]>
Commit | Line | Data |
---|---|---|
b75a7d8f A |
1 | /* |
2 | ******************************************************************************* | |
3 | * | |
2ca993e8 | 4 | * Copyright (C) 1998-2016, International Business Machines |
b75a7d8f A |
5 | * Corporation and others. All Rights Reserved. |
6 | * | |
7 | ******************************************************************************* | |
8 | * | |
2ca993e8 | 9 | * File genrb.cpp |
b75a7d8f A |
10 | * |
11 | * Modification History: | |
12 | * | |
13 | * Date Name Description | |
14 | * 05/25/99 stephen Creation. | |
15 | * 5/10/01 Ram removed ustdio dependency | |
16 | ******************************************************************************* | |
17 | */ | |
18 | ||
2ca993e8 | 19 | #include <assert.h> |
b75a7d8f | 20 | #include "genrb.h" |
2ca993e8 | 21 | #include "unicode/localpointer.h" |
374ca955 | 22 | #include "unicode/uclean.h" |
2ca993e8 A |
23 | #include "unicode/utf16.h" |
24 | #include "charstr.h" | |
25 | #include "cmemory.h" | |
26 | #include "reslist.h" | |
729e4ab9 A |
27 | #include "ucmndata.h" /* TODO: for reading the pool bundle */ |
28 | ||
2ca993e8 A |
29 | U_NAMESPACE_USE |
30 | ||
b75a7d8f | 31 | /* Protos */ |
4388f060 | 32 | void processFile(const char *filename, const char* cp, const char *inputDir, const char *outputDir, |
2ca993e8 A |
33 | const char *packageName, |
34 | SRBRoot *newPoolBundle, UBool omitBinaryCollation, UErrorCode &status); | |
374ca955 | 35 | static char *make_res_filename(const char *filename, const char *outputDir, |
2ca993e8 | 36 | const char *packageName, UErrorCode &status); |
b75a7d8f A |
37 | |
38 | /* File suffixes */ | |
39 | #define RES_SUFFIX ".res" | |
40 | #define COL_SUFFIX ".col" | |
41 | ||
2ca993e8 | 42 | const char *gCurrentFileName = NULL; |
b75a7d8f A |
43 | #ifdef XP_MAC_CONSOLE |
44 | #include <console.h> | |
45 | #endif | |
46 | ||
2ca993e8 A |
47 | void ResFile::close() { |
48 | delete[] fBytes; | |
49 | fBytes = NULL; | |
50 | delete fStrings; | |
51 | fStrings = NULL; | |
52 | } | |
53 | ||
b75a7d8f A |
54 | enum |
55 | { | |
56 | HELP1, | |
57 | HELP2, | |
58 | VERBOSE, | |
59 | QUIET, | |
60 | VERSION, | |
61 | SOURCEDIR, | |
62 | DESTDIR, | |
63 | ENCODING, | |
64 | ICUDATADIR, | |
65 | WRITE_JAVA, | |
66 | COPYRIGHT, | |
51004dcb | 67 | JAVA_PACKAGE, |
b75a7d8f | 68 | BUNDLE_NAME, |
374ca955 | 69 | WRITE_XLIFF, |
b75a7d8f A |
70 | STRICT, |
71 | NO_BINARY_COLLATION, | |
46f4442e | 72 | LANGUAGE, |
729e4ab9 A |
73 | NO_COLLATION_RULES, |
74 | FORMAT_VERSION, | |
75 | WRITE_POOL_BUNDLE, | |
76 | USE_POOL_BUNDLE, | |
77 | INCLUDE_UNIHAN_COLL | |
b75a7d8f A |
78 | }; |
79 | ||
80 | UOption options[]={ | |
81 | UOPTION_HELP_H, | |
82 | UOPTION_HELP_QUESTION_MARK, | |
83 | UOPTION_VERBOSE, | |
84 | UOPTION_QUIET, | |
85 | UOPTION_VERSION, | |
86 | UOPTION_SOURCEDIR, | |
87 | UOPTION_DESTDIR, | |
88 | UOPTION_ENCODING, | |
89 | UOPTION_ICUDATADIR, | |
90 | UOPTION_WRITE_JAVA, | |
91 | UOPTION_COPYRIGHT, | |
51004dcb | 92 | UOPTION_DEF("java-package", '\x01', UOPT_REQUIRES_ARG), |
b75a7d8f | 93 | UOPTION_BUNDLE_NAME, |
729e4ab9 A |
94 | UOPTION_DEF("write-xliff", 'x', UOPT_OPTIONAL_ARG), |
95 | UOPTION_DEF("strict", 'k', UOPT_NO_ARG), /* 14 */ | |
96 | UOPTION_DEF("noBinaryCollation", 'C', UOPT_NO_ARG),/* 15 */ | |
97 | UOPTION_DEF("language", 'l', UOPT_REQUIRES_ARG), /* 16 */ | |
98 | UOPTION_DEF("omitCollationRules", 'R', UOPT_NO_ARG),/* 17 */ | |
99 | UOPTION_DEF("formatVersion", '\x01', UOPT_REQUIRES_ARG),/* 18 */ | |
100 | UOPTION_DEF("writePoolBundle", '\x01', UOPT_NO_ARG),/* 19 */ | |
101 | UOPTION_DEF("usePoolBundle", '\x01', UOPT_OPTIONAL_ARG),/* 20 */ | |
102 | UOPTION_DEF("includeUnihanColl", '\x01', UOPT_NO_ARG),/* 21 */ /* temporary, don't display in usage info */ | |
b75a7d8f A |
103 | }; |
104 | ||
105 | static UBool write_java = FALSE; | |
374ca955 | 106 | static UBool write_xliff = FALSE; |
b75a7d8f | 107 | static const char* outputEnc =""; |
729e4ab9 | 108 | |
2ca993e8 | 109 | static ResFile poolBundle; |
729e4ab9 | 110 | |
b75a7d8f A |
111 | /*added by Jing*/ |
112 | static const char* language = NULL; | |
374ca955 | 113 | static const char* xliffOutputFileName = NULL; |
b75a7d8f A |
114 | int |
115 | main(int argc, | |
116 | char* argv[]) | |
117 | { | |
118 | UErrorCode status = U_ZERO_ERROR; | |
119 | const char *arg = NULL; | |
120 | const char *outputDir = NULL; /* NULL = no output directory, use current */ | |
121 | const char *inputDir = NULL; | |
122 | const char *encoding = ""; | |
123 | int i; | |
51004dcb | 124 | UBool illegalArg = FALSE; |
374ca955 | 125 | |
b75a7d8f A |
126 | U_MAIN_INIT_ARGS(argc, argv); |
127 | ||
51004dcb A |
128 | options[JAVA_PACKAGE].value = "com.ibm.icu.impl.data"; |
129 | options[BUNDLE_NAME].value = "LocaleElements"; | |
2ca993e8 | 130 | argc = u_parseArgs(argc, argv, UPRV_LENGTHOF(options), options); |
b75a7d8f A |
131 | |
132 | /* error handling, printing usage message */ | |
133 | if(argc<0) { | |
134 | fprintf(stderr, "%s: error in command line argument \"%s\"\n", argv[0], argv[-argc]); | |
2ca993e8 | 135 | illegalArg = TRUE; |
b75a7d8f | 136 | } else if(argc<2) { |
2ca993e8 | 137 | illegalArg = TRUE; |
b75a7d8f | 138 | } |
729e4ab9 A |
139 | if(options[WRITE_POOL_BUNDLE].doesOccur && options[USE_POOL_BUNDLE].doesOccur) { |
140 | fprintf(stderr, "%s: cannot combine --writePoolBundle and --usePoolBundle\n", argv[0]); | |
2ca993e8 | 141 | illegalArg = TRUE; |
729e4ab9 A |
142 | } |
143 | if(options[FORMAT_VERSION].doesOccur) { | |
144 | const char *s = options[FORMAT_VERSION].value; | |
2ca993e8 | 145 | if(uprv_strlen(s) != 1 || (s[0] < '1' && '3' < s[0])) { |
729e4ab9 | 146 | fprintf(stderr, "%s: unsupported --formatVersion %s\n", argv[0], s); |
2ca993e8 | 147 | illegalArg = TRUE; |
729e4ab9 A |
148 | } else if(s[0] == '1' && |
149 | (options[WRITE_POOL_BUNDLE].doesOccur || options[USE_POOL_BUNDLE].doesOccur) | |
150 | ) { | |
151 | fprintf(stderr, "%s: cannot combine --formatVersion 1 with --writePoolBundle or --usePoolBundle\n", argv[0]); | |
2ca993e8 | 152 | illegalArg = TRUE; |
729e4ab9 A |
153 | } else { |
154 | setFormatVersion(s[0] - '0'); | |
155 | } | |
156 | } | |
b75a7d8f | 157 | |
2ca993e8 A |
158 | if((options[JAVA_PACKAGE].doesOccur || options[BUNDLE_NAME].doesOccur) && |
159 | !options[WRITE_JAVA].doesOccur) { | |
51004dcb A |
160 | fprintf(stderr, |
161 | "%s error: command line argument --java-package or --bundle-name " | |
162 | "without --write-java\n", | |
163 | argv[0]); | |
164 | illegalArg = TRUE; | |
165 | } | |
166 | ||
2ca993e8 A |
167 | if(options[VERSION].doesOccur) { |
168 | fprintf(stderr, | |
169 | "%s version %s (ICU version %s).\n" | |
170 | "%s\n", | |
171 | argv[0], GENRB_VERSION, U_ICU_VERSION, U_COPYRIGHT_STRING); | |
172 | if(!illegalArg) { | |
173 | return U_ZERO_ERROR; | |
174 | } | |
175 | } | |
176 | ||
51004dcb | 177 | if(illegalArg || options[HELP1].doesOccur || options[HELP2].doesOccur) { |
b75a7d8f | 178 | /* |
729e4ab9 | 179 | * Broken into chunks because the C89 standard says the minimum |
b75a7d8f A |
180 | * required supported string length is 509 bytes. |
181 | */ | |
182 | fprintf(stderr, | |
183 | "Usage: %s [OPTIONS] [FILES]\n" | |
184 | "\tReads the list of resource bundle source files and creates\n" | |
4388f060 | 185 | "\tbinary version of resource bundles (.res files)\n", |
b75a7d8f A |
186 | argv[0]); |
187 | fprintf(stderr, | |
188 | "Options:\n" | |
189 | "\t-h or -? or --help this usage text\n" | |
190 | "\t-q or --quiet do not display warnings\n" | |
191 | "\t-v or --verbose print extra information when processing files\n" | |
192 | "\t-V or --version prints out version number and exits\n" | |
193 | "\t-c or --copyright include copyright notice\n"); | |
194 | fprintf(stderr, | |
195 | "\t-e or --encoding encoding of source files\n" | |
196 | "\t-d of --destdir destination directory, followed by the path, defaults to %s\n" | |
197 | "\t-s or --sourcedir source directory for files followed by path, defaults to %s\n" | |
198 | "\t-i or --icudatadir directory for locating any needed intermediate data files,\n" | |
199 | "\t followed by path, defaults to %s\n", | |
200 | u_getDataDirectory(), u_getDataDirectory(), u_getDataDirectory()); | |
201 | fprintf(stderr, | |
202 | "\t-j or --write-java write a Java ListResourceBundle for ICU4J, followed by optional encoding\n" | |
51004dcb A |
203 | "\t defaults to ASCII and \\uXXXX format.\n" |
204 | "\t --java-package For --write-java: package name for writing the ListResourceBundle,\n" | |
205 | "\t defaults to com.ibm.icu.impl.data\n"); | |
b75a7d8f | 206 | fprintf(stderr, |
51004dcb | 207 | "\t-b or --bundle-name For --write-java: root resource bundle name for writing the ListResourceBundle,\n" |
b75a7d8f | 208 | "\t defaults to LocaleElements\n" |
46f4442e A |
209 | "\t-x or --write-xliff write an XLIFF file for the resource bundle. Followed by\n" |
210 | "\t an optional output file name.\n" | |
b75a7d8f A |
211 | "\t-k or --strict use pedantic parsing of syntax\n" |
212 | /*added by Jing*/ | |
46f4442e A |
213 | "\t-l or --language for XLIFF: language code compliant with BCP 47.\n"); |
214 | fprintf(stderr, | |
215 | "\t-C or --noBinaryCollation do not generate binary collation image;\n" | |
216 | "\t makes .res file smaller but collator instantiation much slower;\n" | |
217 | "\t maintains ability to get tailoring rules\n" | |
218 | "\t-R or --omitCollationRules do not include collation (tailoring) rules;\n" | |
219 | "\t makes .res file smaller and maintains collator instantiation speed\n" | |
220 | "\t but tailoring rules will not be available (they are rarely used)\n"); | |
729e4ab9 A |
221 | fprintf(stderr, |
222 | "\t --formatVersion write a .res file compatible with the requested formatVersion (single digit);\n" | |
223 | "\t for example, --formatVersion 1\n"); | |
224 | fprintf(stderr, | |
225 | "\t --writePoolBundle write a pool.res file with all of the keys of all input bundles\n" | |
226 | "\t --usePoolBundle [path-to-pool.res] point to keys from the pool.res keys pool bundle if they are available there;\n" | |
227 | "\t makes .res files smaller but dependent on the pool bundle\n" | |
228 | "\t (--writePoolBundle and --usePoolBundle cannot be combined)\n"); | |
b75a7d8f | 229 | |
51004dcb | 230 | return illegalArg ? U_ILLEGAL_ARGUMENT_ERROR : U_ZERO_ERROR; |
b75a7d8f A |
231 | } |
232 | ||
233 | if(options[VERBOSE].doesOccur) { | |
234 | setVerbose(TRUE); | |
235 | } | |
236 | ||
237 | if(options[QUIET].doesOccur) { | |
238 | setShowWarning(FALSE); | |
239 | } | |
240 | if(options[STRICT].doesOccur) { | |
241 | setStrict(TRUE); | |
242 | } | |
243 | if(options[COPYRIGHT].doesOccur){ | |
244 | setIncludeCopyright(TRUE); | |
245 | } | |
246 | ||
247 | if(options[SOURCEDIR].doesOccur) { | |
248 | inputDir = options[SOURCEDIR].value; | |
249 | } | |
250 | ||
251 | if(options[DESTDIR].doesOccur) { | |
252 | outputDir = options[DESTDIR].value; | |
253 | } | |
b75a7d8f | 254 | |
b75a7d8f A |
255 | if(options[ENCODING].doesOccur) { |
256 | encoding = options[ENCODING].value; | |
257 | } | |
258 | ||
259 | if(options[ICUDATADIR].doesOccur) { | |
260 | u_setDataDirectory(options[ICUDATADIR].value); | |
261 | } | |
374ca955 A |
262 | /* Initialize ICU */ |
263 | u_init(&status); | |
264 | if (U_FAILURE(status) && status != U_FILE_ACCESS_ERROR) { | |
265 | /* Note: u_init() will try to open ICU property data. | |
266 | * failures here are expected when building ICU from scratch. | |
267 | * ignore them. | |
268 | */ | |
269 | fprintf(stderr, "%s: can not initialize ICU. status = %s\n", | |
270 | argv[0], u_errorName(status)); | |
271 | exit(1); | |
272 | } | |
273 | status = U_ZERO_ERROR; | |
b75a7d8f A |
274 | if(options[WRITE_JAVA].doesOccur) { |
275 | write_java = TRUE; | |
276 | outputEnc = options[WRITE_JAVA].value; | |
277 | } | |
278 | ||
374ca955 A |
279 | if(options[WRITE_XLIFF].doesOccur) { |
280 | write_xliff = TRUE; | |
281 | if(options[WRITE_XLIFF].value != NULL){ | |
282 | xliffOutputFileName = options[WRITE_XLIFF].value; | |
283 | } | |
b75a7d8f A |
284 | } |
285 | ||
57a6839d | 286 | initParser(); |
374ca955 | 287 | |
b75a7d8f A |
288 | /*added by Jing*/ |
289 | if(options[LANGUAGE].doesOccur) { | |
290 | language = options[LANGUAGE].value; | |
291 | } | |
292 | ||
2ca993e8 | 293 | LocalPointer<SRBRoot> newPoolBundle; |
729e4ab9 | 294 | if(options[WRITE_POOL_BUNDLE].doesOccur) { |
2ca993e8 | 295 | newPoolBundle.adoptInsteadAndCheckErrorCode(new SRBRoot(NULL, TRUE, status), status); |
729e4ab9 A |
296 | if(U_FAILURE(status)) { |
297 | fprintf(stderr, "unable to create an empty bundle for the pool keys: %s\n", u_errorName(status)); | |
298 | return status; | |
299 | } else { | |
300 | const char *poolResName = "pool.res"; | |
2ca993e8 | 301 | char *nameWithoutSuffix = static_cast<char *>(uprv_malloc(uprv_strlen(poolResName) + 1)); |
729e4ab9 A |
302 | if (nameWithoutSuffix == NULL) { |
303 | fprintf(stderr, "out of memory error\n"); | |
304 | return U_MEMORY_ALLOCATION_ERROR; | |
305 | } | |
306 | uprv_strcpy(nameWithoutSuffix, poolResName); | |
307 | *uprv_strrchr(nameWithoutSuffix, '.') = 0; | |
308 | newPoolBundle->fLocale = nameWithoutSuffix; | |
309 | } | |
310 | } | |
311 | ||
312 | if(options[USE_POOL_BUNDLE].doesOccur) { | |
313 | const char *poolResName = "pool.res"; | |
314 | FileStream *poolFile; | |
315 | int32_t poolFileSize; | |
316 | int32_t indexLength; | |
317 | /* | |
318 | * TODO: Consolidate inputDir/filename handling from main() and processFile() | |
319 | * into a common function, and use it here as well. | |
320 | * Try to create toolutil functions for dealing with dir/filenames and | |
321 | * loading ICU data files without udata_open(). | |
322 | * Share code with icupkg? | |
323 | * Also, make_res_filename() seems to be unused. Review and remove. | |
324 | */ | |
2ca993e8 | 325 | CharString poolFileName; |
729e4ab9 | 326 | if (options[USE_POOL_BUNDLE].value!=NULL) { |
2ca993e8 | 327 | poolFileName.append(options[USE_POOL_BUNDLE].value, status); |
729e4ab9 | 328 | } else if (inputDir) { |
2ca993e8 A |
329 | poolFileName.append(inputDir, status); |
330 | } | |
331 | poolFileName.appendPathPart(poolResName, status); | |
332 | if (U_FAILURE(status)) { | |
333 | return status; | |
729e4ab9 | 334 | } |
2ca993e8 | 335 | poolFile = T_FileStream_open(poolFileName.data(), "rb"); |
729e4ab9 | 336 | if (poolFile == NULL) { |
2ca993e8 | 337 | fprintf(stderr, "unable to open pool bundle file %s\n", poolFileName.data()); |
729e4ab9 A |
338 | return 1; |
339 | } | |
340 | poolFileSize = T_FileStream_size(poolFile); | |
341 | if (poolFileSize < 32) { | |
2ca993e8 | 342 | fprintf(stderr, "the pool bundle file %s is too small\n", poolFileName.data()); |
729e4ab9 A |
343 | return 1; |
344 | } | |
2ca993e8 | 345 | poolBundle.fBytes = new uint8_t[(poolFileSize + 15) & ~15]; |
729e4ab9 | 346 | if (poolFileSize > 0 && poolBundle.fBytes == NULL) { |
2ca993e8 | 347 | fprintf(stderr, "unable to allocate memory for the pool bundle file %s\n", poolFileName.data()); |
729e4ab9 | 348 | return U_MEMORY_ALLOCATION_ERROR; |
729e4ab9 | 349 | } |
2ca993e8 A |
350 | |
351 | UDataSwapper *ds; | |
352 | const DataHeader *header; | |
353 | int32_t bytesRead = T_FileStream_read(poolFile, poolBundle.fBytes, poolFileSize); | |
354 | if (bytesRead != poolFileSize) { | |
355 | fprintf(stderr, "unable to read the pool bundle file %s\n", poolFileName.data()); | |
356 | return 1; | |
357 | } | |
358 | /* | |
359 | * Swap the pool bundle so that a single checked-in file can be used. | |
360 | * The swapper functions also test that the data looks like | |
361 | * a well-formed .res file. | |
362 | */ | |
363 | ds = udata_openSwapperForInputData(poolBundle.fBytes, bytesRead, | |
364 | U_IS_BIG_ENDIAN, U_CHARSET_FAMILY, &status); | |
365 | if (U_FAILURE(status)) { | |
366 | fprintf(stderr, "udata_openSwapperForInputData(pool bundle %s) failed: %s\n", | |
367 | poolFileName.data(), u_errorName(status)); | |
368 | return status; | |
369 | } | |
370 | ures_swap(ds, poolBundle.fBytes, bytesRead, poolBundle.fBytes, &status); | |
371 | udata_closeSwapper(ds); | |
372 | if (U_FAILURE(status)) { | |
373 | fprintf(stderr, "ures_swap(pool bundle %s) failed: %s\n", | |
374 | poolFileName.data(), u_errorName(status)); | |
375 | return status; | |
376 | } | |
377 | header = (const DataHeader *)poolBundle.fBytes; | |
378 | if (header->info.formatVersion[0] < 2) { | |
379 | fprintf(stderr, "invalid format of pool bundle file %s\n", poolFileName.data()); | |
380 | return U_INVALID_FORMAT_ERROR; | |
381 | } | |
382 | const int32_t *pRoot = (const int32_t *)( | |
383 | (const char *)header + header->dataHeader.headerSize); | |
384 | poolBundle.fIndexes = pRoot + 1; | |
385 | indexLength = poolBundle.fIndexes[URES_INDEX_LENGTH] & 0xff; | |
386 | if (indexLength <= URES_INDEX_POOL_CHECKSUM) { | |
387 | fprintf(stderr, "insufficient indexes[] in pool bundle file %s\n", poolFileName.data()); | |
388 | return U_INVALID_FORMAT_ERROR; | |
389 | } | |
390 | int32_t keysBottom = 1 + indexLength; | |
391 | int32_t keysTop = poolBundle.fIndexes[URES_INDEX_KEYS_TOP]; | |
392 | poolBundle.fKeys = (const char *)(pRoot + keysBottom); | |
393 | poolBundle.fKeysLength = (keysTop - keysBottom) * 4; | |
394 | poolBundle.fChecksum = poolBundle.fIndexes[URES_INDEX_POOL_CHECKSUM]; | |
395 | ||
729e4ab9 A |
396 | for (i = 0; i < poolBundle.fKeysLength; ++i) { |
397 | if (poolBundle.fKeys[i] == 0) { | |
398 | ++poolBundle.fKeysCount; | |
399 | } | |
400 | } | |
2ca993e8 A |
401 | |
402 | // 16BitUnits[] begins with strings-v2. | |
403 | // The strings-v2 may optionally be terminated by what looks like | |
404 | // an explicit string length that exceeds the number of remaining 16-bit units. | |
405 | int32_t stringUnitsLength = (poolBundle.fIndexes[URES_INDEX_16BIT_TOP] - keysTop) * 2; | |
406 | if (stringUnitsLength >= 2 && getFormatVersion() >= 3) { | |
407 | poolBundle.fStrings = new PseudoListResource(NULL, status); | |
408 | if (poolBundle.fStrings == NULL) { | |
409 | fprintf(stderr, "unable to allocate memory for the pool bundle strings %s\n", | |
410 | poolFileName.data()); | |
411 | return U_MEMORY_ALLOCATION_ERROR; | |
412 | } | |
413 | // The PseudoListResource constructor call did not allocate further memory. | |
414 | assert(U_SUCCESS(status)); | |
415 | const UChar *p = (const UChar *)(pRoot + keysTop); | |
416 | int32_t remaining = stringUnitsLength; | |
417 | do { | |
418 | int32_t first = *p; | |
419 | int8_t numCharsForLength; | |
420 | int32_t length; | |
421 | if (!U16_IS_TRAIL(first)) { | |
422 | // NUL-terminated | |
423 | numCharsForLength = 0; | |
424 | for (length = 0; | |
425 | length < remaining && p[length] != 0; | |
426 | ++length) {} | |
427 | } else if (first < 0xdfef) { | |
428 | numCharsForLength = 1; | |
429 | length = first & 0x3ff; | |
430 | } else if (first < 0xdfff && remaining >= 2) { | |
431 | numCharsForLength = 2; | |
432 | length = ((first - 0xdfef) << 16) | p[1]; | |
433 | } else if (first == 0xdfff && remaining >= 3) { | |
434 | numCharsForLength = 3; | |
435 | length = ((int32_t)p[1] << 16) | p[2]; | |
436 | } else { | |
437 | break; // overrun | |
438 | } | |
439 | // Check for overrun before changing remaining, | |
440 | // so that it is always accurate after the loop body. | |
441 | if ((numCharsForLength + length) >= remaining || | |
442 | p[numCharsForLength + length] != 0) { | |
443 | break; // overrun or explicitly terminated | |
444 | } | |
445 | int32_t poolStringIndex = stringUnitsLength - remaining; | |
446 | // Maximum pool string index when suffix-sharing the last character. | |
447 | int32_t maxStringIndex = poolStringIndex + numCharsForLength + length - 1; | |
448 | if (maxStringIndex >= RES_MAX_OFFSET) { | |
449 | // pool string index overrun | |
450 | break; | |
451 | } | |
452 | p += numCharsForLength; | |
453 | remaining -= numCharsForLength; | |
454 | if (length != 0) { | |
455 | StringResource *sr = | |
456 | new StringResource(poolStringIndex, numCharsForLength, | |
457 | p, length, status); | |
458 | if (sr == NULL) { | |
459 | fprintf(stderr, "unable to allocate memory for a pool bundle string %s\n", | |
460 | poolFileName.data()); | |
461 | return U_MEMORY_ALLOCATION_ERROR; | |
462 | } | |
463 | poolBundle.fStrings->add(sr); | |
464 | poolBundle.fStringIndexLimit = maxStringIndex + 1; | |
465 | // The StringResource constructor did not allocate further memory. | |
466 | assert(U_SUCCESS(status)); | |
467 | } | |
468 | p += length + 1; | |
469 | remaining -= length + 1; | |
470 | } while (remaining > 0); | |
471 | if (poolBundle.fStrings->fCount == 0) { | |
472 | delete poolBundle.fStrings; | |
473 | poolBundle.fStrings = NULL; | |
474 | } | |
475 | } | |
476 | ||
729e4ab9 A |
477 | T_FileStream_close(poolFile); |
478 | setUsePoolBundle(TRUE); | |
2ca993e8 A |
479 | if (isVerbose() && poolBundle.fStrings != NULL) { |
480 | printf("number of shared strings: %d\n", (int)poolBundle.fStrings->fCount); | |
481 | int32_t length = poolBundle.fStringIndexLimit + 1; // incl. last NUL | |
482 | printf("16-bit units for strings: %6d = %6d bytes\n", | |
483 | (int)length, (int)length * 2); | |
484 | } | |
485 | } | |
486 | ||
487 | if(!options[FORMAT_VERSION].doesOccur && getFormatVersion() == 3 && | |
488 | poolBundle.fStrings == NULL && | |
489 | !options[WRITE_POOL_BUNDLE].doesOccur) { | |
490 | // If we just default to formatVersion 3 | |
491 | // but there are no pool bundle strings to share | |
492 | // and we do not write a pool bundle, | |
493 | // then write formatVersion 2 which is just as good. | |
494 | setFormatVersion(2); | |
729e4ab9 A |
495 | } |
496 | ||
497 | if(options[INCLUDE_UNIHAN_COLL].doesOccur) { | |
b331163b A |
498 | puts("genrb option --includeUnihanColl ignored: \n" |
499 | "CLDR 26/ICU 54 unihan data is small, except\n" | |
500 | "the ucadata-unihan.icu version of the collation root data\n" | |
501 | "is about 300kB larger than the ucadata-implicithan.icu version."); | |
729e4ab9 A |
502 | } |
503 | ||
504 | if((argc-1)!=1) { | |
505 | printf("genrb number of files: %d\n", argc - 1); | |
506 | } | |
b75a7d8f A |
507 | /* generate the binary files */ |
508 | for(i = 1; i < argc; ++i) { | |
509 | status = U_ZERO_ERROR; | |
510 | arg = getLongPathname(argv[i]); | |
374ca955 | 511 | |
2ca993e8 | 512 | CharString theCurrentFileName; |
b75a7d8f | 513 | if (inputDir) { |
2ca993e8 A |
514 | theCurrentFileName.append(inputDir, status); |
515 | } | |
516 | theCurrentFileName.appendPathPart(arg, status); | |
517 | if (U_FAILURE(status)) { | |
518 | break; | |
b75a7d8f | 519 | } |
b75a7d8f | 520 | |
2ca993e8 | 521 | gCurrentFileName = theCurrentFileName.data(); |
b75a7d8f | 522 | if (isVerbose()) { |
2ca993e8 | 523 | printf("Processing file \"%s\"\n", theCurrentFileName.data()); |
b75a7d8f | 524 | } |
51004dcb | 525 | processFile(arg, encoding, inputDir, outputDir, NULL, |
2ca993e8 A |
526 | newPoolBundle.getAlias(), |
527 | options[NO_BINARY_COLLATION].doesOccur, status); | |
b75a7d8f A |
528 | } |
529 | ||
2ca993e8 | 530 | poolBundle.close(); |
729e4ab9 | 531 | |
2ca993e8 | 532 | if(U_SUCCESS(status) && options[WRITE_POOL_BUNDLE].doesOccur) { |
729e4ab9 | 533 | char outputFileName[256]; |
2ca993e8 | 534 | newPoolBundle->write(outputDir, NULL, outputFileName, sizeof(outputFileName), status); |
729e4ab9 A |
535 | if(U_FAILURE(status)) { |
536 | fprintf(stderr, "unable to write the pool bundle: %s\n", u_errorName(status)); | |
537 | } | |
538 | } | |
539 | ||
51004dcb A |
540 | u_cleanup(); |
541 | ||
46f4442e | 542 | /* Dont return warnings as a failure */ |
729e4ab9 | 543 | if (U_SUCCESS(status)) { |
46f4442e A |
544 | return 0; |
545 | } | |
546 | ||
b75a7d8f A |
547 | return status; |
548 | } | |
549 | ||
550 | /* Process a file */ | |
729e4ab9 | 551 | void |
2ca993e8 A |
552 | processFile(const char *filename, const char *cp, |
553 | const char *inputDir, const char *outputDir, const char *packageName, | |
554 | SRBRoot *newPoolBundle, | |
555 | UBool omitBinaryCollation, UErrorCode &status) { | |
556 | LocalPointer<SRBRoot> data; | |
b75a7d8f A |
557 | UCHARBUF *ucbuf = NULL; |
558 | char *rbname = NULL; | |
559 | char *openFileName = NULL; | |
560 | char *inputDirBuf = NULL; | |
561 | ||
562 | char outputFileName[256]; | |
374ca955 | 563 | |
b75a7d8f A |
564 | int32_t dirlen = 0; |
565 | int32_t filelen = 0; | |
374ca955 | 566 | |
2ca993e8 | 567 | if (U_FAILURE(status)) { |
b75a7d8f A |
568 | return; |
569 | } | |
570 | if(filename==NULL){ | |
2ca993e8 | 571 | status=U_ILLEGAL_ARGUMENT_ERROR; |
b75a7d8f A |
572 | return; |
573 | }else{ | |
574 | filelen = (int32_t)uprv_strlen(filename); | |
575 | } | |
4388f060 | 576 | |
b75a7d8f A |
577 | if(inputDir == NULL) { |
578 | const char *filenameBegin = uprv_strrchr(filename, U_FILE_SEP_CHAR); | |
579 | openFileName = (char *) uprv_malloc(dirlen + filelen + 2); | |
580 | openFileName[0] = '\0'; | |
581 | if (filenameBegin != NULL) { | |
582 | /* | |
374ca955 | 583 | * When a filename ../../../data/root.txt is specified, |
b75a7d8f A |
584 | * we presume that the input directory is ../../../data |
585 | * This is very important when the resource file includes | |
586 | * another file, like UCARules.txt or thaidict.brk. | |
587 | */ | |
374ca955 | 588 | int32_t filenameSize = (int32_t)(filenameBegin - filename + 1); |
b75a7d8f A |
589 | inputDirBuf = uprv_strncpy((char *)uprv_malloc(filenameSize), filename, filenameSize); |
590 | ||
591 | /* test for NULL */ | |
592 | if(inputDirBuf == NULL) { | |
2ca993e8 | 593 | status = U_MEMORY_ALLOCATION_ERROR; |
b75a7d8f A |
594 | goto finish; |
595 | } | |
596 | ||
597 | inputDirBuf[filenameSize - 1] = 0; | |
598 | inputDir = inputDirBuf; | |
599 | dirlen = (int32_t)uprv_strlen(inputDir); | |
600 | } | |
601 | }else{ | |
602 | dirlen = (int32_t)uprv_strlen(inputDir); | |
603 | ||
604 | if(inputDir[dirlen-1] != U_FILE_SEP_CHAR) { | |
605 | openFileName = (char *) uprv_malloc(dirlen + filelen + 2); | |
606 | ||
607 | /* test for NULL */ | |
608 | if(openFileName == NULL) { | |
2ca993e8 | 609 | status = U_MEMORY_ALLOCATION_ERROR; |
b75a7d8f A |
610 | goto finish; |
611 | } | |
612 | ||
613 | openFileName[0] = '\0'; | |
614 | /* | |
374ca955 | 615 | * append the input dir to openFileName if the first char in |
b75a7d8f A |
616 | * filename is not file seperation char and the last char input directory is not '.'. |
617 | * This is to support : | |
618 | * genrb -s. /home/icu/data | |
619 | * genrb -s. icu/data | |
620 | * The user cannot mix notations like | |
621 | * genrb -s. /icu/data --- the absolute path specified. -s redundant | |
622 | * user should use | |
623 | * genrb -s. icu/data --- start from CWD and look in icu/data dir | |
624 | */ | |
625 | if( (filename[0] != U_FILE_SEP_CHAR) && (inputDir[dirlen-1] !='.')){ | |
626 | uprv_strcpy(openFileName, inputDir); | |
627 | openFileName[dirlen] = U_FILE_SEP_CHAR; | |
628 | } | |
629 | openFileName[dirlen + 1] = '\0'; | |
630 | } else { | |
631 | openFileName = (char *) uprv_malloc(dirlen + filelen + 1); | |
632 | ||
633 | /* test for NULL */ | |
634 | if(openFileName == NULL) { | |
2ca993e8 | 635 | status = U_MEMORY_ALLOCATION_ERROR; |
b75a7d8f A |
636 | goto finish; |
637 | } | |
638 | ||
639 | uprv_strcpy(openFileName, inputDir); | |
374ca955 | 640 | |
b75a7d8f A |
641 | } |
642 | } | |
643 | ||
644 | uprv_strcat(openFileName, filename); | |
645 | ||
2ca993e8 A |
646 | ucbuf = ucbuf_open(openFileName, &cp,getShowWarning(),TRUE, &status); |
647 | if(status == U_FILE_ACCESS_ERROR) { | |
374ca955 | 648 | |
b75a7d8f A |
649 | fprintf(stderr, "couldn't open file %s\n", openFileName == NULL ? filename : openFileName); |
650 | goto finish; | |
651 | } | |
2ca993e8 A |
652 | if (ucbuf == NULL || U_FAILURE(status)) { |
653 | fprintf(stderr, "An error occured processing file %s. Error: %s\n", | |
654 | openFileName == NULL ? filename : openFileName, u_errorName(status)); | |
b75a7d8f A |
655 | goto finish; |
656 | } | |
657 | /* auto detected popular encodings? */ | |
374ca955 | 658 | if (cp!=NULL && isVerbose()) { |
b75a7d8f A |
659 | printf("autodetected encoding %s\n", cp); |
660 | } | |
661 | /* Parse the data into an SRBRoot */ | |
2ca993e8 A |
662 | data.adoptInstead(parse(ucbuf, inputDir, outputDir, filename, |
663 | !omitBinaryCollation, options[NO_COLLATION_RULES].doesOccur, &status)); | |
b75a7d8f | 664 | |
2ca993e8 A |
665 | if (data.isNull() || U_FAILURE(status)) { |
666 | fprintf(stderr, "couldn't parse the file %s. Error:%s\n", filename, u_errorName(status)); | |
b75a7d8f A |
667 | goto finish; |
668 | } | |
729e4ab9 | 669 | if(options[WRITE_POOL_BUNDLE].doesOccur) { |
2ca993e8 A |
670 | data->fWritePoolBundle = newPoolBundle; |
671 | data->compactKeys(status); | |
729e4ab9 | 672 | int32_t newKeysLength; |
2ca993e8 A |
673 | const char *newKeys = data->getKeyBytes(&newKeysLength); |
674 | newPoolBundle->addKeyBytes(newKeys, newKeysLength, status); | |
675 | if(U_FAILURE(status)) { | |
729e4ab9 | 676 | fprintf(stderr, "bundle_compactKeys(%s) or bundle_getKeyBytes() failed: %s\n", |
2ca993e8 | 677 | filename, u_errorName(status)); |
729e4ab9 A |
678 | goto finish; |
679 | } | |
680 | /* count the number of just-added key strings */ | |
2ca993e8 | 681 | for(const char *newKeysLimit = newKeys + newKeysLength; newKeys < newKeysLimit; ++newKeys) { |
729e4ab9 A |
682 | if(*newKeys == 0) { |
683 | ++newPoolBundle->fKeysCount; | |
684 | } | |
685 | } | |
686 | } | |
687 | ||
688 | if(options[USE_POOL_BUNDLE].doesOccur) { | |
2ca993e8 | 689 | data->fUsePoolBundle = &poolBundle; |
729e4ab9 | 690 | } |
b75a7d8f A |
691 | |
692 | /* Determine the target rb filename */ | |
693 | rbname = make_res_filename(filename, outputDir, packageName, status); | |
2ca993e8 A |
694 | if(U_FAILURE(status)) { |
695 | fprintf(stderr, "couldn't make the res fileName for bundle %s. Error:%s\n", | |
696 | filename, u_errorName(status)); | |
b75a7d8f A |
697 | goto finish; |
698 | } | |
699 | if(write_java== TRUE){ | |
2ca993e8 A |
700 | bundle_write_java(data.getAlias(), outputDir, outputEnc, |
701 | outputFileName, sizeof(outputFileName), | |
702 | options[JAVA_PACKAGE].value, options[BUNDLE_NAME].value, &status); | |
374ca955 | 703 | }else if(write_xliff ==TRUE){ |
2ca993e8 A |
704 | bundle_write_xml(data.getAlias(), outputDir, outputEnc, |
705 | filename, outputFileName, sizeof(outputFileName), | |
706 | language, xliffOutputFileName, &status); | |
b75a7d8f A |
707 | }else{ |
708 | /* Write the data to the file */ | |
2ca993e8 | 709 | data->write(outputDir, packageName, outputFileName, sizeof(outputFileName), status); |
b75a7d8f | 710 | } |
2ca993e8 A |
711 | if (U_FAILURE(status)) { |
712 | fprintf(stderr, "couldn't write bundle %s. Error:%s\n", outputFileName, u_errorName(status)); | |
b75a7d8f | 713 | } |
b75a7d8f A |
714 | |
715 | finish: | |
716 | ||
717 | if (inputDirBuf != NULL) { | |
718 | uprv_free(inputDirBuf); | |
719 | } | |
720 | ||
721 | if (openFileName != NULL) { | |
722 | uprv_free(openFileName); | |
723 | } | |
724 | ||
725 | if(ucbuf) { | |
726 | ucbuf_close(ucbuf); | |
727 | } | |
728 | ||
729 | if (rbname) { | |
730 | uprv_free(rbname); | |
731 | } | |
732 | } | |
733 | ||
734 | /* Generate the target .res file name from the input file name */ | |
735 | static char* | |
736 | make_res_filename(const char *filename, | |
737 | const char *outputDir, | |
738 | const char *packageName, | |
2ca993e8 | 739 | UErrorCode &status) { |
b75a7d8f A |
740 | char *basename; |
741 | char *dirname; | |
742 | char *resName; | |
743 | ||
744 | int32_t pkgLen = 0; /* length of package prefix */ | |
745 | ||
4388f060 | 746 | |
2ca993e8 | 747 | if (U_FAILURE(status)) { |
b75a7d8f A |
748 | return 0; |
749 | } | |
750 | ||
751 | if(packageName != NULL) | |
752 | { | |
374ca955 | 753 | pkgLen = (int32_t)(1 + uprv_strlen(packageName)); |
b75a7d8f A |
754 | } |
755 | ||
756 | /* setup */ | |
757 | basename = dirname = resName = 0; | |
758 | ||
759 | /* determine basename, and compiled file names */ | |
760 | basename = (char*) uprv_malloc(sizeof(char) * (uprv_strlen(filename) + 1)); | |
761 | if(basename == 0) { | |
2ca993e8 | 762 | status = U_MEMORY_ALLOCATION_ERROR; |
b75a7d8f A |
763 | goto finish; |
764 | } | |
765 | ||
766 | get_basename(basename, filename); | |
767 | ||
768 | dirname = (char*) uprv_malloc(sizeof(char) * (uprv_strlen(filename) + 1)); | |
769 | if(dirname == 0) { | |
2ca993e8 | 770 | status = U_MEMORY_ALLOCATION_ERROR; |
b75a7d8f A |
771 | goto finish; |
772 | } | |
773 | ||
774 | get_dirname(dirname, filename); | |
775 | ||
776 | if (outputDir == NULL) { | |
777 | /* output in same dir as .txt */ | |
778 | resName = (char*) uprv_malloc(sizeof(char) * (uprv_strlen(dirname) | |
779 | + pkgLen | |
780 | + uprv_strlen(basename) | |
781 | + uprv_strlen(RES_SUFFIX) + 8)); | |
782 | if(resName == 0) { | |
2ca993e8 | 783 | status = U_MEMORY_ALLOCATION_ERROR; |
b75a7d8f A |
784 | goto finish; |
785 | } | |
786 | ||
787 | uprv_strcpy(resName, dirname); | |
788 | ||
789 | if(packageName != NULL) | |
790 | { | |
791 | uprv_strcat(resName, packageName); | |
792 | uprv_strcat(resName, "_"); | |
793 | } | |
794 | ||
795 | uprv_strcat(resName, basename); | |
796 | ||
797 | } else { | |
798 | int32_t dirlen = (int32_t)uprv_strlen(outputDir); | |
799 | int32_t basenamelen = (int32_t)uprv_strlen(basename); | |
800 | ||
801 | resName = (char*) uprv_malloc(sizeof(char) * (dirlen + pkgLen + basenamelen + 8)); | |
802 | ||
803 | if (resName == NULL) { | |
2ca993e8 | 804 | status = U_MEMORY_ALLOCATION_ERROR; |
b75a7d8f A |
805 | goto finish; |
806 | } | |
807 | ||
808 | uprv_strcpy(resName, outputDir); | |
809 | ||
810 | if(outputDir[dirlen] != U_FILE_SEP_CHAR) { | |
811 | resName[dirlen] = U_FILE_SEP_CHAR; | |
812 | resName[dirlen + 1] = '\0'; | |
813 | } | |
814 | ||
815 | if(packageName != NULL) | |
816 | { | |
817 | uprv_strcat(resName, packageName); | |
818 | uprv_strcat(resName, "_"); | |
819 | } | |
820 | ||
821 | uprv_strcat(resName, basename); | |
822 | } | |
823 | ||
824 | finish: | |
825 | uprv_free(basename); | |
826 | uprv_free(dirname); | |
827 | ||
828 | return resName; | |
829 | } | |
830 | ||
831 | /* | |
832 | * Local Variables: | |
833 | * indent-tabs-mode: nil | |
834 | * End: | |
835 | */ |