]> diplodocus.org Git - nmh/blob - uip/mhfixmsg.c
projects / nmh / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Alter HasSuffixC()'s char * to be const.
[nmh] / uip / mhfixmsg.c
1 /*
2 * mhfixmsg.c -- rewrite a message with various transformations
3 *
4 * This code is Copyright (c) 2002 and 2013, by the authors of nmh.
5 * See the COPYRIGHT file in the root directory of the nmh
6 * distribution for complete copyright information.
7 */
8
9 #include <h/mh.h>
10 #include <h/mime.h>
11 #include <h/mhparse.h>
12 #include <h/utils.h>
13 #include <h/signals.h>
14 #include <fcntl.h>
15
16 #define MHFIXMSG_SWITCHES \
17 X("decodetext 8bit|7bit|binary", 0, DECODETEXTSW) \
18 X("nodecodetext", 0, NDECODETEXTSW) \
19 X("decodetypes", 0, DECODETYPESW) \
20 X("crlflinebreaks", 0, CRLFLINEBREAKSSW) \
21 X("nocrlflinebreaks", 0, NCRLFLINEBREAKSSW) \
22 X("textcharset", 0, TEXTCHARSETSW) \
23 X("notextcharset", 0, NTEXTCHARSETSW) \
24 X("reformat", 0, REFORMATSW) \
25 X("noreformat", 0, NREFORMATSW) \
26 X("replacetextplain", 0, REPLACETEXTPLAINSW) \
27 X("noreplacetextplain", 0, NREPLACETEXTPLAINSW) \
28 X("fixboundary", 0, FIXBOUNDARYSW) \
29 X("nofixboundary", 0, NFIXBOUNDARYSW) \
30 X("fixcte", 0, FIXCOMPOSITECTESW) \
31 X("nofixcte", 0, NFIXCOMPOSITECTESW) \
32 X("fixtype mimetype", 0, FIXTYPESW) \
33 X("file file", 0, FILESW) \
34 X("outfile file", 0, OUTFILESW) \
35 X("rmmproc program", 0, RPROCSW) \
36 X("normmproc", 0, NRPRCSW) \
37 X("changecur", 0, CHGSW) \
38 X("nochangecur", 0, NCHGSW) \
39 X("verbose", 0, VERBSW) \
40 X("noverbose", 0, NVERBSW) \
41 X("version", 0, VERSIONSW) \
42 X("help", 0, HELPSW) \
43
44 #define X(sw, minchars, id) id,
45 DEFINE_SWITCH_ENUM(MHFIXMSG);
46 #undef X
47
48 #define X(sw, minchars, id) { sw, minchars, id },
49 DEFINE_SWITCH_ARRAY(MHFIXMSG, switches);
50 #undef X
51
52
53 int verbosw;
54 int debugsw; /* Needed by mhparse.c. */
55
56 #define quitser pipeser
57
58 /* mhparse.c */
59 extern int skip_mp_cte_check; /* flag to InitMultiPart */
60 extern int suppress_bogus_mp_content_warning; /* flag to InitMultiPart */
61 extern int bogus_mp_content; /* flag from InitMultiPart */
62 /* flags to/from parse_header_attrs */
63 extern int suppress_extraneous_trailing_semicolon_warning;
64 extern int extraneous_trailing_semicolon;
65
66 /* mhoutsbr.c */
67 int output_message (CT, char *);
68
69 /* mhmisc.c */
70 void flush_errors (void);
71
72 /* mhfree.c */
73 extern CT *cts;
74 void freects_done (int) NORETURN;
75
76 /*
77 * static prototypes
78 */
79 typedef struct fix_transformations {
80 int fixboundary;
81 int fixcompositecte;
82 svector_t fixtypes;
83 int reformat;
84 int replacetextplain;
85 int decodetext;
86 char *decodetypes;
87 /* Whether to use CRLF linebreaks, per RFC 2046 Sec. 4.1.1, par.1. */
88 int lf_line_endings;
89 char *textcharset;
90 } fix_transformations;
91
92 int mhfixmsgsbr (CT *, const fix_transformations *, char *);
93 static int fix_boundary (CT *, int *);
94 static int copy_input_to_output (const char *, const char *);
95 static int get_multipart_boundary (CT, char **);
96 static int replace_boundary (CT, char *, char *);
97 static int fix_types (CT, svector_t, int *);
98 static char *replace_substring (char **, const char *, const char *);
99 static char *remove_parameter (char *, const char *);
100 static int fix_composite_cte (CT, int *);
101 static int set_ce (CT, int);
102 static int ensure_text_plain (CT *, CT, int *, int);
103 static int find_textplain_sibling (CT, int, int *);
104 static int insert_new_text_plain_part (CT, int, CT);
105 static CT build_text_plain_part (CT);
106 static int insert_into_new_mp_alt (CT *, int *);
107 static CT divide_part (CT);
108 static void copy_ctinfo (CI, CI);
109 static int decode_part (CT);
110 static int reformat_part (CT, char *, char *, char *, int);
111 static CT build_multipart_alt (CT, CT, int, int);
112 static int boundary_in_content (FILE **, char *, const char *);
113 static void transfer_noncontent_headers (CT, CT);
114 static int set_ct_type (CT, int type, int subtype, int encoding);
115 static int decode_text_parts (CT, int, const char *, int *);
116 static int should_decode(const char *, const char *, const char *);
117 static int content_encoding (CT, const char **);
118 static int strip_crs (CT, int *);
119 static void update_cte (CT);
120 static int least_restrictive_encoding (CT);
121 static int less_restrictive (int, int);
122 static int convert_charsets (CT, char *, int *);
123 static int fix_always (CT, int *);
124 static int fix_filename_param (char *, char *, PM *, PM *);
125 static int fix_filename_encoding (CT);
126 static int write_content (CT, const char *, char *, int, int);
127 static void set_text_ctparams(CT, char *, int);
128 static int remove_file (const char *);
129 static void report (char *, char *, char *, char *, ...);
130 static void pipeser (int);
131
132
133 int
134 main (int argc, char **argv) {
135 int msgnum;
136 char *cp, *file = NULL, *folder = NULL;
137 char *maildir, buf[100], *outfile = NULL;
138 char **argp, **arguments;
139 struct msgs_array msgs = { 0, 0, NULL };
140 struct msgs *mp = NULL;
141 CT *ctp;
142 FILE *fp;
143 int using_stdin = 0;
144 int chgflag = 1;
145 int status = OK;
146 fix_transformations fx;
147 fx.reformat = fx.fixcompositecte = fx.fixboundary = 1;
148 fx.fixtypes = NULL;
149 fx.replacetextplain = 0;
150 fx.decodetext = CE_8BIT;
151 fx.decodetypes = "text,application/ics"; /* Default, per man page. */
152 fx.lf_line_endings = 0;
153 fx.textcharset = NULL;
154
155 if (nmh_init(argv[0], 2)) { return 1; }
156
157 done = freects_done;
158
159 arguments = getarguments (invo_name, argc, argv, 1);
160 argp = arguments;
161
162 /*
163 * Parse arguments
164 */
165 while ((cp = *argp++)) {
166 if (*cp == '-') {
167 switch (smatch (++cp, switches)) {
168 case AMBIGSW:
169 ambigsw (cp, switches);
170 done (1);
171 case UNKWNSW:
172 adios (NULL, "-%s unknown", cp);
173
174 case HELPSW:
175 snprintf (buf, sizeof buf, "%s [+folder] [msgs] [switches]",
176 invo_name);
177 print_help (buf, switches, 1);
178 done (0);
179 case VERSIONSW:
180 print_version(invo_name);
181 done (0);
182
183 case DECODETEXTSW:
184 if (! (cp = *argp++) || *cp == '-') {
185 adios (NULL, "missing argument to %s", argp[-2]);
186 }
187 if (! strcasecmp (cp, "8bit")) {
188 fx.decodetext = CE_8BIT;
189 } else if (! strcasecmp (cp, "7bit")) {
190 fx.decodetext = CE_7BIT;
191 } else if (! strcasecmp (cp, "binary")) {
192 fx.decodetext = CE_BINARY;
193 } else {
194 adios (NULL, "invalid argument to %s", argp[-2]);
195 }
196 continue;
197 case NDECODETEXTSW:
198 fx.decodetext = 0;
199 continue;
200 case DECODETYPESW:
201 if (! (cp = *argp++) || *cp == '-') {
202 adios (NULL, "missing argument to %s", argp[-2]);
203 }
204 fx.decodetypes = cp;
205 continue;
206 case CRLFLINEBREAKSSW:
207 fx.lf_line_endings = 0;
208 continue;
209 case NCRLFLINEBREAKSSW:
210 fx.lf_line_endings = 1;
211 continue;
212 case TEXTCHARSETSW:
213 if (! (cp = *argp++) || (*cp == '-' && cp[1])) {
214 adios (NULL, "missing argument to %s", argp[-2]);
215 }
216 fx.textcharset = cp;
217 continue;
218 case NTEXTCHARSETSW:
219 fx.textcharset = 0;
220 continue;
221 case FIXBOUNDARYSW:
222 fx.fixboundary = 1;
223 continue;
224 case NFIXBOUNDARYSW:
225 fx.fixboundary = 0;
226 continue;
227 case FIXCOMPOSITECTESW:
228 fx.fixcompositecte = 1;
229 continue;
230 case NFIXCOMPOSITECTESW:
231 fx.fixcompositecte = 0;
232 continue;
233 case FIXTYPESW:
234 if (! (cp = *argp++) || (*cp == '-' && cp[1])) {
235 adios (NULL, "missing argument to %s", argp[-2]);
236 }
237 if (! strncasecmp (cp, "multipart/", 10) ||
238 ! strncasecmp (cp, "message/", 8)) {
239 adios (NULL, "-fixtype %s not allowed", cp);
240 } else if (! strchr (cp, '/')) {
241 adios (NULL, "-fixtype requires type/subtype");
242 }
243 if (fx.fixtypes == NULL) { fx.fixtypes = svector_create (10); }
244 svector_push_back (fx.fixtypes, cp);
245 continue;
246 case REFORMATSW:
247 fx.reformat = 1;
248 continue;
249 case NREFORMATSW:
250 fx.reformat = 0;
251 continue;
252 case REPLACETEXTPLAINSW:
253 fx.replacetextplain = 1;
254 continue;
255 case NREPLACETEXTPLAINSW:
256 fx.replacetextplain = 0;
257 continue;
258 case FILESW:
259 if (! (cp = *argp++) || (*cp == '-' && cp[1])) {
260 adios (NULL, "missing argument to %s", argp[-2]);
261 }
262 file = *cp == '-' ? add (cp, NULL) : path (cp, TFILE);
263 continue;
264 case OUTFILESW:
265 if (! (cp = *argp++) || (*cp == '-' && cp[1])) {
266 adios (NULL, "missing argument to %s", argp[-2]);
267 }
268 outfile = *cp == '-' ? add (cp, NULL) : path (cp, TFILE);
269 continue;
270 case RPROCSW:
271 if (!(rmmproc = *argp++) || *rmmproc == '-') {
272 adios (NULL, "missing argument to %s", argp[-2]);
273 }
274 continue;
275 case NRPRCSW:
276 rmmproc = NULL;
277 continue;
278 case CHGSW:
279 chgflag = 1;
280 continue;
281 case NCHGSW:
282 chgflag = 0;
283 continue;
284 case VERBSW:
285 verbosw = 1;
286 continue;
287 case NVERBSW:
288 verbosw = 0;
289 continue;
290 }
291 }
292 if (*cp == '+' || *cp == '@') {
293 if (folder) {
294 adios (NULL, "only one folder at a time!");
295 } else {
296 folder = pluspath (cp);
297 }
298 } else {
299 if (*cp == '/') {
300 /* Interpret a full path as a filename, not a message. */
301 file = add (cp, NULL);
302 } else {
303 app_msgarg (&msgs, cp);
304 }
305 }
306 }
307
308 SIGNAL (SIGQUIT, quitser);
309 SIGNAL (SIGPIPE, pipeser);
310
311 /*
312 * Read the standard profile setup
313 */
314 if ((fp = fopen (cp = etcpath ("mhn.defaults"), "r"))) {
315 readconfig ((struct node **) 0, fp, cp, 0);
316 fclose (fp);
317 }
318
319 suppress_bogus_mp_content_warning = skip_mp_cte_check = 1;
320 suppress_extraneous_trailing_semicolon_warning = 1;
321
322 if (! context_find ("path")) {
323 free (path ("./", TFOLDER));
324 }
325
326 if (file && msgs.size) {
327 adios (NULL, "cannot specify msg and file at same time!");
328 }
329
330 /*
331 * check if message is coming from file
332 */
333 if (file) {
334 /* If file is stdin, create a tmp file name before parse_mime()
335 has a chance, because it might put in on a different
336 filesystem than the output file. Instead, put it in the
337 user's preferred tmp directory. */
338 CT ct;
339
340 if (! strcmp ("-", file)) {
341 int fd;
342 char *cp;
343
344 using_stdin = 1;
345
346 if ((cp = m_mktemp2 (NULL, invo_name, &fd, NULL)) == NULL) {
347 adios (NULL, "unable to create temporary file in %s",
348 get_temp_dir());
349 } else {
350 free (file);
351 file = add (cp, NULL);
352 cpydata (STDIN_FILENO, fd, "-", file);
353 }
354
355 if (close (fd)) {
356 (void) m_unlink (file);
357 adios (NULL, "failed to write temporary file");
358 }
359 }
360
361 cts = mh_xcalloc(2, sizeof *cts);
362 ctp = cts;
363
364 if ((ct = parse_mime (file))) {
365 set_text_ctparams(ct, fx.decodetypes, fx.lf_line_endings);
366 *ctp++ = ct;
367 } else {
368 advise (NULL, "unable to parse message from file %s", file);
369 status = NOTOK;
370
371 /* If there's an outfile, pass the input message unchanged, so the message won't
372 get dropped from a pipeline. */
373 if (outfile) {
374 /* Something went wrong. Output might be expected, such as if this were run
375 as a filter. Just copy the input to the output. */
376 if (copy_input_to_output (file, outfile) != OK) {
377 advise (NULL, "unable to copy message to %s, it might be lost\n", outfile);
378 }
379 }
380 }
381 } else {
382 /*
383 * message(s) are coming from a folder
384 */
385 CT ct;
386
387 if (! msgs.size) {
388 app_msgarg(&msgs, "cur");
389 }
390 if (! folder) {
391 folder = getfolder (1);
392 }
393 maildir = m_maildir (folder);
394
395 if (chdir (maildir) == NOTOK) {
396 adios (maildir, "unable to change directory to");
397 }
398
399 /* read folder and create message structure */
400 if (! (mp = folder_read (folder, 1))) {
401 adios (NULL, "unable to read folder %s", folder);
402 }
403
404 /* check for empty folder */
405 if (mp->nummsg == 0) {
406 adios (NULL, "no messages in %s", folder);
407 }
408
409 /* parse all the message ranges/sequences and set SELECTED */
410 for (msgnum = 0; msgnum < msgs.size; msgnum++)
411 if (! m_convert (mp, msgs.msgs[msgnum])) {
412 done (1);
413 }
414 seq_setprev (mp); /* set the previous-sequence */
415
416 cts = mh_xcalloc(mp->numsel + 1, sizeof *cts);
417 ctp = cts;
418
419 for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
420 if (is_selected(mp, msgnum)) {
421 char *msgnam;
422
423 msgnam = m_name (msgnum);
424 if ((ct = parse_mime (msgnam))) {
425 set_text_ctparams(ct, fx.decodetypes, fx.lf_line_endings);
426 *ctp++ = ct;
427 } else {
428 advise (NULL, "unable to parse message %s", msgnam);
429 status = NOTOK;
430
431 /* If there's an outfile, pass the input message unchanged, so the message won't
432 get dropped from a pipeline. */
433 if (outfile) {
434 /* Something went wrong. Output might be expected, such as if this were run
435 as a filter. Just copy the input to the output. */
436 const char *input_filename = path (msgnam, TFILE);
437
438 if (copy_input_to_output (input_filename, outfile) != OK) {
439 advise (NULL, "unable to copy message to %s, it might be lost\n", outfile);
440 }
441 }
442 }
443 }
444 }
445
446 if (chgflag) {
447 seq_setcur (mp, mp->hghsel); /* update current message */
448 }
449 seq_save (mp); /* synchronize sequences */
450 context_replace (pfolder, folder);/* update current folder */
451 context_save (); /* save the context file */
452 }
453
454 if (*cts) {
455 for (ctp = cts; *ctp; ++ctp) {
456 status += mhfixmsgsbr (ctp, &fx, outfile);
457
458 if (using_stdin) {
459 (void) m_unlink (file);
460
461 if (! outfile) {
462 /* Just calling m_backup() unlinks the backup file. */
463 (void) m_backup (file);
464 }
465 }
466 }
467 } else {
468 status = 1;
469 }
470
471 if (fx.fixtypes != NULL) { svector_free (fx.fixtypes); }
472 free (outfile);
473 free (file);
474 free (folder);
475 free (arguments);
476
477 /* done is freects_done, which will clean up all of cts. */
478 done (status);
479 return NOTOK;
480 }
481
482
483 /*
484 * Apply transformations to one message.
485 */
486 int
487 mhfixmsgsbr (CT *ctp, const fix_transformations *fx, char *outfile) {
488 /* Store input filename in case one of the transformations, i.e.,
489 fix_boundary(), rewrites to a tmp file. */
490 char *input_filename = add ((*ctp)->c_file, NULL);
491 int modify_inplace = 0;
492 int message_mods = 0;
493 int status = OK;
494
495 if (outfile == NULL) {
496 modify_inplace = 1;
497
498 if ((*ctp)->c_file) {
499 char *tempfile;
500 if ((tempfile = m_mktemp2 (NULL, invo_name, NULL, NULL)) == NULL) {
501 adios (NULL, "unable to create temporary file in %s",
502 get_temp_dir());
503 }
504 outfile = add (tempfile, NULL);
505 } else {
506 adios (NULL, "missing both input and output filenames\n");
507 }
508 }
509
510 reverse_alternative_parts (*ctp);
511 status = fix_always (*ctp, &message_mods);
512 if (status == OK && fx->fixboundary) {
513 status = fix_boundary (ctp, &message_mods);
514 }
515 if (status == OK && fx->fixtypes != NULL) {
516 status = fix_types (*ctp, fx->fixtypes, &message_mods);
517 }
518 if (status == OK && fx->fixcompositecte) {
519 status = fix_composite_cte (*ctp, &message_mods);
520 }
521 if (status == OK && fx->reformat) {
522 status =
523 ensure_text_plain (ctp, NULL, &message_mods, fx->replacetextplain);
524 }
525 if (status == OK && fx->decodetext) {
526 status = decode_text_parts (*ctp, fx->decodetext, fx->decodetypes,
527 &message_mods);
528 update_cte (*ctp);
529 }
530 if (status == OK && fx->textcharset != NULL) {
531 status = convert_charsets (*ctp, fx->textcharset, &message_mods);
532 }
533
534 if (status == OK && ! (*ctp)->c_umask) {
535 /* Set the umask for the contents file. This currently
536 isn't used but just in case it is in the future. */
537 struct stat st;
538
539 if (stat ((*ctp)->c_file, &st) != NOTOK) {
540 (*ctp)->c_umask = ~(st.st_mode & 0777);
541 } else {
542 (*ctp)->c_umask = ~m_gmprot();
543 }
544 }
545
546 /*
547 * Write the content to a file
548 */
549 if (status == OK) {
550 status = write_content (*ctp, input_filename, outfile, modify_inplace,
551 message_mods);
552 } else if (! modify_inplace) {
553 /* Something went wrong. Output might be expected, such
554 as if this were run as a filter. Just copy the input
555 to the output. */
556 if (copy_input_to_output (input_filename, outfile) != OK) {
557 advise (NULL, "unable to copy message to %s, it might be lost\n", outfile);
558 }
559 }
560
561 if (modify_inplace) {
562 if (status != OK) { (void) m_unlink (outfile); }
563 free (outfile);
564 outfile = NULL;
565 }
566
567 free (input_filename);
568
569 return status;
570 }
571
572
573 /*
574 * Copy input message to output. Assumes not modifying in place, so this
575 * might be running as part of a pipeline.
576 */
577 static int
578 copy_input_to_output (const char *input_filename, const char *output_filename) {
579 int in = open (input_filename, O_RDONLY);
580 int out = strcmp (output_filename, "-")
581 ? open (output_filename, O_WRONLY | O_CREAT, m_gmprot ())
582 : STDOUT_FILENO;
583 int status = OK;
584
585 if (in != -1 && out != -1) {
586 cpydata (in, out, input_filename, output_filename);
587 } else {
588 status = NOTOK;
589 }
590
591 close (out);
592 close (in);
593
594 return status;
595 }
596
597
598 /*
599 * Fix mismatched outer level boundary.
600 */
601 static int
602 fix_boundary (CT *ct, int *message_mods) {
603 struct multipart *mp;
604 int status = OK;
605
606 if (ct && (*ct)->c_type == CT_MULTIPART && bogus_mp_content) {
607 mp = (struct multipart *) (*ct)->c_ctparams;
608
609 /*
610 * 1) Get boundary at end of part.
611 * 2) Get boundary at beginning of part and compare to the end-of-part
612 * boundary.
613 * 3) Write out contents of ct to tmp file, replacing boundary in
614 * header with boundary from part. Set c_unlink to 1.
615 * 4) Free ct.
616 * 5) Call parse_mime() on the tmp file, replacing ct.
617 */
618
619 if (mp && mp->mp_start) {
620 char *part_boundary;
621
622 if (get_multipart_boundary (*ct, &part_boundary) == OK) {
623 char *fixed;
624
625 if ((fixed = m_mktemp2 (NULL, invo_name, NULL, &(*ct)->c_fp))) {
626 if (replace_boundary (*ct, fixed, part_boundary) == OK) {
627 char *filename = add ((*ct)->c_file, NULL);
628 CT fixed_ct;
629
630 free_content (*ct);
631 if ((fixed_ct = parse_mime (fixed))) {
632 *ct = fixed_ct;
633 (*ct)->c_unlink = 1;
634
635 ++*message_mods;
636 if (verbosw) {
637 report (NULL, NULL, filename,
638 "fix multipart boundary");
639 }
640 } else {
641 *ct = NULL;
642 advise (NULL, "unable to parse fixed part");
643 status = NOTOK;
644 }
645 free (filename);
646 } else {
647 advise (NULL, "unable to replace broken boundary");
648 status = NOTOK;
649 }
650 } else {
651 advise (NULL, "unable to create temporary file in %s",
652 get_temp_dir());
653 status = NOTOK;
654 }
655
656 free (part_boundary);
657 } else {
658 /* Couldn't fix the boundary. Report failure so that mhfixmsg
659 doesn't modify the message. */
660 status = NOTOK;
661 }
662 } else {
663 /* No multipart struct, even though the content type is
664 CT_MULTIPART. Report failure so that mhfixmsg doesn't modify
665 the message. */
666 status = NOTOK;
667 }
668 }
669
670 return status;
671 }
672
673
674 /*
675 * Find boundary at end of multipart.
676 */
677 static int
678 get_multipart_boundary (CT ct, char **part_boundary) {
679 char buffer[BUFSIZ];
680 char *end_boundary = NULL;
681 off_t begin = (off_t) ct->c_end > (off_t) (ct->c_begin + sizeof buffer)
682 ? (off_t) (ct->c_end - sizeof buffer)
683 : (off_t) ct->c_begin;
684 size_t bytes_read;
685 int status = OK;
686
687 /* This will fail if the boundary spans fread() calls. BUFSIZ should
688 be big enough, even if it's just 1024, to make that unlikely. */
689
690 /* free_content() will close ct->c_fp. */
691 if (! ct->c_fp && (ct->c_fp = fopen (ct->c_file, "r")) == NULL) {
692 advise (ct->c_file, "unable to open for reading");
693 return NOTOK;
694 }
695
696 /* Get boundary at end of multipart. */
697 while (begin >= (off_t) ct->c_begin) {
698 fseeko (ct->c_fp, begin, SEEK_SET);
699 while ((bytes_read = fread (buffer, 1, sizeof buffer, ct->c_fp)) > 0) {
700 char *cp = rfind_str (buffer, bytes_read, "--");
701
702 if (cp) {
703 char *end;
704
705 /* Trim off trailing "--" and anything beyond. */
706 *cp-- = '\0';
707 if ((end = rfind_str (buffer, cp - buffer, "\n"))) {
708 if (strlen (end) > 3 && *end++ == '\n' &&
709 *end++ == '-' && *end++ == '-') {
710 end_boundary = add (end, NULL);
711 break;
712 }
713 }
714 }
715 }
716
717 if (! end_boundary && begin > (off_t) (ct->c_begin + sizeof buffer)) {
718 begin -= sizeof buffer;
719 } else {
720 break;
721 }
722 }
723
724 /* Get boundary at beginning of multipart. */
725 if (end_boundary) {
726 fseeko (ct->c_fp, ct->c_begin, SEEK_SET);
727 while ((bytes_read = fread (buffer, 1, sizeof buffer, ct->c_fp)) > 0) {
728 if (bytes_read >= strlen (end_boundary)) {
729 char *cp = find_str (buffer, bytes_read, end_boundary);
730
731 if (cp && cp - buffer >= 2 && *--cp == '-' &&
732 *--cp == '-' && (cp > buffer && *--cp == '\n')) {
733 status = OK;
734 break;
735 }
736 } else {
737 /* The start and end boundaries didn't match, or the
738 start boundary doesn't begin with "\n--" (or "--"
739 if at the beginning of buffer). Keep trying. */
740 status = NOTOK;
741 }
742 }
743 } else {
744 status = NOTOK;
745 }
746
747 if (status == OK) {
748 *part_boundary = end_boundary;
749 } else {
750 *part_boundary = NULL;
751 free (end_boundary);
752 }
753
754 return status;
755 }
756
757
758 /*
759 * Open and copy ct->c_file to file, replacing the multipart boundary.
760 */
761 static int
762 replace_boundary (CT ct, char *file, char *boundary) {
763 FILE *fpin, *fpout;
764 int compnum, state;
765 char buf[BUFSIZ], name[NAMESZ];
766 char *np, *vp;
767 m_getfld_state_t gstate = 0;
768 int status = OK;
769
770 if (ct->c_file == NULL) {
771 advise (NULL, "missing input filename");
772 return NOTOK;
773 }
774
775 if ((fpin = fopen (ct->c_file, "r")) == NULL) {
776 advise (ct->c_file, "unable to open for reading");
777 return NOTOK;
778 }
779
780 if ((fpout = fopen (file, "w")) == NULL) {
781 fclose (fpin);
782 advise (file, "unable to open for writing");
783 return NOTOK;
784 }
785
786 for (compnum = 1;;) {
787 int bufsz = (int) sizeof buf;
788
789 switch (state = m_getfld (&gstate, name, buf, &bufsz, fpin)) {
790 case FLD:
791 case FLDPLUS:
792 compnum++;
793
794 /* get copies of the buffers */
795 np = add (name, NULL);
796 vp = add (buf, NULL);
797
798 /* if necessary, get rest of field */
799 while (state == FLDPLUS) {
800 bufsz = sizeof buf;
801 state = m_getfld (&gstate, name, buf, &bufsz, fpin);
802 vp = add (buf, vp); /* add to previous value */
803 }
804
805 if (strcasecmp (TYPE_FIELD, np)) {
806 fprintf (fpout, "%s:%s", np, vp);
807 } else {
808 char *new_ctline, *new_params;
809
810 replace_param(&ct->c_ctinfo.ci_first_pm,
811 &ct->c_ctinfo.ci_last_pm, "boundary",
812 boundary, 0);
813
814 new_ctline = concat(" ", ct->c_ctinfo.ci_type, "/",
815 ct->c_ctinfo.ci_subtype, NULL);
816 new_params = output_params(strlen(TYPE_FIELD) +
817 strlen(new_ctline) + 1,
818 ct->c_ctinfo.ci_first_pm, NULL, 0);
819 fprintf (fpout, "%s:%s%s\n", np, new_ctline,
820 new_params ? new_params : "");
821 free(new_ctline);
822 mh_xfree(new_params);
823 }
824
825 free (vp);
826 free (np);
827
828 continue;
829
830 case BODY:
831 putc('\n', fpout);
832 /* buf will have a terminating NULL, skip it. */
833 if ((int) fwrite (buf, 1, bufsz-1, fpout) < bufsz-1) {
834 advise (file, "fwrite");
835 }
836 continue;
837
838 case FILEEOF:
839 break;
840
841 case LENERR:
842 case FMTERR:
843 advise (NULL, "message format error in component #%d", compnum);
844 status = NOTOK;
845 break;
846
847 default:
848 advise (NULL, "getfld() returned %d", state);
849 status = NOTOK;
850 break;
851 }
852
853 break;
854 }
855
856 m_getfld_state_destroy (&gstate);
857 fclose (fpout);
858 fclose (fpin);
859
860 return status;
861 }
862
863
864 /*
865 * Fix Content-Type header to reflect the content of its part.
866 */
867 static int
868 fix_types (CT ct, svector_t fixtypes, int *message_mods) {
869 int status = OK;
870
871 switch (ct->c_type) {
872 case CT_MULTIPART: {
873 struct multipart *m = (struct multipart *) ct->c_ctparams;
874 struct part *part;
875
876 for (part = m->mp_parts; status == OK && part; part = part->mp_next) {
877 status = fix_types (part->mp_part, fixtypes, message_mods);
878 }
879 break;
880 }
881
882 case CT_MESSAGE:
883 if (ct->c_subtype == MESSAGE_EXTERNAL) {
884 struct exbody *e = (struct exbody *) ct->c_ctparams;
885
886 status = fix_types (e->eb_content, fixtypes, message_mods);
887 }
888 break;
889
890 default: {
891 char **typep, *type;
892
893 if (ct->c_ctinfo.ci_type && ct->c_ctinfo.ci_subtype) {
894 for (typep = svector_strs (fixtypes);
895 typep && (type = *typep);
896 ++typep) {
897 char *type_subtype =
898 concat (ct->c_ctinfo.ci_type, "/", ct->c_ctinfo.ci_subtype,
899 NULL);
900
901 if (! strcasecmp (type, type_subtype) &&
902 decode_part (ct) == OK &&
903 ct->c_cefile.ce_file != NULL) {
904 char *ct_type_subtype = mime_type (ct->c_cefile.ce_file);
905 char *cp;
906
907 if ((cp = strchr (ct_type_subtype, ';'))) {
908 /* Truncate to remove any parameter list from
909 mime_type () result. */
910 *cp = '\0';
911 }
912
913 if (strcasecmp (type, ct_type_subtype)) {
914 char *ct_type, *ct_subtype;
915 HF hf;
916
917 /* The Content-Type header does not match the
918 content, so update these struct Content
919 fields to match:
920 * c_type, c_subtype
921 * c_ctinfo.ci_type, c_ctinfo.ci_subtype
922 * c_ctline
923 */
924 /* Extract type and subtype from type/subtype. */
925 ct_type = mh_xstrdup(ct_type_subtype);
926 if ((cp = strchr (ct_type, '/'))) {
927 *cp = '\0';
928 ct_subtype = mh_xstrdup(++cp);
929 } else {
930 advise (NULL, "missing / in MIME type of %s %s",
931 ct->c_file, ct->c_partno);
932 free (ct_type);
933 return NOTOK;
934 }
935
936 ct->c_type = ct_str_type (ct_type);
937 ct->c_subtype = ct_str_subtype (ct->c_type, ct_subtype);
938
939 free (ct->c_ctinfo.ci_type);
940 ct->c_ctinfo.ci_type = ct_type;
941 free (ct->c_ctinfo.ci_subtype);
942 ct->c_ctinfo.ci_subtype = ct_subtype;
943 if (! replace_substring (&ct->c_ctline, type,
944 ct_type_subtype)) {
945 advise (NULL, "did not find %s in %s",
946 type, ct->c_ctline);
947 }
948
949 /* Update Content-Type header field. */
950 for (hf = ct->c_first_hf; hf; hf = hf->next) {
951 if (! strcasecmp (TYPE_FIELD, hf->name)) {
952 if (replace_substring (&hf->value, type,
953 ct_type_subtype)) {
954 ++*message_mods;
955 if (verbosw) {
956 report (NULL, ct->c_partno, ct->c_file,
957 "change Content-Type in header "
958 "from %s to %s",
959 type, ct_type_subtype);
960 }
961 break;
962 } else {
963 advise (NULL, "did not find %s in %s",
964 type, hf->value);
965 }
966 }
967 }
968 }
969 free (ct_type_subtype);
970 }
971 free (type_subtype);
972 }
973 }
974 }}
975
976 return status;
977 }
978
979
980 /*
981 * Replace a substring, allocating space to hold the new one.
982 */
983 char *
984 replace_substring (char **str, const char *old, const char *new) {
985 char *cp;
986
987 if ((cp = strstr (*str, old))) {
988 char *remainder = cp + strlen (old);
989 char *prefix, *new_str;
990
991 if (cp - *str) {
992 prefix = mh_xstrdup(*str);
993 *(prefix + (cp - *str)) = '\0';
994 new_str = concat (prefix, new, remainder, NULL);
995 free (prefix);
996 } else {
997 new_str = concat (new, remainder, NULL);
998 }
999
1000 free (*str);
1001
1002 return *str = new_str;
1003 }
1004
1005 return NULL;
1006 }
1007
1008
1009 /*
1010 * Remove a name=value parameter, given just its name, from a header value.
1011 */
1012 char *
1013 remove_parameter (char *str, const char *name) {
1014 /* It looks to me, based on the BNF in RFC 2045, than there can't
1015 be whitespace betwwen the parameter name and the "=", or
1016 between the "=" and the parameter value. */
1017 char *param_name = concat (name, "=", NULL);
1018 char *cp;
1019
1020 if ((cp = strstr (str, param_name))) {
1021 char *start, *end;
1022 size_t count = 1;
1023
1024 /* Remove any leading spaces, before the parameter name. */
1025 for (start = cp;
1026 start > str && isspace ((unsigned char) *(start-1));
1027 --start) {
1028 continue;
1029 }
1030 /* Remove a leading semicolon. */
1031 if (start > str && *(start-1) == ';') { --start; }
1032
1033 end = cp + strlen (name) + 1;
1034 if (*end == '"') {
1035 /* Skip past the quoted value, and then the final quote. */
1036 for (++end ; *end && *end != '"'; ++end) { continue; }
1037 ++end;
1038 } else {
1039 /* Skip past the value. */
1040 for (++end ; *end && ! isspace ((unsigned char) *end); ++end) {}
1041 }
1042
1043 /* Count how many characters need to be moved. Include
1044 trailing null, which is accounted for by the
1045 initialization of count to 1. */
1046 for (cp = end; *cp; ++cp) { ++count; }
1047 (void) memmove (start, end, count);
1048 }
1049
1050 free (param_name);
1051
1052 return str;
1053 }
1054
1055
1056 /*
1057 * Fix Content-Transfer-Encoding of composite,, e.g., message or multipart, part.
1058 * According to RFC 2045 Sec. 6.4, it must be 7bit, 8bit, or binary. Set it to
1059 * 8 bit.
1060 */
1061 static int
1062 fix_composite_cte (CT ct, int *message_mods) {
1063 int status = OK;
1064
1065 if (ct->c_type == CT_MESSAGE || ct->c_type == CT_MULTIPART) {
1066 if (ct->c_encoding != CE_7BIT && ct->c_encoding != CE_8BIT &&
1067 ct->c_encoding != CE_BINARY) {
1068 HF hf;
1069
1070 for (hf = ct->c_first_hf; hf; hf = hf->next) {
1071 char *name = hf->name;
1072 for (; *name && isspace ((unsigned char) *name); ++name) {
1073 continue;
1074 }
1075
1076 if (! strncasecmp (name, ENCODING_FIELD,
1077 strlen (ENCODING_FIELD))) {
1078 char *prefix = "Nmh-REPLACED-INVALID-";
1079 HF h;
1080
1081 NEW(h);
1082 h->name = add (hf->name, NULL);
1083 h->hf_encoding = hf->hf_encoding;
1084 h->next = hf->next;
1085 hf->next = h;
1086
1087 /* Retain old header but prefix its name. */
1088 free (hf->name);
1089 hf->name = concat (prefix, h->name, NULL);
1090
1091 ++*message_mods;
1092 if (verbosw) {
1093 char *encoding = cpytrim (hf->value);
1094 report (NULL, ct->c_partno, ct->c_file,
1095 "replace Content-Transfer-Encoding of %s "
1096 "with 8 bit", encoding);
1097 free (encoding);
1098 }
1099
1100 h->value = add (" 8bit\n", NULL);
1101
1102 /* Don't need to warn for multiple C-T-E header
1103 fields, parse_mime() already does that. But
1104 if there are any, fix them all as necessary. */
1105 hf = h;
1106 }
1107 }
1108
1109 set_ce (ct, CE_8BIT);
1110 }
1111
1112 if (ct->c_type == CT_MULTIPART) {
1113 struct multipart *m;
1114 struct part *part;
1115
1116 m = (struct multipart *) ct->c_ctparams;
1117 for (part = m->mp_parts; part; part = part->mp_next) {
1118 if (fix_composite_cte (part->mp_part, message_mods) != OK) {
1119 status = NOTOK;
1120 break;
1121 }
1122 }
1123 }
1124 }
1125
1126 return status;
1127 }
1128
1129
1130 /*
1131 * Set content encoding.
1132 */
1133 static int
1134 set_ce (CT ct, int encoding) {
1135 const char *ce = ce_str (encoding);
1136 const struct str2init *ctinit = get_ce_method (ce);
1137
1138 if (ctinit) {
1139 char *cte = concat (" ", ce, "\n", NULL);
1140 int found_cte = 0;
1141 HF hf;
1142 /* Decoded contents might be in ct->c_cefile.ce_file, if the
1143 caller is decode_text_parts (). Save because we'll
1144 overwrite below. */
1145 struct cefile decoded_content_info = ct->c_cefile;
1146
1147 ct->c_encoding = encoding;
1148
1149 ct->c_ctinitfnx = ctinit->si_init;
1150 /* This will assign ct->c_cefile with an all-0 struct, which
1151 is what we want. */
1152 (*ctinit->si_init) (ct);
1153 /* After returning, the caller should set
1154 ct->c_cefile.ce_file to the name of the file containing
1155 the contents. */
1156
1157 if (ct->c_ceclosefnx) {
1158 (*ct->c_ceclosefnx) (ct);
1159 }
1160
1161 /* Restore the cefile. */
1162 ct->c_cefile = decoded_content_info;
1163
1164 /* Update/add Content-Transfer-Encoding header field. */
1165 for (hf = ct->c_first_hf; hf; hf = hf->next) {
1166 if (! strcasecmp (ENCODING_FIELD, hf->name)) {
1167 found_cte = 1;
1168 free (hf->value);
1169 hf->value = cte;
1170 }
1171 }
1172 if (! found_cte) {
1173 add_header (ct, add (ENCODING_FIELD, NULL), cte);
1174 }
1175
1176 /* Update c_celine. It's used only by mhlist -debug. */
1177 free (ct->c_celine);
1178 ct->c_celine = add (cte, NULL);
1179
1180 return OK;
1181 }
1182
1183 return NOTOK;
1184 }
1185
1186
1187 /*
1188 * Make sure each text part has a corresponding text/plain part.
1189 */
1190 static int
1191 ensure_text_plain (CT *ct, CT parent, int *message_mods, int replacetextplain) {
1192 int status = OK;
1193
1194 switch ((*ct)->c_type) {
1195 case CT_TEXT: {
1196 /* Nothing to do for text/plain. */
1197 if ((*ct)->c_subtype == TEXT_PLAIN) { return OK; }
1198
1199 if (parent && parent->c_type == CT_MULTIPART &&
1200 parent->c_subtype == MULTI_ALTERNATE) {
1201 int new_subpart_number = 1;
1202 int has_text_plain =
1203 find_textplain_sibling (parent, replacetextplain,
1204 &new_subpart_number);
1205
1206 if (! has_text_plain) {
1207 /* Parent is a multipart/alternative. Insert a new
1208 text/plain subpart. */
1209 const int inserted =
1210 insert_new_text_plain_part (*ct, new_subpart_number,
1211 parent);
1212 if (inserted) {
1213 ++*message_mods;
1214 if (verbosw) {
1215 report (NULL, parent->c_partno, parent->c_file,
1216 "insert text/plain part");
1217 }
1218 } else {
1219 status = NOTOK;
1220 }
1221 }
1222 } else if (parent && parent->c_type == CT_MULTIPART &&
1223 parent->c_subtype == MULTI_RELATED) {
1224 char *type_subtype =
1225 concat ((*ct)->c_ctinfo.ci_type, "/",
1226 (*ct)->c_ctinfo.ci_subtype, NULL);
1227 const char *parent_type =
1228 get_param (parent->c_ctinfo.ci_first_pm, "type", '?', 1);
1229 int new_subpart_number = 1;
1230 int has_text_plain = 0;
1231
1232 /* Have to do string comparison on the subtype because we
1233 don't enumerate all of them in c_subtype values.
1234 parent_type will be NULL if the multipart/related part
1235 doesn't have a type parameter. The type parameter must
1236 be specified according to RFC 2387 Sec. 3.1 but not all
1237 messages comply. */
1238 if (parent_type && strcasecmp (type_subtype, parent_type) == 0) {
1239 /* The type of this part matches the root type of the
1240 parent multipart/related. Look to see if there's
1241 text/plain sibling. */
1242 has_text_plain =
1243 find_textplain_sibling (parent, replacetextplain,
1244 &new_subpart_number);
1245 }
1246
1247 free (type_subtype);
1248
1249 if (! has_text_plain) {
1250 struct multipart *mp = (struct multipart *) parent->c_ctparams;
1251 struct part *part;
1252 int siblings = 0;
1253
1254 for (part = mp->mp_parts; part; part = part->mp_next) {
1255 if (*ct != part->mp_part) {
1256 ++siblings;
1257 }
1258 }
1259
1260 if (siblings) {
1261 /* Parent is a multipart/related. Insert a new
1262 text/plain subpart in a new multipart/alternative. */
1263 if (insert_into_new_mp_alt (ct, message_mods)) {
1264 /* Not an error if text/plain couldn't be added. */
1265 }
1266 } else {
1267 /* There are no siblings, so insert a new text/plain
1268 subpart, and change the parent type from
1269 multipart/related to multipart/alternative. */
1270 const int inserted =
1271 insert_new_text_plain_part (*ct, new_subpart_number,
1272 parent);
1273
1274 if (inserted) {
1275 HF hf;
1276
1277 parent->c_subtype = MULTI_ALTERNATE;
1278 parent->c_ctinfo.ci_subtype = mh_xstrdup("alternative");
1279 if (! replace_substring (&parent->c_ctline, "/related",
1280 "/alternative")) {
1281 advise (NULL,
1282 "did not find multipart/related in %s",
1283 parent->c_ctline);
1284 }
1285
1286 /* Update Content-Type header field. */
1287 for (hf = parent->c_first_hf; hf; hf = hf->next) {
1288 if (! strcasecmp (TYPE_FIELD, hf->name)) {
1289 if (replace_substring (&hf->value, "/related",
1290 "/alternative")) {
1291 ++*message_mods;
1292 if (verbosw) {
1293 report (NULL, parent->c_partno,
1294 parent->c_file,
1295 "insert text/plain part");
1296 }
1297
1298 /* Remove, e.g., type="text/html" from
1299 multipart/alternative. */
1300 remove_parameter (hf->value, "type");
1301 break;
1302 } else {
1303 advise (NULL, "did not find multipart/"
1304 "related in header %s",
1305 hf->value);
1306 }
1307 }
1308 }
1309 } else {
1310 /* Not an error if text/plain couldn't be inserted. */
1311 }
1312 }
1313 }
1314 } else {
1315 if (insert_into_new_mp_alt (ct, message_mods)) {
1316 status = NOTOK;
1317 }
1318 }
1319 break;
1320 }
1321
1322 case CT_MULTIPART: {
1323 struct multipart *mp = (struct multipart *) (*ct)->c_ctparams;
1324 struct part *part;
1325
1326 for (part = mp->mp_parts; status == OK && part; part = part->mp_next) {
1327 if ((*ct)->c_type == CT_MULTIPART) {
1328 status = ensure_text_plain (&part->mp_part, *ct, message_mods,
1329 replacetextplain);
1330 }
1331 }
1332 break;
1333 }
1334
1335 case CT_MESSAGE:
1336 if ((*ct)->c_subtype == MESSAGE_EXTERNAL) {
1337 struct exbody *e = (struct exbody *) (*ct)->c_ctparams;
1338
1339 status = ensure_text_plain (&e->eb_content, *ct, message_mods,
1340 replacetextplain);
1341 }
1342 break;
1343 }
1344
1345 return status;
1346 }
1347
1348
1349 /*
1350 * See if there is a sibling text/plain, and return its subpart number.
1351 */
1352 static int
1353 find_textplain_sibling (CT parent, int replacetextplain,
1354 int *new_subpart_number) {
1355 struct multipart *mp = (struct multipart *) parent->c_ctparams;
1356 struct part *part, *prev;
1357 int has_text_plain = 0;
1358
1359 for (prev = part = mp->mp_parts; part; part = part->mp_next) {
1360 ++*new_subpart_number;
1361 if (part->mp_part->c_type == CT_TEXT &&
1362 part->mp_part->c_subtype == TEXT_PLAIN) {
1363 if (replacetextplain) {
1364 struct part *old_part;
1365 if (part == mp->mp_parts) {
1366 old_part = mp->mp_parts;
1367 mp->mp_parts = part->mp_next;
1368 } else {
1369 old_part = prev->mp_next;
1370 prev->mp_next = part->mp_next;
1371 }
1372 if (verbosw) {
1373 report (NULL, parent->c_partno, parent->c_file,
1374 "remove text/plain part %s",
1375 old_part->mp_part->c_partno);
1376 }
1377 free_content (old_part->mp_part);
1378 free (old_part);
1379 } else {
1380 has_text_plain = 1;
1381 }
1382 break;
1383 }
1384 prev = part;
1385 }
1386
1387 return has_text_plain;
1388 }
1389
1390
1391 /*
1392 * Insert a new text/plain part.
1393 */
1394 static int
1395 insert_new_text_plain_part (CT ct, int new_subpart_number, CT parent) {
1396 struct multipart *mp = (struct multipart *) parent->c_ctparams;
1397 struct part *new_part;
1398
1399 NEW(new_part);
1400 if ((new_part->mp_part = build_text_plain_part (ct))) {
1401 char buffer[16];
1402 snprintf (buffer, sizeof buffer, "%d", new_subpart_number);
1403
1404 new_part->mp_next = mp->mp_parts;
1405 mp->mp_parts = new_part;
1406 new_part->mp_part->c_partno =
1407 concat (parent->c_partno ? parent->c_partno : "1", ".",
1408 buffer, NULL);
1409
1410 return 1;
1411 }
1412
1413 free_content (new_part->mp_part);
1414 free (new_part);
1415
1416 return 0;
1417 }
1418
1419
1420 /*
1421 * Create a text/plain part to go along with non-plain sibling part.
1422 */
1423 static CT
1424 build_text_plain_part (CT encoded_part) {
1425 CT tp_part = divide_part (encoded_part);
1426 char *tmp_plain_file = NULL;
1427
1428 if (decode_part (tp_part) == OK) {
1429 /* Now, tp_part->c_cefile.ce_file is the name of the tmp file that
1430 contains the decoded contents. And the decoding function, such
1431 as openQuoted, will have set ...->ce_unlink to 1 so that it will
1432 be unlinked by free_content (). */
1433 char *tempfile;
1434
1435 if ((tempfile = m_mktemp2 (NULL, invo_name, NULL, NULL)) == NULL) {
1436 advise (NULL, "unable to create temporary file in %s",
1437 get_temp_dir());
1438 } else {
1439 tmp_plain_file = add (tempfile, NULL);
1440 if (reformat_part (tp_part, tmp_plain_file,
1441 tp_part->c_ctinfo.ci_type,
1442 tp_part->c_ctinfo.ci_subtype,
1443 tp_part->c_type) == OK) {
1444 return tp_part;
1445 }
1446 }
1447 }
1448
1449 free_content (tp_part);
1450 if (tmp_plain_file) { (void) m_unlink (tmp_plain_file); }
1451 free (tmp_plain_file);
1452
1453 return NULL;
1454 }
1455
1456
1457 /*
1458 * Slip new text/plain part into a new multipart/alternative.
1459 */
1460 static int
1461 insert_into_new_mp_alt (CT *ct, int *message_mods) {
1462 CT tp_part = build_text_plain_part (*ct);
1463 int status = OK;
1464
1465 if (tp_part) {
1466 CT mp_alt = build_multipart_alt (*ct, tp_part, CT_MULTIPART,
1467 MULTI_ALTERNATE);
1468 if (mp_alt) {
1469 struct multipart *mp = (struct multipart *) mp_alt->c_ctparams;
1470
1471 if (mp && mp->mp_parts) {
1472 mp->mp_parts->mp_part = tp_part;
1473 /* Make the new multipart/alternative the parent. */
1474 *ct = mp_alt;
1475
1476 ++*message_mods;
1477 if (verbosw) {
1478 report (NULL, (*ct)->c_partno, (*ct)->c_file,
1479 "insert text/plain part");
1480 }
1481 } else {
1482 free_content (tp_part);
1483 free_content (mp_alt);
1484 status = NOTOK;
1485 }
1486 } else {
1487 status = NOTOK;
1488 }
1489 } else {
1490 /* Not an error if text/plain couldn't be built. */
1491 }
1492
1493 return status;
1494 }
1495
1496
1497 /*
1498 * Clone a MIME part.
1499 */
1500 static CT
1501 divide_part (CT ct) {
1502 CT new_part;
1503
1504 NEW0(new_part);
1505 /* Just copy over what is needed for decoding. c_vrsn and
1506 c_celine aren't necessary. */
1507 new_part->c_file = add (ct->c_file, NULL);
1508 new_part->c_begin = ct->c_begin;
1509 new_part->c_end = ct->c_end;
1510 copy_ctinfo (&new_part->c_ctinfo, &ct->c_ctinfo);
1511 new_part->c_type = ct->c_type;
1512 new_part->c_cefile = ct->c_cefile;
1513 new_part->c_encoding = ct->c_encoding;
1514 new_part->c_ctinitfnx = ct->c_ctinitfnx;
1515 new_part->c_ceopenfnx = ct->c_ceopenfnx;
1516 new_part->c_ceclosefnx = ct->c_ceclosefnx;
1517 new_part->c_cesizefnx = ct->c_cesizefnx;
1518
1519 /* c_ctline is used by reformat__part(), so it can preserve
1520 anything after the type/subtype. */
1521 new_part->c_ctline = add (ct->c_ctline, NULL);
1522
1523 return new_part;
1524 }
1525
1526
1527 /*
1528 * Copy the content info from one part to another.
1529 */
1530 static void
1531 copy_ctinfo (CI dest, CI src) {
1532 PM s_pm, d_pm;
1533
1534 dest->ci_type = src->ci_type ? add (src->ci_type, NULL) : NULL;
1535 dest->ci_subtype = src->ci_subtype ? add (src->ci_subtype, NULL) : NULL;
1536
1537 for (s_pm = src->ci_first_pm; s_pm; s_pm = s_pm->pm_next) {
1538 d_pm = add_param(&dest->ci_first_pm, &dest->ci_last_pm, s_pm->pm_name,
1539 s_pm->pm_value, 0);
1540 if (s_pm->pm_charset) {
1541 d_pm->pm_charset = mh_xstrdup(s_pm->pm_charset);
1542 }
1543 if (s_pm->pm_lang) {
1544 d_pm->pm_lang = mh_xstrdup(s_pm->pm_lang);
1545 }
1546 }
1547
1548 dest->ci_comment = src->ci_comment ? add (src->ci_comment, NULL) : NULL;
1549 dest->ci_magic = src->ci_magic ? add (src->ci_magic, NULL) : NULL;
1550 }
1551
1552
1553 /*
1554 * Decode content.
1555 */
1556 static int
1557 decode_part (CT ct) {
1558 char *tmp_decoded;
1559 int status;
1560 char *tempfile;
1561
1562 if ((tempfile = m_mktemp2 (NULL, invo_name, NULL, NULL)) == NULL) {
1563 adios (NULL, "unable to create temporary file in %s", get_temp_dir());
1564 }
1565 tmp_decoded = add (tempfile, NULL);
1566 /* The following call will load ct->c_cefile.ce_file with the tmp
1567 filename of the decoded content. tmp_decoded will contain the
1568 encoded output, get rid of that. */
1569 status = output_message (ct, tmp_decoded);
1570 (void) m_unlink (tmp_decoded);
1571 free (tmp_decoded);
1572
1573 return status;
1574 }
1575
1576
1577 /*
1578 * Reformat content as plain text.
1579 * Some of the arguments aren't really needed now, but maybe will
1580 * be in the future for other than text types.
1581 */
1582 static int
1583 reformat_part (CT ct, char *file, char *type, char *subtype, int c_type) {
1584 int output_subtype, output_encoding;
1585 const char *reason = NULL;
1586 char *cp, *cf;
1587 int status;
1588
1589 /* Hacky: this redirects the output from whatever command is used
1590 to show the part to a file. So, the user can't have any output
1591 redirection in that command.
1592 Could show_multi() in mhshowsbr.c avoid this? */
1593
1594 /* Check for invo_name-format-type/subtype. */
1595 if ((cf = context_find_by_type ("format", type, subtype)) == NULL) {
1596 if (verbosw) {
1597 advise (NULL, "Don't know how to convert %s, there is no "
1598 "%s-format-%s/%s profile entry",
1599 ct->c_file, invo_name, type, subtype);
1600 }
1601 return NOTOK;
1602 }
1603 if (strchr (cf, '>')) {
1604 advise (NULL, "'>' prohibited in \"%s\",\nplease fix your "
1605 "%s-format-%s/%s profile entry", cf, invo_name, type,
1606 subtype ? subtype : "");
1607
1608 return NOTOK;
1609 }
1610
1611 cp = concat (cf, " >", file, NULL);
1612 status = show_content_aux (ct, 0, cp, NULL, NULL);
1613 free (cp);
1614
1615 /* Unlink decoded content tmp file and free its filename to avoid
1616 leaks. The file stream should already have been closed. */
1617 if (ct->c_cefile.ce_unlink) {
1618 (void) m_unlink (ct->c_cefile.ce_file);
1619 free (ct->c_cefile.ce_file);
1620 ct->c_cefile.ce_file = NULL;
1621 ct->c_cefile.ce_unlink = 0;
1622 }
1623
1624 if (c_type == CT_TEXT) {
1625 output_subtype = TEXT_PLAIN;
1626 } else {
1627 /* Set subtype to 0, which is always an UNKNOWN subtype. */
1628 output_subtype = 0;
1629 }
1630
1631 output_encoding = content_encoding (ct, &reason);
1632 if (set_ct_type (ct, c_type, output_subtype, output_encoding) == OK) {
1633 ct->c_cefile.ce_file = file;
1634 ct->c_cefile.ce_unlink = 1;
1635 } else {
1636 ct->c_cefile.ce_unlink = 0;
1637 status = NOTOK;
1638 }
1639
1640 return status;
1641 }
1642
1643
1644 /*
1645 * Fill in a multipart/alternative part.
1646 */
1647 static CT
1648 build_multipart_alt (CT first_alt, CT new_part, int type, int subtype) {
1649 char *boundary_prefix = "----=_nmh-multipart";
1650 char *boundary = concat (boundary_prefix, first_alt->c_partno, NULL);
1651 char *boundary_indicator = "; boundary=";
1652 char *typename, *subtypename, *name;
1653 CT ct;
1654 struct part *p;
1655 struct multipart *m;
1656 const struct str2init *ctinit;
1657
1658 NEW0(ct);
1659
1660 /* Set up the multipart/alternative part. These fields of *ct were
1661 initialized to 0 by mh_xcalloc():
1662 c_fp, c_unlink, c_begin, c_end,
1663 c_vrsn, c_ctline, c_celine,
1664 c_id, c_descr, c_dispo, c_partno,
1665 c_ctinfo.ci_comment, c_ctinfo.ci_magic,
1666 c_cefile, c_encoding,
1667 c_digested, c_digest[16], c_ctexbody,
1668 c_ctinitfnx, c_ceopenfnx, c_ceclosefnx, c_cesizefnx,
1669 c_umask, c_rfc934,
1670 c_showproc, c_termproc, c_storeproc, c_storage, c_folder
1671 */
1672
1673 ct->c_file = add (first_alt->c_file, NULL);
1674 ct->c_type = type;
1675 ct->c_subtype = subtype;
1676
1677 ctinit = get_ct_init (ct->c_type);
1678
1679 typename = ct_type_str (type);
1680 subtypename = ct_subtype_str (type, subtype);
1681
1682 {
1683 int serial = 0;
1684 int found_boundary = 1;
1685
1686 while (found_boundary && serial < 1000000) {
1687 found_boundary = 0;
1688
1689 /* Ensure that the boundary doesn't appear in the decoded
1690 content. */
1691 if (new_part->c_cefile.ce_file) {
1692 if ((found_boundary =
1693 boundary_in_content (&new_part->c_cefile.ce_fp,
1694 new_part->c_cefile.ce_file,
1695 boundary)) == -1) {
1696 free (ct);
1697 return NULL;
1698 }
1699 }
1700
1701 /* Ensure that the boundary doesn't appear in the encoded
1702 content. */
1703 if (! found_boundary && new_part->c_file) {
1704 if ((found_boundary = boundary_in_content (&new_part->c_fp,
1705 new_part->c_file,
1706 boundary)) == -1) {
1707 free (ct);
1708 return NULL;
1709 }
1710 }
1711
1712 if (found_boundary) {
1713 /* Try a slightly different boundary. */
1714 char buffer2[16];
1715
1716 free (boundary);
1717 ++serial;
1718 snprintf (buffer2, sizeof buffer2, "%d", serial);
1719 boundary =
1720 concat (boundary_prefix,
1721 first_alt->c_partno ? first_alt->c_partno : "",
1722 "-", buffer2, NULL);
1723 }
1724 }
1725
1726 if (found_boundary) {
1727 advise (NULL, "giving up trying to find a unique boundary");
1728 free (ct);
1729 return NULL;
1730 }
1731 }
1732
1733 name = concat (" ", typename, "/", subtypename, boundary_indicator, "\"",
1734 boundary, "\"", NULL);
1735
1736 /* Load c_first_hf and c_last_hf. */
1737 transfer_noncontent_headers (first_alt, ct);
1738 add_header (ct, add (TYPE_FIELD, NULL), concat (name, "\n", NULL));
1739 free (name);
1740
1741 /* Load c_partno. */
1742 if (first_alt->c_partno) {
1743 ct->c_partno = add (first_alt->c_partno, NULL);
1744 free (first_alt->c_partno);
1745 first_alt->c_partno = concat (ct->c_partno, ".1", NULL);
1746 new_part->c_partno = concat (ct->c_partno, ".2", NULL);
1747 } else {
1748 first_alt->c_partno = add ("1", NULL);
1749 new_part->c_partno = add ("2", NULL);
1750 }
1751
1752 if (ctinit) {
1753 ct->c_ctinfo.ci_type = add (typename, NULL);
1754 ct->c_ctinfo.ci_subtype = add (subtypename, NULL);
1755 }
1756
1757 add_param(&ct->c_ctinfo.ci_first_pm, &ct->c_ctinfo.ci_last_pm,
1758 "boundary", boundary, 0);
1759
1760 NEW(p);
1761 NEW(p->mp_next);
1762 p->mp_next->mp_next = NULL;
1763 p->mp_next->mp_part = first_alt;
1764
1765 NEW0(m);
1766 m->mp_start = concat (boundary, "\n", NULL);
1767 m->mp_stop = concat (boundary, "--\n", NULL);
1768 m->mp_parts = p;
1769 ct->c_ctparams = m;
1770
1771 free (boundary);
1772
1773 return ct;
1774 }
1775
1776
1777 /*
1778 * Check that the boundary does not appear in the content.
1779 */
1780 static int
1781 boundary_in_content (FILE **fp, char *file, const char *boundary) {
1782 char buffer[BUFSIZ];
1783 size_t bytes_read;
1784 int found_boundary = 0;
1785
1786 /* free_content() will close *fp if we fopen it here. */
1787 if (! *fp && (*fp = fopen (file, "r")) == NULL) {
1788 advise (file, "unable to open %s for reading", file);
1789 return NOTOK;
1790 }
1791
1792 fseeko (*fp, 0L, SEEK_SET);
1793 while ((bytes_read = fread (buffer, 1, sizeof buffer, *fp)) > 0) {
1794 if (find_str (buffer, bytes_read, boundary)) {
1795 found_boundary = 1;
1796 break;
1797 }
1798 }
1799
1800 return found_boundary;
1801 }
1802
1803
1804 /*
1805 * Remove all non-Content headers.
1806 */
1807 static void
1808 transfer_noncontent_headers (CT old, CT new) {
1809 HF hp, hp_prev;
1810
1811 hp_prev = hp = old->c_first_hf;
1812 while (hp) {
1813 HF next = hp->next;
1814
1815 if (strncasecmp (XXX_FIELD_PRF, hp->name, strlen (XXX_FIELD_PRF))) {
1816 if (hp == old->c_last_hf) {
1817 if (hp == old->c_first_hf) {
1818 old->c_last_hf = old->c_first_hf = NULL;
1819 } else {
1820 hp_prev->next = NULL;
1821 old->c_last_hf = hp_prev;
1822 }
1823 } else {
1824 if (hp == old->c_first_hf) {
1825 old->c_first_hf = next;
1826 } else {
1827 hp_prev->next = next;
1828 }
1829 }
1830
1831 /* Put node hp in the new CT. */
1832 if (new->c_first_hf == NULL) {
1833 new->c_first_hf = hp;
1834 } else {
1835 new->c_last_hf->next = hp;
1836 }
1837 new->c_last_hf = hp;
1838 } else {
1839 /* A Content- header, leave in old. */
1840 hp_prev = hp;
1841 }
1842
1843 hp = next;
1844 }
1845 }
1846
1847
1848 /*
1849 * Set content type.
1850 */
1851 static int
1852 set_ct_type (CT ct, int type, int subtype, int encoding) {
1853 char *typename = ct_type_str (type);
1854 char *subtypename = ct_subtype_str (type, subtype);
1855 /* E.g, " text/plain" */
1856 char *type_subtypename = concat (" ", typename, "/", subtypename, NULL);
1857 /* E.g, " text/plain\n" */
1858 char *name_plus_nl = concat (type_subtypename, "\n", NULL);
1859 int found_content_type = 0;
1860 HF hf;
1861 const char *cp = NULL;
1862 char *ctline;
1863 int status;
1864
1865 /* Update/add Content-Type header field. */
1866 for (hf = ct->c_first_hf; hf; hf = hf->next) {
1867 if (! strcasecmp (TYPE_FIELD, hf->name)) {
1868 found_content_type = 1;
1869 free (hf->value);
1870 hf->value = (cp = strchr (ct->c_ctline, ';'))
1871 ? concat (type_subtypename, cp, "\n", NULL)
1872 : add (name_plus_nl, NULL);
1873 }
1874 }
1875 if (! found_content_type) {
1876 add_header (ct, add (TYPE_FIELD, NULL),
1877 (cp = strchr (ct->c_ctline, ';'))
1878 ? concat (type_subtypename, cp, "\n", NULL)
1879 : add (name_plus_nl, NULL));
1880 }
1881
1882 /* Some of these might not be used, but set them anyway. */
1883 ctline = cp
1884 ? concat (type_subtypename, cp, NULL)
1885 : concat (type_subtypename, NULL);
1886 free (ct->c_ctline);
1887 ct->c_ctline = ctline;
1888 /* Leave other ctinfo members as they were. */
1889 free (ct->c_ctinfo.ci_type);
1890 ct->c_ctinfo.ci_type = add (typename, NULL);
1891 free (ct->c_ctinfo.ci_subtype);
1892 ct->c_ctinfo.ci_subtype = add (subtypename, NULL);
1893 ct->c_type = type;
1894 ct->c_subtype = subtype;
1895
1896 free (name_plus_nl);
1897 free (type_subtypename);
1898
1899 status = set_ce (ct, encoding);
1900
1901 return status;
1902 }
1903
1904
1905 /*
1906 * It's not necessary to update the charset parameter of a Content-Type
1907 * header for a text part. According to RFC 2045 Sec. 6.4, the body
1908 * (content) was originally in the specified charset, "and will be in
1909 * that character set again after decoding."
1910 */
1911 static int
1912 decode_text_parts (CT ct, int encoding, const char *decodetypes,
1913 int *message_mods) {
1914 int status = OK;
1915 int lf_line_endings = 0;
1916
1917 switch (ct->c_type) {
1918 case CT_MULTIPART: {
1919 struct multipart *m = (struct multipart *) ct->c_ctparams;
1920 struct part *part;
1921
1922 /* Should check to see if the body for this part is encoded?
1923 For now, it gets passed along as-is by InitMultiPart(). */
1924 for (part = m->mp_parts; status == OK && part; part = part->mp_next) {
1925 status = decode_text_parts (part->mp_part, encoding, decodetypes,
1926 message_mods);
1927 }
1928 break;
1929 }
1930
1931 case CT_MESSAGE:
1932 if (ct->c_subtype == MESSAGE_EXTERNAL) {
1933 struct exbody *e = (struct exbody *) ct->c_ctparams;
1934
1935 status = decode_text_parts (e->eb_content, encoding, decodetypes,
1936 message_mods);
1937 }
1938 break;
1939
1940 default:
1941 if (! should_decode(decodetypes, ct->c_ctinfo.ci_type, ct->c_ctinfo.ci_subtype)) {
1942 break;
1943 }
1944
1945 lf_line_endings =
1946 ct->c_ctparams && ((struct text *) ct->c_ctparams)->lf_line_endings;
1947
1948 switch (ct->c_encoding) {
1949 case CE_BASE64:
1950 case CE_QUOTED: {
1951 int ct_encoding;
1952
1953 if (decode_part (ct) == OK && ct->c_cefile.ce_file) {
1954 const char *reason = NULL;
1955
1956 if ((ct_encoding = content_encoding (ct, &reason)) == CE_BINARY
1957 && encoding != CE_BINARY) {
1958 /* The decoding isn't acceptable so discard it.
1959 Leave status as OK to allow other transformations. */
1960 if (verbosw) {
1961 report (NULL, ct->c_partno, ct->c_file,
1962 "will not decode%s because it is binary (%s)",
1963 ct->c_partno ? ""
1964 : ct->c_ctline ? ct->c_ctline
1965 : "",
1966 reason);
1967 }
1968 (void) m_unlink (ct->c_cefile.ce_file);
1969 free (ct->c_cefile.ce_file);
1970 ct->c_cefile.ce_file = NULL;
1971 } else if (ct->c_encoding == CE_QUOTED &&
1972 ct_encoding == CE_8BIT && encoding == CE_7BIT) {
1973 /* The decoding isn't acceptable so discard it.
1974 Leave status as OK to allow other transformations. */
1975 if (verbosw) {
1976 report (NULL, ct->c_partno, ct->c_file,
1977 "will not decode%s because it is 8bit",
1978 ct->c_partno ? ""
1979 : ct->c_ctline ? ct->c_ctline
1980 : "");
1981 }
1982 (void) m_unlink (ct->c_cefile.ce_file);
1983 free (ct->c_cefile.ce_file);
1984 ct->c_cefile.ce_file = NULL;
1985 } else {
1986 int enc;
1987
1988 if (ct_encoding == CE_BINARY) {
1989 enc = CE_BINARY;
1990 } else if (ct_encoding == CE_8BIT && encoding == CE_7BIT) {
1991 enc = CE_QUOTED;
1992 } else {
1993 enc = ct_encoding;
1994 }
1995 if (set_ce (ct, enc) == OK) {
1996 ++*message_mods;
1997 if (verbosw) {
1998 report (NULL, ct->c_partno, ct->c_file, "decode%s",
1999 ct->c_ctline ? ct->c_ctline : "");
2000 }
2001 if (lf_line_endings) {
2002 strip_crs (ct, message_mods);
2003 }
2004 } else {
2005 status = NOTOK;
2006 }
2007 }
2008 } else {
2009 status = NOTOK;
2010 }
2011 break;
2012 }
2013 case CE_8BIT:
2014 case CE_7BIT:
2015 if (lf_line_endings) {
2016 strip_crs (ct, message_mods);
2017 }
2018 break;
2019 default:
2020 break;
2021 }
2022
2023 break;
2024 }
2025
2026 return status;
2027 }
2028
2029
2030 /*
2031 * Determine if the part with type[/subtype] should be decoded, according to
2032 * decodetypes (which came from the -decodetypes switch).
2033 */
2034 static int
2035 should_decode(const char *decodetypes, const char *type, const char *subtype) {
2036 /* Quick search for matching type[/subtype] in decodetypes: bracket
2037 decodetypes with commas, then search for ,type, and ,type/subtype, in
2038 it. */
2039
2040 int found_match = 0;
2041 char *delimited_decodetypes = concat(",", decodetypes, ",", NULL);
2042 char *delimited_type = concat(",", type, ",", NULL);
2043
2044 if (nmh_strcasestr(delimited_decodetypes, delimited_type)) {
2045 found_match = 1;
2046 } else if (subtype != NULL) {
2047 char *delimited_type_subtype =
2048 concat(",", type, "/", subtype, ",", NULL);
2049
2050 if (nmh_strcasestr(delimited_decodetypes, delimited_type_subtype)) {
2051 found_match = 1;
2052 }
2053 free(delimited_type_subtype);
2054 }
2055
2056 free(delimited_type);
2057 free(delimited_decodetypes);
2058
2059 return found_match;
2060 }
2061
2062
2063 /*
2064 * See if the decoded content is 7bit, 8bit, or binary. It's binary
2065 * if it has any NUL characters, a CR not followed by a LF, or lines
2066 * greater than 998 characters in length. If binary, reason is set
2067 * to a string explaining why.
2068 */
2069 static int
2070 content_encoding (CT ct, const char **reason) {
2071 CE ce = &ct->c_cefile;
2072 int encoding = CE_7BIT;
2073
2074 if (ce->ce_file) {
2075 size_t line_len = 0;
2076 char buffer[BUFSIZ];
2077 size_t inbytes;
2078
2079 if (! ce->ce_fp && (ce->ce_fp = fopen (ce->ce_file, "r")) == NULL) {
2080 advise (ce->ce_file, "unable to open for reading");
2081 return CE_UNKNOWN;
2082 }
2083
2084 fseeko (ce->ce_fp, 0L, SEEK_SET);
2085 while (encoding != CE_BINARY &&
2086 (inbytes = fread (buffer, 1, sizeof buffer, ce->ce_fp)) > 0) {
2087 char *cp;
2088 size_t i;
2089 int last_char_was_cr = 0;
2090
2091 for (i = 0, cp = buffer; i < inbytes; ++i, ++cp) {
2092 if (*cp == '\0' || ++line_len > 998 ||
2093 (*cp != '\n' && last_char_was_cr)) {
2094 encoding = CE_BINARY;
2095 if (*cp == '\0') {
2096 *reason = "null character";
2097 } else if (line_len > 998) {
2098 *reason = "line length > 998";
2099 } else if (*cp != '\n' && last_char_was_cr) {
2100 *reason = "CR not followed by LF";
2101 } else {
2102 /* Should not reach this. */
2103 *reason = "";
2104 }
2105 break;
2106 } else if (*cp == '\n') {
2107 line_len = 0;
2108 } else if (! isascii ((unsigned char) *cp)) {
2109 encoding = CE_8BIT;
2110 }
2111
2112 last_char_was_cr = *cp == '\r' ? 1 : 0;
2113 }
2114 }
2115
2116 fclose (ce->ce_fp);
2117 ce->ce_fp = NULL;
2118 } /* else should never happen */
2119
2120 return encoding;
2121 }
2122
2123
2124 /*
2125 * Strip carriage returns from content.
2126 */
2127 static int
2128 strip_crs (CT ct, int *message_mods) {
2129 char *charset = content_charset (ct);
2130 int status = OK;
2131
2132 /* Only strip carriage returns if content is ASCII or another
2133 charset that has the same readily recognizable CR followed by a
2134 LF. We can include UTF-8 here because if the high-order bit of
2135 a UTF-8 byte is 0, then it must be a single-byte ASCII
2136 character. */
2137 if (! strcasecmp (charset, "US-ASCII") ||
2138 ! strcasecmp (charset, "UTF-8") ||
2139 ! strncasecmp (charset, "ISO-8859-", 9) ||
2140 ! strncasecmp (charset, "WINDOWS-12", 10)) {
2141 char **file = NULL;
2142 FILE **fp = NULL;
2143 size_t begin;
2144 size_t end;
2145 int has_crs = 0;
2146 int opened_input_file = 0;
2147
2148 if (ct->c_cefile.ce_file) {
2149 file = &ct->c_cefile.ce_file;
2150 fp = &ct->c_cefile.ce_fp;
2151 begin = end = 0;
2152 } else if (ct->c_file) {
2153 file = &ct->c_file;
2154 fp = &ct->c_fp;
2155 begin = (size_t) ct->c_begin;
2156 end = (size_t) ct->c_end;
2157 } /* else don't know where the content is */
2158
2159 if (file && *file && fp) {
2160 if (! *fp) {
2161 if ((*fp = fopen (*file, "r")) == NULL) {
2162 advise (*file, "unable to open for reading");
2163 status = NOTOK;
2164 } else {
2165 opened_input_file = 1;
2166 }
2167 }
2168 }
2169
2170 if (fp && *fp) {
2171 char buffer[BUFSIZ];
2172 size_t bytes_read;
2173 size_t bytes_to_read =
2174 end > 0 && end > begin ? end - begin : sizeof buffer;
2175
2176 fseeko (*fp, begin, SEEK_SET);
2177 while ((bytes_read = fread (buffer, 1,
2178 min (bytes_to_read, sizeof buffer),
2179 *fp)) > 0) {
2180 /* Look for CR followed by a LF. This is supposed to
2181 be text so there should be LF's. If not, don't
2182 modify the content. */
2183 char *cp;
2184 size_t i;
2185 int last_char_was_cr = 0;
2186
2187 if (end > 0) { bytes_to_read -= bytes_read; }
2188
2189 for (i = 0, cp = buffer; i < bytes_read; ++i, ++cp) {
2190 if (*cp == '\n' && last_char_was_cr) {
2191 has_crs = 1;
2192 break;
2193 }
2194
2195 last_char_was_cr = *cp == '\r' ? 1 : 0;
2196 }
2197 }
2198
2199 if (has_crs) {
2200 int fd;
2201 char *stripped_content_file;
2202 char *tempfile = m_mktemp2 (NULL, invo_name, &fd, NULL);
2203
2204 if (tempfile == NULL) {
2205 adios (NULL, "unable to create temporary file in %s",
2206 get_temp_dir());
2207 }
2208 stripped_content_file = add (tempfile, NULL);
2209
2210 /* Strip each CR before a LF from the content. */
2211 fseeko (*fp, begin, SEEK_SET);
2212 while ((bytes_read = fread (buffer, 1, sizeof buffer, *fp)) >
2213 0) {
2214 char *cp;
2215 size_t i;
2216 int last_char_was_cr = 0;
2217
2218 for (i = 0, cp = buffer; i < bytes_read; ++i, ++cp) {
2219 if (*cp == '\r') {
2220 last_char_was_cr = 1;
2221 } else if (last_char_was_cr) {
2222 if (*cp != '\n') {
2223 if (write (fd, "\r", 1) < 0) {
2224 advise (tempfile, "CR write");
2225 }
2226 }
2227 if (write (fd, cp, 1) < 0) {
2228 advise (tempfile, "write");
2229 }
2230 last_char_was_cr = 0;
2231 } else {
2232 if (write (fd, cp, 1) < 0) {
2233 advise (tempfile, "write");
2234 }
2235 last_char_was_cr = 0;
2236 }
2237 }
2238 }
2239
2240 if (close (fd)) {
2241 admonish (NULL, "unable to write temporary file %s",
2242 stripped_content_file);
2243 (void) m_unlink (stripped_content_file);
2244 status = NOTOK;
2245 } else {
2246 /* Replace the decoded file with the converted one. */
2247 if (ct->c_cefile.ce_file && ct->c_cefile.ce_unlink)
2248 (void) m_unlink (ct->c_cefile.ce_file);
2249
2250 mh_xfree(ct->c_cefile.ce_file);
2251 ct->c_cefile.ce_file = stripped_content_file;
2252 ct->c_cefile.ce_unlink = 1;
2253
2254 ++*message_mods;
2255 if (verbosw) {
2256 report (NULL, ct->c_partno,
2257 begin == 0 && end == 0 ? "" : *file,
2258 "stripped CRs");
2259 }
2260 }
2261 }
2262
2263 if (opened_input_file) {
2264 fclose (*fp);
2265 *fp = NULL;
2266 }
2267 }
2268 }
2269
2270 free (charset);
2271
2272 return status;
2273 }
2274
2275
2276 /*
2277 * Add/update, if necessary, the message C-T-E, based on the least restrictive
2278 * of the part C-T-E's.
2279 */
2280 static void
2281 update_cte (CT ct) {
2282 const int least_restrictive_enc = least_restrictive_encoding (ct);
2283
2284 if (least_restrictive_enc != CE_UNKNOWN &&
2285 least_restrictive_enc != CE_7BIT) {
2286 char *cte = concat (" ", ce_str (least_restrictive_enc), "\n", NULL);
2287 HF hf;
2288 int found_cte = 0;
2289
2290 /* Update/add Content-Transfer-Encoding header field. */
2291 for (hf = ct->c_first_hf; hf; hf = hf->next) {
2292 if (! strcasecmp (ENCODING_FIELD, hf->name)) {
2293 found_cte = 1;
2294 free (hf->value);
2295 hf->value = cte;
2296 }
2297 }
2298 if (! found_cte) {
2299 add_header (ct, add (ENCODING_FIELD, NULL), cte);
2300 }
2301 }
2302 }
2303
2304
2305 /*
2306 * Find the least restrictive encoding (7bit, 8bit, binary) of the parts
2307 * within a message.
2308 */
2309 static int
2310 least_restrictive_encoding (CT ct) {
2311 int encoding = CE_UNKNOWN;
2312
2313 switch (ct->c_type) {
2314 case CT_MULTIPART: {
2315 struct multipart *m = (struct multipart *) ct->c_ctparams;
2316 struct part *part;
2317
2318 for (part = m->mp_parts; part; part = part->mp_next) {
2319 const int part_encoding =
2320 least_restrictive_encoding (part->mp_part);
2321
2322 if (less_restrictive (encoding, part_encoding)) {
2323 encoding = part_encoding;
2324 }
2325 }
2326 break;
2327 }
2328
2329 case CT_MESSAGE:
2330 if (ct->c_subtype == MESSAGE_EXTERNAL) {
2331 struct exbody *e = (struct exbody *) ct->c_ctparams;
2332 const int part_encoding =
2333 least_restrictive_encoding (e->eb_content);
2334
2335 if (less_restrictive (encoding, part_encoding)) {
2336 encoding = part_encoding;
2337 }
2338 }
2339 break;
2340
2341 default: {
2342 if (less_restrictive (encoding, ct->c_encoding)) {
2343 encoding = ct->c_encoding;
2344 }
2345 }}
2346
2347 return encoding;
2348 }
2349
2350
2351 /*
2352 * Return whether the second encoding is less restrictive than the first, where
2353 * "less restrictive" is in the sense used by RFC 2045 Secs. 6.1 and 6.4. So,
2354 * CE_BINARY is less restrictive than CE_8BIT and
2355 * CE_8BIT is less restrictive than CE_7BIT.
2356 */
2357 static int
2358 less_restrictive (int encoding, int second_encoding) {
2359 switch (second_encoding) {
2360 case CE_BINARY:
2361 return encoding != CE_BINARY;
2362 case CE_8BIT:
2363 return encoding != CE_BINARY && encoding != CE_8BIT;
2364 case CE_7BIT:
2365 return encoding != CE_BINARY && encoding != CE_8BIT &&
2366 encoding != CE_7BIT;
2367 default :
2368 return 0;
2369 }
2370 }
2371
2372
2373 /*
2374 * Convert character set of each part.
2375 */
2376 static int
2377 convert_charsets (CT ct, char *dest_charset, int *message_mods) {
2378 int status = OK;
2379
2380 switch (ct->c_type) {
2381 case CT_TEXT:
2382 if (ct->c_subtype == TEXT_PLAIN) {
2383 status = convert_charset (ct, dest_charset, message_mods);
2384 if (status == OK) {
2385 if (verbosw) {
2386 char *ct_charset = content_charset (ct);
2387
2388 report (NULL, ct->c_partno, ct->c_file,
2389 "convert %s to %s", ct_charset, dest_charset);
2390 free (ct_charset);
2391 }
2392 } else {
2393 char *ct_charset = content_charset (ct);
2394
2395 report ("iconv", ct->c_partno, ct->c_file,
2396 "failed to convert %s to %s", ct_charset, dest_charset);
2397 free (ct_charset);
2398 }
2399 }
2400 break;
2401
2402 case CT_MULTIPART: {
2403 struct multipart *m = (struct multipart *) ct->c_ctparams;
2404 struct part *part;
2405
2406 /* Should check to see if the body for this part is encoded?
2407 For now, it gets passed along as-is by InitMultiPart(). */
2408 for (part = m->mp_parts; status == OK && part; part = part->mp_next) {
2409 status =
2410 convert_charsets (part->mp_part, dest_charset, message_mods);
2411 }
2412 break;
2413 }
2414
2415 case CT_MESSAGE:
2416 if (ct->c_subtype == MESSAGE_EXTERNAL) {
2417 struct exbody *e = (struct exbody *) ct->c_ctparams;
2418
2419 status =
2420 convert_charsets (e->eb_content, dest_charset, message_mods);
2421 }
2422 break;
2423
2424 default:
2425 break;
2426 }
2427
2428 return status;
2429 }
2430
2431
2432 /*
2433 * Fix various problems that aren't handled elsewhere. These
2434 * are fixed unconditionally: there are no switches to disable
2435 * them. Currently, "problems" are these:
2436 * 1) remove extraneous semicolon at the end of a header parameter list
2437 * 2) replace RFC 2047 encoding with RFC 2231 encoding of name and
2438 * filename parameters in Content-Type and Content-Disposition
2439 * headers, respectively.
2440 */
2441 static int
2442 fix_always (CT ct, int *message_mods) {
2443 int status = OK;
2444
2445 switch (ct->c_type) {
2446 case CT_MULTIPART: {
2447 struct multipart *m = (struct multipart *) ct->c_ctparams;
2448 struct part *part;
2449
2450 for (part = m->mp_parts; status == OK && part; part = part->mp_next) {
2451 status = fix_always (part->mp_part, message_mods);
2452 }
2453 break;
2454 }
2455
2456 case CT_MESSAGE:
2457 if (ct->c_subtype == MESSAGE_EXTERNAL) {
2458 struct exbody *e = (struct exbody *) ct->c_ctparams;
2459
2460 status = fix_always (e->eb_content, message_mods);
2461 }
2462 break;
2463
2464 default: {
2465 HF hf;
2466
2467 if (ct->c_first_hf) {
2468 fix_filename_encoding (ct);
2469 }
2470
2471 for (hf = ct->c_first_hf; hf; hf = hf->next) {
2472 size_t len = strlen (hf->value);
2473
2474 if (strcasecmp (hf->name, TYPE_FIELD) != 0 &&
2475 strcasecmp (hf->name, DISPO_FIELD) != 0) {
2476 /* Only do this for Content-Type and
2477 Content-Disposition fields because those are the
2478 only headers that parse_mime() warns about. */
2479 continue;
2480 }
2481
2482 /* whitespace following a trailing ';' will be nuked as well */
2483 if (hf->value[len - 1] == '\n') {
2484 while (isspace((unsigned char)(hf->value[len - 2]))) {
2485 if (len-- == 0) { break; }
2486 }
2487 }
2488
2489 if (hf->value[len - 2] == ';') {
2490 /* Remove trailing ';' from parameter value. */
2491 hf->value[len - 2] = '\n';
2492 hf->value[len - 1] = '\0';
2493
2494 /* Also, if Content-Type parameter, remove trailing ';'
2495 from ct->c_ctline. This probably isn't necessary
2496 but can't hurt. */
2497 if (strcasecmp(hf->name, TYPE_FIELD) == 0 && ct->c_ctline) {
2498 size_t l = strlen(ct->c_ctline) - 1;
2499 while (isspace((unsigned char)(ct->c_ctline[l])) ||
2500 ct->c_ctline[l] == ';') {
2501 ct->c_ctline[l--] = '\0';
2502 if (l == 0) { break; }
2503 }
2504 }
2505
2506 ++*message_mods;
2507 if (verbosw) {
2508 report (NULL, ct->c_partno, ct->c_file,
2509 "remove trailing ; from %s parameter value",
2510 hf->name);
2511 }
2512 }
2513 }
2514 }}
2515
2516 return status;
2517 }
2518
2519
2520 /*
2521 * Factor out common code for loops in fix_filename_encoding().
2522 */
2523 static int
2524 fix_filename_param (char *name, char *value, PM *first_pm, PM *last_pm) {
2525 size_t value_len;
2526 int fixed = 0;
2527
2528 if (((value_len = strlen (value)) > 0) &&
2529 strncmp (value, "=?", 2) == 0 &&
2530 strncmp (&value[value_len - 2], "?=", 2) == 0) {
2531 /* Looks like an RFC 2047 encoded parameter. */
2532 char decoded[PATH_MAX + 1];
2533
2534 if (decode_rfc2047 (value, decoded, sizeof decoded)) {
2535 /* Encode using RFC 2231. */
2536 replace_param (first_pm, last_pm, name, decoded, 0);
2537 fixed = 1;
2538 } else {
2539 advise (NULL, "failed to decode %s parameter %s", name, value);
2540 }
2541 }
2542
2543 return fixed;
2544 }
2545
2546
2547 /*
2548 * Replace RFC 2047 encoding with RFC 2231 encoding of name and
2549 * filename parameters in Content-Type and Content-Disposition
2550 * headers, respectively.
2551 */
2552 static int
2553 fix_filename_encoding (CT ct) {
2554 PM pm;
2555 HF hf;
2556 int fixed = 0;
2557
2558 for (pm = ct->c_ctinfo.ci_first_pm; pm; pm = pm->pm_next) {
2559 if (pm->pm_name && pm->pm_value &&
2560 strcasecmp (pm->pm_name, "name") == 0) {
2561 fixed = fix_filename_param (pm->pm_name, pm->pm_value,
2562 &ct->c_ctinfo.ci_first_pm,
2563 &ct->c_ctinfo.ci_last_pm);
2564 }
2565 }
2566
2567 for (pm = ct->c_dispo_first; pm; pm = pm->pm_next) {
2568 if (pm->pm_name && pm->pm_value &&
2569 strcasecmp (pm->pm_name, "filename") == 0) {
2570 fixed = fix_filename_param (pm->pm_name, pm->pm_value,
2571 &ct->c_dispo_first,
2572 &ct->c_dispo_last);
2573 }
2574 }
2575
2576 /* Fix hf values to correspond. */
2577 for (hf = ct->c_first_hf; fixed && hf; hf = hf->next) {
2578 enum { OTHER, TYPE_HEADER, DISPO_HEADER } field = OTHER;
2579
2580 if (strcasecmp (hf->name, TYPE_FIELD) == 0) {
2581 field = TYPE_HEADER;
2582 } else if (strcasecmp (hf->name, DISPO_FIELD) == 0) {
2583 field = DISPO_HEADER;
2584 }
2585
2586 if (field != OTHER) {
2587 const char *const semicolon_loc = strchr (hf->value, ';');
2588
2589 if (semicolon_loc) {
2590 const size_t len =
2591 strlen (hf->name) + 1 + semicolon_loc - hf->value;
2592 const char *const params =
2593 output_params (len,
2594 field == TYPE_HEADER
2595 ? ct->c_ctinfo.ci_first_pm
2596 : ct->c_dispo_first,
2597 NULL, 0);
2598 const char *const new_params = concat (params, "\n", NULL);
2599
2600 replace_substring (&hf->value, semicolon_loc, new_params);
2601 free((void *)new_params); /* Cast away const. Sigh. */
2602 free((void *)params);
2603 } else {
2604 advise (NULL, "did not find semicolon in %s:%s\n",
2605 hf->name, hf->value);
2606 }
2607 }
2608 }
2609
2610 return OK;
2611 }
2612
2613
2614 /*
2615 * Output content in input file to output file.
2616 */
2617 static int
2618 write_content (CT ct, const char *input_filename, char *outfile, int modify_inplace,
2619 int message_mods) {
2620 int status = OK;
2621
2622 if (modify_inplace) {
2623 if (message_mods > 0) {
2624 if ((status = output_message (ct, outfile)) == OK) {
2625 char *infile = input_filename
2626 ? add (input_filename, NULL)
2627 : add (ct->c_file ? ct->c_file : "-", NULL);
2628
2629 if (remove_file (infile) == OK) {
2630 if (rename (outfile, infile)) {
2631 /* Rename didn't work, possibly because of an
2632 attempt to rename across filesystems. Try
2633 brute force copy. */
2634 int old = open (outfile, O_RDONLY);
2635 int new =
2636 open (infile, O_WRONLY | O_CREAT, m_gmprot ());
2637 int i = -1;
2638
2639 if (old != -1 && new != -1) {
2640 char buffer[BUFSIZ];
2641
2642 while ((i = read (old, buffer, sizeof buffer)) >
2643 0) {
2644 if (write (new, buffer, i) != i) {
2645 i = -1;
2646 break;
2647 }
2648 }
2649 }
2650 if (new != -1) { close (new); }
2651 if (old != -1) { close (old); }
2652 (void) m_unlink (outfile);
2653
2654 if (i < 0) {
2655 /* The -file argument processing used path() to
2656 expand filename to absolute path. */
2657 int file = ct->c_file && ct->c_file[0] == '/';
2658
2659 admonish (NULL, "unable to rename %s %s to %s",
2660 file ? "file" : "message", outfile,
2661 infile);
2662 status = NOTOK;
2663 }
2664 }
2665 } else {
2666 admonish (NULL, "unable to remove input file %s, "
2667 "not modifying it", infile);
2668 (void) m_unlink (outfile);
2669 status = NOTOK;
2670 }
2671
2672 free (infile);
2673 } else {
2674 status = NOTOK;
2675 }
2676 } else {
2677 /* No modifications and didn't need the tmp outfile. */
2678 (void) m_unlink (outfile);
2679 }
2680 } else {
2681 /* Output is going to some file. Produce it whether or not
2682 there were modifications. */
2683 status = output_message (ct, outfile);
2684 }
2685
2686 flush_errors ();
2687 return status;
2688 }
2689
2690
2691 /*
2692 * parse_mime() does not set lf_line_endings in struct text, so use this
2693 * function to do it. It touches the parts the decodetypes identifies.
2694 */
2695 static void
2696 set_text_ctparams(CT ct, char *decodetypes, int lf_line_endings) {
2697 switch (ct->c_type) {
2698 case CT_MULTIPART: {
2699 struct multipart *m = (struct multipart *) ct->c_ctparams;
2700 struct part *part;
2701
2702 for (part = m->mp_parts; part; part = part->mp_next) {
2703 set_text_ctparams(part->mp_part, decodetypes, lf_line_endings);
2704 }
2705 break;
2706 }
2707
2708 case CT_MESSAGE:
2709 if (ct->c_subtype == MESSAGE_EXTERNAL) {
2710 struct exbody *e = (struct exbody *) ct->c_ctparams;
2711
2712 set_text_ctparams(e->eb_content, decodetypes, lf_line_endings);
2713 }
2714 break;
2715
2716 default:
2717 if (should_decode(decodetypes, ct->c_ctinfo.ci_type, ct->c_ctinfo.ci_subtype)) {
2718 if (ct->c_ctparams == NULL) {
2719 ct->c_ctparams = mh_xcalloc(1, sizeof (struct text));
2720 }
2721 ((struct text *) ct->c_ctparams)->lf_line_endings = lf_line_endings;
2722 }
2723 }
2724 }
2725
2726
2727 /*
2728 * If "rmmproc" is defined, call that to remove the file. Otherwise,
2729 * use the standard MH backup file.
2730 */
2731 static int
2732 remove_file (const char *file) {
2733 if (rmmproc) {
2734 char *rmm_command = concat (rmmproc, " ", file, NULL);
2735 int status = system (rmm_command);
2736
2737 free (rmm_command);
2738 return WIFEXITED (status) ? WEXITSTATUS (status) : NOTOK;
2739 }
2740 /* This is OK for a non-message file, it still uses the
2741 BACKUP_PREFIX form. The backup file will be in the same
2742 directory as file. */
2743 return rename (file, m_backup (file));
2744 }
2745
2746
2747 /*
2748 * Output formatted message to user.
2749 */
2750 static void
2751 report (char *what, char *partno, char *filename, char *message, ...) {
2752 va_list args;
2753 char *fmt;
2754
2755 if (verbosw) {
2756 va_start (args, message);
2757 fmt = concat (filename, partno ? " part " : ", ",
2758 partno ? partno : "", partno ? ", " : "", message, NULL);
2759
2760 advertise (what, NULL, fmt, args);
2761
2762 free (fmt);
2763 va_end (args);
2764 }
2765 }
2766
2767
2768 static void
2769 pipeser (int i)
2770 {
2771 if (i == SIGQUIT) {
2772 fflush (stdout);
2773 fprintf (stderr, "\n");
2774 fflush (stderr);
2775 }
2776
2777 done (1);
2778 /* NOTREACHED */
2779 }
Fork of https://git.savannah.nongnu.org/cgit/nmh.git