Brought over the updated MIME parser from Citadel.
[citadel.git] / webcit / mime_parser.c
1 /*
2  * $Id$
3  */
4 /**
5  * \defgroup MIME This is the MIME parser for Citadel.
6  *
7  * Copyright (c) 1998-2005 by Art Cancro
8  * This code is distributed under the terms of the GNU General Public License.
9  * \ingroup WebcitHttpServer
10  */
11 /*@{*/
12 #include "webcit.h"
13 #include "webserver.h"
14 #include "mime_parser.h"
15
16 void extract_key(char *target, char *source, char *key)
17 {
18         int a, b;
19
20         strcpy(target, source);
21         for (a = 0; a < strlen(target); ++a) {
22                 if ((!strncasecmp(&target[a], key, strlen(key)))
23                     && (target[a + strlen(key)] == '=')) {
24                         strcpy(target, &target[a + strlen(key) + 1]);
25                         if (target[0] == 34)
26                                 strcpy(target, &target[1]);
27                         for (b = 0; b < strlen(target); ++b)
28                                 if (target[b] == 34)
29                                         target[b] = 0;
30                         return;
31                 }
32         }
33         strcpy(target, "");
34 }
35
36
37 /*
38  * For non-multipart messages, we need to generate a quickie partnum of "1"
39  * to return to callback functions.  Some callbacks demand it.
40  */
41 char *fixed_partnum(char *supplied_partnum) {
42         if (supplied_partnum == NULL) return "1";
43         if (strlen(supplied_partnum)==0) return "1";
44         return supplied_partnum;
45 }
46
47
48
49 /*
50  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
51  * according to RFC2045 section 6.7
52  */
53 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
54         unsigned int ch;
55         int decoded_length = 0;
56         int pos = 0;
57
58         while (pos < sourcelen)
59         {
60                 if (!strncmp(&encoded[pos], "=\r\n", 3))
61                 {
62                         pos += 3;
63                 }
64                 else if (!strncmp(&encoded[pos], "=\n", 2))
65                 {
66                         pos += 2;
67                 }
68                 else if (encoded[pos] == '=')
69                 {
70                         ch = 0;
71                         sscanf(&encoded[pos+1], "%02x", &ch);
72                         pos += 3;
73                         decoded[decoded_length++] = ch;
74                 }
75                 else
76                 {
77                         decoded[decoded_length++] = encoded[pos];
78                         pos += 1;
79                 }
80         }
81         decoded[decoded_length] = 0;
82         return(decoded_length);
83 }
84
85
86 /*
87  * Given a message or message-part body and a length, handle any necessary
88  * decoding and pass the request up the stack.
89  */
90 void mime_decode(char *partnum,
91                  char *part_start, size_t length,
92                  char *content_type, char *charset, char *encoding,
93                  char *disposition,
94                  char *name, char *filename,
95                  void (*CallBack)
96                   (char *cbname,
97                    char *cbfilename,
98                    char *cbpartnum,
99                    char *cbdisp,
100                    void *cbcontent,
101                    char *cbtype,
102                    char *cbcharset,
103                    size_t cblength,
104                    char *cbencoding,
105                    void *cbuserdata),
106                  void (*PreMultiPartCallBack)
107                   (char *cbname,
108                    char *cbfilename,
109                    char *cbpartnum,
110                    char *cbdisp,
111                    void *cbcontent,
112                    char *cbtype,
113                    char *cbcharset,
114                    size_t cblength,
115                    char *cbencoding,
116                    void *cbuserdata),
117                  void (*PostMultiPartCallBack)
118                   (char *cbname,
119                    char *cbfilename,
120                    char *cbpartnum,
121                    char *cbdisp,
122                    void *cbcontent,
123                    char *cbtype,
124                    char *cbcharset,
125                    size_t cblength,
126                    char *cbencoding,
127                    void *cbuserdata),
128                   void *userdata,
129                   int dont_decode
130 )
131 {
132
133         char *decoded;
134         size_t bytes_decoded = 0;
135
136         /* Some encodings aren't really encodings */
137         if (!strcasecmp(encoding, "7bit"))
138                 strcpy(encoding, "");
139         if (!strcasecmp(encoding, "8bit"))
140                 strcpy(encoding, "");
141         if (!strcasecmp(encoding, "binary"))
142                 strcpy(encoding, "");
143
144         /* If this part is not encoded, send as-is */
145         if ( (strlen(encoding) == 0) || (dont_decode)) {
146                 if (CallBack != NULL) {
147                         CallBack(name, filename, fixed_partnum(partnum),
148                                 disposition, part_start,
149                                 content_type, charset, length, encoding, userdata);
150                         }
151                 return;
152         }
153         
154         /* Fail silently if we hit an unknown encoding. */
155         if ((strcasecmp(encoding, "base64"))
156             && (strcasecmp(encoding, "quoted-printable"))) {
157                 return;
158         }
159
160         /*
161          * Allocate a buffer for the decoded data.  The output buffer is slightly
162          * larger than the input buffer; this assumes that the decoded data
163          * will never be significantly larger than the encoded data.  This is a
164          * safe assumption with base64, uuencode, and quoted-printable.
165          */
166         decoded = malloc(length + 32768);
167         if (decoded == NULL) {
168                 return;
169         }
170
171         if (!strcasecmp(encoding, "base64")) {
172                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
173         }
174         else if (!strcasecmp(encoding, "quoted-printable")) {
175                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
176         }
177
178         if (bytes_decoded > 0) if (CallBack != NULL) {
179                 CallBack(name, filename, fixed_partnum(partnum),
180                         disposition, decoded,
181                         content_type, charset, bytes_decoded, "binary", userdata);
182         }
183
184         free(decoded);
185 }
186
187 /*
188  * Break out the components of a multipart message
189  * (This function expects to be fed HEADERS + CONTENT)
190  * Note: NULL can be supplied as content_end; in this case, the message is
191  * considered to have ended when the parser encounters a 0x00 byte.
192  */
193 void the_mime_parser(char *partnum,
194                      char *content_start, char *content_end,
195                      void (*CallBack)
196                       (char *cbname,
197                        char *cbfilename,
198                        char *cbpartnum,
199                        char *cbdisp,
200                        void *cbcontent,
201                        char *cbtype,
202                        char *cbcharset,
203                        size_t cblength,
204                        char *cbencoding,
205                        void *cbuserdata),
206                      void (*PreMultiPartCallBack)
207                       (char *cbname,
208                        char *cbfilename,
209                        char *cbpartnum,
210                        char *cbdisp,
211                        void *cbcontent,
212                        char *cbtype,
213                        char *cbcharset,
214                        size_t cblength,
215                        char *cbencoding,
216                        void *cbuserdata),
217                      void (*PostMultiPartCallBack)
218                       (char *cbname,
219                        char *cbfilename,
220                        char *cbpartnum,
221                        char *cbdisp,
222                        void *cbcontent,
223                        char *cbtype,
224                        char *cbcharset,
225                        size_t cblength,
226                        char *cbencoding,
227                        void *cbuserdata),
228                       void *userdata,
229                       int dont_decode
230 )
231 {
232
233         char *ptr;
234         char *srch = NULL;
235         char *part_start, *part_end = NULL;
236         char buf[SIZ];
237         char *header;
238         char *boundary;
239         char *startary;
240         size_t startary_len = 0;
241         char *endary;
242         char *next_boundary;
243         char *content_type;
244         char *charset;
245         size_t content_length;
246         char *encoding;
247         char *disposition;
248         char *name = NULL;
249         char *content_type_name;
250         char *content_disposition_name;
251         char *filename;
252         int is_multipart;
253         int part_seq = 0;
254         int i;
255         size_t length;
256         char nested_partnum[256];
257         int crlf_in_use = 0;
258         char *evaluate_crlf_ptr = NULL;
259
260         ptr = content_start;
261         content_length = 0;
262
263         boundary = malloc(SIZ);
264         memset(boundary, 0, SIZ);
265
266         startary = malloc(SIZ);
267         memset(startary, 0, SIZ);
268
269         endary = malloc(SIZ);
270         memset(endary, 0, SIZ);
271
272         header = malloc(SIZ);
273         memset(header, 0, SIZ);
274
275         content_type = malloc(SIZ);
276         memset(content_type, 0, SIZ);
277
278         charset = malloc(SIZ);
279         memset(charset, 0, SIZ);
280
281         encoding = malloc(SIZ);
282         memset(encoding, 0, SIZ);
283
284         content_type_name = malloc(SIZ);
285         memset(content_type_name, 0, SIZ);
286
287         content_disposition_name = malloc(SIZ);
288         memset(content_disposition_name, 0, SIZ);
289
290         filename = malloc(SIZ);
291         memset(filename, 0, SIZ);
292
293         disposition = malloc(SIZ);
294         memset(disposition, 0, SIZ);
295
296         /* If the caller didn't supply an endpointer, generate one by measure */
297         if (content_end == NULL) {
298                 content_end = &content_start[strlen(content_start)];
299         }
300
301         /* Learn interesting things from the headers */
302         strcpy(header, "");
303         do {
304                 ptr = memreadline(ptr, buf, SIZ);
305                 if (ptr >= content_end) {
306                         goto end_parser;
307                 }
308
309                 for (i = 0; i < strlen(buf); ++i) {
310                         if (isspace(buf[i])) {
311                                 buf[i] = ' ';
312                         }
313                 }
314
315                 if (!isspace(buf[0])) {
316                         if (!strncasecmp(header, "Content-type: ", 14)) {
317                                 strcpy(content_type, &header[14]);
318                                 extract_key(content_type_name, content_type, "name");
319                                 extract_key(charset, content_type, "charset");
320                                 /* Deal with weird headers */
321                                 if (strchr(content_type, ' '))
322                                         *(strchr(content_type, ' ')) = '\0';
323                                 if (strchr(content_type, ';'))
324                                         *(strchr(content_type, ';')) = '\0';
325                         }
326                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
327                                 strcpy(disposition, &header[21]);
328                                 extract_key(content_disposition_name, disposition, "name");
329                                 extract_key(filename, disposition, "filename");
330                         }
331                         if (!strncasecmp(header, "Content-length: ", 16)) {
332                                 content_length = (size_t) atol(&header[16]);
333                         }
334                         if (!strncasecmp(header,
335                                       "Content-transfer-encoding: ", 27))
336                                 strcpy(encoding, &header[27]);
337                         if (strlen(boundary) == 0)
338                                 extract_key(boundary, header, "boundary");
339                         strcpy(header, "");
340                 }
341                 if ((strlen(header) + strlen(buf) + 2) < SIZ)
342                         strcat(header, buf);
343         } while ((strlen(buf) > 0) && (*ptr != 0));
344
345         if (strchr(disposition, ';'))
346                 *(strchr(disposition, ';')) = '\0';
347         striplt(disposition);
348         if (strchr(content_type, ';'))
349                 *(strchr(content_type, ';')) = '\0';
350         striplt(content_type);
351
352         if (strlen(boundary) > 0) {
353                 is_multipart = 1;
354         } else {
355                 is_multipart = 0;
356         }
357
358         /* If this is a multipart message, then recursively process it */
359         part_start = NULL;
360         if (is_multipart) {
361
362                 /* Tell the client about this message's multipartedness */
363                 if (PreMultiPartCallBack != NULL) {
364                         PreMultiPartCallBack("", "", partnum, "",
365                                 NULL, content_type, charset,
366                                 0, encoding, userdata);
367                 }
368
369                 /* Figure out where the boundaries are */
370                 snprintf(startary, SIZ, "--%s", boundary);
371                 snprintf(endary, SIZ, "--%s--", boundary);
372                 startary_len = strlen(startary);
373
374                 part_start = NULL;
375                 do {
376                         next_boundary = NULL;
377                         for (srch=ptr; srch<content_end; ++srch) {
378                                 if (!memcmp(srch, startary, startary_len)) {
379                                         next_boundary = srch;
380                                         srch = content_end;
381                                 }
382                         }
383
384                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
385                                 part_end = next_boundary;
386                                 --part_end;             /* omit the trailing LF */
387                                 if (crlf_in_use) {
388                                         --part_end;     /* omit the trailing CR */
389                                 }
390
391                                 if (strlen(partnum) > 0) {
392                                         snprintf(nested_partnum,
393                                                  sizeof nested_partnum,
394                                                  "%s.%d", partnum,
395                                                  ++part_seq);
396                                 }
397                                 else {
398                                         snprintf(nested_partnum,
399                                                  sizeof nested_partnum,
400                                                  "%d", ++part_seq);
401                                 }
402                                 the_mime_parser(nested_partnum,
403                                             part_start, part_end,
404                                                 CallBack,
405                                                 PreMultiPartCallBack,
406                                                 PostMultiPartCallBack,
407                                                 userdata,
408                                                 dont_decode);
409                         }
410
411                         if (next_boundary != NULL) {
412                                 /* If we pass out of scope, don't attempt to
413                                  * read past the end boundary. */
414                                 if (!strcmp(next_boundary, endary)) {
415                                         ptr = content_end;
416                                 }
417                                 else {
418                                         /* Set up for the next part. */
419                                         part_start = strstr(next_boundary, "\n");
420                                         
421                                         /* Determine whether newlines are LF or CRLF */
422                                         evaluate_crlf_ptr = part_start;
423                                         --evaluate_crlf_ptr;
424                                         if (!memcmp(evaluate_crlf_ptr, "\r\n", 2)) {
425                                                 crlf_in_use = 1;
426                                         }
427                                         else {
428                                                 crlf_in_use = 0;
429                                         }
430
431                                         /* Advance past the LF ... now we're in the next part */
432                                         ++part_start;
433                                         ptr = part_start;
434                                 }
435                         }
436                         else {
437                                 /* Invalid end of multipart.  Bail out! */
438                                 ptr = content_end;
439                         }
440                 } while ( (ptr < content_end) && (next_boundary != NULL) );
441
442                 if (PostMultiPartCallBack != NULL) {
443                         PostMultiPartCallBack("", "", partnum, "", NULL,
444                                 content_type, charset, 0, encoding, userdata);
445                 }
446                 goto end_parser;
447         }
448
449         /* If it's not a multipart message, then do something with it */
450         if (!is_multipart) {
451                 part_start = ptr;
452                 length = 0;
453                 while (ptr < content_end) {
454                         ++ptr;
455                         ++length;
456                 }
457                 part_end = content_end;
458
459                 /******
460                  * I thought there was an off-by-one error here, but there isn't.
461                  * This probably means that there's an off-by-one error somewhere
462                  * else ... or maybe only in certain messages?
463                 --part_end;
464                 --length;
465                 ******/
466                 
467                 /* Truncate if the header told us to */
468                 if ( (content_length > 0) && (length > content_length) ) {
469                         length = content_length;
470                 }
471
472                 /* Sometimes the "name" field is tacked on to Content-type,
473                  * and sometimes it's tacked on to Content-disposition.  Use
474                  * whichever one we have.
475                  */
476                 if (strlen(content_disposition_name) > strlen(content_type_name)) {
477                         name = content_disposition_name;
478                 }
479                 else {
480                         name = content_type_name;
481                 }
482         
483                 /* lprintf(CTDL_DEBUG, "mime_decode part=%s, len=%d, type=%s, charset=%s, encoding=%s\n",
484                         partnum, length, content_type, charset, encoding); */
485
486                 /* Ok, we've got a non-multipart part here, so do something with it.
487                  */
488                 mime_decode(partnum,
489                         part_start, length,
490                         content_type, charset, encoding, disposition,
491                         name, filename,
492                         CallBack, NULL, NULL,
493                         userdata, dont_decode
494                 );
495
496                 /*
497                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
498                  */
499                 if (!strcasecmp(content_type, "message/rfc822")) {
500
501                         if (PreMultiPartCallBack != NULL) {
502                                 PreMultiPartCallBack("", "", partnum, "",
503                                         NULL, content_type, charset,
504                                         0, encoding, userdata);
505                         }
506                         if (CallBack != NULL) {
507                                 if (strlen(partnum) > 0) {
508                                         snprintf(nested_partnum,
509                                                  sizeof nested_partnum,
510                                                  "%s.%d", partnum,
511                                                  ++part_seq);
512                                 }
513                                 else {
514                                         snprintf(nested_partnum,
515                                                  sizeof nested_partnum,
516                                                  "%d", ++part_seq);
517                                 }
518                                 the_mime_parser(nested_partnum,
519                                         part_start, part_end,
520                                         CallBack,
521                                         PreMultiPartCallBack,
522                                         PostMultiPartCallBack,
523                                         userdata,
524                                         dont_decode
525                                 );
526                         }
527                         if (PostMultiPartCallBack != NULL) {
528                                 PostMultiPartCallBack("", "", partnum, "", NULL,
529                                         content_type, charset, 0, encoding, userdata);
530                         }
531
532
533                 }
534
535         }
536
537 end_parser:     /* free the buffers!  end the oppression!! */
538         free(boundary);
539         free(startary);
540         free(endary);   
541         free(header);
542         free(content_type);
543         free(charset);
544         free(encoding);
545         free(content_type_name);
546         free(content_disposition_name);
547         free(filename);
548         free(disposition);
549 }
550
551
552
553 /*
554  * Entry point for the MIME parser.
555  * (This function expects to be fed HEADERS + CONTENT)
556  * Note: NULL can be supplied as content_end; in this case, the message is
557  * considered to have ended when the parser encounters a 0x00 byte.
558  */
559 void mime_parser(char *content_start,
560                 char *content_end,
561
562                  void (*CallBack)
563                   (char *cbname,
564                    char *cbfilename,
565                    char *cbpartnum,
566                    char *cbdisp,
567                    void *cbcontent,
568                    char *cbtype,
569                    char *cbcharset,
570                    size_t cblength,
571                    char *cbencoding,
572                    void *cbuserdata),
573
574                  void (*PreMultiPartCallBack)
575                   (char *cbname,
576                    char *cbfilename,
577                    char *cbpartnum,
578                    char *cbdisp,
579                    void *cbcontent,
580                    char *cbtype,
581                    char *cbcharset,
582                    size_t cblength,
583                    char *cbencoding,
584                    void *cbuserdata),
585
586                  void (*PostMultiPartCallBack)
587                   (char *cbname,
588                    char *cbfilename,
589                    char *cbpartnum,
590                    char *cbdisp,
591                    void *cbcontent,
592                    char *cbtype,
593                    char *cbcharset,
594                    size_t cblength,
595                    char *cbencoding,
596                    void *cbuserdata),
597
598                   void *userdata,
599                   int dont_decode
600 )
601 {
602
603         the_mime_parser("", content_start, content_end,
604                         CallBack,
605                         PreMultiPartCallBack,
606                         PostMultiPartCallBack,
607                         userdata, dont_decode);
608 }
609
610 /*@}*/