4 * This is the MIME parser for Citadel.
6 * Copyright (c) 1998-2006 by Art Cancro
7 * This code is distributed under the GNU General Public License v2.
15 #include <sys/types.h>
23 #include "serv_extensions.h"
24 #include "sysdep_decls.h"
27 #include "mime_parser.h"
30 void extract_key(char *target, char *source, char *key)
34 strcpy(target, source);
35 for (a = 0; a < strlen(target); ++a) {
36 if ((!strncasecmp(&target[a], key, strlen(key)))
37 && (target[a + strlen(key)] == '=')) {
38 strcpy(target, &target[a + strlen(key) + 1]);
40 strcpy(target, &target[1]);
41 for (b = 0; b < strlen(target); ++b)
52 * For non-multipart messages, we need to generate a quickie partnum of "1"
53 * to return to callback functions. Some callbacks demand it.
55 char *fixed_partnum(char *supplied_partnum) {
56 if (supplied_partnum == NULL) return "1";
57 if (strlen(supplied_partnum)==0) return "1";
58 return supplied_partnum;
64 * Convert "quoted-printable" to binary. Returns number of bytes decoded.
66 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
75 decoded[0] = encoded[0];
76 while (destpos <= sourcelen){
77 check = &decoded[destpos];
78 decoded[destpos] = encoded[sourcepos];
79 if ((ignore_last == 0) && (decoded[destpos-1] == '='))
81 if ((*check == '\0') ||
85 decoded[destpos - 1] = '\r';
86 decoded[destpos] = '\n';
88 else if (sourcelen - sourcepos > 2)
90 sscanf(&encoded[sourcepos], "%02x", &ch);
91 decoded[destpos - 1] = ch;
103 decoded[destpos] = 0;
109 * Given a message or message-part body and a length, handle any necessary
110 * decoding and pass the request up the stack.
112 void mime_decode(char *partnum,
113 char *part_start, size_t length,
114 char *content_type, char *charset, char *encoding,
116 char *name, char *filename,
128 void (*PreMultiPartCallBack)
139 void (*PostMultiPartCallBack)
156 size_t bytes_decoded = 0;
158 /* Some encodings aren't really encodings */
159 if (!strcasecmp(encoding, "7bit"))
160 strcpy(encoding, "");
161 if (!strcasecmp(encoding, "8bit"))
162 strcpy(encoding, "");
163 if (!strcasecmp(encoding, "binary"))
164 strcpy(encoding, "");
166 /* If this part is not encoded, send as-is */
167 if ( (strlen(encoding) == 0) || (dont_decode)) {
168 if (CallBack != NULL) {
169 CallBack(name, filename, fixed_partnum(partnum),
170 disposition, part_start,
171 content_type, charset, length, encoding, userdata);
176 /* Fail silently if we hit an unknown encoding. */
177 if ((strcasecmp(encoding, "base64"))
178 && (strcasecmp(encoding, "quoted-printable"))) {
183 * Allocate a buffer for the decoded data. The output buffer is slightly
184 * larger than the input buffer; this assumes that the decoded data
185 * will never be significantly larger than the encoded data. This is a
186 * safe assumption with base64, uuencode, and quoted-printable.
188 decoded = malloc(length + 32768);
189 if (decoded == NULL) {
193 if (!strcasecmp(encoding, "base64")) {
194 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
196 else if (!strcasecmp(encoding, "quoted-printable")) {
197 bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
200 if (bytes_decoded > 0) if (CallBack != NULL) {
201 CallBack(name, filename, fixed_partnum(partnum),
202 disposition, decoded,
203 content_type, charset, bytes_decoded, "binary", userdata);
210 * Break out the components of a multipart message
211 * (This function expects to be fed HEADERS + CONTENT)
212 * Note: NULL can be supplied as content_end; in this case, the message is
213 * considered to have ended when the parser encounters a 0x00 byte.
215 void the_mime_parser(char *partnum,
216 char *content_start, char *content_end,
228 void (*PreMultiPartCallBack)
239 void (*PostMultiPartCallBack)
257 char *part_start, *part_end = NULL;
262 size_t startary_len = 0;
267 size_t content_length;
271 char *content_type_name;
272 char *content_disposition_name;
278 char nested_partnum[SIZ];
283 boundary = malloc(SIZ);
284 memset(boundary, 0, SIZ);
286 startary = malloc(SIZ);
287 memset(startary, 0, SIZ);
289 endary = malloc(SIZ);
290 memset(endary, 0, SIZ);
292 header = malloc(SIZ);
293 memset(header, 0, SIZ);
295 content_type = malloc(SIZ);
296 memset(content_type, 0, SIZ);
298 charset = malloc(SIZ);
299 memset(charset, 0, SIZ);
301 encoding = malloc(SIZ);
302 memset(encoding, 0, SIZ);
304 content_type_name = malloc(SIZ);
305 memset(content_type_name, 0, SIZ);
307 content_disposition_name = malloc(SIZ);
308 memset(content_disposition_name, 0, SIZ);
310 filename = malloc(SIZ);
311 memset(filename, 0, SIZ);
313 disposition = malloc(SIZ);
314 memset(disposition, 0, SIZ);
316 /* If the caller didn't supply an endpointer, generate one by measure */
317 if (content_end == NULL) {
318 content_end = &content_start[strlen(content_start)];
321 /* Learn interesting things from the headers */
324 ptr = memreadline(ptr, buf, SIZ);
325 if (ptr >= content_end) {
329 for (i = 0; i < strlen(buf); ++i) {
330 if (isspace(buf[i])) {
335 if (!isspace(buf[0])) {
336 if (!strncasecmp(header, "Content-type: ", 14)) {
337 strcpy(content_type, &header[14]);
338 extract_key(content_type_name, content_type, "name");
339 extract_key(charset, content_type, "charset");
340 /* Deal with weird headers */
341 if (strchr(content_type, ' '))
342 *(strchr(content_type, ' ')) = '\0';
343 if (strchr(content_type, ';'))
344 *(strchr(content_type, ';')) = '\0';
346 if (!strncasecmp(header, "Content-Disposition: ", 21)) {
347 strcpy(disposition, &header[21]);
348 extract_key(content_disposition_name, disposition, "name");
349 extract_key(filename, disposition, "filename");
351 if (!strncasecmp(header, "Content-length: ", 16)) {
352 content_length = (size_t) atol(&header[16]);
354 if (!strncasecmp(header,
355 "Content-transfer-encoding: ", 27))
356 strcpy(encoding, &header[27]);
357 if (strlen(boundary) == 0)
358 extract_key(boundary, header, "boundary");
361 if ((strlen(header) + strlen(buf) + 2) < SIZ)
363 } while ((strlen(buf) > 0) && (*ptr != 0));
365 if (strchr(disposition, ';'))
366 *(strchr(disposition, ';')) = '\0';
367 striplt(disposition);
368 if (strchr(content_type, ';'))
369 *(strchr(content_type, ';')) = '\0';
370 striplt(content_type);
372 if (strlen(boundary) > 0) {
378 /* If this is a multipart message, then recursively process it */
382 /* Tell the client about this message's multipartedness */
383 if (PreMultiPartCallBack != NULL) {
384 PreMultiPartCallBack("", "", partnum, "",
385 NULL, content_type, charset,
386 0, encoding, userdata);
389 /* Figure out where the boundaries are */
390 snprintf(startary, SIZ, "--%s", boundary);
391 snprintf(endary, SIZ, "--%s--", boundary);
392 startary_len = strlen(startary);
396 next_boundary = NULL;
397 for (srch=ptr; srch<content_end; ++srch) {
398 if (!memcmp(srch, startary, startary_len)) {
399 next_boundary = srch;
404 if ( (part_start != NULL) && (next_boundary != NULL) ) {
405 part_end = next_boundary;
408 if (strlen(partnum) > 0) {
409 snprintf(nested_partnum,
410 sizeof nested_partnum,
415 snprintf(nested_partnum,
416 sizeof nested_partnum,
419 the_mime_parser(nested_partnum,
420 part_start, part_end,
422 PreMultiPartCallBack,
423 PostMultiPartCallBack,
428 if (next_boundary != NULL) {
429 /* If we pass out of scope, don't attempt to
430 * read past the end boundary. */
431 if (!strcmp(next_boundary, endary)) {
435 /* Set up for the next part. */
436 part_start = strstr(next_boundary, "\n");
442 /* Invalid end of multipart. Bail out! */
445 } while ( (ptr < content_end) && (next_boundary != NULL) );
447 if (PostMultiPartCallBack != NULL) {
448 PostMultiPartCallBack("", "", partnum, "", NULL,
449 content_type, charset, 0, encoding, userdata);
454 /* If it's not a multipart message, then do something with it */
458 while (ptr < content_end) {
462 part_end = content_end;
463 /* fix an off-by-one error */
467 /* Truncate if the header told us to */
468 if ( (content_length > 0) && (length > content_length) ) {
469 length = content_length;
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.
476 if (strlen(content_disposition_name) > strlen(content_type_name)) {
477 name = content_disposition_name;
480 name = content_type_name;
483 /* lprintf(CTDL_DEBUG, "mime_decode part=%s, len=%d, type=%s, charset=%s, encoding=%s\n",
484 partnum, length, content_type, charset, encoding); */
486 /* Ok, we've got a non-multipart part here, so do something with it.
490 content_type, charset, encoding, disposition,
492 CallBack, NULL, NULL,
493 userdata, dont_decode
497 * Now if it's an encapsulated message/rfc822 then we have to recurse into it
499 if (!strcasecmp(content_type, "message/rfc822")) {
501 if (PreMultiPartCallBack != NULL) {
502 PreMultiPartCallBack("", "", partnum, "",
503 NULL, content_type, charset,
504 0, encoding, userdata);
506 if (CallBack != NULL) {
507 if (strlen(partnum) > 0) {
508 snprintf(nested_partnum,
509 sizeof nested_partnum,
514 snprintf(nested_partnum,
515 sizeof nested_partnum,
518 the_mime_parser(nested_partnum,
519 part_start, part_end,
521 PreMultiPartCallBack,
522 PostMultiPartCallBack,
527 if (PostMultiPartCallBack != NULL) {
528 PostMultiPartCallBack("", "", partnum, "", NULL,
529 content_type, charset, 0, encoding, userdata);
537 end_parser: /* free the buffers! end the oppression!! */
545 free(content_type_name);
546 free(content_disposition_name);
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.
559 void mime_parser(char *content_start,
574 void (*PreMultiPartCallBack)
586 void (*PostMultiPartCallBack)
603 the_mime_parser("", content_start, content_end,
605 PreMultiPartCallBack,
606 PostMultiPartCallBack,
607 userdata, dont_decode);