4 * This is the MIME parser for Citadel. Sometimes it actually works.
6 * Copyright (c) 1998-2003 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 *encoding,
132 char *name, char *filename,
143 void (*PreMultiPartCallBack)
153 void (*PostMultiPartCallBack)
169 size_t bytes_decoded = 0;
171 /* Some encodings aren't really encodings */
172 if (!strcasecmp(encoding, "7bit"))
173 strcpy(encoding, "");
174 if (!strcasecmp(encoding, "8bit"))
175 strcpy(encoding, "");
176 if (!strcasecmp(encoding, "binary"))
177 strcpy(encoding, "");
179 /* If this part is not encoded, send as-is */
180 if ( (strlen(encoding) == 0) || (dont_decode)) {
181 if (CallBack != NULL) {
182 CallBack(name, filename, fixed_partnum(partnum),
183 disposition, part_start,
184 content_type, length, encoding, userdata);
189 if ((strcasecmp(encoding, "base64"))
190 && (strcasecmp(encoding, "quoted-printable"))) {
194 * Allocate a buffer for the decoded data. The output buffer is the
195 * same size as the input buffer; this assumes that the decoded data
196 * will never be larger than the encoded data. This is a safe
197 * assumption with base64, uuencode, and quoted-printable.
199 decoded = mallok(length+2048);
200 if (decoded == NULL) {
204 if (!strcasecmp(encoding, "base64")) {
205 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
207 else if (!strcasecmp(encoding, "quoted-printable")) {
208 bytes_decoded = CtdlDecodeQuotedPrintable(decoded,
212 if (bytes_decoded > 0) if (CallBack != NULL) {
213 CallBack(name, filename, fixed_partnum(partnum),
214 disposition, decoded,
215 content_type, bytes_decoded, "binary", userdata);
222 * Break out the components of a multipart message
223 * (This function expects to be fed HEADERS + CONTENT)
224 * Note: NULL can be supplied as content_end; in this case, the message is
225 * considered to have ended when the parser encounters a 0x00 byte.
227 void the_mime_parser(char *partnum,
228 char *content_start, char *content_end,
239 void (*PreMultiPartCallBack)
249 void (*PostMultiPartCallBack)
265 char *part_start, *part_end = NULL;
272 size_t content_length;
276 char *content_type_name;
277 char *content_disposition_name;
283 char nested_partnum[SIZ];
288 boundary = mallok(SIZ);
289 memset(boundary, 0, SIZ);
291 startary = mallok(SIZ);
292 memset(startary, 0, SIZ);
294 endary = mallok(SIZ);
295 memset(endary, 0, SIZ);
297 header = mallok(SIZ);
298 memset(header, 0, SIZ);
300 content_type = mallok(SIZ);
301 memset(content_type, 0, SIZ);
303 encoding = mallok(SIZ);
304 memset(encoding, 0, SIZ);
306 content_type_name = mallok(SIZ);
307 memset(content_type_name, 0, SIZ);
309 content_disposition_name = mallok(SIZ);
310 memset(content_disposition_name, 0, SIZ);
312 filename = mallok(SIZ);
313 memset(filename, 0, SIZ);
315 disposition = mallok(SIZ);
316 memset(disposition, 0, SIZ);
318 /* If the caller didn't supply an endpointer, generate one by measure */
319 if (content_end == NULL) {
320 content_end = &content_start[strlen(content_start)];
323 /* Learn interesting things from the headers */
326 ptr = memreadline(ptr, buf, SIZ);
327 if (ptr >= content_end) {
331 for (i = 0; i < strlen(buf); ++i)
334 if (!isspace(buf[0])) {
335 if (!strncasecmp(header, "Content-type: ", 14)) {
336 strcpy(content_type, &header[14]);
337 extract_key(content_type_name, content_type, "name");
338 /* Deal with weird headers */
339 if (strchr(content_type, ' '))
340 *(strchr(content_type, ' ')) = '\0';
341 if (strchr(content_type, ';'))
342 *(strchr(content_type, ';')) = '\0';
344 if (!strncasecmp(header, "Content-Disposition: ", 21)) {
345 strcpy(disposition, &header[21]);
346 extract_key(content_disposition_name, disposition, "name");
347 extract_key(filename, disposition, "filename");
349 if (!strncasecmp(header, "Content-length: ", 16)) {
350 content_length = (size_t) atol(&header[16]);
352 if (!strncasecmp(header,
353 "Content-transfer-encoding: ", 27))
354 strcpy(encoding, &header[27]);
355 if (strlen(boundary) == 0)
356 extract_key(boundary, header, "boundary");
359 if ((strlen(header) + strlen(buf) + 2) < SIZ)
361 } while ((strlen(buf) > 0) && (*ptr != 0));
363 if (strchr(disposition, ';'))
364 *(strchr(disposition, ';')) = '\0';
365 striplt(disposition);
366 if (strchr(content_type, ';'))
367 *(strchr(content_type, ';')) = '\0';
368 striplt(content_type);
370 if (strlen(boundary) > 0) {
376 /* If this is a multipart message, then recursively process it */
380 /* Tell the client about this message's multipartedness */
381 if (PreMultiPartCallBack != NULL) {
382 PreMultiPartCallBack("", "", partnum, "",
384 0, encoding, userdata);
387 /* Figure out where the boundaries are */
388 snprintf(startary, SIZ, "--%s", boundary);
389 snprintf(endary, SIZ, "--%s--", boundary);
391 if ( (!strncasecmp(ptr, startary, strlen(startary)))
392 || (!strncasecmp(ptr, endary, strlen(endary))) ) {
393 if (part_start != NULL) {
394 if (strlen(partnum) > 0) {
395 snprintf(nested_partnum,
396 sizeof nested_partnum,
401 snprintf(nested_partnum,
402 sizeof nested_partnum,
405 the_mime_parser(nested_partnum,
406 part_start, part_end,
408 PreMultiPartCallBack,
409 PostMultiPartCallBack,
413 ptr = memreadline(ptr, buf, SIZ);
420 /* If we pass out of scope in the MIME multipart (by
421 * hitting the end boundary), force the pointer out
422 * of scope so this loop ends.
424 if (ptr < content_end) {
425 if (!strcasecmp(ptr, endary)) {
429 } while (ptr <= content_end);
430 if (PostMultiPartCallBack != NULL) {
431 PostMultiPartCallBack("", "", partnum, "", NULL,
432 content_type, 0, encoding, userdata);
437 /* If it's not a multipart message, then do something with it */
441 while (ptr < content_end) {
445 part_end = content_end;
446 /* fix an off-by-one error */
450 /* Truncate if the header told us to */
451 if ( (content_length > 0) && (length > content_length) ) {
452 length = content_length;
455 /* Sometimes the "name" field is tacked on to Content-type,
456 * and sometimes it's tacked on to Content-disposition. Use
457 * whichever one we have.
459 if (strlen(content_disposition_name) > strlen(content_type_name)) {
460 name = content_disposition_name;
463 name = content_type_name;
468 content_type, encoding, disposition,
470 CallBack, NULL, NULL,
471 userdata, dont_decode);
474 end_parser: /* free the buffers! end the oppression!! */
481 phree(content_type_name);
482 phree(content_disposition_name);
490 * Entry point for the MIME parser.
491 * (This function expects to be fed HEADERS + CONTENT)
492 * Note: NULL can be supplied as content_end; in this case, the message is
493 * considered to have ended when the parser encounters a 0x00 byte.
495 void mime_parser(char *content_start,
509 void (*PreMultiPartCallBack)
520 void (*PostMultiPartCallBack)
536 the_mime_parser("", content_start, content_end,
538 PreMultiPartCallBack,
539 PostMultiPartCallBack,
540 userdata, dont_decode);