Annotation of libwww/Library/src/HTMIME.c, revision 2.59
2.15 frystyk 1: /* HTMIME.c
2: ** MIME MESSAGE PARSE
3: **
2.22 frystyk 4: ** (c) COPYRIGHT MIT 1995.
2.15 frystyk 5: ** Please first read the full copyright statement in the file COPYRIGH.
2.1 timbl 6: **
7: ** This is RFC 1341-specific code.
8: ** The input stream pushed into this parser is assumed to be
9: ** stripped on CRs, ie lines end with LF, not CR LF.
10: ** (It is easy to change this except for the body part where
11: ** conversion can be slow.)
12: **
13: ** History:
14: ** Feb 92 Written Tim Berners-Lee, CERN
2.13 duns 15: ** 8 Jul 94 FM Insulate free() from _free structure element.
2.18 frystyk 16: ** 14 Mar 95 HFN Now using anchor for storing data. No more `\n',
17: ** static buffers etc.
2.1 timbl 18: */
2.17 frystyk 19:
20: /* Library include files */
2.57 frystyk 21: #include "sysdep.h"
2.17 frystyk 22: #include "HTUtils.h"
23: #include "HTString.h"
2.9 luotonen 24: #include "HTFormat.h"
2.27 frystyk 25: #include "HTCache.h"
26: #include "HTAlert.h"
2.42 frystyk 27: #include "HTAncMan.h"
2.18 frystyk 28: #include "HTChunk.h"
2.26 frystyk 29: #include "HTMethod.h"
2.36 frystyk 30: #include "HTHeader.h"
2.24 frystyk 31: #include "HTSocket.h"
2.39 frystyk 32: #include "HTWWWStr.h"
2.17 frystyk 33: #include "HTFWrite.h"
2.32 frystyk 34: #include "HTNetMan.h"
2.31 frystyk 35: #include "HTReqMan.h"
2.14 frystyk 36: #include "HTMIME.h" /* Implemented here */
2.1 timbl 37:
38: /* MIME Object
39: ** -----------
40: */
41: typedef enum _MIME_state {
2.23 frystyk 42: BEGINNING_OF_LINE=0,
2.18 frystyk 43: CHECK, /* check against check_pointer */
44: UNKNOWN, /* Unknown header */
45: JUNK_LINE, /* Ignore rest of header */
46:
2.32 frystyk 47: CON, /* Intermediate states */
48: CONTENT,
2.45 frystyk 49: FIRSTLETTER_A,
2.18 frystyk 50: FIRSTLETTER_D,
51: FIRSTLETTER_L,
52: CONTENTLETTER_L,
53: CONTENTLETTER_T,
54:
2.45 frystyk 55: ACCEPT_TYPE, /* Headers supported */
56: ACCEPT_CHARSET,
57: ACCEPT_ENCODING,
58: ACCEPT_LANGUAGE,
59: ALLOW,
2.18 frystyk 60: AUTHENTICATE,
2.32 frystyk 61: CONNECTION,
2.18 frystyk 62: CONTENT_ENCODING,
63: CONTENT_LANGUAGE,
64: CONTENT_LENGTH,
2.14 frystyk 65: CONTENT_TRANSFER_ENCODING,
66: CONTENT_TYPE,
2.56 frystyk 67: MESSAGE_DIGEST,
2.23 frystyk 68: MIME_DATE,
2.18 frystyk 69: DERIVED_FROM,
70: EXPIRES,
71: LAST_MODIFIED,
72: LINK,
2.14 frystyk 73: LOCATION,
2.18 frystyk 74: PUBLIC_METHODS,
75: RETRY_AFTER,
76: TITLE,
77: URI_HEADER,
78: VERSION
2.1 timbl 79: } MIME_state;
80:
81: struct _HTStream {
2.57 frystyk 82: const HTStreamClass * isa;
2.18 frystyk 83: HTRequest * request;
2.32 frystyk 84: HTNet * net;
85: HTParentAnchor * anchor;
2.18 frystyk 86: HTStream * target;
87: HTFormat target_format;
88: HTChunk * buffer;
2.59 ! frystyk 89: HTEOLState EOLstate;
2.18 frystyk 90: BOOL transparent;
2.48 frystyk 91: BOOL head_only;
2.35 frystyk 92: BOOL nntp;
2.1 timbl 93: };
94:
2.18 frystyk 95: /* ------------------------------------------------------------------------- */
2.1 timbl 96:
2.18 frystyk 97: /*
2.1 timbl 98: ** This is a FSM parser which is tolerant as it can be of all
99: ** syntax errors. It ignores field names it does not understand,
100: ** and resynchronises on line beginnings.
101: */
2.36 frystyk 102: PRIVATE int parseheader (HTStream * me, HTRequest * request,
103: HTParentAnchor * anchor)
2.18 frystyk 104: {
105: MIME_state state = BEGINNING_OF_LINE;
106: MIME_state ok_state; /* got this state if match */
107: char *ptr = me->buffer->data-1; /* We dont change the data in length */
108: char *stop = ptr+me->buffer->size; /* When to stop */
109: char *header = ptr; /* For diagnostics */
2.57 frystyk 110: const char * check_pointer; /* checking input */
2.18 frystyk 111: char *value;
2.27 frystyk 112:
113: /* In case we get an empty header consisting of a CRLF, we fall thru */
2.18 frystyk 114: while (ptr < stop) {
115: switch (state) {
116: case BEGINNING_OF_LINE:
117: header = ++ptr;
118: switch (TOLOWER(*ptr)) {
2.50 frystyk 119: case '\0':
120: state = BEGINNING_OF_LINE; /* Empty line */
121: continue;
122:
2.18 frystyk 123: case 'a':
2.45 frystyk 124: state = FIRSTLETTER_A;
2.18 frystyk 125: break;
126:
127: case 'c':
2.32 frystyk 128: check_pointer = "on";
129: ok_state = CON;
2.18 frystyk 130: state = CHECK;
131: break;
132:
133: case 'd':
134: state = FIRSTLETTER_D;
135: break;
136:
137: case 'e':
138: check_pointer = "xpires";
139: ok_state = EXPIRES;
140: state = CHECK;
141: break;
142:
2.32 frystyk 143: case 'k':
2.33 frystyk 144: check_pointer = "eep-alive";
145: ok_state = JUNK_LINE; /* We don't use this but recognize it */
146: state = CHECK;
2.32 frystyk 147: break;
148:
2.18 frystyk 149: case 'l':
150: state = FIRSTLETTER_L;
151: break;
152:
153: case 'm':
154: check_pointer = "ime-version";
155: ok_state = JUNK_LINE; /* We don't use this but recognize it */
156: state = CHECK;
157: break;
158:
2.35 frystyk 159: case 'n':
160: check_pointer = "ewsgroups";
161: me->nntp = YES; /* Due to news brain damage */
162: ok_state = JUNK_LINE; /* We don't use this but recognize it */
163: state = CHECK;
164: break;
165:
2.18 frystyk 166: case 'r':
167: check_pointer = "etry-after";
168: ok_state = RETRY_AFTER;
169: state = CHECK;
170: break;
171:
172: case 's':
173: check_pointer = "erver";
174: ok_state = JUNK_LINE; /* We don't use this but recognize it */
175: state = CHECK;
176: break;
2.1 timbl 177:
2.18 frystyk 178: case 't':
179: check_pointer = "itle";
180: ok_state = TITLE;
181: state = CHECK;
182: break;
183:
184: case 'u':
185: check_pointer = "ri";
186: ok_state = URI_HEADER;
187: state = CHECK;
188: break;
189:
190: case 'v':
191: check_pointer = "ersion";
192: ok_state = VERSION;
193: state = CHECK;
194: break;
195:
196: case 'w':
197: check_pointer = "ww-authenticate";
198: ok_state = AUTHENTICATE;
199: state = CHECK;
200: break;
2.1 timbl 201:
2.18 frystyk 202: default:
203: state = UNKNOWN;
204: break;
205: }
206: ptr++;
2.1 timbl 207: break;
208:
2.45 frystyk 209: case FIRSTLETTER_A:
210: if (!strncasecomp(ptr, "llow", 4)) {
211: state = ALLOW;
212: ptr += 4;
213: } else if (!strncasecomp(ptr, "ccept-language", 14)) {
214: state = ACCEPT_LANGUAGE;
215: ptr += 14;
216: } else if (!strncasecomp(ptr, "ccept-charset", 13)) {
217: state = ACCEPT_CHARSET;
218: ptr += 13;
219: } else if (!strncasecomp(ptr, "ccept", 5)) {
220: state = ACCEPT_TYPE;
221: ptr += 5;
222: } else
223: state = UNKNOWN;
224: ptr++;
225: break;
226:
2.18 frystyk 227: case FIRSTLETTER_D:
228: switch (TOLOWER(*ptr)) {
229: case 'a':
230: check_pointer = "te";
2.23 frystyk 231: ok_state = MIME_DATE;
2.18 frystyk 232: state = CHECK;
233: break;
234:
235: case 'e':
236: check_pointer = "rived-from";
237: ok_state = DERIVED_FROM;
238: state = CHECK;
239: break;
240:
2.56 frystyk 241: case 'i':
242: check_pointer = "gest-MessageDigest";
243: ok_state = MESSAGE_DIGEST;
244: state = CHECK;
245: break;
246:
2.18 frystyk 247: default:
248: state = UNKNOWN;
249: break;
250: }
251: ptr++;
252: break;
253:
254: case FIRSTLETTER_L:
255: switch (TOLOWER(*ptr)) {
256: case 'a':
257: check_pointer = "st-modified";
258: ok_state = LAST_MODIFIED;
259: state = CHECK;
260: break;
261:
262: case 'i':
263: check_pointer = "nk";
264: ok_state = LINK;
265: state = CHECK;
266: break;
267:
268: case 'o':
269: check_pointer = "cation";
270: ok_state = LOCATION;
271: state = CHECK;
272: break;
273:
274: default:
275: state = UNKNOWN;
276: break;
277: }
278: ptr++;
279: break;
280:
2.32 frystyk 281: case CON:
282: switch (TOLOWER(*ptr)) {
283: case 'n':
284: check_pointer = "ection";
285: ok_state = CONNECTION;
286: state = CHECK;
287: break;
288:
289: case 't':
290: check_pointer = "ent-";
291: ok_state = CONTENT;
292: state = CHECK;
293: break;
294:
295: default:
296: state = UNKNOWN;
297: break;
298: }
299: ptr++;
300: break;
301:
2.18 frystyk 302: case CONTENT:
303: switch (TOLOWER(*ptr)) {
304: case 'e':
305: check_pointer = "ncoding";
306: ok_state = CONTENT_ENCODING;
307: state = CHECK;
308: break;
309:
310: case 'l':
311: state = CONTENTLETTER_L;
312: break;
313:
314: case 't':
315: state = CONTENTLETTER_T;
316: break;
317:
318: default:
319: state = UNKNOWN;
320: break;
321: }
322: ptr++;
2.1 timbl 323: break;
2.14 frystyk 324:
2.18 frystyk 325: case CONTENTLETTER_L:
326: switch (TOLOWER(*ptr)) {
327: case 'a':
328: check_pointer = "nguage";
329: ok_state = CONTENT_LANGUAGE;
330: state = CHECK;
331: break;
332:
333: case 'e':
334: check_pointer = "ngth";
335: ok_state = CONTENT_LENGTH;
336: state = CHECK;
337: break;
338:
339: default:
340: state = UNKNOWN;
341: break;
342: }
343: ptr++;
2.14 frystyk 344: break;
345:
2.18 frystyk 346: case CONTENTLETTER_T:
347: switch (TOLOWER(*ptr)) {
348: case 'r':
349: check_pointer = "ansfer-encoding";
350: ok_state = CONTENT_TRANSFER_ENCODING;
351: state = CHECK;
352: break;
353:
354: case 'y':
355: check_pointer = "pe";
356: ok_state = CONTENT_TYPE;
357: state = CHECK;
358: break;
359:
360: default:
361: state = UNKNOWN;
362: break;
363: }
364: ptr++;
2.14 frystyk 365: break;
366:
2.18 frystyk 367: case CHECK: /* Check against string */
2.45 frystyk 368: while (TOLOWER(*ptr) == *check_pointer) ptr++, check_pointer++;
369: if (!*check_pointer) {
2.18 frystyk 370: state = ok_state;
371: while (*ptr && (WHITE(*ptr) || *ptr==':')) /* Spool to value */
372: ptr++;
373: } else
374: state = UNKNOWN;
2.14 frystyk 375: break;
376:
2.45 frystyk 377: case ACCEPT_TYPE: /* @@@ */
378: state = JUNK_LINE;
379: break;
380:
381: case ACCEPT_CHARSET: /* @@@ */
382: state = JUNK_LINE;
383: break;
384:
385: case ACCEPT_ENCODING: /* @@@ */
386: state = JUNK_LINE;
387: break;
388:
389: case ACCEPT_LANGUAGE: /* @@@ */
390: state = JUNK_LINE;
391: break;
392:
393: case ALLOW:
2.20 frystyk 394: while ((value = HTNextField(&ptr)) != NULL) {
395: HTMethod new_method;
2.26 frystyk 396: /* We treat them as case-insensitive! */
2.20 frystyk 397: if ((new_method = HTMethod_enum(value)) != METHOD_INVALID)
398: anchor->methods += new_method;
2.1 timbl 399: }
2.18 frystyk 400: if (STREAM_TRACE)
2.55 eric 401: HTTrace("MIMEParser.. Methods allowed: %d\n",
2.18 frystyk 402: anchor->methods);
403: state = JUNK_LINE;
2.1 timbl 404: break;
2.18 frystyk 405:
406: case AUTHENTICATE:
2.56 frystyk 407: if (!request->challenge) request->challenge = HTAssocList_new();
408:
2.59 ! frystyk 409: StrAllocCopy(request->scheme, "basic"); /* @@@@@@@@@ */
2.20 frystyk 410:
2.56 frystyk 411: HTAssocList_add(request->challenge, "WWW-authenticate", ptr);
2.18 frystyk 412: state = JUNK_LINE;
413: break;
414:
2.32 frystyk 415: case CONNECTION:
416: if ((value = HTNextField(&ptr)) != NULL) {
417: if (!strcasecomp(value, "keep-alive")) {
418: if (STREAM_TRACE)
2.55 eric 419: HTTrace("MIMEParser.. Persistent Connection\n");
2.59 ! frystyk 420: HTNet_setPersistent(me->net, YES);
2.32 frystyk 421: }
422: }
423: state = JUNK_LINE;
424: break;
425:
2.18 frystyk 426: case CONTENT_ENCODING:
427: if ((value = HTNextField(&ptr)) != NULL) {
428: char *lc = value;
2.20 frystyk 429: while ((*lc = TOLOWER(*lc))) lc++;
2.18 frystyk 430: anchor->content_encoding = HTAtom_for(value);
431: }
432: state = JUNK_LINE;
433: break;
434:
2.21 frystyk 435: case CONTENT_LANGUAGE: /* @@@ SHOULD BE A LIST @@@ */
436: if ((value = HTNextField(&ptr)) != NULL) {
437: char *lc = value;
438: while ((*lc = TOLOWER(*lc))) lc++;
439: anchor->content_language = HTAtom_for(value);
440: }
441: state = JUNK_LINE;
2.18 frystyk 442: break;
443:
444: case CONTENT_LENGTH:
445: if ((value = HTNextField(&ptr)) != NULL)
446: anchor->content_length = atol(value);
447: state = JUNK_LINE;
448: break;
449:
450: case CONTENT_TRANSFER_ENCODING:
451: if ((value = HTNextField(&ptr)) != NULL) {
452: char *lc = value;
2.20 frystyk 453: while ((*lc = TOLOWER(*lc))) lc++;
2.18 frystyk 454: anchor->cte = HTAtom_for(value);
455: }
456: state = JUNK_LINE;
457: break;
458:
459: case CONTENT_TYPE:
460: if ((value = HTNextField(&ptr)) != NULL) {
461: char *lc = value;
462: while ((*lc = TOLOWER(*lc))) lc++;
463: anchor->content_type = HTAtom_for(value);
2.38 frystyk 464: while ((value = HTNextField(&ptr)) != NULL) {
2.20 frystyk 465: if (!strcasecomp(value, "charset")) {
466: if ((value = HTNextField(&ptr)) != NULL) {
467: lc = value;
468: while ((*lc = TOLOWER(*lc))) lc++;
469: anchor->charset = HTAtom_for(value);
470: }
2.38 frystyk 471: } else if (!strcasecomp(value, "level")) {
2.20 frystyk 472: if ((value = HTNextField(&ptr)) != NULL) {
473: lc = value;
474: while ((*lc = TOLOWER(*lc))) lc++;
475: anchor->level = HTAtom_for(value);
476: }
2.38 frystyk 477: } else if (!strcasecomp(value, "boundary")) {
478: if ((value = HTNextField(&ptr)) != NULL) {
479: StrAllocCopy(request->boundary, value);
480: }
2.20 frystyk 481: }
482: }
2.1 timbl 483: }
2.20 frystyk 484: state = JUNK_LINE;
2.18 frystyk 485: break;
486:
2.56 frystyk 487: case MESSAGE_DIGEST:
488: if (!request->challenge) request->challenge = HTAssocList_new();
489: HTAssocList_add(request->challenge, "Digest-MessageDigest", ptr);
490: state = JUNK_LINE;
491: break;
492:
2.23 frystyk 493: case MIME_DATE:
2.18 frystyk 494: anchor->date = HTParseTime(ptr);
495: state = JUNK_LINE;
496: break;
497:
498: case DERIVED_FROM:
499: if ((value = HTNextField(&ptr)) != NULL)
500: StrAllocCopy(anchor->derived_from, value);
501: state = JUNK_LINE;
502: break;
503:
504: case EXPIRES:
505: anchor->expires = HTParseTime(ptr);
506: state = JUNK_LINE;
507: break;
508:
509: case LAST_MODIFIED:
510: anchor->last_modified = HTParseTime(ptr);
511: state = JUNK_LINE;
512: break;
513:
514: case LINK:
2.20 frystyk 515: state = UNKNOWN; /* @@@@@@@@@@@ */
2.18 frystyk 516: break;
517:
518: case LOCATION:
2.46 frystyk 519: request->redirectionAnchor = HTAnchor_findAddress(HTStrip(ptr));
2.18 frystyk 520: state = JUNK_LINE;
521: break;
522:
523: case PUBLIC_METHODS:
2.20 frystyk 524: state = UNKNOWN; /* @@@@@@@@@@@ */
2.18 frystyk 525: break;
526:
527: case RETRY_AFTER:
2.19 frystyk 528: request->retry_after = HTParseTime(ptr);
529: state = JUNK_LINE;
2.18 frystyk 530: break;
531:
532: case TITLE: /* Can't reuse buffer as HTML version might differ */
533: if ((value = HTNextField(&ptr)) != NULL)
534: StrAllocCopy(anchor->title, value);
535: state = JUNK_LINE;
536: break;
537:
538: case URI_HEADER:
539: state = LOCATION; /* @@@ Need extended parsing */
540: break;
541:
542: case VERSION:
543: if ((value = HTNextField(&ptr)) != NULL)
544: StrAllocCopy(anchor->version, value);
545: state = JUNK_LINE;
546: break;
547:
548: case UNKNOWN:
2.40 frystyk 549: {
2.36 frystyk 550: HTList * list;
551: HTParserCallback *cbf;
552: int status;
553: BOOL override;
554: if (STREAM_TRACE)
2.55 eric 555: HTTrace("MIMEParser.. Unknown `%s\'\n", header);
2.36 frystyk 556: if ((list = HTRequest_parser(request, &override)) &&
2.40 frystyk 557: (cbf = HTParser_find(list, header)) &&
2.58 eric 558: ((status = (*cbf)(request, header)) != HT_OK)) {
2.36 frystyk 559: return status;
560: } else if (!override &&
561: (list = HTHeader_parser()) &&
2.40 frystyk 562: (cbf = HTParser_find(list, header)) &&
2.58 eric 563: ((status = (*cbf)(request, header)) != HT_OK)) {
2.36 frystyk 564: return status;
565: }
566: }
2.18 frystyk 567:
568: case JUNK_LINE:
569: while (*ptr) ptr++;
570: state = BEGINNING_OF_LINE;
571: break;
2.1 timbl 572: }
2.18 frystyk 573: }
2.48 frystyk 574: me->transparent = YES; /* Pump rest of data right through */
2.50 frystyk 575: #if 0
576: HTChunk_clear(me->buffer); /* Get ready for next header */
577: #endif
2.27 frystyk 578:
2.48 frystyk 579: /* If this request us a source in PostWeb then pause here */
580: if (me->head_only || HTRequest_isSource(request)) return HT_PAUSE;
2.47 frystyk 581:
2.48 frystyk 582: /* If HEAD method then we just stop here */
2.47 frystyk 583: if (request->method == METHOD_HEAD) return HT_LOADED;
2.43 frystyk 584:
2.35 frystyk 585: /* News server almost never send content type or content length */
586: if (anchor->content_type != WWW_UNKNOWN || me->nntp) {
2.55 eric 587: if (STREAM_TRACE) HTTrace("MIMEParser.. Convert %s to %s\n",
2.48 frystyk 588: HTAtom_name(anchor->content_type),
589: HTAtom_name(me->target_format));
590: me->target = HTStreamStack(anchor->content_type, me->target_format,
591: me->target, request, YES);
2.18 frystyk 592: }
2.49 frystyk 593: if (!me->target) me->target = HTErrorStream();
2.27 frystyk 594: return HT_OK;
2.1 timbl 595: }
596:
597:
2.18 frystyk 598: /*
599: ** Header is terminated by CRCR, LFLF, CRLFLF, CRLFCRLF
600: ** Folding is either of CF LWS, LF LWS, CRLF LWS
601: */
2.57 frystyk 602: PRIVATE int HTMIME_put_block (HTStream * me, const char * b, int l)
2.18 frystyk 603: {
2.57 frystyk 604: const char * start = b;
605: const char * end = start;
2.18 frystyk 606: while (!me->transparent && l-- > 0) {
607: if (me->EOLstate == EOL_FCR) {
2.27 frystyk 608: if (*b == CR) { /* End of header */
2.53 frystyk 609: int status;
610: HTChunk_putb(me->buffer, start, end-start);
611: status = parseheader(me, me->request, me->anchor);
2.46 frystyk 612: HTNet_setBytesRead(me->net, l);
2.27 frystyk 613: if (status != HT_OK)
614: return status;
615: } else if (*b == LF) /* CRLF */
2.18 frystyk 616: me->EOLstate = EOL_FLF;
617: else if (WHITE(*b)) { /* Folding: CR SP */
618: me->EOLstate = EOL_BEGIN;
2.53 frystyk 619: HTChunk_putb(me->buffer, start, end-start);
2.44 frystyk 620: HTChunk_putc(me->buffer, ' ');
2.53 frystyk 621: start=b, end=b+1;
2.18 frystyk 622: } else { /* New line */
623: me->EOLstate = EOL_BEGIN;
2.53 frystyk 624: HTChunk_putb(me->buffer, start, end-start);
2.44 frystyk 625: HTChunk_putc(me->buffer, '\0');
2.53 frystyk 626: start=b, end=b+1;
2.18 frystyk 627: }
628: } else if (me->EOLstate == EOL_FLF) {
629: if (*b == CR) /* LF CR or CR LF CR */
630: me->EOLstate = EOL_SCR;
2.27 frystyk 631: else if (*b == LF) { /* End of header */
2.53 frystyk 632: int status;
633: HTChunk_putb(me->buffer, start, end-start);
2.54 frystyk 634: status = parseheader(me, me->request, me->anchor);
2.46 frystyk 635: HTNet_setBytesRead(me->net, l);
2.27 frystyk 636: if (status != HT_OK)
637: return status;
638: } else if (WHITE(*b)) { /* Folding: LF SP or CR LF SP */
2.18 frystyk 639: me->EOLstate = EOL_BEGIN;
2.53 frystyk 640: HTChunk_putb(me->buffer, start, end-start);
2.44 frystyk 641: HTChunk_putc(me->buffer, ' ');
2.53 frystyk 642: start=b, end=b+1;
2.18 frystyk 643: } else { /* New line */
644: me->EOLstate = EOL_BEGIN;
2.53 frystyk 645: HTChunk_putb(me->buffer, start, end-start);
2.44 frystyk 646: HTChunk_putc(me->buffer, '\0');
2.53 frystyk 647: start=b, end=b+1;
2.18 frystyk 648: }
649: } else if (me->EOLstate == EOL_SCR) {
2.27 frystyk 650: if (*b==CR || *b==LF) { /* End of header */
2.53 frystyk 651: int status;
652: HTChunk_putb(me->buffer, start, end-start);
653: status = parseheader(me, me->request, me->anchor);
2.46 frystyk 654: HTNet_setBytesRead(me->net, l);
2.27 frystyk 655: if (status != HT_OK)
656: return status;
657: } else if (WHITE(*b)) { /* Folding: LF CR SP or CR LF CR SP */
2.18 frystyk 658: me->EOLstate = EOL_BEGIN;
2.53 frystyk 659: HTChunk_putb(me->buffer, start, end-start);
2.44 frystyk 660: HTChunk_putc(me->buffer, ' ');
2.53 frystyk 661: start=b, end=b+1;
2.18 frystyk 662: } else { /* New line */
663: me->EOLstate = EOL_BEGIN;
2.53 frystyk 664: HTChunk_putb(me->buffer, start, end-start);
2.44 frystyk 665: HTChunk_putc(me->buffer, '\0');
2.53 frystyk 666: start=b, end=b+1;
2.18 frystyk 667: }
668: } else if (*b == CR) {
669: me->EOLstate = EOL_FCR;
670: } else if (*b == LF) {
671: me->EOLstate = EOL_FLF; /* Line found */
672: } else
2.53 frystyk 673: end++;
2.18 frystyk 674: b++;
675: }
2.32 frystyk 676:
677: /*
678: ** Put the rest down the stream without touching the data but make sure
679: ** that we get the correct content length of data
680: */
2.48 frystyk 681: if (me->transparent) {
2.47 frystyk 682: int status = (*me->target->isa->put_block)(me->target, b, l);
2.48 frystyk 683: if (status == HT_OK) {
2.47 frystyk 684: /* Check if CL at all - thanks to jwei@hal.com (John Wei) */
2.48 frystyk 685: long cl = HTAnchor_length(me->anchor);
686: return (cl>=0 && HTNet_bytesRead(me->net)>=cl) ? HT_LOADED : HT_OK;
687: } else
2.47 frystyk 688: return status;
2.48 frystyk 689: }
690: return HT_OK;
2.18 frystyk 691: }
692:
693:
694: /* Character handling
695: ** ------------------
696: */
2.36 frystyk 697: PRIVATE int HTMIME_put_character (HTStream * me, char c)
2.18 frystyk 698: {
699: return HTMIME_put_block(me, &c, 1);
700: }
701:
2.1 timbl 702:
703: /* String handling
704: ** ---------------
705: */
2.57 frystyk 706: PRIVATE int HTMIME_put_string (HTStream * me, const char * s)
2.1 timbl 707: {
2.18 frystyk 708: return HTMIME_put_block(me, s, (int) strlen(s));
2.1 timbl 709: }
710:
711:
2.18 frystyk 712: /* Flush an stream object
713: ** ---------------------
2.1 timbl 714: */
2.36 frystyk 715: PRIVATE int HTMIME_flush (HTStream * me)
2.1 timbl 716: {
2.47 frystyk 717: return me->target ? (*me->target->isa->flush)(me->target) : HT_OK;
2.1 timbl 718: }
719:
2.18 frystyk 720: /* Free a stream object
721: ** --------------------
2.1 timbl 722: */
2.36 frystyk 723: PRIVATE int HTMIME_free (HTStream * me)
2.1 timbl 724: {
2.18 frystyk 725: int status = HT_OK;
2.51 frystyk 726: if (!me->transparent) parseheader(me, me->request, me->anchor);
2.25 frystyk 727: if (me->target) {
728: if ((status = (*me->target->isa->_free)(me->target))==HT_WOULD_BLOCK)
729: return HT_WOULD_BLOCK;
730: }
2.26 frystyk 731: if (PROT_TRACE)
2.55 eric 732: HTTrace("MIME........ FREEING....\n");
2.44 frystyk 733: HTChunk_delete(me->buffer);
2.52 frystyk 734: HT_FREE(me);
2.18 frystyk 735: return status;
2.1 timbl 736: }
737:
738: /* End writing
739: */
2.38 frystyk 740: PRIVATE int HTMIME_abort (HTStream * me, HTList * e)
2.1 timbl 741: {
2.18 frystyk 742: int status = HT_ERROR;
2.41 frystyk 743: if (me->target) status = (*me->target->isa->abort)(me->target, e);
2.26 frystyk 744: if (PROT_TRACE)
2.55 eric 745: HTTrace("MIME........ ABORTING...\n");
2.44 frystyk 746: HTChunk_delete(me->buffer);
2.52 frystyk 747: HT_FREE(me);
2.18 frystyk 748: return status;
2.1 timbl 749: }
750:
751:
752:
753: /* Structured Object Class
754: ** -----------------------
755: */
2.57 frystyk 756: PRIVATE const HTStreamClass HTMIME =
2.1 timbl 757: {
758: "MIMEParser",
2.18 frystyk 759: HTMIME_flush,
2.1 timbl 760: HTMIME_free,
2.6 timbl 761: HTMIME_abort,
762: HTMIME_put_character,
763: HTMIME_put_string,
2.18 frystyk 764: HTMIME_put_block
2.1 timbl 765: };
766:
767:
2.48 frystyk 768: /* MIME header parser stream.
2.1 timbl 769: ** -------------------------
2.48 frystyk 770: ** This stream parses a complete MIME header and if a content type header
771: ** is found then the stream stack is called. Any left over data is pumped
772: ** right through the stream
2.1 timbl 773: */
2.36 frystyk 774: PUBLIC HTStream* HTMIMEConvert (HTRequest * request,
775: void * param,
776: HTFormat input_format,
777: HTFormat output_format,
778: HTStream * output_stream)
2.1 timbl 779: {
780: HTStream* me;
2.52 frystyk 781: if ((me = (HTStream *) HT_CALLOC(1, sizeof(* me))) == NULL)
782: HT_OUTOFMEM("HTMIMEConvert");
2.1 timbl 783: me->isa = &HTMIME;
2.18 frystyk 784: me->request = request;
2.32 frystyk 785: me->anchor = request->anchor;
786: me->net = request->net;
2.49 frystyk 787: me->target = output_stream;
2.18 frystyk 788: me->target_format = output_format;
2.44 frystyk 789: me->buffer = HTChunk_new(512);
2.18 frystyk 790: me->EOLstate = EOL_BEGIN;
2.1 timbl 791: return me;
792: }
2.32 frystyk 793:
2.48 frystyk 794: /* MIME header ONLY parser stream
795: ** ------------------------------
796: ** This stream parses a complete MIME header and then returnes HT_PAUSE.
797: ** It does not set up any streams and resting data stays in the buffer.
798: ** This can be used if you only want to parse the headers before you
799: ** decide what to do next. This is for example the case in a server app.
800: */
801: PUBLIC HTStream * HTMIMEHeader (HTRequest * request,
802: void * param,
803: HTFormat input_format,
804: HTFormat output_format,
805: HTStream * output_stream)
806: {
807: HTStream * me;
2.52 frystyk 808: if ((me = (HTStream *) HT_CALLOC(1, sizeof(HTStream))) == NULL)
809: HT_OUTOFMEM("HTMIMEConvert");
2.48 frystyk 810: me->isa = &HTMIME;
811: me->request = request;
812: me->anchor = request->anchor;
813: me->net = request->net;
2.49 frystyk 814: me->target = output_stream;
2.48 frystyk 815: me->target_format = output_format;
816: me->buffer = HTChunk_new(512);
817: me->EOLstate = EOL_BEGIN;
818: me->head_only = YES; /* We want to pause after header */
819: return me;
820: }
Webmaster