1// © 2016 and later: Unicode, Inc. and others.
2// License & terms of use: http://www.unicode.org/copyright.html
3/*
4*******************************************************************************
5* Copyright (C) 2011-2013, International Business Machines
6* Corporation and others. All Rights Reserved.
7*******************************************************************************
8* file name: messagepattern.h
9* encoding: UTF-8
10* tab size: 8 (not used)
11* indentation:4
12*
13* created on: 2011mar14
14* created by: Markus W. Scherer
15*/
16
17#ifndef __MESSAGEPATTERN_H__
18#define __MESSAGEPATTERN_H__
19
20/**
21 * \file
22 * \brief C++ API: MessagePattern class: Parses and represents ICU MessageFormat patterns.
23 */
24
25#include "unicode/utypes.h"
26
27#if U_SHOW_CPLUSPLUS_API
28
29#if !UCONFIG_NO_FORMATTING
30
31#include "unicode/parseerr.h"
32#include "unicode/unistr.h"
33
34/**
35 * Mode for when an apostrophe starts quoted literal text for MessageFormat output.
36 * The default is DOUBLE_OPTIONAL unless overridden via uconfig.h
37 * (UCONFIG_MSGPAT_DEFAULT_APOSTROPHE_MODE).
38 * <p>
39 * A pair of adjacent apostrophes always results in a single apostrophe in the output,
40 * even when the pair is between two single, text-quoting apostrophes.
41 * <p>
42 * The following table shows examples of desired MessageFormat.format() output
43 * with the pattern strings that yield that output.
44 * <p>
45 * <table>
46 * <tr>
47 * <th>Desired output</th>
48 * <th>DOUBLE_OPTIONAL</th>
49 * <th>DOUBLE_REQUIRED</th>
50 * </tr>
51 * <tr>
52 * <td>I see {many}</td>
53 * <td>I see '{many}'</td>
54 * <td>(same)</td>
55 * </tr>
56 * <tr>
57 * <td>I said {'Wow!'}</td>
58 * <td>I said '{''Wow!''}'</td>
59 * <td>(same)</td>
60 * </tr>
61 * <tr>
62 * <td>I don't know</td>
63 * <td>I don't know OR<br> I don''t know</td>
64 * <td>I don''t know</td>
65 * </tr>
66 * </table>
67 * @stable ICU 4.8
68 * @see UCONFIG_MSGPAT_DEFAULT_APOSTROPHE_MODE
69 */
70enum UMessagePatternApostropheMode {
71 /**
72 * A literal apostrophe is represented by
73 * either a single or a double apostrophe pattern character.
74 * Within a MessageFormat pattern, a single apostrophe only starts quoted literal text
75 * if it immediately precedes a curly brace {},
76 * or a pipe symbol | if inside a choice format,
77 * or a pound symbol # if inside a plural format.
78 * <p>
79 * This is the default behavior starting with ICU 4.8.
80 * @stable ICU 4.8
81 */
82 UMSGPAT_APOS_DOUBLE_OPTIONAL,
83 /**
84 * A literal apostrophe must be represented by
85 * a double apostrophe pattern character.
86 * A single apostrophe always starts quoted literal text.
87 * <p>
88 * This is the behavior of ICU 4.6 and earlier, and of the JDK.
89 * @stable ICU 4.8
90 */
91 UMSGPAT_APOS_DOUBLE_REQUIRED
92};
93/**
94 * @stable ICU 4.8
95 */
96typedef enum UMessagePatternApostropheMode UMessagePatternApostropheMode;
97
98/**
99 * MessagePattern::Part type constants.
100 * @stable ICU 4.8
101 */
102enum UMessagePatternPartType {
103 /**
104 * Start of a message pattern (main or nested).
105 * The length is 0 for the top-level message
106 * and for a choice argument sub-message, otherwise 1 for the '{'.
107 * The value indicates the nesting level, starting with 0 for the main message.
108 * <p>
109 * There is always a later MSG_LIMIT part.
110 * @stable ICU 4.8
111 */
112 UMSGPAT_PART_TYPE_MSG_START,
113 /**
114 * End of a message pattern (main or nested).
115 * The length is 0 for the top-level message and
116 * the last sub-message of a choice argument,
117 * otherwise 1 for the '}' or (in a choice argument style) the '|'.
118 * The value indicates the nesting level, starting with 0 for the main message.
119 * @stable ICU 4.8
120 */
121 UMSGPAT_PART_TYPE_MSG_LIMIT,
122 /**
123 * Indicates a substring of the pattern string which is to be skipped when formatting.
124 * For example, an apostrophe that begins or ends quoted text
125 * would be indicated with such a part.
126 * The value is undefined and currently always 0.
127 * @stable ICU 4.8
128 */
129 UMSGPAT_PART_TYPE_SKIP_SYNTAX,
130 /**
131 * Indicates that a syntax character needs to be inserted for auto-quoting.
132 * The length is 0.
133 * The value is the character code of the insertion character. (U+0027=APOSTROPHE)
134 * @stable ICU 4.8
135 */
136 UMSGPAT_PART_TYPE_INSERT_CHAR,
137 /**
138 * Indicates a syntactic (non-escaped) # symbol in a plural variant.
139 * When formatting, replace this part's substring with the
140 * (value-offset) for the plural argument value.
141 * The value is undefined and currently always 0.
142 * @stable ICU 4.8
143 */
144 UMSGPAT_PART_TYPE_REPLACE_NUMBER,
145 /**
146 * Start of an argument.
147 * The length is 1 for the '{'.
148 * The value is the ordinal value of the ArgType. Use getArgType().
149 * <p>
150 * This part is followed by either an ARG_NUMBER or ARG_NAME,
151 * followed by optional argument sub-parts (see UMessagePatternArgType constants)
152 * and finally an ARG_LIMIT part.
153 * @stable ICU 4.8
154 */
155 UMSGPAT_PART_TYPE_ARG_START,
156 /**
157 * End of an argument.
158 * The length is 1 for the '}'.
159 * The value is the ordinal value of the ArgType. Use getArgType().
160 * @stable ICU 4.8
161 */
162 UMSGPAT_PART_TYPE_ARG_LIMIT,
163 /**
164 * The argument number, provided by the value.
165 * @stable ICU 4.8
166 */
167 UMSGPAT_PART_TYPE_ARG_NUMBER,
168 /**
169 * The argument name.
170 * The value is undefined and currently always 0.
171 * @stable ICU 4.8
172 */
173 UMSGPAT_PART_TYPE_ARG_NAME,
174 /**
175 * The argument type.
176 * The value is undefined and currently always 0.
177 * @stable ICU 4.8
178 */
179 UMSGPAT_PART_TYPE_ARG_TYPE,
180 /**
181 * The argument style text.
182 * The value is undefined and currently always 0.
183 * @stable ICU 4.8
184 */
185 UMSGPAT_PART_TYPE_ARG_STYLE,
186 /**
187 * A selector substring in a "complex" argument style.
188 * The value is undefined and currently always 0.
189 * @stable ICU 4.8
190 */
191 UMSGPAT_PART_TYPE_ARG_SELECTOR,
192 /**
193 * An integer value, for example the offset or an explicit selector value
194 * in a PluralFormat style.
195 * The part value is the integer value.
196 * @stable ICU 4.8
197 */
198 UMSGPAT_PART_TYPE_ARG_INT,
199 /**
200 * A numeric value, for example the offset or an explicit selector value
201 * in a PluralFormat style.
202 * The part value is an index into an internal array of numeric values;
203 * use getNumericValue().
204 * @stable ICU 4.8
205 */
206 UMSGPAT_PART_TYPE_ARG_DOUBLE
207};
208/**
209 * @stable ICU 4.8
210 */
211typedef enum UMessagePatternPartType UMessagePatternPartType;
212
213/**
214 * Argument type constants.
215 * Returned by Part.getArgType() for ARG_START and ARG_LIMIT parts.
216 *
217 * Messages nested inside an argument are each delimited by MSG_START and MSG_LIMIT,
218 * with a nesting level one greater than the surrounding message.
219 * @stable ICU 4.8
220 */
221enum UMessagePatternArgType {
222 /**
223 * The argument has no specified type.
224 * @stable ICU 4.8
225 */
226 UMSGPAT_ARG_TYPE_NONE,
227 /**
228 * The argument has a "simple" type which is provided by the ARG_TYPE part.
229 * An ARG_STYLE part might follow that.
230 * @stable ICU 4.8
231 */
232 UMSGPAT_ARG_TYPE_SIMPLE,
233 /**
234 * The argument is a ChoiceFormat with one or more
235 * ((ARG_INT | ARG_DOUBLE), ARG_SELECTOR, message) tuples.
236 * @stable ICU 4.8
237 */
238 UMSGPAT_ARG_TYPE_CHOICE,
239 /**
240 * The argument is a cardinal-number PluralFormat with an optional ARG_INT or ARG_DOUBLE offset
241 * (e.g., offset:1)
242 * and one or more (ARG_SELECTOR [explicit-value] message) tuples.
243 * If the selector has an explicit value (e.g., =2), then
244 * that value is provided by the ARG_INT or ARG_DOUBLE part preceding the message.
245 * Otherwise the message immediately follows the ARG_SELECTOR.
246 * @stable ICU 4.8
247 */
248 UMSGPAT_ARG_TYPE_PLURAL,
249 /**
250 * The argument is a SelectFormat with one or more (ARG_SELECTOR, message) pairs.
251 * @stable ICU 4.8
252 */
253 UMSGPAT_ARG_TYPE_SELECT,
254 /**
255 * The argument is an ordinal-number PluralFormat
256 * with the same style parts sequence and semantics as UMSGPAT_ARG_TYPE_PLURAL.
257 * @stable ICU 50
258 */
259 UMSGPAT_ARG_TYPE_SELECTORDINAL
260};
261/**
262 * @stable ICU 4.8
263 */
264typedef enum UMessagePatternArgType UMessagePatternArgType;
265
266/**
267 * \def UMSGPAT_ARG_TYPE_HAS_PLURAL_STYLE
268 * Returns TRUE if the argument type has a plural style part sequence and semantics,
269 * for example UMSGPAT_ARG_TYPE_PLURAL and UMSGPAT_ARG_TYPE_SELECTORDINAL.
270 * @stable ICU 50
271 */
272#define UMSGPAT_ARG_TYPE_HAS_PLURAL_STYLE(argType) \
273 ((argType)==UMSGPAT_ARG_TYPE_PLURAL || (argType)==UMSGPAT_ARG_TYPE_SELECTORDINAL)
274
275enum {
276 /**
277 * Return value from MessagePattern.validateArgumentName() for when
278 * the string is a valid "pattern identifier" but not a number.
279 * @stable ICU 4.8
280 */
281 UMSGPAT_ARG_NAME_NOT_NUMBER=-1,
282
283 /**
284 * Return value from MessagePattern.validateArgumentName() for when
285 * the string is invalid.
286 * It might not be a valid "pattern identifier",
287 * or it have only ASCII digits but there is a leading zero or the number is too large.
288 * @stable ICU 4.8
289 */
290 UMSGPAT_ARG_NAME_NOT_VALID=-2
291};
292
293/**
294 * Special value that is returned by getNumericValue(Part) when no
295 * numeric value is defined for a part.
296 * @see MessagePattern.getNumericValue()
297 * @stable ICU 4.8
298 */
299#define UMSGPAT_NO_NUMERIC_VALUE ((double)(-123456789))
300
301U_NAMESPACE_BEGIN
302
303class MessagePatternDoubleList;
304class MessagePatternPartsList;
305
306/**
307 * Parses and represents ICU MessageFormat patterns.
308 * Also handles patterns for ChoiceFormat, PluralFormat and SelectFormat.
309 * Used in the implementations of those classes as well as in tools
310 * for message validation, translation and format conversion.
311 * <p>
312 * The parser handles all syntax relevant for identifying message arguments.
313 * This includes "complex" arguments whose style strings contain
314 * nested MessageFormat pattern substrings.
315 * For "simple" arguments (with no nested MessageFormat pattern substrings),
316 * the argument style is not parsed any further.
317 * <p>
318 * The parser handles named and numbered message arguments and allows both in one message.
319 * <p>
320 * Once a pattern has been parsed successfully, iterate through the parsed data
321 * with countParts(), getPart() and related methods.
322 * <p>
323 * The data logically represents a parse tree, but is stored and accessed
324 * as a list of "parts" for fast and simple parsing and to minimize object allocations.
325 * Arguments and nested messages are best handled via recursion.
326 * For every _START "part", MessagePattern.getLimitPartIndex() efficiently returns
327 * the index of the corresponding _LIMIT "part".
328 * <p>
329 * List of "parts":
330 * <pre>
331 * message = MSG_START (SKIP_SYNTAX | INSERT_CHAR | REPLACE_NUMBER | argument)* MSG_LIMIT
332 * argument = noneArg | simpleArg | complexArg
333 * complexArg = choiceArg | pluralArg | selectArg
334 *
335 * noneArg = ARG_START.NONE (ARG_NAME | ARG_NUMBER) ARG_LIMIT.NONE
336 * simpleArg = ARG_START.SIMPLE (ARG_NAME | ARG_NUMBER) ARG_TYPE [ARG_STYLE] ARG_LIMIT.SIMPLE
337 * choiceArg = ARG_START.CHOICE (ARG_NAME | ARG_NUMBER) choiceStyle ARG_LIMIT.CHOICE
338 * pluralArg = ARG_START.PLURAL (ARG_NAME | ARG_NUMBER) pluralStyle ARG_LIMIT.PLURAL
339 * selectArg = ARG_START.SELECT (ARG_NAME | ARG_NUMBER) selectStyle ARG_LIMIT.SELECT
340 *
341 * choiceStyle = ((ARG_INT | ARG_DOUBLE) ARG_SELECTOR message)+
342 * pluralStyle = [ARG_INT | ARG_DOUBLE] (ARG_SELECTOR [ARG_INT | ARG_DOUBLE] message)+
343 * selectStyle = (ARG_SELECTOR message)+
344 * </pre>
345 * <ul>
346 * <li>Literal output text is not represented directly by "parts" but accessed
347 * between parts of a message, from one part's getLimit() to the next part's getIndex().
348 * <li><code>ARG_START.CHOICE</code> stands for an ARG_START Part with ArgType CHOICE.
349 * <li>In the choiceStyle, the ARG_SELECTOR has the '<', the '#' or
350 * the less-than-or-equal-to sign (U+2264).
351 * <li>In the pluralStyle, the first, optional numeric Part has the "offset:" value.
352 * The optional numeric Part between each (ARG_SELECTOR, message) pair
353 * is the value of an explicit-number selector like "=2",
354 * otherwise the selector is a non-numeric identifier.
355 * <li>The REPLACE_NUMBER Part can occur only in an immediate sub-message of the pluralStyle.
356 * </ul>
357 * <p>
358 * This class is not intended for public subclassing.
359 *
360 * @stable ICU 4.8
361 */
362class U_COMMON_API MessagePattern : public UObject {
363public:
364 /**
365 * Constructs an empty MessagePattern with default UMessagePatternApostropheMode.
366 * @param errorCode Standard ICU error code. Its input value must
367 * pass the U_SUCCESS() test, or else the function returns
368 * immediately. Check for U_FAILURE() on output or use with
369 * function chaining. (See User Guide for details.)
370 * @stable ICU 4.8
371 */
372 MessagePattern(UErrorCode &errorCode);
373
374 /**
375 * Constructs an empty MessagePattern.
376 * @param mode Explicit UMessagePatternApostropheMode.
377 * @param errorCode Standard ICU error code. Its input value must
378 * pass the U_SUCCESS() test, or else the function returns
379 * immediately. Check for U_FAILURE() on output or use with
380 * function chaining. (See User Guide for details.)
381 * @stable ICU 4.8
382 */
383 MessagePattern(UMessagePatternApostropheMode mode, UErrorCode &errorCode);
384
385 /**
386 * Constructs a MessagePattern with default UMessagePatternApostropheMode and
387 * parses the MessageFormat pattern string.
388 * @param pattern a MessageFormat pattern string
389 * @param parseError Struct to receive information on the position
390 * of an error within the pattern.
391 * Can be NULL.
392 * @param errorCode Standard ICU error code. Its input value must
393 * pass the U_SUCCESS() test, or else the function returns
394 * immediately. Check for U_FAILURE() on output or use with
395 * function chaining. (See User Guide for details.)
396 * TODO: turn @throws into UErrorCode specifics?
397 * @throws IllegalArgumentException for syntax errors in the pattern string
398 * @throws IndexOutOfBoundsException if certain limits are exceeded
399 * (e.g., argument number too high, argument name too long, etc.)
400 * @throws NumberFormatException if a number could not be parsed
401 * @stable ICU 4.8
402 */
403 MessagePattern(const UnicodeString &pattern, UParseError *parseError, UErrorCode &errorCode);
404
405 /**
406 * Copy constructor.
407 * @param other Object to copy.
408 * @stable ICU 4.8
409 */
410 MessagePattern(const MessagePattern &other);
411
412 /**
413 * Assignment operator.
414 * @param other Object to copy.
415 * @return *this=other
416 * @stable ICU 4.8
417 */
418 MessagePattern &operator=(const MessagePattern &other);
419
420 /**
421 * Destructor.
422 * @stable ICU 4.8
423 */
424 virtual ~MessagePattern();
425
426 /**
427 * Parses a MessageFormat pattern string.
428 * @param pattern a MessageFormat pattern string
429 * @param parseError Struct to receive information on the position
430 * of an error within the pattern.
431 * Can be NULL.
432 * @param errorCode Standard ICU error code. Its input value must
433 * pass the U_SUCCESS() test, or else the function returns
434 * immediately. Check for U_FAILURE() on output or use with
435 * function chaining. (See User Guide for details.)
436 * @return *this
437 * @throws IllegalArgumentException for syntax errors in the pattern string
438 * @throws IndexOutOfBoundsException if certain limits are exceeded
439 * (e.g., argument number too high, argument name too long, etc.)
440 * @throws NumberFormatException if a number could not be parsed
441 * @stable ICU 4.8
442 */
443 MessagePattern &parse(const UnicodeString &pattern,
444 UParseError *parseError, UErrorCode &errorCode);
445
446 /**
447 * Parses a ChoiceFormat pattern string.
448 * @param pattern a ChoiceFormat pattern string
449 * @param parseError Struct to receive information on the position
450 * of an error within the pattern.
451 * Can be NULL.
452 * @param errorCode Standard ICU error code. Its input value must
453 * pass the U_SUCCESS() test, or else the function returns
454 * immediately. Check for U_FAILURE() on output or use with
455 * function chaining. (See User Guide for details.)
456 * @return *this
457 * @throws IllegalArgumentException for syntax errors in the pattern string
458 * @throws IndexOutOfBoundsException if certain limits are exceeded
459 * (e.g., argument number too high, argument name too long, etc.)
460 * @throws NumberFormatException if a number could not be parsed
461 * @stable ICU 4.8
462 */
463 MessagePattern &parseChoiceStyle(const UnicodeString &pattern,
464 UParseError *parseError, UErrorCode &errorCode);
465
466 /**
467 * Parses a PluralFormat pattern string.
468 * @param pattern a PluralFormat pattern string
469 * @param parseError Struct to receive information on the position
470 * of an error within the pattern.
471 * Can be NULL.
472 * @param errorCode Standard ICU error code. Its input value must
473 * pass the U_SUCCESS() test, or else the function returns
474 * immediately. Check for U_FAILURE() on output or use with
475 * function chaining. (See User Guide for details.)
476 * @return *this
477 * @throws IllegalArgumentException for syntax errors in the pattern string
478 * @throws IndexOutOfBoundsException if certain limits are exceeded
479 * (e.g., argument number too high, argument name too long, etc.)
480 * @throws NumberFormatException if a number could not be parsed
481 * @stable ICU 4.8
482 */
483 MessagePattern &parsePluralStyle(const UnicodeString &pattern,
484 UParseError *parseError, UErrorCode &errorCode);
485
486 /**
487 * Parses a SelectFormat pattern string.
488 * @param pattern a SelectFormat pattern string
489 * @param parseError Struct to receive information on the position
490 * of an error within the pattern.
491 * Can be NULL.
492 * @param errorCode Standard ICU error code. Its input value must
493 * pass the U_SUCCESS() test, or else the function returns
494 * immediately. Check for U_FAILURE() on output or use with
495 * function chaining. (See User Guide for details.)
496 * @return *this
497 * @throws IllegalArgumentException for syntax errors in the pattern string
498 * @throws IndexOutOfBoundsException if certain limits are exceeded
499 * (e.g., argument number too high, argument name too long, etc.)
500 * @throws NumberFormatException if a number could not be parsed
501 * @stable ICU 4.8
502 */
503 MessagePattern &parseSelectStyle(const UnicodeString &pattern,
504 UParseError *parseError, UErrorCode &errorCode);
505
506 /**
507 * Clears this MessagePattern.
508 * countParts() will return 0.
509 * @stable ICU 4.8
510 */
511 void clear();
512
513 /**
514 * Clears this MessagePattern and sets the UMessagePatternApostropheMode.
515 * countParts() will return 0.
516 * @param mode The new UMessagePatternApostropheMode.
517 * @stable ICU 4.8
518 */
519 void clearPatternAndSetApostropheMode(UMessagePatternApostropheMode mode) {
520 clear();
521 aposMode=mode;
522 }
523
524 /**
525 * @param other another object to compare with.
526 * @return TRUE if this object is equivalent to the other one.
527 * @stable ICU 4.8
528 */
529 UBool operator==(const MessagePattern &other) const;
530
531 /**
532 * @param other another object to compare with.
533 * @return FALSE if this object is equivalent to the other one.
534 * @stable ICU 4.8
535 */
536 inline UBool operator!=(const MessagePattern &other) const {
537 return !operator==(other);
538 }
539
540 /**
541 * @return A hash code for this object.
542 * @stable ICU 4.8
543 */
544 int32_t hashCode() const;
545
546 /**
547 * @return this instance's UMessagePatternApostropheMode.
548 * @stable ICU 4.8
549 */
550 UMessagePatternApostropheMode getApostropheMode() const {
551 return aposMode;
552 }
553
554 // Java has package-private jdkAposMode() here.
555 // In C++, this is declared in the MessageImpl class.
556
557 /**
558 * @return the parsed pattern string (null if none was parsed).
559 * @stable ICU 4.8
560 */
561 const UnicodeString &getPatternString() const {
562 return msg;
563 }
564
565 /**
566 * Does the parsed pattern have named arguments like {first_name}?
567 * @return TRUE if the parsed pattern has at least one named argument.
568 * @stable ICU 4.8
569 */
570 UBool hasNamedArguments() const {
571 return hasArgNames;
572 }
573
574 /**
575 * Does the parsed pattern have numbered arguments like {2}?
576 * @return TRUE if the parsed pattern has at least one numbered argument.
577 * @stable ICU 4.8
578 */
579 UBool hasNumberedArguments() const {
580 return hasArgNumbers;
581 }
582
583 /**
584 * Validates and parses an argument name or argument number string.
585 * An argument name must be a "pattern identifier", that is, it must contain
586 * no Unicode Pattern_Syntax or Pattern_White_Space characters.
587 * If it only contains ASCII digits, then it must be a small integer with no leading zero.
588 * @param name Input string.
589 * @return &gt;=0 if the name is a valid number,
590 * ARG_NAME_NOT_NUMBER (-1) if it is a "pattern identifier" but not all ASCII digits,
591 * ARG_NAME_NOT_VALID (-2) if it is neither.
592 * @stable ICU 4.8
593 */
594 static int32_t validateArgumentName(const UnicodeString &name);
595
596 /**
597 * Returns a version of the parsed pattern string where each ASCII apostrophe
598 * is doubled (escaped) if it is not already, and if it is not interpreted as quoting syntax.
599 * <p>
600 * For example, this turns "I don't '{know}' {gender,select,female{h''er}other{h'im}}."
601 * into "I don''t '{know}' {gender,select,female{h''er}other{h''im}}."
602 * @return the deep-auto-quoted version of the parsed pattern string.
603 * @see MessageFormat.autoQuoteApostrophe()
604 * @stable ICU 4.8
605 */
606 UnicodeString autoQuoteApostropheDeep() const;
607
608 class Part;
609
610 /**
611 * Returns the number of "parts" created by parsing the pattern string.
612 * Returns 0 if no pattern has been parsed or clear() was called.
613 * @return the number of pattern parts.
614 * @stable ICU 4.8
615 */
616 int32_t countParts() const {
617 return partsLength;
618 }
619
620 /**
621 * Gets the i-th pattern "part".
622 * @param i The index of the Part data. (0..countParts()-1)
623 * @return the i-th pattern "part".
624 * @stable ICU 4.8
625 */
626 const Part &getPart(int32_t i) const {
627 return parts[i];
628 }
629
630 /**
631 * Returns the UMessagePatternPartType of the i-th pattern "part".
632 * Convenience method for getPart(i).getType().
633 * @param i The index of the Part data. (0..countParts()-1)
634 * @return The UMessagePatternPartType of the i-th Part.
635 * @stable ICU 4.8
636 */
637 UMessagePatternPartType getPartType(int32_t i) const {
638 return getPart(i).type;
639 }
640
641 /**
642 * Returns the pattern index of the specified pattern "part".
643 * Convenience method for getPart(partIndex).getIndex().
644 * @param partIndex The index of the Part data. (0..countParts()-1)
645 * @return The pattern index of this Part.
646 * @stable ICU 4.8
647 */
648 int32_t getPatternIndex(int32_t partIndex) const {
649 return getPart(partIndex).index;
650 }
651
652 /**
653 * Returns the substring of the pattern string indicated by the Part.
654 * Convenience method for getPatternString().substring(part.getIndex(), part.getLimit()).
655 * @param part a part of this MessagePattern.
656 * @return the substring associated with part.
657 * @stable ICU 4.8
658 */
659 UnicodeString getSubstring(const Part &part) const {
660 return msg.tempSubString(part.index, part.length);
661 }
662
663 /**
664 * Compares the part's substring with the input string s.
665 * @param part a part of this MessagePattern.
666 * @param s a string.
667 * @return TRUE if getSubstring(part).equals(s).
668 * @stable ICU 4.8
669 */
670 UBool partSubstringMatches(const Part &part, const UnicodeString &s) const {
671 return 0==msg.compare(part.index, part.length, s);
672 }
673
674 /**
675 * Returns the numeric value associated with an ARG_INT or ARG_DOUBLE.
676 * @param part a part of this MessagePattern.
677 * @return the part's numeric value, or UMSGPAT_NO_NUMERIC_VALUE if this is not a numeric part.
678 * @stable ICU 4.8
679 */
680 double getNumericValue(const Part &part) const;
681
682 /**
683 * Returns the "offset:" value of a PluralFormat argument, or 0 if none is specified.
684 * @param pluralStart the index of the first PluralFormat argument style part. (0..countParts()-1)
685 * @return the "offset:" value.
686 * @stable ICU 4.8
687 */
688 double getPluralOffset(int32_t pluralStart) const;
689
690 /**
691 * Returns the index of the ARG|MSG_LIMIT part corresponding to the ARG|MSG_START at start.
692 * @param start The index of some Part data (0..countParts()-1);
693 * this Part should be of Type ARG_START or MSG_START.
694 * @return The first i>start where getPart(i).getType()==ARG|MSG_LIMIT at the same nesting level,
695 * or start itself if getPartType(msgStart)!=ARG|MSG_START.
696 * @stable ICU 4.8
697 */
698 int32_t getLimitPartIndex(int32_t start) const {
699 int32_t limit=getPart(start).limitPartIndex;
700 if(limit<start) {
701 return start;
702 }
703 return limit;
704 }
705
706 /**
707 * A message pattern "part", representing a pattern parsing event.
708 * There is a part for the start and end of a message or argument,
709 * for quoting and escaping of and with ASCII apostrophes,
710 * and for syntax elements of "complex" arguments.
711 * @stable ICU 4.8
712 */
713 class Part : public UMemory {
714 public:
715 /**
716 * Default constructor, do not use.
717 * @internal
718 */
719 Part() {}
720
721 /**
722 * Returns the type of this part.
723 * @return the part type.
724 * @stable ICU 4.8
725 */
726 UMessagePatternPartType getType() const {
727 return type;
728 }
729
730 /**
731 * Returns the pattern string index associated with this Part.
732 * @return this part's pattern string index.
733 * @stable ICU 4.8
734 */
735 int32_t getIndex() const {
736 return index;
737 }
738
739 /**
740 * Returns the length of the pattern substring associated with this Part.
741 * This is 0 for some parts.
742 * @return this part's pattern substring length.
743 * @stable ICU 4.8
744 */
745 int32_t getLength() const {
746 return length;
747 }
748
749 /**
750 * Returns the pattern string limit (exclusive-end) index associated with this Part.
751 * Convenience method for getIndex()+getLength().
752 * @return this part's pattern string limit index, same as getIndex()+getLength().
753 * @stable ICU 4.8
754 */
755 int32_t getLimit() const {
756 return index+length;
757 }
758
759 /**
760 * Returns a value associated with this part.
761 * See the documentation of each part type for details.
762 * @return the part value.
763 * @stable ICU 4.8
764 */
765 int32_t getValue() const {
766 return value;
767 }
768
769 /**
770 * Returns the argument type if this part is of type ARG_START or ARG_LIMIT,
771 * otherwise UMSGPAT_ARG_TYPE_NONE.
772 * @return the argument type for this part.
773 * @stable ICU 4.8
774 */
775 UMessagePatternArgType getArgType() const {
776 UMessagePatternPartType msgType=getType();
777 if(msgType ==UMSGPAT_PART_TYPE_ARG_START || msgType ==UMSGPAT_PART_TYPE_ARG_LIMIT) {
778 return (UMessagePatternArgType)value;
779 } else {
780 return UMSGPAT_ARG_TYPE_NONE;
781 }
782 }
783
784 /**
785 * Indicates whether the Part type has a numeric value.
786 * If so, then that numeric value can be retrieved via MessagePattern.getNumericValue().
787 * @param type The Part type to be tested.
788 * @return TRUE if the Part type has a numeric value.
789 * @stable ICU 4.8
790 */
791 static UBool hasNumericValue(UMessagePatternPartType type) {
792 return type==UMSGPAT_PART_TYPE_ARG_INT || type==UMSGPAT_PART_TYPE_ARG_DOUBLE;
793 }
794
795 /**
796 * @param other another object to compare with.
797 * @return TRUE if this object is equivalent to the other one.
798 * @stable ICU 4.8
799 */
800 UBool operator==(const Part &other) const;
801
802 /**
803 * @param other another object to compare with.
804 * @return FALSE if this object is equivalent to the other one.
805 * @stable ICU 4.8
806 */
807 inline UBool operator!=(const Part &other) const {
808 return !operator==(other);
809 }
810
811 /**
812 * @return A hash code for this object.
813 * @stable ICU 4.8
814 */
815 int32_t hashCode() const {
816 return ((type*37+index)*37+length)*37+value;
817 }
818
819 private:
820 friend class MessagePattern;
821
822 static const int32_t MAX_LENGTH=0xffff;
823 static const int32_t MAX_VALUE=0x7fff;
824
825 // Some fields are not final because they are modified during pattern parsing.
826 // After pattern parsing, the parts are effectively immutable.
827 UMessagePatternPartType type;
828 int32_t index;
829 uint16_t length;
830 int16_t value;
831 int32_t limitPartIndex;
832 };
833
834private:
835 void preParse(const UnicodeString &pattern, UParseError *parseError, UErrorCode &errorCode);
836
837 void postParse();
838
839 int32_t parseMessage(int32_t index, int32_t msgStartLength,
840 int32_t nestingLevel, UMessagePatternArgType parentType,
841 UParseError *parseError, UErrorCode &errorCode);
842
843 int32_t parseArg(int32_t index, int32_t argStartLength, int32_t nestingLevel,
844 UParseError *parseError, UErrorCode &errorCode);
845
846 int32_t parseSimpleStyle(int32_t index, UParseError *parseError, UErrorCode &errorCode);
847
848 int32_t parseChoiceStyle(int32_t index, int32_t nestingLevel,
849 UParseError *parseError, UErrorCode &errorCode);
850
851 int32_t parsePluralOrSelectStyle(UMessagePatternArgType argType, int32_t index, int32_t nestingLevel,
852 UParseError *parseError, UErrorCode &errorCode);
853
854 /**
855 * Validates and parses an argument name or argument number string.
856 * This internal method assumes that the input substring is a "pattern identifier".
857 * @return &gt;=0 if the name is a valid number,
858 * ARG_NAME_NOT_NUMBER (-1) if it is a "pattern identifier" but not all ASCII digits,
859 * ARG_NAME_NOT_VALID (-2) if it is neither.
860 * @see #validateArgumentName(String)
861 */
862 static int32_t parseArgNumber(const UnicodeString &s, int32_t start, int32_t limit);
863
864 int32_t parseArgNumber(int32_t start, int32_t limit) {
865 return parseArgNumber(msg, start, limit);
866 }
867
868 /**
869 * Parses a number from the specified message substring.
870 * @param start start index into the message string
871 * @param limit limit index into the message string, must be start<limit
872 * @param allowInfinity TRUE if U+221E is allowed (for ChoiceFormat)
873 * @param parseError
874 * @param errorCode
875 */
876 void parseDouble(int32_t start, int32_t limit, UBool allowInfinity,
877 UParseError *parseError, UErrorCode &errorCode);
878
879 // Java has package-private appendReducedApostrophes() here.
880 // In C++, this is declared in the MessageImpl class.
881
882 int32_t skipWhiteSpace(int32_t index);
883
884 int32_t skipIdentifier(int32_t index);
885
886 /**
887 * Skips a sequence of characters that could occur in a double value.
888 * Does not fully parse or validate the value.
889 */
890 int32_t skipDouble(int32_t index);
891
892 static UBool isArgTypeChar(UChar32 c);
893
894 UBool isChoice(int32_t index);
895
896 UBool isPlural(int32_t index);
897
898 UBool isSelect(int32_t index);
899
900 UBool isOrdinal(int32_t index);
901
902 /**
903 * @return TRUE if we are inside a MessageFormat (sub-)pattern,
904 * as opposed to inside a top-level choice/plural/select pattern.
905 */
906 UBool inMessageFormatPattern(int32_t nestingLevel);
907
908 /**
909 * @return TRUE if we are in a MessageFormat sub-pattern
910 * of a top-level ChoiceFormat pattern.
911 */
912 UBool inTopLevelChoiceMessage(int32_t nestingLevel, UMessagePatternArgType parentType);
913
914 void addPart(UMessagePatternPartType type, int32_t index, int32_t length,
915 int32_t value, UErrorCode &errorCode);
916
917 void addLimitPart(int32_t start,
918 UMessagePatternPartType type, int32_t index, int32_t length,
919 int32_t value, UErrorCode &errorCode);
920
921 void addArgDoublePart(double numericValue, int32_t start, int32_t length, UErrorCode &errorCode);
922
923 void setParseError(UParseError *parseError, int32_t index);
924
925 UBool init(UErrorCode &errorCode);
926 UBool copyStorage(const MessagePattern &other, UErrorCode &errorCode);
927
928 UMessagePatternApostropheMode aposMode;
929 UnicodeString msg;
930 // ArrayList<Part> parts=new ArrayList<Part>();
931 MessagePatternPartsList *partsList;
932 Part *parts;
933 int32_t partsLength;
934 // ArrayList<Double> numericValues;
935 MessagePatternDoubleList *numericValuesList;
936 double *numericValues;
937 int32_t numericValuesLength;
938 UBool hasArgNames;
939 UBool hasArgNumbers;
940 UBool needsAutoQuoting;
941};
942
943U_NAMESPACE_END
944
945#endif // !UCONFIG_NO_FORMATTING
946
947#endif /* U_SHOW_CPLUSPLUS_API */
948
949#endif // __MESSAGEPATTERN_H__
950