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