]>
Commit | Line | Data |
---|---|---|
b75a7d8f A |
1 | /* |
2 | ******************************************************************************* | |
3 | * | |
4388f060 | 4 | * Copyright (C) 1998-2011, International Business Machines |
b75a7d8f A |
5 | * Corporation and others. All Rights Reserved. |
6 | * | |
7 | ******************************************************************************* | |
8 | * | |
9 | * File ucbuf.c | |
10 | * | |
11 | * Modification History: | |
12 | * | |
13 | * Date Name Description | |
14 | * 05/10/01 Ram Creation. | |
15 | ******************************************************************************* | |
16 | */ | |
17 | ||
18 | #include "unicode/utypes.h" | |
374ca955 | 19 | #include "unicode/putil.h" |
4388f060 | 20 | #include "unicode/uchar.h" |
b75a7d8f A |
21 | #include "unicode/ucnv.h" |
22 | #include "unicode/ucnv_err.h" | |
4388f060 A |
23 | #include "unicode/ustring.h" |
24 | #include "unicode/utf16.h" | |
b75a7d8f A |
25 | #include "filestrm.h" |
26 | #include "cstring.h" | |
27 | #include "cmemory.h" | |
28 | #include "ustrfmt.h" | |
b75a7d8f A |
29 | #include "ucbuf.h" |
30 | #include <stdio.h> | |
31 | ||
73c04bcf A |
32 | #if !UCONFIG_NO_CONVERSION |
33 | ||
34 | ||
b75a7d8f A |
35 | #define MAX_IN_BUF 1000 |
36 | #define MAX_U_BUF 1500 | |
46f4442e | 37 | #define CONTEXT_LEN 20 |
b75a7d8f A |
38 | |
39 | struct UCHARBUF { | |
40 | UChar* buffer; | |
41 | UChar* currentPos; | |
42 | UChar* bufLimit; | |
43 | int32_t bufCapacity; | |
44 | int32_t remaining; | |
45 | int32_t signatureLength; | |
46 | FileStream* in; | |
47 | UConverter* conv; | |
48 | UBool showWarning; /* makes this API not produce any errors */ | |
49 | UBool isBuffered; | |
50 | }; | |
51 | ||
52 | U_CAPI UBool U_EXPORT2 | |
53 | ucbuf_autodetect_fs(FileStream* in, const char** cp, UConverter** conv, int32_t* signatureLength, UErrorCode* error){ | |
54 | char start[8]; | |
55 | int32_t numRead; | |
56 | ||
57 | UChar target[1]={ 0 }; | |
58 | UChar* pTarget; | |
59 | const char* pStart; | |
60 | ||
61 | /* read a few bytes */ | |
62 | numRead=T_FileStream_read(in, start, sizeof(start)); | |
63 | ||
64 | *cp = ucnv_detectUnicodeSignature(start, numRead, signatureLength, error); | |
65 | ||
66 | /* unread the bytes beyond what was consumed for U+FEFF */ | |
67 | T_FileStream_rewind(in); | |
68 | if (*signatureLength > 0) { | |
4388f060 | 69 | T_FileStream_read(in, start, *signatureLength); |
b75a7d8f A |
70 | } |
71 | ||
72 | if(*cp==NULL){ | |
73 | *conv =NULL; | |
74 | return FALSE; | |
75 | } | |
76 | ||
77 | /* open the converter for the detected Unicode charset */ | |
78 | *conv = ucnv_open(*cp,error); | |
79 | ||
80 | /* convert and ignore initial U+FEFF, and the buffer overflow */ | |
81 | pTarget = target; | |
82 | pStart = start; | |
83 | ucnv_toUnicode(*conv, &pTarget, target+1, &pStart, start+*signatureLength, NULL, FALSE, error); | |
374ca955 | 84 | *signatureLength = (int32_t)(pStart - start); |
b75a7d8f A |
85 | if(*error==U_BUFFER_OVERFLOW_ERROR) { |
86 | *error=U_ZERO_ERROR; | |
87 | } | |
88 | ||
89 | /* verify that we successfully read exactly U+FEFF */ | |
90 | if(U_SUCCESS(*error) && (pTarget!=(target+1) || target[0]!=0xfeff)) { | |
91 | *error=U_INTERNAL_PROGRAM_ERROR; | |
92 | } | |
93 | ||
94 | ||
b75a7d8f A |
95 | return TRUE; |
96 | } | |
97 | static UBool ucbuf_isCPKnown(const char* cp){ | |
98 | if(ucnv_compareNames("UTF-8",cp)==0){ | |
99 | return TRUE; | |
100 | } | |
101 | if(ucnv_compareNames("UTF-16BE",cp)==0){ | |
102 | return TRUE; | |
103 | } | |
104 | if(ucnv_compareNames("UTF-16LE",cp)==0){ | |
105 | return TRUE; | |
106 | } | |
107 | if(ucnv_compareNames("UTF-16",cp)==0){ | |
108 | return TRUE; | |
109 | } | |
110 | if(ucnv_compareNames("UTF-32",cp)==0){ | |
111 | return TRUE; | |
112 | } | |
113 | if(ucnv_compareNames("UTF-32BE",cp)==0){ | |
114 | return TRUE; | |
115 | } | |
116 | if(ucnv_compareNames("UTF-32LE",cp)==0){ | |
117 | return TRUE; | |
118 | } | |
b75a7d8f A |
119 | if(ucnv_compareNames("SCSU",cp)==0){ |
120 | return TRUE; | |
121 | } | |
374ca955 | 122 | if(ucnv_compareNames("BOCU-1",cp)==0){ |
b75a7d8f A |
123 | return TRUE; |
124 | } | |
125 | if(ucnv_compareNames("UTF-7",cp)==0){ | |
126 | return TRUE; | |
127 | } | |
128 | return FALSE; | |
129 | } | |
130 | ||
131 | U_CAPI FileStream * U_EXPORT2 | |
132 | ucbuf_autodetect(const char* fileName, const char** cp,UConverter** conv, int32_t* signatureLength,UErrorCode* error){ | |
133 | FileStream* in=NULL; | |
134 | if(error==NULL || U_FAILURE(*error)){ | |
135 | return NULL; | |
136 | } | |
137 | if(conv==NULL || cp==NULL || fileName==NULL){ | |
138 | *error = U_ILLEGAL_ARGUMENT_ERROR; | |
139 | return NULL; | |
140 | } | |
141 | /* open the file */ | |
142 | in= T_FileStream_open(fileName,"rb"); | |
143 | ||
144 | if(in == NULL){ | |
145 | *error=U_FILE_ACCESS_ERROR; | |
146 | return NULL; | |
147 | } | |
148 | ||
149 | if(ucbuf_autodetect_fs(in,cp,conv,signatureLength,error)) { | |
150 | return in; | |
151 | } else { | |
152 | ucnv_close(*conv); | |
153 | *conv=NULL; | |
154 | T_FileStream_close(in); | |
155 | return NULL; | |
156 | } | |
157 | } | |
158 | ||
159 | /* fill the uchar buffer */ | |
160 | static UCHARBUF* | |
161 | ucbuf_fillucbuf( UCHARBUF* buf,UErrorCode* error){ | |
162 | UChar* pTarget=NULL; | |
163 | UChar* target=NULL; | |
164 | const char* source=NULL; | |
165 | char carr[MAX_IN_BUF] = {'\0'}; | |
166 | char* cbuf = carr; | |
167 | int32_t inputRead=0; | |
168 | int32_t outputWritten=0; | |
169 | int32_t offset=0; | |
170 | const char* sourceLimit =NULL; | |
171 | int32_t cbufSize=0; | |
172 | pTarget = buf->buffer; | |
173 | /* check if we arrived here without exhausting the buffer*/ | |
174 | if(buf->currentPos<buf->bufLimit){ | |
175 | offset = (int32_t)(buf->bufLimit-buf->currentPos); | |
176 | memmove(buf->buffer,buf->currentPos,offset* sizeof(UChar)); | |
177 | } | |
178 | ||
179 | #if DEBUG | |
180 | memset(pTarget+offset,0xff,sizeof(UChar)*(MAX_IN_BUF-offset)); | |
181 | #endif | |
182 | if(buf->isBuffered){ | |
183 | cbufSize = MAX_IN_BUF; | |
184 | /* read the file */ | |
185 | inputRead=T_FileStream_read(buf->in,cbuf,cbufSize-offset); | |
186 | buf->remaining-=inputRead; | |
187 | ||
188 | }else{ | |
189 | cbufSize = T_FileStream_size(buf->in); | |
190 | cbuf = (char*)uprv_malloc(cbufSize); | |
46f4442e A |
191 | if (cbuf == NULL) { |
192 | *error = U_MEMORY_ALLOCATION_ERROR; | |
193 | return NULL; | |
194 | } | |
b75a7d8f A |
195 | inputRead= T_FileStream_read(buf->in,cbuf,cbufSize); |
196 | buf->remaining-=inputRead; | |
197 | } | |
198 | ||
199 | /* just to be sure...*/ | |
200 | if ( 0 == inputRead ) | |
201 | buf->remaining = 0; | |
202 | ||
203 | target=pTarget; | |
204 | /* convert the bytes */ | |
205 | if(buf->conv){ | |
206 | /* set the callback to stop */ | |
207 | UConverterToUCallback toUOldAction ; | |
208 | void* toUOldContext; | |
209 | void* toUNewContext=NULL; | |
210 | ucnv_setToUCallBack(buf->conv, | |
211 | UCNV_TO_U_CALLBACK_STOP, | |
212 | toUNewContext, | |
213 | &toUOldAction, | |
214 | (const void**)&toUOldContext, | |
215 | error); | |
216 | /* since state is saved in the converter we add offset to source*/ | |
217 | target = pTarget+offset; | |
218 | source = cbuf; | |
219 | sourceLimit = source + inputRead; | |
220 | ucnv_toUnicode(buf->conv,&target,target+(buf->bufCapacity-offset), | |
221 | &source,sourceLimit,NULL, | |
222 | (UBool)(buf->remaining==0),error); | |
223 | ||
224 | if(U_FAILURE(*error)){ | |
46f4442e A |
225 | char context[CONTEXT_LEN+1]; |
226 | char preContext[CONTEXT_LEN+1]; | |
227 | char postContext[CONTEXT_LEN+1]; | |
b75a7d8f A |
228 | int8_t len = CONTEXT_LEN; |
229 | int32_t start=0; | |
230 | int32_t stop =0; | |
231 | int32_t pos =0; | |
232 | /* use erro1 to preserve the error code */ | |
233 | UErrorCode error1 =U_ZERO_ERROR; | |
234 | ||
235 | if( buf->showWarning==TRUE){ | |
236 | fprintf(stderr,"\n###WARNING: Encountered abnormal bytes while" | |
237 | " converting input stream to target encoding: %s\n", | |
238 | u_errorName(*error)); | |
239 | } | |
240 | ||
241 | ||
242 | /* now get the context chars */ | |
243 | ucnv_getInvalidChars(buf->conv,context,&len,&error1); | |
244 | context[len]= 0 ; /* null terminate the buffer */ | |
245 | ||
246 | pos = (int32_t)(source - cbuf - len); | |
247 | ||
248 | /* for pre-context */ | |
249 | start = (pos <=CONTEXT_LEN)? 0 : (pos - (CONTEXT_LEN-1)); | |
250 | stop = pos-len; | |
251 | ||
252 | memcpy(preContext,cbuf+start,stop-start); | |
253 | /* null terminate the buffer */ | |
254 | preContext[stop-start] = 0; | |
255 | ||
256 | /* for post-context */ | |
257 | start = pos+len; | |
258 | stop = (int32_t)(((pos+CONTEXT_LEN)<= (sourceLimit-cbuf) )? (pos+(CONTEXT_LEN-1)) : (sourceLimit-cbuf)); | |
259 | ||
260 | memcpy(postContext,source,stop-start); | |
261 | /* null terminate the buffer */ | |
262 | postContext[stop-start] = 0; | |
263 | ||
264 | if(buf->showWarning ==TRUE){ | |
265 | /* print out the context */ | |
266 | fprintf(stderr,"\tPre-context: %s\n",preContext); | |
267 | fprintf(stderr,"\tContext: %s\n",context); | |
268 | fprintf(stderr,"\tPost-context: %s\n", postContext); | |
269 | } | |
270 | ||
271 | /* reset the converter */ | |
272 | ucnv_reset(buf->conv); | |
273 | ||
274 | /* set the call back to substitute | |
275 | * and restart conversion | |
276 | */ | |
277 | ucnv_setToUCallBack(buf->conv, | |
278 | UCNV_TO_U_CALLBACK_SUBSTITUTE, | |
279 | toUNewContext, | |
280 | &toUOldAction, | |
281 | (const void**)&toUOldContext, | |
282 | &error1); | |
283 | ||
284 | /* reset source and target start positions */ | |
285 | target = pTarget+offset; | |
286 | source = cbuf; | |
287 | ||
288 | /* re convert */ | |
289 | ucnv_toUnicode(buf->conv,&target,target+(buf->bufCapacity-offset), | |
290 | &source,sourceLimit,NULL, | |
291 | (UBool)(buf->remaining==0),&error1); | |
292 | ||
293 | } | |
294 | outputWritten = (int32_t)(target - pTarget); | |
295 | ||
296 | ||
297 | #if DEBUG | |
298 | { | |
299 | int i; | |
300 | target = pTarget; | |
301 | for(i=0;i<numRead;i++){ | |
302 | /* printf("%c", (char)(*target++));*/ | |
303 | } | |
304 | } | |
305 | #endif | |
306 | ||
307 | }else{ | |
308 | u_charsToUChars(cbuf,target+offset,inputRead); | |
309 | outputWritten=((buf->remaining>cbufSize)? cbufSize:inputRead+offset); | |
310 | } | |
311 | buf->currentPos = pTarget; | |
312 | buf->bufLimit=pTarget+outputWritten; | |
374ca955 | 313 | *buf->bufLimit=0; /*NUL terminate*/ |
b75a7d8f A |
314 | if(cbuf!=carr){ |
315 | uprv_free(cbuf); | |
316 | } | |
317 | return buf; | |
318 | } | |
319 | ||
320 | ||
321 | ||
322 | /* get a UChar from the stream*/ | |
323 | U_CAPI int32_t U_EXPORT2 | |
324 | ucbuf_getc(UCHARBUF* buf,UErrorCode* error){ | |
325 | if(error==NULL || U_FAILURE(*error)){ | |
326 | return FALSE; | |
327 | } | |
328 | if(buf->currentPos>=buf->bufLimit){ | |
329 | if(buf->remaining==0){ | |
330 | return U_EOF; | |
331 | } | |
332 | buf=ucbuf_fillucbuf(buf,error); | |
333 | if(U_FAILURE(*error)){ | |
334 | return U_EOF; | |
335 | } | |
336 | } | |
337 | ||
338 | return *(buf->currentPos++); | |
339 | } | |
340 | ||
341 | /* get a UChar32 from the stream*/ | |
342 | U_CAPI int32_t U_EXPORT2 | |
343 | ucbuf_getc32(UCHARBUF* buf,UErrorCode* error){ | |
344 | int32_t retVal = (int32_t)U_EOF; | |
345 | if(error==NULL || U_FAILURE(*error)){ | |
346 | return FALSE; | |
347 | } | |
348 | if(buf->currentPos+1>=buf->bufLimit){ | |
349 | if(buf->remaining==0){ | |
350 | return U_EOF; | |
351 | } | |
352 | buf=ucbuf_fillucbuf(buf,error); | |
353 | if(U_FAILURE(*error)){ | |
354 | return U_EOF; | |
355 | } | |
356 | } | |
4388f060 A |
357 | if(U16_IS_LEAD(*(buf->currentPos))){ |
358 | retVal=U16_GET_SUPPLEMENTARY(buf->currentPos[0],buf->currentPos[1]); | |
46f4442e | 359 | buf->currentPos+=2; |
b75a7d8f A |
360 | }else{ |
361 | retVal = *(buf->currentPos++); | |
362 | } | |
363 | return retVal; | |
364 | } | |
365 | ||
366 | /* u_unescapeAt() callback to return a UChar*/ | |
367 | static UChar U_CALLCONV | |
368 | _charAt(int32_t offset, void *context) { | |
369 | return ((UCHARBUF*) context)->currentPos[offset]; | |
370 | } | |
371 | ||
372 | /* getc and escape it */ | |
373 | U_CAPI int32_t U_EXPORT2 | |
374 | ucbuf_getcx32(UCHARBUF* buf,UErrorCode* error) { | |
375 | int32_t length; | |
376 | int32_t offset; | |
377 | UChar32 c32,c1,c2; | |
378 | if(error==NULL || U_FAILURE(*error)){ | |
379 | return FALSE; | |
380 | } | |
381 | /* Fill the buffer if it is empty */ | |
382 | if (buf->currentPos >=buf->bufLimit-2) { | |
383 | ucbuf_fillucbuf(buf,error); | |
384 | } | |
385 | ||
386 | /* Get the next character in the buffer */ | |
387 | if (buf->currentPos < buf->bufLimit) { | |
388 | c1 = *(buf->currentPos)++; | |
389 | } else { | |
390 | c1 = U_EOF; | |
391 | } | |
392 | ||
393 | c2 = *(buf->currentPos); | |
394 | ||
395 | /* If it isn't a backslash, return it */ | |
396 | if (c1 != 0x005C) { | |
397 | return c1; | |
398 | } | |
399 | ||
400 | /* Determine the amount of data in the buffer */ | |
401 | length = (int32_t)(buf->bufLimit - buf->currentPos); | |
402 | ||
403 | /* The longest escape sequence is \Uhhhhhhhh; make sure | |
404 | we have at least that many characters */ | |
405 | if (length < 10) { | |
406 | ||
407 | /* fill the buffer */ | |
408 | ucbuf_fillucbuf(buf,error); | |
409 | length = (int32_t)(buf->bufLimit - buf->buffer); | |
410 | } | |
411 | ||
412 | /* Process the escape */ | |
413 | offset = 0; | |
414 | c32 = u_unescapeAt(_charAt, &offset, length, (void*)buf); | |
415 | ||
416 | /* check if u_unescapeAt unescaped and converted | |
417 | * to c32 or not | |
418 | */ | |
419 | if(c32==0xFFFFFFFF){ | |
374ca955 | 420 | if(buf->showWarning) { |
46f4442e A |
421 | char context[CONTEXT_LEN+1]; |
422 | int32_t len = CONTEXT_LEN; | |
423 | if(length < len) { | |
374ca955 A |
424 | len = length; |
425 | } | |
426 | context[len]= 0 ; /* null terminate the buffer */ | |
427 | u_UCharsToChars( buf->currentPos, context, len); | |
428 | fprintf(stderr,"Bad escape: [%c%s]...\n", (int)c1, context); | |
429 | } | |
b75a7d8f A |
430 | *error= U_ILLEGAL_ESCAPE_SEQUENCE; |
431 | return c1; | |
432 | }else if(c32!=c2 || (c32==0x0075 && c2==0x0075 && c1==0x005C) /* for \u0075 c2=0x0075 and c32==0x0075*/){ | |
433 | /* Update the current buffer position */ | |
434 | buf->currentPos += offset; | |
435 | }else{ | |
436 | /* unescaping failed so we just return | |
437 | * c1 and not consume the buffer | |
438 | * this is useful for rules with escapes | |
439 | * in resouce bundles | |
440 | * eg: \' \\ \" | |
441 | */ | |
442 | return c1; | |
443 | } | |
444 | ||
445 | return c32; | |
446 | } | |
447 | ||
448 | U_CAPI UCHARBUF* U_EXPORT2 | |
449 | ucbuf_open(const char* fileName,const char** cp,UBool showWarning, UBool buffered, UErrorCode* error){ | |
450 | ||
451 | FileStream* in = NULL; | |
452 | int32_t fileSize=0; | |
453 | const char* knownCp; | |
454 | if(error==NULL || U_FAILURE(*error)){ | |
455 | return NULL; | |
456 | } | |
457 | if(cp==NULL || fileName==NULL){ | |
458 | *error = U_ILLEGAL_ARGUMENT_ERROR; | |
459 | return FALSE; | |
460 | } | |
461 | if (!uprv_strcmp(fileName, "-")) { | |
462 | in = T_FileStream_stdin(); | |
463 | }else{ | |
464 | in = T_FileStream_open(fileName, "rb"); | |
465 | } | |
466 | ||
467 | if(in!=NULL){ | |
468 | UCHARBUF* buf =(UCHARBUF*) uprv_malloc(sizeof(UCHARBUF)); | |
469 | fileSize = T_FileStream_size(in); | |
73c04bcf A |
470 | if(buf == NULL){ |
471 | *error = U_MEMORY_ALLOCATION_ERROR; | |
472 | T_FileStream_close(in); | |
473 | return NULL; | |
474 | } | |
475 | buf->in=in; | |
476 | buf->conv=NULL; | |
477 | buf->showWarning = showWarning; | |
478 | buf->isBuffered = buffered; | |
479 | buf->signatureLength=0; | |
480 | if(*cp==NULL || **cp=='\0'){ | |
481 | /* don't have code page name... try to autodetect */ | |
482 | ucbuf_autodetect_fs(in,cp,&buf->conv,&buf->signatureLength,error); | |
483 | }else if(ucbuf_isCPKnown(*cp)){ | |
484 | /* discard BOM */ | |
485 | ucbuf_autodetect_fs(in,&knownCp,&buf->conv,&buf->signatureLength,error); | |
486 | } | |
487 | if(U_SUCCESS(*error) && buf->conv==NULL) { | |
488 | buf->conv=ucnv_open(*cp,error); | |
489 | } | |
490 | if(U_FAILURE(*error)){ | |
491 | ucnv_close(buf->conv); | |
492 | uprv_free(buf); | |
493 | T_FileStream_close(in); | |
494 | return NULL; | |
495 | } | |
496 | ||
497 | if((buf->conv==NULL) && (buf->showWarning==TRUE)){ | |
498 | fprintf(stderr,"###WARNING: No converter defined. Using codepage of system.\n"); | |
499 | } | |
500 | buf->remaining=fileSize-buf->signatureLength; | |
501 | if(buf->isBuffered){ | |
502 | buf->bufCapacity=MAX_U_BUF; | |
b75a7d8f | 503 | }else{ |
73c04bcf A |
504 | buf->bufCapacity=buf->remaining+buf->signatureLength+1/*for terminating nul*/; |
505 | } | |
506 | buf->buffer=(UChar*) uprv_malloc(U_SIZEOF_UCHAR * buf->bufCapacity ); | |
507 | if (buf->buffer == NULL) { | |
b75a7d8f | 508 | *error = U_MEMORY_ALLOCATION_ERROR; |
46f4442e | 509 | ucbuf_close(buf); |
b75a7d8f A |
510 | return NULL; |
511 | } | |
73c04bcf A |
512 | buf->currentPos=buf->buffer; |
513 | buf->bufLimit=buf->buffer; | |
514 | if(U_FAILURE(*error)){ | |
515 | fprintf(stderr, "Could not open codepage [%s]: %s\n", *cp, u_errorName(*error)); | |
46f4442e A |
516 | ucbuf_close(buf); |
517 | return NULL; | |
518 | } | |
519 | ucbuf_fillucbuf(buf,error); | |
520 | if(U_FAILURE(*error)){ | |
521 | ucbuf_close(buf); | |
73c04bcf A |
522 | return NULL; |
523 | } | |
73c04bcf | 524 | return buf; |
b75a7d8f A |
525 | } |
526 | *error =U_FILE_ACCESS_ERROR; | |
527 | return NULL; | |
528 | } | |
529 | ||
530 | ||
531 | ||
532 | /* TODO: this method will fail if at the | |
533 | * begining of buffer and the uchar to unget | |
534 | * is from the previous buffer. Need to implement | |
535 | * system to take care of that situation. | |
536 | */ | |
537 | U_CAPI void U_EXPORT2 | |
538 | ucbuf_ungetc(int32_t c,UCHARBUF* buf){ | |
539 | /* decrement currentPos pointer | |
540 | * if not at the begining of buffer | |
541 | */ | |
b75a7d8f A |
542 | if(buf->currentPos!=buf->buffer){ |
543 | if(*(buf->currentPos-1)==c){ | |
544 | buf->currentPos--; | |
46f4442e A |
545 | } else { |
546 | /* ungetc failed - did not match. */ | |
b75a7d8f | 547 | } |
46f4442e A |
548 | } else { |
549 | /* ungetc failed - beginning of buffer. */ | |
b75a7d8f A |
550 | } |
551 | } | |
552 | ||
553 | /* frees the resources of UChar* buffer */ | |
554 | static void | |
555 | ucbuf_closebuf(UCHARBUF* buf){ | |
556 | uprv_free(buf->buffer); | |
557 | buf->buffer = NULL; | |
558 | } | |
559 | ||
560 | /* close the buf and release resources*/ | |
561 | U_CAPI void U_EXPORT2 | |
562 | ucbuf_close(UCHARBUF* buf){ | |
563 | if(buf!=NULL){ | |
564 | if(buf->conv){ | |
565 | ucnv_close(buf->conv); | |
566 | } | |
567 | T_FileStream_close(buf->in); | |
568 | ucbuf_closebuf(buf); | |
569 | uprv_free(buf); | |
570 | } | |
571 | } | |
572 | ||
573 | /* rewind the buf and file stream */ | |
574 | U_CAPI void U_EXPORT2 | |
575 | ucbuf_rewind(UCHARBUF* buf,UErrorCode* error){ | |
576 | if(error==NULL || U_FAILURE(*error)){ | |
577 | return; | |
578 | } | |
579 | if(buf){ | |
580 | buf->currentPos=buf->buffer; | |
581 | buf->bufLimit=buf->buffer; | |
582 | T_FileStream_rewind(buf->in); | |
583 | buf->remaining=T_FileStream_size(buf->in)-buf->signatureLength; | |
584 | ||
585 | ucnv_resetToUnicode(buf->conv); | |
586 | if(buf->signatureLength>0) { | |
587 | UChar target[1]={ 0 }; | |
588 | UChar* pTarget; | |
589 | char start[8]; | |
590 | const char* pStart; | |
591 | int32_t numRead; | |
592 | ||
593 | /* read the signature bytes */ | |
594 | numRead=T_FileStream_read(buf->in, start, buf->signatureLength); | |
595 | ||
596 | /* convert and ignore initial U+FEFF, and the buffer overflow */ | |
597 | pTarget = target; | |
598 | pStart = start; | |
599 | ucnv_toUnicode(buf->conv, &pTarget, target+1, &pStart, start+numRead, NULL, FALSE, error); | |
600 | if(*error==U_BUFFER_OVERFLOW_ERROR) { | |
601 | *error=U_ZERO_ERROR; | |
602 | } | |
603 | ||
604 | /* verify that we successfully read exactly U+FEFF */ | |
605 | if(U_SUCCESS(*error) && (numRead!=buf->signatureLength || pTarget!=(target+1) || target[0]!=0xfeff)) { | |
606 | *error=U_INTERNAL_PROGRAM_ERROR; | |
607 | } | |
608 | } | |
609 | } | |
610 | } | |
611 | ||
612 | ||
613 | U_CAPI int32_t U_EXPORT2 | |
614 | ucbuf_size(UCHARBUF* buf){ | |
615 | if(buf){ | |
616 | if(buf->isBuffered){ | |
617 | return (T_FileStream_size(buf->in)-buf->signatureLength)/ucnv_getMinCharSize(buf->conv); | |
618 | }else{ | |
374ca955 | 619 | return (int32_t)(buf->bufLimit - buf->buffer); |
b75a7d8f A |
620 | } |
621 | } | |
622 | return 0; | |
623 | } | |
624 | ||
625 | U_CAPI const UChar* U_EXPORT2 | |
626 | ucbuf_getBuffer(UCHARBUF* buf,int32_t* len,UErrorCode* error){ | |
627 | if(error==NULL || U_FAILURE(*error)){ | |
628 | return NULL; | |
629 | } | |
630 | if(buf==NULL || len==NULL){ | |
631 | *error = U_ILLEGAL_ARGUMENT_ERROR; | |
632 | return NULL; | |
633 | } | |
374ca955 | 634 | *len = (int32_t)(buf->bufLimit - buf->buffer); |
b75a7d8f A |
635 | return buf->buffer; |
636 | } | |
637 | ||
638 | U_CAPI const char* U_EXPORT2 | |
639 | ucbuf_resolveFileName(const char* inputDir, const char* fileName, char* target, int32_t* len, UErrorCode* status){ | |
640 | int32_t requiredLen = 0; | |
641 | int32_t dirlen = 0; | |
642 | int32_t filelen = 0; | |
643 | if(status==NULL || U_FAILURE(*status)){ | |
644 | return NULL; | |
645 | } | |
646 | ||
647 | if(inputDir == NULL || fileName == NULL || len==NULL || (target==NULL && *len>0)){ | |
648 | *status = U_ILLEGAL_ARGUMENT_ERROR; | |
649 | return NULL; | |
650 | } | |
651 | ||
652 | ||
653 | dirlen = (int32_t)uprv_strlen(inputDir); | |
654 | filelen = (int32_t)uprv_strlen(fileName); | |
655 | if(inputDir[dirlen-1] != U_FILE_SEP_CHAR) { | |
656 | requiredLen = dirlen + filelen + 2; | |
657 | if((*len < requiredLen) || target==NULL){ | |
658 | *len = requiredLen; | |
659 | *status = U_BUFFER_OVERFLOW_ERROR; | |
660 | return NULL; | |
661 | } | |
662 | ||
663 | target[0] = '\0'; | |
664 | /* | |
665 | * append the input dir to openFileName if the first char in | |
666 | * filename is not file seperation char and the last char input directory is not '.'. | |
667 | * This is to support : | |
668 | * genrb -s. /home/icu/data | |
669 | * genrb -s. icu/data | |
670 | * The user cannot mix notations like | |
671 | * genrb -s. /icu/data --- the absolute path specified. -s redundant | |
672 | * user should use | |
673 | * genrb -s. icu/data --- start from CWD and look in icu/data dir | |
674 | */ | |
675 | if( (fileName[0] != U_FILE_SEP_CHAR) && (inputDir[dirlen-1] !='.')){ | |
676 | uprv_strcpy(target, inputDir); | |
677 | target[dirlen] = U_FILE_SEP_CHAR; | |
678 | } | |
679 | target[dirlen + 1] = '\0'; | |
680 | } else { | |
681 | requiredLen = dirlen + filelen + 1; | |
682 | if((*len < requiredLen) || target==NULL){ | |
683 | *len = requiredLen; | |
684 | *status = U_BUFFER_OVERFLOW_ERROR; | |
685 | return NULL; | |
686 | } | |
687 | ||
688 | uprv_strcpy(target, inputDir); | |
689 | } | |
690 | ||
691 | uprv_strcat(target, fileName); | |
692 | return target; | |
693 | } | |
694 | /* | |
695 | * Unicode TR 13 says any of the below chars is | |
696 | * a new line char in a readline function in addition | |
697 | * to CR+LF combination which needs to be | |
698 | * handled seperately | |
699 | */ | |
700 | static UBool ucbuf_isCharNewLine(UChar c){ | |
701 | switch(c){ | |
702 | case 0x000A: /* LF */ | |
703 | case 0x000D: /* CR */ | |
704 | case 0x000C: /* FF */ | |
705 | case 0x0085: /* NEL */ | |
706 | case 0x2028: /* LS */ | |
707 | case 0x2029: /* PS */ | |
708 | return TRUE; | |
709 | default: | |
710 | return FALSE; | |
711 | } | |
712 | } | |
713 | ||
714 | U_CAPI const UChar* U_EXPORT2 | |
715 | ucbuf_readline(UCHARBUF* buf,int32_t* len,UErrorCode* err){ | |
716 | UChar* temp = buf->currentPos; | |
717 | UChar* savePos =NULL; | |
718 | UChar c=0x0000; | |
719 | if(buf->isBuffered){ | |
374ca955 A |
720 | /* The input is buffered we have to do more |
721 | * for returning a pointer U_TRUNCATED_CHAR_FOUND | |
b75a7d8f A |
722 | */ |
723 | for(;;){ | |
724 | c = *temp++; | |
725 | if(buf->remaining==0){ | |
374ca955 | 726 | return NULL; /* end of file is reached return NULL */ |
b75a7d8f A |
727 | } |
728 | if(temp>=buf->bufLimit && buf->currentPos == buf->buffer){ | |
729 | *err= U_TRUNCATED_CHAR_FOUND; | |
730 | return NULL; | |
731 | }else{ | |
732 | ucbuf_fillucbuf(buf,err); | |
733 | if(U_FAILURE(*err)){ | |
374ca955 | 734 | return NULL; |
b75a7d8f A |
735 | } |
736 | } | |
737 | /* | |
738 | * Accoding to TR 13 readLine functions must interpret | |
739 | * CR, CR+LF, LF, NEL, PS, LS or FF as line seperators | |
740 | */ | |
741 | /* Windows CR LF */ | |
742 | if(c ==0x0d && temp+1<=buf->bufLimit && *(temp+1) == 0x0a ){ | |
374ca955 | 743 | *len = (int32_t)(temp++ - buf->currentPos); |
b75a7d8f A |
744 | savePos = buf->currentPos; |
745 | buf->currentPos = temp; | |
746 | return savePos; | |
747 | } | |
748 | /* else */ | |
749 | ||
750 | if (temp>=buf->bufLimit|| ucbuf_isCharNewLine(c)){ /* Unipad inserts 2028 line separators! */ | |
374ca955 | 751 | *len = (int32_t)(temp - buf->currentPos); |
b75a7d8f A |
752 | savePos = buf->currentPos; |
753 | buf->currentPos = temp; | |
754 | return savePos; | |
755 | } | |
756 | } | |
757 | }else{ | |
758 | /* we know that all input is read into the internal | |
759 | * buffer so we can safely return pointers | |
760 | */ | |
761 | for(;;){ | |
762 | c = *temp++; | |
763 | ||
764 | if(buf->currentPos==buf->bufLimit){ | |
374ca955 | 765 | return NULL; /* end of file is reached return NULL */ |
b75a7d8f A |
766 | } |
767 | /* Windows CR LF */ | |
768 | if(c ==0x0d && temp+1<=buf->bufLimit && *(temp+1) == 0x0a ){ | |
374ca955 | 769 | *len = (int32_t)(temp++ - buf->currentPos); |
b75a7d8f A |
770 | savePos = buf->currentPos; |
771 | buf->currentPos = temp; | |
772 | return savePos; | |
773 | } | |
774 | /* else */ | |
775 | if (temp>=buf->bufLimit|| ucbuf_isCharNewLine(c)) { /* Unipad inserts 2028 line separators! */ | |
374ca955 | 776 | *len = (int32_t)(temp - buf->currentPos); |
b75a7d8f A |
777 | savePos = buf->currentPos; |
778 | buf->currentPos = temp; | |
779 | return savePos; | |
780 | } | |
781 | } | |
782 | } | |
783 | /* not reached */ | |
784 | /* A compiler warning will appear if all paths don't contain a return statement. */ | |
785 | /* return NULL;*/ | |
786 | } | |
73c04bcf | 787 | #endif |