4 * This is the MIME parser for Citadel. Sometimes it actually works.
6 * Copyright (c) 1998-2005 by Art Cancro
7 * This code is distributed under the terms of the GNU General Public License.
16 #include <sys/types.h>
24 #include "mime_parser.h"
27 void extract_key(char *target, char *source, char *key)
31 strcpy(target, source);
32 for (a = 0; a < strlen(target); ++a) {
33 if ((!strncasecmp(&target[a], key, strlen(key)))
34 && (target[a + strlen(key)] == '=')) {
35 strcpy(target, &target[a + strlen(key) + 1]);
37 strcpy(target, &target[1]);
38 for (b = 0; b < strlen(target); ++b)
49 * For non-multipart messages, we need to generate a quickie partnum of "1"
50 * to return to callback functions. Some callbacks demand it.
52 char *fixed_partnum(char *supplied_partnum) {
53 if (supplied_partnum == NULL) return "1";
54 if (strlen(supplied_partnum)==0) return "1";
55 return supplied_partnum;
61 * Convert "quoted-printable" to binary. Returns number of bytes decoded.
63 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
66 int soft_line_break = 0;
68 int decoded_length = 0;
76 for (i = 0; i < sourcelen; ++i) {
78 buf[buf_length++] = encoded[i];
80 if ( (encoded[i] == '\n')
82 || (i == (sourcelen-1)) ) {
83 buf[buf_length++] = 0;
85 /*** begin -- process one line ***/
87 if (buf[strlen(buf)-1] == '\n') {
88 buf[strlen(buf)-1] = 0;
90 if (buf[strlen(buf)-1] == '\r') {
91 buf[strlen(buf)-1] = 0;
93 while (isspace(buf[strlen(buf)-1])) {
94 buf[strlen(buf)-1] = 0;
98 while (strlen(buf) > 0) {
99 if (!strcmp(buf, "=")) {
102 } else if ((strlen(buf)>=3) && (buf[0]=='=')) {
103 sscanf(&buf[1], "%02x", &ch);
104 decoded[decoded_length++] = ch;
105 strcpy(buf, &buf[3]);
107 decoded[decoded_length++] = buf[0];
108 strcpy(buf, &buf[1]);
111 if (soft_line_break == 0) {
112 decoded[decoded_length++] = '\r';
113 decoded[decoded_length++] = '\n';
116 /*** end -- process one line ***/
120 decoded[decoded_length++] = 0;
121 return(decoded_length);
125 * Given a message or message-part body and a length, handle any necessary
126 * decoding and pass the request up the stack.
128 void mime_decode(char *partnum,
129 char *part_start, size_t length,
130 char *content_type, char *charset, char *encoding,
132 char *name, char *filename,
144 void (*PreMultiPartCallBack)
155 void (*PostMultiPartCallBack)
172 size_t bytes_decoded = 0;
174 /* Some encodings aren't really encodings */
175 if (!strcasecmp(encoding, "7bit"))
176 strcpy(encoding, "");
177 if (!strcasecmp(encoding, "8bit"))
178 strcpy(encoding, "");
179 if (!strcasecmp(encoding, "binary"))
180 strcpy(encoding, "");
182 /* If this part is not encoded, send as-is */
183 if ( (strlen(encoding) == 0) || (dont_decode)) {
184 if (CallBack != NULL) {
185 CallBack(name, filename, fixed_partnum(partnum),
186 disposition, part_start,
187 content_type, charset, length, encoding, userdata);
192 if ((strcasecmp(encoding, "base64"))
193 && (strcasecmp(encoding, "quoted-printable"))) {
197 * Allocate a buffer for the decoded data. The output buffer is the
198 * same size as the input buffer; this assumes that the decoded data
199 * will never be larger than the encoded data. This is a safe
200 * assumption with base64, uuencode, and quoted-printable.
202 decoded = malloc(length+2048);
203 if (decoded == NULL) {
207 if (!strcasecmp(encoding, "base64")) {
208 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
210 else if (!strcasecmp(encoding, "quoted-printable")) {
211 bytes_decoded = CtdlDecodeQuotedPrintable(decoded,
215 if (bytes_decoded > 0) if (CallBack != NULL) {
216 CallBack(name, filename, fixed_partnum(partnum),
217 disposition, decoded,
218 content_type, charset, bytes_decoded, "binary", userdata);
225 * Break out the components of a multipart message
226 * (This function expects to be fed HEADERS + CONTENT)
227 * Note: NULL can be supplied as content_end; in this case, the message is
228 * considered to have ended when the parser encounters a 0x00 byte.
230 void the_mime_parser(char *partnum,
231 char *content_start, char *content_end,
243 void (*PreMultiPartCallBack)
254 void (*PostMultiPartCallBack)
271 char *part_start, *part_end = NULL;
280 size_t content_length;
284 char *content_type_name;
285 char *content_disposition_name;
291 char nested_partnum[SIZ];
296 boundary = malloc(SIZ);
297 memset(boundary, 0, SIZ);
299 startary = malloc(SIZ);
300 memset(startary, 0, SIZ);
302 endary = malloc(SIZ);
303 memset(endary, 0, SIZ);
305 header = malloc(SIZ);
306 memset(header, 0, SIZ);
308 content_type = malloc(SIZ);
309 memset(content_type, 0, SIZ);
311 charset = malloc(SIZ);
312 memset(charset, 0, SIZ);
314 encoding = malloc(SIZ);
315 memset(encoding, 0, SIZ);
317 content_type_name = malloc(SIZ);
318 memset(content_type_name, 0, SIZ);
320 content_disposition_name = malloc(SIZ);
321 memset(content_disposition_name, 0, SIZ);
323 filename = malloc(SIZ);
324 memset(filename, 0, SIZ);
326 disposition = malloc(SIZ);
327 memset(disposition, 0, SIZ);
329 /* If the caller didn't supply an endpointer, generate one by measure */
330 if (content_end == NULL) {
331 content_end = &content_start[strlen(content_start)];
334 /* Learn interesting things from the headers */
337 ptr = memreadline(ptr, buf, SIZ);
338 if (ptr >= content_end) {
342 for (i = 0; i < strlen(buf); ++i) {
343 if (isspace(buf[i])) {
348 if (!isspace(buf[0])) {
349 if (!strncasecmp(header, "Content-type: ", 14)) {
350 strcpy(content_type, &header[14]);
351 extract_key(content_type_name, content_type, "name");
352 extract_key(charset, content_type, "charset");
353 /* Deal with weird headers */
354 if (strchr(content_type, ' '))
355 *(strchr(content_type, ' ')) = '\0';
356 if (strchr(content_type, ';'))
357 *(strchr(content_type, ';')) = '\0';
359 if (!strncasecmp(header, "Content-Disposition: ", 21)) {
360 strcpy(disposition, &header[21]);
361 extract_key(content_disposition_name, disposition, "name");
362 extract_key(filename, disposition, "filename");
364 if (!strncasecmp(header, "Content-length: ", 16)) {
365 content_length = (size_t) atol(&header[16]);
367 if (!strncasecmp(header,
368 "Content-transfer-encoding: ", 27))
369 strcpy(encoding, &header[27]);
370 if (strlen(boundary) == 0)
371 extract_key(boundary, header, "boundary");
374 if ((strlen(header) + strlen(buf) + 2) < SIZ)
376 } while ((strlen(buf) > 0) && (*ptr != 0));
378 if (strchr(disposition, ';'))
379 *(strchr(disposition, ';')) = '\0';
380 striplt(disposition);
381 if (strchr(content_type, ';'))
382 *(strchr(content_type, ';')) = '\0';
383 striplt(content_type);
385 if (strlen(boundary) > 0) {
391 /* If this is a multipart message, then recursively process it */
395 /* Tell the client about this message's multipartedness */
396 if (PreMultiPartCallBack != NULL) {
397 PreMultiPartCallBack("", "", partnum, "",
398 NULL, content_type, charset,
399 0, encoding, userdata);
402 /* Figure out where the boundaries are */
403 snprintf(startary, SIZ, "--%s", boundary);
404 snprintf(endary, SIZ, "--%s--", boundary);
408 next_boundary = bmstrstr(ptr, startary, strncmp);
409 if ( (part_start != NULL) && (next_boundary != NULL) ) {
410 part_end = next_boundary;
413 if (strlen(partnum) > 0) {
414 snprintf(nested_partnum,
415 sizeof nested_partnum,
420 snprintf(nested_partnum,
421 sizeof nested_partnum,
424 the_mime_parser(nested_partnum,
425 part_start, part_end,
427 PreMultiPartCallBack,
428 PostMultiPartCallBack,
433 if (next_boundary != NULL) {
434 /* If we pass out of scope, don't attempt to read
435 * past the end boundary. */
436 if (!strcmp(next_boundary, endary)) {
440 /* Set up for the next part. */
441 part_start = strstr(next_boundary, "\n");
447 /* Invalid end of multipart. Bail out! */
450 } while ( (ptr < content_end) && (next_boundary != NULL) );
452 if (PostMultiPartCallBack != NULL) {
453 PostMultiPartCallBack("", "", partnum, "", NULL,
454 content_type, charset, 0, encoding, userdata);
459 /* If it's not a multipart message, then do something with it */
463 while (ptr < content_end) {
467 part_end = content_end;
468 /* fix an off-by-one error */
472 /* Truncate if the header told us to */
473 if ( (content_length > 0) && (length > content_length) ) {
474 length = content_length;
477 /* Sometimes the "name" field is tacked on to Content-type,
478 * and sometimes it's tacked on to Content-disposition. Use
479 * whichever one we have.
481 if (strlen(content_disposition_name) > strlen(content_type_name)) {
482 name = content_disposition_name;
485 name = content_type_name;
490 content_type, charset, encoding, disposition,
492 CallBack, NULL, NULL,
493 userdata, dont_decode);
496 end_parser: /* free the buffers! end the oppression!! */
504 free(content_type_name);
505 free(content_disposition_name);
513 * Entry point for the MIME parser.
514 * (This function expects to be fed HEADERS + CONTENT)
515 * Note: NULL can be supplied as content_end; in this case, the message is
516 * considered to have ended when the parser encounters a 0x00 byte.
518 void mime_parser(char *content_start,
533 void (*PreMultiPartCallBack)
545 void (*PostMultiPartCallBack)
562 the_mime_parser("", content_start, content_end,
564 PreMultiPartCallBack,
565 PostMultiPartCallBack,
566 userdata, dont_decode);