * @stable ICU 2.0
1087      */
1088     static Transliterator* U_EXPORT2 createInstance(const UnicodeString& ID,
1089                                           UTransDirection dir,
1090                                           UErrorCode& status);
1091 
1092     /**
1093      * Returns a <code>Transliterator</code> object constructed from
1094      * the given rule string.  This will be a rule-based Transliterator,
1095      * if the rule string contains only rules, or a
1096      * compound Transliterator, if it contains ID blocks, or a
1097      * null Transliterator, if it contains ID blocks which parse as
1098      * empty for the given direction.
1099      *
1100      * @param ID            the id for the transliterator.
1101      * @param rules         rules, separated by ';'
1102      * @param dir           either FORWARD or REVERSE.
1103      * @param parseError    Struct to receive information on position
1104      *                      of error if an error is encountered
1105      * @param status        Output param set to success/failure code.
1106      * @return a newly created Transliterator
1107      * @stable ICU 2.0
1108      */
1109     static Transliterator* U_EXPORT2 createFromRules(const UnicodeString& ID,
1110                                            const UnicodeString& rules,
1111                                            UTransDirection dir,
1112                                            UParseError& parseError,
1113                                            UErrorCode& status);
1114 
1115     /**
1116      * Create a rule string that can be passed to createFromRules()
1117      * to recreate this transliterator.
1118      * @param result the string to receive the rules.  Previous
1119      * contents will be deleted.
1120      * @param escapeUnprintable if TRUE then convert unprintable
1121      * character to their hex escape representations, \\uxxxx or
1122      * \\Uxxxxxxxx.  Unprintable characters are those other than
1123      * U+000A, U+0020..U+007E.
1124      * @stable ICU 2.0
1125      */
1126     virtual UnicodeString& toRules(UnicodeString& result,
1127                                    UBool escapeUnprintable) const;
1128 
1129     /**
1130      * Return the number of elements that make up this transliterator.
1131      * For example, if the transliterator "NFD;Jamo-Latin;Latin-Greek"
1132      * were created, the return value of this method would be 3.
1133      *
1134      * <p>If this transliterator is not composed of other
1135      * transliterators, then this method returns 1.
1136      * @return the number of transliterators that compose this
1137      * transliterator, or 1 if this transliterator is not composed of
1138      * multiple transliterators
1139      * @stable ICU 3.0
1140      */
1141     int32_t countElements() const;
1142 
1143     /**
1144      * Return an element that makes up this transliterator.  For
1145      * example, if the transliterator "NFD;Jamo-Latin;Latin-Greek"
1146      * were created, the return value of this method would be one
1147      * of the three transliterator objects that make up that
1148      * transliterator: [NFD, Jamo-Latin, Latin-Greek].
1149      *
1150      * <p>If this transliterator is not composed of other
1151      * transliterators, then this method will return a reference to
1152      * this transliterator when given the index 0.
1153      * @param index a value from 0..countElements()-1 indicating the
1154      * transliterator to return
1155      * @param ec input-output error code
1156      * @return one of the transliterators that makes up this
1157      * transliterator, if this transliterator is made up of multiple
1158      * transliterators, otherwise a reference to this object if given
1159      * an index of 0
1160      * @stable ICU 3.0
1161      */
1162     const Transliterator& getElement(int32_t index, UErrorCode& ec) const;
1163 
1164     /**
1165      * Returns the set of all characters that may be modified in the
1166      * input text by this Transliterator.  This incorporates this
1167      * object's current filter; if the filter is changed, the return
1168      * value of this function will change.  The default implementation
1169      * returns an empty set.  Some subclasses may override {@link
1170      * #handleGetSourceSet } to return a more precise result.  The
1171      * return result is approximate in any case and is intended for
1172      * use by tests, tools, or utilities.
1173      * @param result receives result set; previous contents lost
1174      * @return a reference to result
1175      * @see #getTargetSet
1176      * @see #handleGetSourceSet
1177      * @stable ICU 2.4
1178      */
1179     UnicodeSet& getSourceSet(UnicodeSet& result) const;
1180 
1181     /**
1182      * Framework method that returns the set of all characters that
1183      * may be modified in the input text by this Transliterator,
1184      * ignoring the effect of this object's filter.  The base class
1185      * implementation returns the empty set.  Subclasses that wish to
1186      * implement this should override this method.
1187      * @return the set of characters that this transliterator may
1188      * modify.  The set may be modified, so subclasses should return a
1189      * newly-created object.
1190      * @param result receives result set; previous contents lost
1191      * @see #getSourceSet
1192      * @see #getTargetSet
1193      * @stable ICU 2.4
1194      */
1195     virtual void handleGetSourceSet(UnicodeSet& result) const;
1196 
1197     /**
1198      * Returns the set of all characters that may be generated as
1199      * replacement text by this transliterator.  The default
1200      * implementation returns the empty set.  Some subclasses may
1201      * override this method to return a more precise result.  The
1202      * return result is approximate in any case and is intended for
1203      * use by tests, tools, or utilities requiring such
1204      * meta-information.
1205      * @param result receives result set; previous contents lost
1206      * @return a reference to result
1207      * @see #getTargetSet
1208      * @stable ICU 2.4
1209      */
1210     virtual UnicodeSet& getTargetSet(UnicodeSet& result) const;
1211 
1212 public:
1213 
1214     /**
1215      * Registers a factory function that creates transliterators of
1216      * a given ID.
1217      *
1218      * Because ICU may choose to cache Transliterators internally, this must
1219      * be called at application startup, prior to any calls to
1220      * Transliterator::createXXX to avoid undefined behavior.
1221      *
1222      * @param id the ID being registered
1223      * @param factory a function pointer that will be copied and
1224      * called later when the given ID is passed to createInstance()
1225      * @param context a context pointer that will be stored and
1226      * later passed to the factory function when an ID matching
1227      * the registration ID is being instantiated with this factory.
1228      * @stable ICU 2.0
1229      */
1230     static void U_EXPORT2 registerFactory(const UnicodeString& id,
1231                                 Factory factory,
1232                                 Token context);
1233 
1234     /**
1235      * Registers an instance <tt>obj</tt> of a subclass of
1236      * <code>Transliterator</code> with the system.  When
1237      * <tt>createInstance()</tt> is called with an ID string that is
1238      * equal to <tt>obj->getID()</tt>, then <tt>obj->clone()</tt> is
1239      * returned.
1240      *
1241      * After this call the Transliterator class owns the adoptedObj
1242      * and will delete it.
1243      *
1244      * Because ICU may choose to cache Transliterators internally, this must
1245      * be called at application startup, prior to any calls to
1246      * Transliterator::createXXX to avoid undefined behavior.
1247      *
1248      * @param adoptedObj an instance of subclass of
1249      * <code>Transliterator</code> that defines <tt>clone()</tt>
1250      * @see #createInstance
1251      * @see #registerFactory
1252      * @see #unregister
1253      * @stable ICU 2.0
1254      */
1255     static void U_EXPORT2 registerInstance(Transliterator* adoptedObj);
1256 
1257     /**
1258      * Registers an ID string as an alias of another ID string.
1259      * That is, after calling this function, <tt>createInstance(aliasID)</tt>
1260      * will return the same thing as <tt>createInstance(realID)</tt>.
1261      * This is generally used to create shorter, more mnemonic aliases
1262      * for long compound IDs.
1263      *
1264      * @param aliasID The new ID being registered.
1265      * @param realID The ID that the new ID is to be an alias for.
1266      * This can be a compound ID and can include filters and should
1267      * refer to transliterators that have already been registered with
1268      * the framework, although this isn't checked.
1269      * @stable ICU 3.6
1270      */
1271      static void U_EXPORT2 registerAlias(const UnicodeString& aliasID,
1272                                          const UnicodeString& realID);
1273 
1274 protected:
1275 
1276 #ifndef U_HIDE_INTERNAL_API
1277     /**
1278      * @param id the ID being registered
1279      * @param factory a function pointer that will be copied and
1280      * called later when the given ID is passed to createInstance()
1281      * @param context a context pointer that will be stored and
1282      * later passed to the factory function when an ID matching
1283      * the registration ID is being instantiated with this factory.
1284      * @internal
1285      */
1286     static void _registerFactory(const UnicodeString& id,
1287                                  Factory factory,
1288                                  Token context);
1289 
1290     /**
1291      * @internal
1292      */
1293     static void _registerInstance(Transliterator* adoptedObj);
1294 
1295     /**
1296      * @internal
1297      */
1298     static void _registerAlias(const UnicodeString& aliasID, const UnicodeString& realID);
1299 
1300     /**
1301      * Register two targets as being inverses of one another.  For
1302      * example, calling registerSpecialInverse("NFC", "NFD", true) causes
1303      * Transliterator to form the following inverse relationships:
1304      *
1305      * <pre>NFC => NFD
1306      * Any-NFC => Any-NFD
1307      * NFD => NFC
1308      * Any-NFD => Any-NFC</pre>
1309      *
1310      * (Without the special inverse registration, the inverse of NFC
1311      * would be NFC-Any.)  Note that NFD is shorthand for Any-NFD, but
1312      * that the presence or absence of "Any-" is preserved.
1313      *
1314      * <p>The relationship is symmetrical; registering (a, b) is
1315      * equivalent to registering (b, a).
1316      *
1317      * <p>The relevant IDs must still be registered separately as
1318      * factories or classes.
1319      *
1320      * <p>Only the targets are specified.  Special inverses always
1321      * have the form Any-Target1 <=> Any-Target2.  The target should
1322      * have canonical casing (the casing desired to be produced when
1323      * an inverse is formed) and should contain no whitespace or other
1324      * extraneous characters.
1325      *
1326      * @param target the target against which to register the inverse
1327      * @param inverseTarget the inverse of target, that is
1328      * Any-target.getInverse() => Any-inverseTarget
1329      * @param bidirectional if true, register the reverse relation
1330      * as well, that is, Any-inverseTarget.getInverse() => Any-target
1331      * @internal
1332      */
1333     static void _registerSpecialInverse(const UnicodeString& target,
1334                                         const UnicodeString& inverseTarget,
1335                                         UBool bidirectional);
1336 #endif  /* U_HIDE_INTERNAL_API */
1337 
1338 public:
1339 
1340     /**
1341      * Unregisters a transliterator or class.  This may be either
1342      * a system transliterator or a user transliterator or class.
1343      * Any attempt to construct an unregistered transliterator based
1344      * on its ID will fail.
1345      *
1346      * Because ICU may choose to cache Transliterators internally, this should
1347      * be called during application shutdown, after all calls to
1348      * Transliterator::createXXX to avoid undefined behavior.
1349      *
1350      * @param ID the ID of the transliterator or class
1351      * @return the <code>Object</code> that was registered with
1352      * <code>ID</code>, or <code>null</code> if none was
1353      * @see #registerInstance
1354      * @see #registerFactory
1355      * @stable ICU 2.0
1356      */
1357     static void U_EXPORT2 unregister(const UnicodeString& ID);
1358 
1359 public:
1360 
1361     /**
1362      * Return a StringEnumeration over the IDs available at the time of the
1363      * call, including user-registered IDs.
1364      * @param ec input-output error code
1365      * @return a newly-created StringEnumeration over the transliterators
1366      * available at the time of the call. The caller should delete this object
1367      * when done using it.
1368      * @stable ICU 3.0
1369      */
1370     static StringEnumeration* U_EXPORT2 getAvailableIDs(UErrorCode& ec);
1371 
1372     /**
1373      * Return the number of registered source specifiers.
1374      * @return the number of registered source specifiers.
1375      * @stable ICU 2.0
1376      */
1377     static int32_t U_EXPORT2 countAvailableSources(void);
1378 
1379     /**
1380      * Return a registered source specifier.
1381      * @param index which specifier to return, from 0 to n-1, where
1382      * n = countAvailableSources()
1383      * @param result fill-in paramter to receive the source specifier.
1384      * If index is out of range, result will be empty.
1385      * @return reference to result
1386      * @stable ICU 2.0
1387      */
1388     static UnicodeString& U_EXPORT2 getAvailableSource(int32_t index,
1389                                              UnicodeString& result);
1390 
1391     /**
1392      * Return the number of registered target specifiers for a given
1393      * source specifier.
1394      * @param source the given source specifier.
1395      * @return the number of registered target specifiers for a given
1396      *         source specifier.
1397      * @stable ICU 2.0
1398      */
1399     static int32_t U_EXPORT2 countAvailableTargets(const UnicodeString& source);
1400 
1401     /**
1402      * Return a registered target specifier for a given source.
1403      * @param index which specifier to return, from 0 to n-1, where
1404      * n = countAvailableTargets(source)
1405      * @param source the source specifier
1406      * @param result fill-in paramter to receive the target specifier.
1407      * If source is invalid or if index is out of range, result will
1408      * be empty.
1409      * @return reference to result
1410      * @stable ICU 2.0
1411      */
1412     static UnicodeString& U_EXPORT2 getAvailableTarget(int32_t index,
1413                                              const UnicodeString& source,
1414                                              UnicodeString& result);
1415 
1416     /**
1417      * Return the number of registered variant specifiers for a given
1418      * source-target pair.
1419      * @param source    the source specifiers.
1420      * @param target    the target specifiers.
1421      * @stable ICU 2.0
1422      */
1423     static int32_t U_EXPORT2 countAvailableVariants(const UnicodeString& source,
1424                                           const UnicodeString& target);
1425 
1426     /**
1427      * Return a registered variant specifier for a given source-target
1428      * pair.
1429      * @param index which specifier to return, from 0 to n-1, where
1430      * n = countAvailableVariants(source, target)
1431      * @param source the source specifier
1432      * @param target the target specifier
1433      * @param result fill-in paramter to receive the variant
1434      * specifier.  If source is invalid or if target is invalid or if
1435      * index is out of range, result will be empty.
1436      * @return reference to result
1437      * @stable ICU 2.0
1438      */
1439     static UnicodeString& U_EXPORT2 getAvailableVariant(int32_t index,
1440                                               const UnicodeString& source,
1441                                               const UnicodeString& target,
1442                                               UnicodeString& result);
1443 
1444 protected:
1445 
1446 #ifndef U_HIDE_INTERNAL_API
1447     /**
1448      * Non-mutexed internal method
1449      * @internal
1450      */
1451     static int32_t _countAvailableSources(void);
1452 
1453     /**
1454      * Non-mutexed internal method
1455      * @internal
1456      */
1457     static UnicodeString& _getAvailableSource(int32_t index,
1458                                               UnicodeString& result);
1459 
1460     /**
1461      * Non-mutexed internal method
1462      * @internal
1463      */
1464     static int32_t _countAvailableTargets(const UnicodeString& source);
1465 
1466     /**
1467      * Non-mutexed internal method
1468      * @internal
1469      */
1470     static UnicodeString& _getAvailableTarget(int32_t index,
1471                                               const UnicodeString& source,
1472                                               UnicodeString& result);
1473 
1474     /**
1475      * Non-mutexed internal method
1476      * @internal
1477      */
1478     static int32_t _countAvailableVariants(const UnicodeString& source,
1479                                            const UnicodeString& target);
1480 
1481     /**
1482      * Non-mutexed internal method
1483      * @internal
1484      */
1485     static UnicodeString& _getAvailableVariant(int32_t index,
1486                                                const UnicodeString& source,
1487                                                const UnicodeString& target,
1488                                                UnicodeString& result);
1489 #endif  /* U_HIDE_INTERNAL_API */
1490 
1491 protected:
1492 
1493     /**
1494      * Set the ID of this transliterators.  Subclasses shouldn't do
1495      * this, unless the underlying script behavior has changed.
1496      * @param id the new id t to be set.
1497      * @stable ICU 2.4
1498      */
1499     void setID(const UnicodeString& id);
1500 
1501 public:
1502 
1503     /**
1504      * Return the class ID for this class.  This is useful only for
1505      * comparing to a return value from getDynamicClassID().
1506      * Note that Transliterator is an abstract base class, and therefor
1507      * no fully constructed object will  have a dynamic
1508      * UCLassID that equals the UClassID returned from
1509      * TRansliterator::getStaticClassID().
1510      * @return       The class ID for class Transliterator.
1511      * @stable ICU 2.0
1512      */
1513     static UClassID U_EXPORT2 getStaticClassID(void);
1514 
1515     /**
1516      * Returns a unique class ID <b>polymorphically</b>.  This method
1517      * is to implement a simple version of RTTI, since not all C++
1518      * compilers support genuine RTTI.  Polymorphic operator==() and
1519      * clone() methods call this method.
1520      *
1521      * <p>Concrete subclasses of Transliterator must use the
1522      *    UOBJECT_DEFINE_RTTI_IMPLEMENTATION macro from
1523      *    uobject.h to provide the RTTI functions.
1524      *
1525      * @return The class ID for this object. All objects of a given
1526      * class have the same class ID.  Objects of other classes have
1527      * different class IDs.
1528      * @stable ICU 2.0
1529      */
1530     virtual UClassID getDynamicClassID(void) const = 0;
1531 
1532 private:
1533     static UBool initializeRegistry(UErrorCode &status);
1534 
1535 public:
1536 #ifndef U_HIDE_OBSOLETE_API
1537     /**
1538      * Return the number of IDs currently registered with the system.
1539      * To retrieve the actual IDs, call getAvailableID(i) with
1540      * i from 0 to countAvailableIDs() - 1.
1541      * @return the number of IDs currently registered with the system.
1542      * @obsolete ICU 3.4 use getAvailableIDs() instead
1543      */
1544     static int32_t U_EXPORT2 countAvailableIDs(void);
1545 
1546     /**
1547      * Return the index-th available ID.  index must be between 0
1548      * and countAvailableIDs() - 1, inclusive.  If index is out of
1549      * range, the result of getAvailableID(0) is returned.
1550      * @param index the given ID index.
1551      * @return      the index-th available ID.  index must be between 0
1552      *              and countAvailableIDs() - 1, inclusive.  If index is out of
1553      *              range, the result of getAvailableID(0) is returned.
1554      * @obsolete ICU 3.4 use getAvailableIDs() instead; this function
1555      * is not thread safe, since it returns a reference to storage that
1556      * may become invalid if another thread calls unregister
1557      */
1558     static const UnicodeString& U_EXPORT2 getAvailableID(int32_t index);
1559 #endif  /* U_HIDE_OBSOLETE_API */
1560 };
1561 
getMaximumContextLength(void)1562 inline int32_t Transliterator::getMaximumContextLength(void) const {
1563     return maximumContextLength;
1564 }
1565 
setID(const UnicodeString & id)1566 inline void Transliterator::setID(const UnicodeString& id) {
1567     ID = id;
1568     // NUL-terminate the ID string, which is a non-aliased copy.
1569     ID.append((char16_t)0);
1570     ID.truncate(ID.length()-1);
1571 }
1572 
1573 #ifndef U_HIDE_INTERNAL_API
integerToken(int32_t i)1574 inline Transliterator::Token Transliterator::integerToken(int32_t i) {
1575     Token t;
1576     t.integer = i;
1577     return t;
1578 }
1579 
pointerToken(void * p)1580 inline Transliterator::Token Transliterator::pointerToken(void* p) {
1581     Token t;
1582     t.pointer = p;
1583     return t;
1584 }
1585 #endif  /* U_HIDE_INTERNAL_API */
1586 
1587 U_NAMESPACE_END
1588 
1589 #endif /* #if !UCONFIG_NO_TRANSLITERATION */
1590 
1591 #endif
1592