Annotation of embedaddon/php/ext/mbstring/libmbfl/mbfl/mbfl_convert.c, revision 1.1.1.1
1.1 misho 1: /*
2: * "streamable kanji code filter and converter"
3: * Copyright (c) 1998-2002 HappySize, Inc. All rights reserved.
4: *
5: * LICENSE NOTICES
6: *
7: * This file is part of "streamable kanji code filter and converter",
8: * which is distributed under the terms of GNU Lesser General Public
9: * License (version 2) as published by the Free Software Foundation.
10: *
11: * This software is distributed in the hope that it will be useful,
12: * but WITHOUT ANY WARRANTY; without even the implied warranty of
13: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14: * GNU Lesser General Public License for more details.
15: *
16: * You should have received a copy of the GNU Lesser General Public
17: * License along with "streamable kanji code filter and converter";
18: * if not, write to the Free Software Foundation, Inc., 59 Temple Place,
19: * Suite 330, Boston, MA 02111-1307 USA
20: *
21: * The author of this file:
22: *
23: */
24: /*
25: * The source code included in this files was separated from mbfilter.c
26: * by Moriyoshi Koizumi <moriyoshi@php.net> on 20 Dec 2002. The file
27: * mbfilter.c is included in this package .
28: *
29: */
30:
31: #ifdef HAVE_CONFIG_H
32: #include "config.h"
33: #endif
34:
35: #ifdef HAVE_STDDEF_H
36: #include <stddef.h>
37: #endif
38:
39: #include "mbfl_encoding.h"
40: #include "mbfl_allocators.h"
41: #include "mbfl_filter_output.h"
42: #include "mbfilter_pass.h"
43: #include "mbfilter_8bit.h"
44: #include "mbfilter_wchar.h"
45:
46: #include "filters/mbfilter_euc_cn.h"
47: #include "filters/mbfilter_hz.h"
48: #include "filters/mbfilter_euc_tw.h"
49: #include "filters/mbfilter_big5.h"
50: #include "filters/mbfilter_uhc.h"
51: #include "filters/mbfilter_euc_kr.h"
52: #include "filters/mbfilter_iso2022_kr.h"
53: #include "filters/mbfilter_sjis.h"
54: #include "filters/mbfilter_sjis_open.h"
55: #include "filters/mbfilter_cp51932.h"
56: #include "filters/mbfilter_jis.h"
57: #include "filters/mbfilter_iso2022_jp_ms.h"
58: #include "filters/mbfilter_euc_jp.h"
59: #include "filters/mbfilter_euc_jp_win.h"
60: #include "filters/mbfilter_ascii.h"
61: #include "filters/mbfilter_koi8r.h"
62: #include "filters/mbfilter_koi8u.h"
63: #include "filters/mbfilter_cp866.h"
64: #include "filters/mbfilter_cp932.h"
65: #include "filters/mbfilter_cp936.h"
66: #include "filters/mbfilter_cp1251.h"
67: #include "filters/mbfilter_cp1252.h"
68: #include "filters/mbfilter_cp1254.h"
69: #include "filters/mbfilter_cp5022x.h"
70: #include "filters/mbfilter_iso8859_1.h"
71: #include "filters/mbfilter_iso8859_2.h"
72: #include "filters/mbfilter_iso8859_3.h"
73: #include "filters/mbfilter_iso8859_4.h"
74: #include "filters/mbfilter_iso8859_5.h"
75: #include "filters/mbfilter_iso8859_6.h"
76: #include "filters/mbfilter_iso8859_7.h"
77: #include "filters/mbfilter_iso8859_8.h"
78: #include "filters/mbfilter_iso8859_9.h"
79: #include "filters/mbfilter_iso8859_10.h"
80: #include "filters/mbfilter_iso8859_13.h"
81: #include "filters/mbfilter_iso8859_14.h"
82: #include "filters/mbfilter_iso8859_15.h"
83: #include "filters/mbfilter_base64.h"
84: #include "filters/mbfilter_qprint.h"
85: #include "filters/mbfilter_uuencode.h"
86: #include "filters/mbfilter_7bit.h"
87: #include "filters/mbfilter_utf7.h"
88: #include "filters/mbfilter_utf7imap.h"
89: #include "filters/mbfilter_utf8.h"
90: #include "filters/mbfilter_utf16.h"
91: #include "filters/mbfilter_utf32.h"
92: #include "filters/mbfilter_byte2.h"
93: #include "filters/mbfilter_byte4.h"
94: #include "filters/mbfilter_ucs4.h"
95: #include "filters/mbfilter_ucs2.h"
96: #include "filters/mbfilter_htmlent.h"
97: #include "filters/mbfilter_armscii8.h"
98: #include "filters/mbfilter_cp850.h"
99:
100: /* hex character table "0123456789ABCDEF" */
101: static char mbfl_hexchar_table[] = {
102: 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0x41,0x42,0x43,0x44,0x45,0x46
103: };
104:
105: const struct mbfl_convert_vtbl *mbfl_convert_filter_list[] = {
106: &vtbl_utf8_wchar,
107: &vtbl_wchar_utf8,
108: &vtbl_eucjp_wchar,
109: &vtbl_wchar_eucjp,
110: &vtbl_sjis_wchar,
111: &vtbl_wchar_sjis,
112: &vtbl_sjis_open_wchar,
113: &vtbl_wchar_sjis_open,
114: &vtbl_cp51932_wchar,
115: &vtbl_wchar_cp51932,
116: &vtbl_jis_wchar,
117: &vtbl_wchar_jis,
118: &vtbl_jis_ms_wchar,
119: &vtbl_wchar_jis_ms,
120: &vtbl_2022jp_wchar,
121: &vtbl_wchar_2022jp,
122: &vtbl_2022jpms_wchar,
123: &vtbl_wchar_2022jpms,
124: &vtbl_eucjpwin_wchar,
125: &vtbl_wchar_eucjpwin,
126: &vtbl_cp932_wchar,
127: &vtbl_wchar_cp932,
128: &vtbl_euccn_wchar,
129: &vtbl_wchar_euccn,
130: &vtbl_cp936_wchar,
131: &vtbl_wchar_cp936,
132: &vtbl_hz_wchar,
133: &vtbl_wchar_hz,
134: &vtbl_euctw_wchar,
135: &vtbl_wchar_euctw,
136: &vtbl_big5_wchar,
137: &vtbl_wchar_big5,
138: &vtbl_euckr_wchar,
139: &vtbl_wchar_euckr,
140: &vtbl_uhc_wchar,
141: &vtbl_wchar_uhc,
142: &vtbl_2022kr_wchar,
143: &vtbl_wchar_2022kr,
144: &vtbl_cp1251_wchar,
145: &vtbl_wchar_cp1251,
146: &vtbl_cp866_wchar,
147: &vtbl_wchar_cp866,
148: &vtbl_koi8r_wchar,
149: &vtbl_wchar_koi8r,
150: &vtbl_koi8u_wchar,
151: &vtbl_wchar_koi8u,
152: &vtbl_cp1252_wchar,
153: &vtbl_wchar_cp1252,
154: &vtbl_cp1254_wchar,
155: &vtbl_wchar_cp1254,
156: &vtbl_cp50220_wchar,
157: &vtbl_wchar_cp50220,
158: &vtbl_cp50220raw_wchar,
159: &vtbl_wchar_cp50220raw,
160: &vtbl_cp50221_wchar,
161: &vtbl_wchar_cp50221,
162: &vtbl_cp50222_wchar,
163: &vtbl_wchar_cp50222,
164: &vtbl_ascii_wchar,
165: &vtbl_wchar_ascii,
166: &vtbl_8859_1_wchar,
167: &vtbl_wchar_8859_1,
168: &vtbl_8859_2_wchar,
169: &vtbl_wchar_8859_2,
170: &vtbl_8859_3_wchar,
171: &vtbl_wchar_8859_3,
172: &vtbl_8859_4_wchar,
173: &vtbl_wchar_8859_4,
174: &vtbl_8859_5_wchar,
175: &vtbl_wchar_8859_5,
176: &vtbl_8859_6_wchar,
177: &vtbl_wchar_8859_6,
178: &vtbl_8859_7_wchar,
179: &vtbl_wchar_8859_7,
180: &vtbl_8859_8_wchar,
181: &vtbl_wchar_8859_8,
182: &vtbl_8859_9_wchar,
183: &vtbl_wchar_8859_9,
184: &vtbl_8859_10_wchar,
185: &vtbl_wchar_8859_10,
186: &vtbl_8859_13_wchar,
187: &vtbl_wchar_8859_13,
188: &vtbl_8859_14_wchar,
189: &vtbl_wchar_8859_14,
190: &vtbl_8859_15_wchar,
191: &vtbl_wchar_8859_15,
192: &vtbl_8bit_b64,
193: &vtbl_b64_8bit,
194: &vtbl_uuencode_8bit,
195: &vtbl_wchar_html,
196: &vtbl_html_wchar,
197: &vtbl_8bit_qprint,
198: &vtbl_qprint_8bit,
199: &vtbl_8bit_7bit,
200: &vtbl_7bit_8bit,
201: &vtbl_utf7_wchar,
202: &vtbl_wchar_utf7,
203: &vtbl_utf7imap_wchar,
204: &vtbl_wchar_utf7imap,
205: &vtbl_utf16_wchar,
206: &vtbl_wchar_utf16,
207: &vtbl_utf16be_wchar,
208: &vtbl_wchar_utf16be,
209: &vtbl_utf16le_wchar,
210: &vtbl_wchar_utf16le,
211: &vtbl_utf32_wchar,
212: &vtbl_wchar_utf32,
213: &vtbl_utf32be_wchar,
214: &vtbl_wchar_utf32be,
215: &vtbl_utf32le_wchar,
216: &vtbl_wchar_utf32le,
217: &vtbl_ucs4_wchar,
218: &vtbl_wchar_ucs4,
219: &vtbl_ucs4be_wchar,
220: &vtbl_wchar_ucs4be,
221: &vtbl_ucs4le_wchar,
222: &vtbl_wchar_ucs4le,
223: &vtbl_ucs2_wchar,
224: &vtbl_wchar_ucs2,
225: &vtbl_ucs2be_wchar,
226: &vtbl_wchar_ucs2be,
227: &vtbl_ucs2le_wchar,
228: &vtbl_wchar_ucs2le,
229: &vtbl_byte4be_wchar,
230: &vtbl_wchar_byte4be,
231: &vtbl_byte4le_wchar,
232: &vtbl_wchar_byte4le,
233: &vtbl_byte2be_wchar,
234: &vtbl_wchar_byte2be,
235: &vtbl_byte2le_wchar,
236: &vtbl_wchar_byte2le,
237: &vtbl_armscii8_wchar,
238: &vtbl_wchar_armscii8,
239: &vtbl_cp850_wchar,
240: &vtbl_wchar_cp850,
241: &vtbl_pass,
242: NULL
243: };
244:
245: static int
246: mbfl_convert_filter_common_init(
247: mbfl_convert_filter *filter,
248: enum mbfl_no_encoding from,
249: enum mbfl_no_encoding to,
250: const struct mbfl_convert_vtbl *vtbl,
251: int (*output_function)(int, void* ),
252: int (*flush_function)(void*),
253: void* data)
254: {
255: /* encoding structure */
256: if ((filter->from = mbfl_no2encoding(from)) == NULL) {
257: return 1;
258: }
259:
260: if ((filter->to = mbfl_no2encoding(to)) == NULL) {
261: return 1;
262: }
263:
264: if (output_function != NULL) {
265: filter->output_function = output_function;
266: } else {
267: filter->output_function = mbfl_filter_output_null;
268: }
269:
270: filter->flush_function = flush_function;
271: filter->data = data;
272: filter->illegal_mode = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
273: filter->illegal_substchar = 0x3f; /* '?' */
274: filter->num_illegalchar = 0;
275: filter->filter_ctor = vtbl->filter_ctor;
276: filter->filter_dtor = vtbl->filter_dtor;
277: filter->filter_function = vtbl->filter_function;
278: filter->filter_flush = vtbl->filter_flush;
279: filter->filter_copy = vtbl->filter_copy;
280:
281: (*filter->filter_ctor)(filter);
282:
283: return 0;
284: }
285:
286:
287: mbfl_convert_filter *
288: mbfl_convert_filter_new(
289: enum mbfl_no_encoding from,
290: enum mbfl_no_encoding to,
291: int (*output_function)(int, void* ),
292: int (*flush_function)(void*),
293: void* data)
294: {
295: mbfl_convert_filter * filter;
296: const struct mbfl_convert_vtbl *vtbl;
297:
298: vtbl = mbfl_convert_filter_get_vtbl(from, to);
299:
300: if (vtbl == NULL) {
301: vtbl = &vtbl_pass;
302: }
303:
304: /* allocate */
305: filter = (mbfl_convert_filter *)mbfl_malloc(sizeof(mbfl_convert_filter));
306: if (filter == NULL) {
307: return NULL;
308: }
309:
310: if (mbfl_convert_filter_common_init(filter, from, to, vtbl,
311: output_function, flush_function, data)) {
312: mbfl_free(filter);
313: return NULL;
314: }
315:
316: return filter;
317: }
318:
319: mbfl_convert_filter *
320: mbfl_convert_filter_new2(
321: const struct mbfl_convert_vtbl *vtbl,
322: int (*output_function)(int, void* ),
323: int (*flush_function)(void*),
324: void* data)
325: {
326: mbfl_convert_filter * filter;
327:
328: if (vtbl == NULL) {
329: vtbl = &vtbl_pass;
330: }
331:
332: /* allocate */
333: filter = (mbfl_convert_filter *)mbfl_malloc(sizeof(mbfl_convert_filter));
334: if (filter == NULL) {
335: return NULL;
336: }
337:
338: if (mbfl_convert_filter_common_init(filter, vtbl->from, vtbl->to, vtbl,
339: output_function, flush_function, data)) {
340: mbfl_free(filter);
341: return NULL;
342: }
343:
344: return filter;
345: }
346:
347: void
348: mbfl_convert_filter_delete(mbfl_convert_filter *filter)
349: {
350: if (filter) {
351: (*filter->filter_dtor)(filter);
352: mbfl_free((void*)filter);
353: }
354: }
355:
356: int
357: mbfl_convert_filter_feed(int c, mbfl_convert_filter *filter)
358: {
359: return (*filter->filter_function)(c, filter);
360: }
361:
362: int
363: mbfl_convert_filter_flush(mbfl_convert_filter *filter)
364: {
365: (*filter->filter_flush)(filter);
366: return (filter->flush_function ? (*filter->flush_function)(filter->data) : 0);
367: }
368:
369: void mbfl_convert_filter_reset(mbfl_convert_filter *filter,
370: enum mbfl_no_encoding from, enum mbfl_no_encoding to)
371: {
372: const struct mbfl_convert_vtbl *vtbl;
373:
374: /* destruct old filter */
375: (*filter->filter_dtor)(filter);
376:
377: vtbl = mbfl_convert_filter_get_vtbl(from, to);
378:
379: if (vtbl == NULL) {
380: vtbl = &vtbl_pass;
381: }
382:
383: mbfl_convert_filter_common_init(filter, from, to, vtbl,
384: filter->output_function, filter->flush_function, filter->data);
385: }
386:
387: void
388: mbfl_convert_filter_copy(
389: mbfl_convert_filter *src,
390: mbfl_convert_filter *dest)
391: {
392: if (src->filter_copy != NULL) {
393: src->filter_copy(src, dest);
394: return;
395: }
396:
397: *dest = *src;
398: }
399:
400: int mbfl_convert_filter_devcat(mbfl_convert_filter *filter, mbfl_memory_device *src)
401: {
402: int n;
403: unsigned char *p;
404:
405: p = src->buffer;
406: n = src->pos;
407: while (n > 0) {
408: if ((*filter->filter_function)(*p++, filter) < 0) {
409: return -1;
410: }
411: n--;
412: }
413:
414: return n;
415: }
416:
417: int mbfl_convert_filter_strcat(mbfl_convert_filter *filter, const unsigned char *p)
418: {
419: int c;
420:
421: while ((c = *p++) != '\0') {
422: if ((*filter->filter_function)(c, filter) < 0) {
423: return -1;
424: }
425: }
426:
427: return 0;
428: }
429:
430: /* illegal character output function for conv-filter */
431: int
432: mbfl_filt_conv_illegal_output(int c, mbfl_convert_filter *filter)
433: {
434: int mode_backup, ret, n, m, r;
435:
436: ret = 0;
437: mode_backup = filter->illegal_mode;
438: filter->illegal_mode = MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE;
439: switch (mode_backup) {
440: case MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR:
441: ret = (*filter->filter_function)(filter->illegal_substchar, filter);
442: break;
443: case MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG:
444: if (c >= 0) {
445: if (c < MBFL_WCSGROUP_UCS4MAX) { /* unicode */
446: ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)"U+");
447: } else {
448: if (c < MBFL_WCSGROUP_WCHARMAX) {
449: m = c & ~MBFL_WCSPLANE_MASK;
450: switch (m) {
451: case MBFL_WCSPLANE_JIS0208:
452: ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)"JIS+");
453: break;
454: case MBFL_WCSPLANE_JIS0212:
455: ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)"JIS2+");
456: break;
457: case MBFL_WCSPLANE_WINCP932:
458: ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)"W932+");
459: break;
460: case MBFL_WCSPLANE_8859_1:
461: ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)"I8859_1+");
462: break;
463: default:
464: ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)"?+");
465: break;
466: }
467: c &= MBFL_WCSPLANE_MASK;
468: } else {
469: ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)"BAD+");
470: c &= MBFL_WCSGROUP_MASK;
471: }
472: }
473: if (ret >= 0) {
474: m = 0;
475: r = 28;
476: while (r >= 0) {
477: n = (c >> r) & 0xf;
478: if (n || m) {
479: m = 1;
480: ret = (*filter->filter_function)(mbfl_hexchar_table[n], filter);
481: if (ret < 0) {
482: break;
483: }
484: }
485: r -= 4;
486: }
487: if (m == 0 && ret >= 0) {
488: ret = (*filter->filter_function)(mbfl_hexchar_table[0], filter);
489: }
490: }
491: }
492: break;
493: case MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY:
494: if (c >= 0) {
495: if (c < MBFL_WCSGROUP_UCS4MAX) { /* unicode */
496: ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)"&#x");
497: if (ret < 0)
498: break;
499:
500: m = 0;
501: r = 28;
502: while (r >= 0) {
503: n = (c >> r) & 0xf;
504: if (n || m) {
505: m = 1;
506: ret = (*filter->filter_function)(mbfl_hexchar_table[n], filter);
507: if (ret < 0) {
508: break;
509: }
510: }
511: r -= 4;
512: }
513: if (ret < 0) {
514: break;
515: }
516: if (m == 0) {
517: ret = (*filter->filter_function)(mbfl_hexchar_table[0], filter);
518: }
519: ret = mbfl_convert_filter_strcat(filter, (const unsigned char *)";");
520: } else {
521: ret = (*filter->filter_function)(filter->illegal_substchar, filter);
522: }
523: }
524: break;
525: default:
526: break;
527: }
528: filter->illegal_mode = mode_backup;
529: filter->num_illegalchar++;
530:
531: return ret;
532: }
533:
534: const struct mbfl_convert_vtbl * mbfl_convert_filter_get_vtbl(enum mbfl_no_encoding from, enum mbfl_no_encoding to)
535: {
536: const struct mbfl_convert_vtbl *vtbl;
537: int i;
538:
539: if (to == mbfl_no_encoding_base64 ||
540: to == mbfl_no_encoding_qprint ||
541: to == mbfl_no_encoding_7bit) {
542: from = mbfl_no_encoding_8bit;
543: } else if (from == mbfl_no_encoding_base64 ||
544: from == mbfl_no_encoding_qprint ||
545: from == mbfl_no_encoding_uuencode) {
546: to = mbfl_no_encoding_8bit;
547: }
548:
549: i = 0;
550: while ((vtbl = mbfl_convert_filter_list[i++]) != NULL){
551: if (vtbl->from == from && vtbl->to == to) {
552: return vtbl;
553: }
554: }
555:
556: return NULL;
557: }
558:
559: /*
560: * commonly used constructor and destructor
561: */
562: void mbfl_filt_conv_common_ctor(mbfl_convert_filter *filter)
563: {
564: filter->status = 0;
565: filter->cache = 0;
566: }
567:
568: int mbfl_filt_conv_common_flush(mbfl_convert_filter *filter)
569: {
570: filter->status = 0;
571: filter->cache = 0;
572:
573: if (filter->flush_function != NULL) {
574: (*filter->flush_function)(filter->data);
575: }
576: return 0;
577: }
578:
579: void mbfl_filt_conv_common_dtor(mbfl_convert_filter *filter)
580: {
581: filter->status = 0;
582: filter->cache = 0;
583: }
584:
585:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>