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