]> diplodocus.org Git - nmh/blob - uip/mhfixmsg.c
projects / nmh / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
man/*.man: Use italic for emphasis, not bold or SHOUTING.
[nmh] / uip / mhfixmsg.c
1 /*
2 * mhfixmsg.c -- rewrite a message with various transformations
3 *
4 * This code is Copyright (c) 2002 and 2013, by the authors of nmh.
5 * See the COPYRIGHT file in the root directory of the nmh
6 * distribution for complete copyright information.
7 */
8
9 #include <h/mh.h>
10 #include <h/mime.h>
11 #include <h/mhparse.h>
12 #include <h/utils.h>
13 #include <h/signals.h>
14 #include <fcntl.h>
15
16 #define MHFIXMSG_SWITCHES \
17 X("decodetext 8bit|7bit|binary", 0, DECODETEXTSW) \
18 X("nodecodetext", 0, NDECODETEXTSW) \
19 X("decodetypes", 0, DECODETYPESW) \
20 X("crlflinebreaks", 0, CRLFLINEBREAKSSW) \
21 X("nocrlflinebreaks", 0, NCRLFLINEBREAKSSW) \
22 X("textcharset", 0, TEXTCHARSETSW) \
23 X("notextcharset", 0, NTEXTCHARSETSW) \
24 X("reformat", 0, REFORMATSW) \
25 X("noreformat", 0, NREFORMATSW) \
26 X("replacetextplain", 0, REPLACETEXTPLAINSW) \
27 X("noreplacetextplain", 0, NREPLACETEXTPLAINSW) \
28 X("fixboundary", 0, FIXBOUNDARYSW) \
29 X("nofixboundary", 0, NFIXBOUNDARYSW) \
30 X("fixcte", 0, FIXCOMPOSITECTESW) \
31 X("nofixcte", 0, NFIXCOMPOSITECTESW) \
32 X("fixtype mimetype", 0, FIXTYPESW) \
33 X("file file", 0, FILESW) \
34 X("outfile file", 0, OUTFILESW) \
35 X("rmmproc program", 0, RPROCSW) \
36 X("normmproc", 0, NRPRCSW) \
37 X("changecur", 0, CHGSW) \
38 X("nochangecur", 0, NCHGSW) \
39 X("verbose", 0, VERBSW) \
40 X("noverbose", 0, NVERBSW) \
41 X("version", 0, VERSIONSW) \
42 X("help", 0, HELPSW) \
43
44 #define X(sw, minchars, id) id,
45 DEFINE_SWITCH_ENUM(MHFIXMSG);
46 #undef X
47
48 #define X(sw, minchars, id) { sw, minchars, id },
49 DEFINE_SWITCH_ARRAY(MHFIXMSG, switches);
50 #undef X
51
52
53 int verbosw;
54 int debugsw; /* Needed by mhparse.c. */
55
56 #define quitser pipeser
57
58 /* mhparse.c */
59 extern int skip_mp_cte_check; /* flag to InitMultiPart */
60 extern int suppress_bogus_mp_content_warning; /* flag to InitMultiPart */
61 extern int bogus_mp_content; /* flag from InitMultiPart */
62 /* flags to/from parse_header_attrs */
63 extern int suppress_extraneous_trailing_semicolon_warning;
64
65 /* mhoutsbr.c */
66 int output_message_fp (CT, FILE *, char *);
67
68 /* mhmisc.c */
69 void flush_errors (void);
70
71 /* mhfree.c */
72 extern CT *cts;
73
74 /*
75 * static prototypes
76 */
77 typedef struct fix_transformations {
78 int fixboundary;
79 int fixcompositecte;
80 svector_t fixtypes;
81 int reformat;
82 int replacetextplain;
83 int decodetext;
84 char *decodetypes;
85 /* Whether to use CRLF linebreaks, per RFC 2046 Sec. 4.1.1, par.1. */
86 int lf_line_endings;
87 char *textcharset;
88 } fix_transformations;
89
90 int mhfixmsgsbr (CT *, char *, const fix_transformations *, FILE **, char *,
91 FILE **);
92 static int fix_boundary (CT *, int *);
93 static int copy_input_to_output (const char *, FILE *, const char *, FILE *);
94 static int get_multipart_boundary (CT, char **);
95 static int replace_boundary (CT, char *, char *);
96 static int fix_types (CT, svector_t, int *);
97 static char *replace_substring (char **, const char *, const char *);
98 static char *remove_parameter (char *, const char *);
99 static int fix_composite_cte (CT, int *);
100 static int set_ce (CT, int);
101 static int ensure_text_plain (CT *, CT, int *, int);
102 static int find_textplain_sibling (CT, int, int *);
103 static int insert_new_text_plain_part (CT, int, CT);
104 static CT build_text_plain_part (CT);
105 static int insert_into_new_mp_alt (CT *, int *);
106 static CT divide_part (CT);
107 static void copy_ctinfo (CI, CI);
108 static int decode_part (CT);
109 static int reformat_part (CT, char *, char *, char *, int);
110 static CT build_multipart_alt (CT, CT, int, int);
111 static int boundary_in_content (FILE **, char *, const char *);
112 static void transfer_noncontent_headers (CT, CT);
113 static int set_ct_type (CT, int type, int subtype, int encoding);
114 static int decode_text_parts (CT, int, const char *, int *);
115 static int should_decode(const char *, const char *, const char *);
116 static int content_encoding (CT, const char **);
117 static int strip_crs (CT, int *);
118 static void update_cte (CT);
119 static int least_restrictive_encoding (CT);
120 static int less_restrictive (int, int);
121 static int convert_charsets (CT, char *, int *);
122 static int fix_always (CT, int *);
123 static int fix_filename_param (char *, char *, PM *, PM *);
124 static int fix_filename_encoding (CT);
125 static int write_content (CT, const char *, char *, FILE *, int, int);
126 static void set_text_ctparams(CT, char *, int);
127 static int remove_file (const char *);
128 static void report (char *, char *, char *, char *, ...);
129 static void pipeser (int);
130
131
132 int
133 main (int argc, char **argv) {
134 int msgnum;
135 char *cp, *file = NULL, *folder = NULL;
136 char *maildir = NULL, buf[100], *outfile = NULL;
137 char **argp, **arguments;
138 struct msgs_array msgs = { 0, 0, NULL };
139 struct msgs *mp = NULL;
140 CT *ctp;
141 FILE *fp, *infp = NULL, *outfp = NULL;
142 int using_stdin = 0;
143 int chgflag = 1;
144 int status = OK;
145 fix_transformations fx;
146 fx.reformat = fx.fixcompositecte = fx.fixboundary = 1;
147 fx.fixtypes = NULL;
148 fx.replacetextplain = 0;
149 fx.decodetext = CE_8BIT;
150 fx.decodetypes = "text,application/ics"; /* Default, per man page. */
151 fx.lf_line_endings = 0;
152 fx.textcharset = NULL;
153
154 if (nmh_init(argv[0], 2)) { return 1; }
155
156 arguments = getarguments (invo_name, argc, argv, 1);
157 argp = arguments;
158
159 /*
160 * Parse arguments
161 */
162 while ((cp = *argp++)) {
163 if (*cp == '-') {
164 switch (smatch (++cp, switches)) {
165 case AMBIGSW:
166 ambigsw (cp, switches);
167 done (1);
168 case UNKWNSW:
169 adios (NULL, "-%s unknown", cp);
170
171 case HELPSW:
172 snprintf (buf, sizeof buf, "%s [+folder] [msgs] [switches]",
173 invo_name);
174 print_help (buf, switches, 1);
175 done (0);
176 case VERSIONSW:
177 print_version(invo_name);
178 done (0);
179
180 case DECODETEXTSW:
181 if (! (cp = *argp++) || *cp == '-') {
182 adios (NULL, "missing argument to %s", argp[-2]);
183 }
184 if (! strcasecmp (cp, "8bit")) {
185 fx.decodetext = CE_8BIT;
186 } else if (! strcasecmp (cp, "7bit")) {
187 fx.decodetext = CE_7BIT;
188 } else if (! strcasecmp (cp, "binary")) {
189 fx.decodetext = CE_BINARY;
190 } else {
191 adios (NULL, "invalid argument to %s", argp[-2]);
192 }
193 continue;
194 case NDECODETEXTSW:
195 fx.decodetext = 0;
196 continue;
197 case DECODETYPESW:
198 if (! (cp = *argp++) || *cp == '-') {
199 adios (NULL, "missing argument to %s", argp[-2]);
200 }
201 fx.decodetypes = cp;
202 continue;
203 case CRLFLINEBREAKSSW:
204 fx.lf_line_endings = 0;
205 continue;
206 case NCRLFLINEBREAKSSW:
207 fx.lf_line_endings = 1;
208 continue;
209 case TEXTCHARSETSW:
210 if (! (cp = *argp++) || (*cp == '-' && cp[1])) {
211 adios (NULL, "missing argument to %s", argp[-2]);
212 }
213 fx.textcharset = cp;
214 continue;
215 case NTEXTCHARSETSW:
216 fx.textcharset = 0;
217 continue;
218 case FIXBOUNDARYSW:
219 fx.fixboundary = 1;
220 continue;
221 case NFIXBOUNDARYSW:
222 fx.fixboundary = 0;
223 continue;
224 case FIXCOMPOSITECTESW:
225 fx.fixcompositecte = 1;
226 continue;
227 case NFIXCOMPOSITECTESW:
228 fx.fixcompositecte = 0;
229 continue;
230 case FIXTYPESW:
231 if (! (cp = *argp++) || (*cp == '-' && cp[1])) {
232 adios (NULL, "missing argument to %s", argp[-2]);
233 }
234 if (! strncasecmp (cp, "multipart/", 10) ||
235 ! strncasecmp (cp, "message/", 8)) {
236 adios (NULL, "-fixtype %s not allowed", cp);
237 } else if (! strchr (cp, '/')) {
238 adios (NULL, "-fixtype requires type/subtype");
239 }
240 if (fx.fixtypes == NULL) { fx.fixtypes = svector_create (10); }
241 svector_push_back (fx.fixtypes, cp);
242 continue;
243 case REFORMATSW:
244 fx.reformat = 1;
245 continue;
246 case NREFORMATSW:
247 fx.reformat = 0;
248 continue;
249 case REPLACETEXTPLAINSW:
250 fx.replacetextplain = 1;
251 continue;
252 case NREPLACETEXTPLAINSW:
253 fx.replacetextplain = 0;
254 continue;
255 case FILESW:
256 if (! (cp = *argp++) || (*cp == '-' && cp[1])) {
257 adios (NULL, "missing argument to %s", argp[-2]);
258 }
259 file = *cp == '-' ? mh_xstrdup (cp) : path (cp, TFILE);
260 continue;
261 case OUTFILESW:
262 if (! (cp = *argp++) || (*cp == '-' && cp[1])) {
263 adios (NULL, "missing argument to %s", argp[-2]);
264 }
265 outfile = *cp == '-' ? mh_xstrdup (cp) : path (cp, TFILE);
266 continue;
267 case RPROCSW:
268 if (!(rmmproc = *argp++) || *rmmproc == '-') {
269 adios (NULL, "missing argument to %s", argp[-2]);
270 }
271 continue;
272 case NRPRCSW:
273 rmmproc = NULL;
274 continue;
275 case CHGSW:
276 chgflag = 1;
277 continue;
278 case NCHGSW:
279 chgflag = 0;
280 continue;
281 case VERBSW:
282 verbosw = 1;
283 continue;
284 case NVERBSW:
285 verbosw = 0;
286 continue;
287 }
288 }
289 if (*cp == '+' || *cp == '@') {
290 if (folder) {
291 adios (NULL, "only one folder at a time!");
292 } else {
293 folder = pluspath (cp);
294 }
295 } else {
296 if (*cp == '/') {
297 /* Interpret a full path as a filename, not a message. */
298 file = mh_xstrdup (cp);
299 } else {
300 app_msgarg (&msgs, cp);
301 }
302 }
303 }
304
305 SIGNAL (SIGQUIT, quitser);
306 SIGNAL (SIGPIPE, pipeser);
307
308 /*
309 * Read the standard profile setup
310 */
311 if ((fp = fopen (cp = etcpath ("mhn.defaults"), "r"))) {
312 readconfig ((struct node **) 0, fp, cp, 0);
313 fclose (fp);
314 }
315
316 suppress_bogus_mp_content_warning = skip_mp_cte_check = 1;
317 suppress_extraneous_trailing_semicolon_warning = 1;
318
319 if (! context_find ("path")) {
320 free (path ("./", TFOLDER));
321 }
322
323 if (file && msgs.size) {
324 adios (NULL, "cannot specify msg and file at same time!");
325 }
326
327 if (outfile) {
328 /* Open the outfile now, so we don't have to risk opening it
329 after running out of fds. */
330 if (strcmp (outfile, "-") == 0) {
331 outfp = stdout;
332 } else if ((outfp = fopen (outfile, "w")) == NULL) {
333 adios (outfile, "unable to open for writing");
334 }
335 }
336
337 /*
338 * check if message is coming from file
339 */
340 if (file) {
341 /* If file is stdin, create a tmp file name before parse_mime()
342 has a chance, because it might put in on a different
343 filesystem than the output file. Instead, put it in the
344 user's preferred tmp directory. */
345 CT ct;
346
347 if (! strcmp ("-", file)) {
348 int fd;
349 char *cp;
350
351 using_stdin = 1;
352
353 if ((cp = m_mktemp2 (NULL, invo_name, &fd, NULL)) == NULL) {
354 adios (NULL, "unable to create temporary file in %s",
355 get_temp_dir());
356 } else {
357 free (file);
358 file = mh_xstrdup (cp);
359 cpydata (STDIN_FILENO, fd, "-", file);
360 }
361
362 if (close (fd)) {
363 (void) m_unlink (file);
364 adios (NULL, "failed to write temporary file");
365 }
366 }
367
368 cts = mh_xcalloc(2, sizeof *cts);
369 ctp = cts;
370
371 if ((ct = parse_mime (file))) {
372 set_text_ctparams(ct, fx.decodetypes, fx.lf_line_endings);
373 *ctp++ = ct;
374 } else {
375 inform("unable to parse message from file %s", file);
376 status = NOTOK;
377
378 /* If there's an outfile, pass the input message unchanged, so the
379 message won't get dropped from a pipeline. */
380 if (outfile) {
381 /* Something went wrong. Output might be expected, such as if
382 this were run as a filter. Just copy the input to the
383 output. */
384 if ((infp = fopen (file, "r")) == NULL) {
385 adios (file, "unable to open for reading");
386 }
387
388 if (copy_input_to_output (file, infp, outfile, outfp) != OK) {
389 inform("unable to copy message to %s, "
390 "it might be lost\n", outfile);
391 }
392
393 fclose (infp);
394 infp = NULL;
395 }
396 }
397 } else {
398 /*
399 * message(s) are coming from a folder
400 */
401 CT ct;
402
403 if (! msgs.size) {
404 app_msgarg(&msgs, "cur");
405 }
406 if (! folder) {
407 folder = getfolder (1);
408 }
409 maildir = mh_xstrdup(m_maildir (folder));
410
411 /* chdir so that error messages, esp. from MIME parser, just
412 refer to the message and not its path. */
413 if (chdir (maildir) == NOTOK) {
414 adios (maildir, "unable to change directory to");
415 }
416
417 /* read folder and create message structure */
418 if (! (mp = folder_read (folder, 1))) {
419 adios (NULL, "unable to read folder %s", folder);
420 }
421
422 /* check for empty folder */
423 if (mp->nummsg == 0) {
424 adios (NULL, "no messages in %s", folder);
425 }
426
427 /* parse all the message ranges/sequences and set SELECTED */
428 for (msgnum = 0; msgnum < msgs.size; msgnum++)
429 if (! m_convert (mp, msgs.msgs[msgnum])) {
430 done (1);
431 }
432 seq_setprev (mp); /* set the previous-sequence */
433
434 cts = mh_xcalloc(mp->numsel + 1, sizeof *cts);
435 ctp = cts;
436
437 for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
438 if (is_selected(mp, msgnum)) {
439 char *msgnam = m_name (msgnum);
440
441 if ((ct = parse_mime (msgnam))) {
442 set_text_ctparams(ct, fx.decodetypes, fx.lf_line_endings);
443 *ctp++ = ct;
444 } else {
445 inform("unable to parse message %s", msgnam);
446 status = NOTOK;
447
448 /* If there's an outfile, pass the input message
449 unchanged, so the message won't get dropped from a
450 pipeline. */
451 if (outfile) {
452 /* Something went wrong. Output might be expected,
453 such as if this were run as a filter. Just copy
454 the input to the output. */
455 /* Can't use path() here because 1) it might have been
456 called before and it caches the pwd, and 2) we call
457 chdir() after that. */
458 char *input_filename =
459 concat (maildir, "/", msgnam, NULL);
460
461 if ((infp = fopen (input_filename, "r")) == NULL) {
462 adios (input_filename,
463 "unable to open for reading");
464 }
465
466 if (copy_input_to_output (input_filename, infp,
467 outfile, outfp) != OK) {
468 inform("unable to copy message to %s, "
469 "it might be lost\n", outfile);
470 }
471
472 fclose (infp);
473 infp = NULL;
474 free (input_filename);
475 }
476 }
477 }
478 }
479
480 if (chgflag) {
481 seq_setcur (mp, mp->hghsel); /* update current message */
482 }
483 seq_save (mp); /* synchronize sequences */
484 context_replace (pfolder, folder);/* update current folder */
485 context_save (); /* save the context file */
486 }
487
488 if (*cts) {
489 for (ctp = cts; *ctp; ++ctp) {
490 status += mhfixmsgsbr (ctp, maildir, &fx, &infp, outfile, &outfp);
491 free_content (*ctp);
492
493 if (using_stdin) {
494 (void) m_unlink (file);
495
496 if (! outfile) {
497 /* Just calling m_backup() unlinks the backup file. */
498 (void) m_backup (file);
499 }
500 }
501 }
502 } else {
503 status = 1;
504 }
505
506 mh_xfree(maildir);
507 free (cts);
508
509 if (fx.fixtypes != NULL) { svector_free (fx.fixtypes); }
510 if (infp) { fclose (infp); } /* even if stdin */
511 if (outfp) { fclose (outfp); } /* even if stdout */
512 free (outfile);
513 free (file);
514 free (folder);
515 free (arguments);
516
517 done (status);
518 return NOTOK;
519 }
520
521
522 /*
523 * Apply transformations to one message.
524 */
525 int
526 mhfixmsgsbr (CT *ctp, char *maildir, const fix_transformations *fx,
527 FILE **infp, char *outfile, FILE **outfp) {
528 /* Store input filename in case one of the transformations, i.e.,
529 fix_boundary(), rewrites to a tmp file. */
530 char *input_filename = maildir
531 ? concat (maildir, "/", (*ctp)->c_file, NULL)
532 : mh_xstrdup ((*ctp)->c_file);
533 int modify_inplace = 0;
534 int message_mods = 0;
535 int status = OK;
536
537 /* Though the input file won't need to be opened if everything goes
538 well, do it here just in case there's a failure, and that failure is
539 running out of file descriptors. */
540 if ((*infp = fopen (input_filename, "r")) == NULL) {
541 adios (input_filename, "unable to open for reading");
542 }
543
544 if (outfile == NULL) {
545 modify_inplace = 1;
546
547 if ((*ctp)->c_file) {
548 char *tempfile;
549 /* outfp will be closed by the caller */
550 if ((tempfile = m_mktemp2 (NULL, invo_name, NULL, outfp)) ==
551 NULL) {
552 adios (NULL, "unable to create temporary file in %s",
553 get_temp_dir());
554 }
555 outfile = mh_xstrdup (tempfile);
556 } else {
557 adios (NULL, "missing both input and output filenames\n");
558 }
559 } /* else *outfp was defined by caller */
560
561 reverse_alternative_parts (*ctp);
562 status = fix_always (*ctp, &message_mods);
563 if (status == OK && fx->fixboundary) {
564 status = fix_boundary (ctp, &message_mods);
565 }
566 if (status == OK && fx->fixtypes != NULL) {
567 status = fix_types (*ctp, fx->fixtypes, &message_mods);
568 }
569 if (status == OK && fx->fixcompositecte) {
570 status = fix_composite_cte (*ctp, &message_mods);
571 }
572 if (status == OK && fx->reformat) {
573 status =
574 ensure_text_plain (ctp, NULL, &message_mods, fx->replacetextplain);
575 }
576 if (status == OK && fx->decodetext) {
577 status = decode_text_parts (*ctp, fx->decodetext, fx->decodetypes,
578 &message_mods);
579 update_cte (*ctp);
580 }
581 if (status == OK && fx->textcharset != NULL) {
582 status = convert_charsets (*ctp, fx->textcharset, &message_mods);
583 }
584
585 if (status == OK && ! (*ctp)->c_umask) {
586 /* Set the umask for the contents file. This currently
587 isn't used but just in case it is in the future. */
588 struct stat st;
589
590 if (stat ((*ctp)->c_file, &st) != NOTOK) {
591 (*ctp)->c_umask = ~(st.st_mode & 0777);
592 } else {
593 (*ctp)->c_umask = ~m_gmprot();
594 }
595 }
596
597 /*
598 * Write the content to a file
599 */
600 if (status == OK) {
601 status = write_content (*ctp, input_filename, outfile, *outfp,
602 modify_inplace, message_mods);
603 } else if (! modify_inplace) {
604 /* Something went wrong. Output might be expected, such
605 as if this were run as a filter. Just copy the input
606 to the output. */
607 if (copy_input_to_output (input_filename, *infp, outfile,
608 *outfp) != OK) {
609 inform("unable to copy message to %s, it might be lost\n",
610 outfile);
611 }
612 }
613
614 if (modify_inplace) {
615 if (status != OK) { (void) m_unlink (outfile); }
616 free (outfile);
617 outfile = NULL;
618 }
619
620 fclose (*infp);
621 *infp = NULL;
622 free (input_filename);
623
624 return status;
625 }
626
627
628 /*
629 * Copy input message to output. Assumes not modifying in place, so this
630 * might be running as part of a pipeline.
631 */
632 static int
633 copy_input_to_output (const char *input_filename, FILE *infp,
634 const char *output_filename, FILE *outfp) {
635 int in = fileno (infp);
636 int out = fileno (outfp);
637 int status = OK;
638
639 if (in != -1 && out != -1) {
640 cpydata (in, out, input_filename, output_filename);
641 } else {
642 status = NOTOK;
643 }
644
645 return status;
646 }
647
648
649 /*
650 * Fix mismatched outer level boundary.
651 */
652 static int
653 fix_boundary (CT *ct, int *message_mods) {
654 struct multipart *mp;
655 int status = OK;
656
657 if (ct && (*ct)->c_type == CT_MULTIPART && bogus_mp_content) {
658 mp = (struct multipart *) (*ct)->c_ctparams;
659
660 /*
661 * 1) Get boundary at end of part.
662 * 2) Get boundary at beginning of part and compare to the end-of-part
663 * boundary.
664 * 3) Write out contents of ct to tmp file, replacing boundary in
665 * header with boundary from part. Set c_unlink to 1.
666 * 4) Free ct.
667 * 5) Call parse_mime() on the tmp file, replacing ct.
668 */
669
670 if (mp && mp->mp_start) {
671 char *part_boundary;
672
673 if (get_multipart_boundary (*ct, &part_boundary) == OK) {
674 char *fixed;
675
676 if ((fixed = m_mktemp2 (NULL, invo_name, NULL, &(*ct)->c_fp))) {
677 if (replace_boundary (*ct, fixed, part_boundary) == OK) {
678 char *filename = mh_xstrdup ((*ct)->c_file);
679 CT fixed_ct;
680
681 free_content (*ct);
682 if ((fixed_ct = parse_mime (fixed))) {
683 *ct = fixed_ct;
684 (*ct)->c_unlink = 1;
685
686 ++*message_mods;
687 if (verbosw) {
688 report (NULL, NULL, filename,
689 "fix multipart boundary");
690 }
691 } else {
692 *ct = NULL;
693 inform("unable to parse fixed part");
694 status = NOTOK;
695 }
696 free (filename);
697 } else {
698 inform("unable to replace broken boundary");
699 status = NOTOK;
700 }
701 } else {
702 inform("unable to create temporary file in %s",
703 get_temp_dir());
704 status = NOTOK;
705 }
706
707 free (part_boundary);
708 } else {
709 /* Couldn't fix the boundary. Report failure so that mhfixmsg
710 doesn't modify the message. */
711 status = NOTOK;
712 }
713 } else {
714 /* No multipart struct, even though the content type is
715 CT_MULTIPART. Report failure so that mhfixmsg doesn't modify
716 the message. */
717 status = NOTOK;
718 }
719 }
720
721 return status;
722 }
723
724
725 /*
726 * Find boundary at end of multipart.
727 */
728 static int
729 get_multipart_boundary (CT ct, char **part_boundary) {
730 char buffer[NMH_BUFSIZ];
731 char *end_boundary = NULL;
732 off_t begin = (off_t) ct->c_end > (off_t) (ct->c_begin + sizeof buffer)
733 ? (off_t) (ct->c_end - sizeof buffer)
734 : (off_t) ct->c_begin;
735 size_t bytes_read;
736 int status = OK;
737
738 /* This will fail if the boundary spans fread() calls. NMH_BUFSIZ should
739 be big enough, even if it's just 1024, to make that unlikely. */
740
741 /* free_content() will close ct->c_fp if bogus MP boundary is fixed. */
742 if (! ct->c_fp && (ct->c_fp = fopen (ct->c_file, "r")) == NULL) {
743 advise (ct->c_file, "unable to open for reading");
744 return NOTOK;
745 }
746
747 /* Get boundary at end of multipart. */
748 while (begin >= (off_t) ct->c_begin) {
749 fseeko (ct->c_fp, begin, SEEK_SET);
750 while ((bytes_read = fread (buffer, 1, sizeof buffer, ct->c_fp)) > 0) {
751 char *cp = rfind_str (buffer, bytes_read, "--");
752
753 if (cp) {
754 char *end;
755
756 /* Trim off trailing "--" and anything beyond. */
757 *cp-- = '\0';
758 if ((end = rfind_str (buffer, cp - buffer, "\n"))) {
759 if (strlen (end) > 3 && *end++ == '\n' &&
760 *end++ == '-' && *end++ == '-') {
761 end_boundary = mh_xstrdup (end);
762 break;
763 }
764 }
765 }
766 }
767
768 if (! end_boundary && begin > (off_t) (ct->c_begin + sizeof buffer)) {
769 begin -= sizeof buffer;
770 } else {
771 break;
772 }
773 }
774
775 /* Get boundary at beginning of multipart. */
776 if (end_boundary) {
777 fseeko (ct->c_fp, ct->c_begin, SEEK_SET);
778 while ((bytes_read = fread (buffer, 1, sizeof buffer, ct->c_fp)) > 0) {
779 if (bytes_read >= strlen (end_boundary)) {
780 char *cp = find_str (buffer, bytes_read, end_boundary);
781
782 if (cp && cp - buffer >= 2 && *--cp == '-' &&
783 *--cp == '-' && (cp > buffer && *--cp == '\n')) {
784 status = OK;
785 break;
786 }
787 } else {
788 /* The start and end boundaries didn't match, or the
789 start boundary doesn't begin with "\n--" (or "--"
790 if at the beginning of buffer). Keep trying. */
791 status = NOTOK;
792 }
793 }
794 } else {
795 status = NOTOK;
796 }
797
798 if (ct->c_fp) {
799 fclose (ct->c_fp);
800 ct->c_fp = NULL;
801 }
802
803 if (status == OK) {
804 *part_boundary = end_boundary;
805 } else {
806 *part_boundary = NULL;
807 free (end_boundary);
808 }
809
810 return status;
811 }
812
813
814 /*
815 * Open and copy ct->c_file to file, replacing the multipart boundary.
816 */
817 static int
818 replace_boundary (CT ct, char *file, char *boundary) {
819 FILE *fpin, *fpout;
820 int compnum, state;
821 char buf[NMH_BUFSIZ], name[NAMESZ];
822 char *np, *vp;
823 m_getfld_state_t gstate = 0;
824 int status = OK;
825
826 if (ct->c_file == NULL) {
827 inform("missing input filename");
828 return NOTOK;
829 }
830
831 if ((fpin = fopen (ct->c_file, "r")) == NULL) {
832 advise (ct->c_file, "unable to open for reading");
833 return NOTOK;
834 }
835
836 if ((fpout = fopen (file, "w")) == NULL) {
837 fclose (fpin);
838 advise (file, "unable to open for writing");
839 return NOTOK;
840 }
841
842 for (compnum = 1;;) {
843 int bufsz = (int) sizeof buf;
844
845 switch (state = m_getfld (&gstate, name, buf, &bufsz, fpin)) {
846 case FLD:
847 case FLDPLUS:
848 compnum++;
849
850 /* get copies of the buffers */
851 np = mh_xstrdup (name);
852 vp = mh_xstrdup (buf);
853
854 /* if necessary, get rest of field */
855 while (state == FLDPLUS) {
856 bufsz = sizeof buf;
857 state = m_getfld (&gstate, name, buf, &bufsz, fpin);
858 vp = add (buf, vp); /* add to previous value */
859 }
860
861 if (strcasecmp (TYPE_FIELD, np)) {
862 fprintf (fpout, "%s:%s", np, vp);
863 } else {
864 char *new_ctline, *new_params;
865
866 replace_param(&ct->c_ctinfo.ci_first_pm,
867 &ct->c_ctinfo.ci_last_pm, "boundary",
868 boundary, 0);
869
870 new_ctline = concat(" ", ct->c_ctinfo.ci_type, "/",
871 ct->c_ctinfo.ci_subtype, NULL);
872 new_params = output_params(strlen(TYPE_FIELD) +
873 strlen(new_ctline) + 1,
874 ct->c_ctinfo.ci_first_pm, NULL, 0);
875 fprintf (fpout, "%s:%s%s\n", np, new_ctline,
876 new_params ? new_params : "");
877 free(new_ctline);
878 mh_xfree(new_params);
879 }
880
881 free (vp);
882 free (np);
883
884 continue;
885
886 case BODY:
887 putc('\n', fpout);
888 /* buf will have a terminating NULL, skip it. */
889 if ((int) fwrite (buf, 1, bufsz-1, fpout) < bufsz-1) {
890 advise (file, "fwrite");
891 }
892 continue;
893
894 case FILEEOF:
895 break;
896
897 case LENERR:
898 case FMTERR:
899 inform("message format error in component #%d", compnum);
900 status = NOTOK;
901 break;
902
903 default:
904 inform("getfld() returned %d", state);
905 status = NOTOK;
906 break;
907 }
908
909 break;
910 }
911
912 m_getfld_state_destroy (&gstate);
913 fclose (fpout);
914 fclose (fpin);
915
916 return status;
917 }
918
919
920 /*
921 * Fix Content-Type header to reflect the content of its part.
922 */
923 static int
924 fix_types (CT ct, svector_t fixtypes, int *message_mods) {
925 int status = OK;
926
927 switch (ct->c_type) {
928 case CT_MULTIPART: {
929 struct multipart *m = (struct multipart *) ct->c_ctparams;
930 struct part *part;
931
932 for (part = m->mp_parts; status == OK && part; part = part->mp_next) {
933 status = fix_types (part->mp_part, fixtypes, message_mods);
934 }
935 break;
936 }
937
938 case CT_MESSAGE:
939 if (ct->c_subtype == MESSAGE_EXTERNAL) {
940 struct exbody *e = (struct exbody *) ct->c_ctparams;
941
942 status = fix_types (e->eb_content, fixtypes, message_mods);
943 }
944 break;
945
946 default: {
947 char **typep, *type;
948
949 if (ct->c_ctinfo.ci_type && ct->c_ctinfo.ci_subtype) {
950 for (typep = svector_strs (fixtypes);
951 typep && (type = *typep);
952 ++typep) {
953 char *type_subtype =
954 concat (ct->c_ctinfo.ci_type, "/", ct->c_ctinfo.ci_subtype,
955 NULL);
956
957 if (! strcasecmp (type, type_subtype) &&
958 decode_part (ct) == OK &&
959 ct->c_cefile.ce_file != NULL) {
960 char *ct_type_subtype = mime_type (ct->c_cefile.ce_file);
961 char *cp;
962
963 if ((cp = strchr (ct_type_subtype, ';'))) {
964 /* Truncate to remove any parameter list from
965 mime_type () result. */
966 *cp = '\0';
967 }
968
969 if (strcasecmp (type, ct_type_subtype)) {
970 char *ct_type, *ct_subtype;
971 HF hf;
972
973 /* The Content-Type header does not match the
974 content, so update these struct Content
975 fields to match:
976 * c_type, c_subtype
977 * c_ctinfo.ci_type, c_ctinfo.ci_subtype
978 * c_ctline
979 */
980 /* Extract type and subtype from type/subtype. */
981 ct_type = mh_xstrdup(ct_type_subtype);
982 if ((cp = strchr (ct_type, '/'))) {
983 *cp = '\0';
984 ct_subtype = mh_xstrdup(++cp);
985 } else {
986 inform("missing / in MIME type of %s %s",
987 ct->c_file, ct->c_partno);
988 free (ct_type);
989 return NOTOK;
990 }
991
992 ct->c_type = ct_str_type (ct_type);
993 ct->c_subtype = ct_str_subtype (ct->c_type, ct_subtype);
994
995 free (ct->c_ctinfo.ci_type);
996 ct->c_ctinfo.ci_type = ct_type;
997 free (ct->c_ctinfo.ci_subtype);
998 ct->c_ctinfo.ci_subtype = ct_subtype;
999 if (! replace_substring (&ct->c_ctline, type,
1000 ct_type_subtype)) {
1001 inform("did not find %s in %s",
1002 type, ct->c_ctline);
1003 }
1004
1005 /* Update Content-Type header field. */
1006 for (hf = ct->c_first_hf; hf; hf = hf->next) {
1007 if (! strcasecmp (TYPE_FIELD, hf->name)) {
1008 if (replace_substring (&hf->value, type,
1009 ct_type_subtype)) {
1010 ++*message_mods;
1011 if (verbosw) {
1012 report (NULL, ct->c_partno, ct->c_file,
1013 "change Content-Type in header "
1014 "from %s to %s",
1015 type, ct_type_subtype);
1016 }
1017 break;
1018 } else {
1019 inform("did not find %s in %s",
1020 type, hf->value);
1021 }
1022 }
1023 }
1024 }
1025 free (ct_type_subtype);
1026 }
1027 free (type_subtype);
1028 }
1029 }
1030 }}
1031
1032 return status;
1033 }
1034
1035
1036 /*
1037 * Replace a substring, allocating space to hold the new one.
1038 */
1039 char *
1040 replace_substring (char **str, const char *old, const char *new) {
1041 char *cp;
1042
1043 if ((cp = strstr (*str, old))) {
1044 char *remainder = cp + strlen (old);
1045 char *prefix, *new_str;
1046
1047 if (cp - *str) {
1048 prefix = mh_xstrdup(*str);
1049 *(prefix + (cp - *str)) = '\0';
1050 new_str = concat (prefix, new, remainder, NULL);
1051 free (prefix);
1052 } else {
1053 new_str = concat (new, remainder, NULL);
1054 }
1055
1056 free (*str);
1057
1058 return *str = new_str;
1059 }
1060
1061 return NULL;
1062 }
1063
1064
1065 /*
1066 * Remove a name=value parameter, given just its name, from a header value.
1067 */
1068 char *
1069 remove_parameter (char *str, const char *name) {
1070 /* It looks to me, based on the BNF in RFC 2045, than there can't
1071 be whitespace betwwen the parameter name and the "=", or
1072 between the "=" and the parameter value. */
1073 char *param_name = concat (name, "=", NULL);
1074 char *cp;
1075
1076 if ((cp = strstr (str, param_name))) {
1077 char *start, *end;
1078 size_t count = 1;
1079
1080 /* Remove any leading spaces, before the parameter name. */
1081 for (start = cp;
1082 start > str && isspace ((unsigned char) *(start-1));
1083 --start) {
1084 continue;
1085 }
1086 /* Remove a leading semicolon. */
1087 if (start > str && *(start-1) == ';') { --start; }
1088
1089 end = cp + strlen (name) + 1;
1090 if (*end == '"') {
1091 /* Skip past the quoted value, and then the final quote. */
1092 for (++end ; *end && *end != '"'; ++end) { continue; }
1093 ++end;
1094 } else {
1095 /* Skip past the value. */
1096 for (++end ; *end && ! isspace ((unsigned char) *end); ++end) {}
1097 }
1098
1099 /* Count how many characters need to be moved. Include
1100 trailing null, which is accounted for by the
1101 initialization of count to 1. */
1102 for (cp = end; *cp; ++cp) { ++count; }
1103 (void) memmove (start, end, count);
1104 }
1105
1106 free (param_name);
1107
1108 return str;
1109 }
1110
1111
1112 /*
1113 * Fix Content-Transfer-Encoding of composite,, e.g., message or multipart, part.
1114 * According to RFC 2045 Sec. 6.4, it must be 7bit, 8bit, or binary. Set it to
1115 * 8 bit.
1116 */
1117 static int
1118 fix_composite_cte (CT ct, int *message_mods) {
1119 int status = OK;
1120
1121 if (ct->c_type == CT_MESSAGE || ct->c_type == CT_MULTIPART) {
1122 if (ct->c_encoding != CE_7BIT && ct->c_encoding != CE_8BIT &&
1123 ct->c_encoding != CE_BINARY) {
1124 HF hf;
1125
1126 for (hf = ct->c_first_hf; hf; hf = hf->next) {
1127 char *name = hf->name;
1128 for (; *name && isspace ((unsigned char) *name); ++name) {
1129 continue;
1130 }
1131
1132 if (! strncasecmp (name, ENCODING_FIELD,
1133 strlen (ENCODING_FIELD))) {
1134 char *prefix = "Nmh-REPLACED-INVALID-";
1135 HF h;
1136
1137 NEW(h);
1138 h->name = mh_xstrdup (hf->name);
1139 h->hf_encoding = hf->hf_encoding;
1140 h->next = hf->next;
1141 hf->next = h;
1142
1143 /* Retain old header but prefix its name. */
1144 free (hf->name);
1145 hf->name = concat (prefix, h->name, NULL);
1146
1147 ++*message_mods;
1148 if (verbosw) {
1149 char *encoding = cpytrim (hf->value);
1150 report (NULL, ct->c_partno, ct->c_file,
1151 "replace Content-Transfer-Encoding of %s "
1152 "with 8 bit", encoding);
1153 free (encoding);
1154 }
1155
1156 h->value = mh_xstrdup (" 8bit\n");
1157
1158 /* Don't need to warn for multiple C-T-E header
1159 fields, parse_mime() already does that. But
1160 if there are any, fix them all as necessary. */
1161 hf = h;
1162 }
1163 }
1164
1165 set_ce (ct, CE_8BIT);
1166 }
1167
1168 if (ct->c_type == CT_MULTIPART) {
1169 struct multipart *m;
1170 struct part *part;
1171
1172 m = (struct multipart *) ct->c_ctparams;
1173 for (part = m->mp_parts; part; part = part->mp_next) {
1174 if (fix_composite_cte (part->mp_part, message_mods) != OK) {
1175 status = NOTOK;
1176 break;
1177 }
1178 }
1179 }
1180 }
1181
1182 return status;
1183 }
1184
1185
1186 /*
1187 * Set content encoding.
1188 */
1189 static int
1190 set_ce (CT ct, int encoding) {
1191 const char *ce = ce_str (encoding);
1192 const struct str2init *ctinit = get_ce_method (ce);
1193
1194 if (ctinit) {
1195 char *cte = concat (" ", ce, "\n", NULL);
1196 int found_cte = 0;
1197 HF hf;
1198 /* Decoded contents might be in ct->c_cefile.ce_file, if the
1199 caller is decode_text_parts (). Save because we'll
1200 overwrite below. */
1201 struct cefile decoded_content_info = ct->c_cefile;
1202
1203 ct->c_encoding = encoding;
1204
1205 ct->c_ctinitfnx = ctinit->si_init;
1206 /* This will assign ct->c_cefile with an all-0 struct, which
1207 is what we want. */
1208 (*ctinit->si_init) (ct);
1209 /* After returning, the caller should set
1210 ct->c_cefile.ce_file to the name of the file containing
1211 the contents. */
1212
1213 if (ct->c_ceclosefnx) {
1214 (*ct->c_ceclosefnx) (ct);
1215 }
1216
1217 /* Restore the cefile. */
1218 ct->c_cefile = decoded_content_info;
1219
1220 /* Update/add Content-Transfer-Encoding header field. */
1221 for (hf = ct->c_first_hf; hf; hf = hf->next) {
1222 if (! strcasecmp (ENCODING_FIELD, hf->name)) {
1223 found_cte = 1;
1224 free (hf->value);
1225 hf->value = cte;
1226 }
1227 }
1228 if (! found_cte) {
1229 add_header (ct, mh_xstrdup (ENCODING_FIELD), cte);
1230 }
1231
1232 /* Update c_celine. It's used only by mhlist -debug. */
1233 free (ct->c_celine);
1234 ct->c_celine = mh_xstrdup (cte);
1235
1236 return OK;
1237 }
1238
1239 return NOTOK;
1240 }
1241
1242
1243 /*
1244 * Make sure each text part has a corresponding text/plain part.
1245 */
1246 static int
1247 ensure_text_plain (CT *ct, CT parent, int *message_mods, int replacetextplain) {
1248 int status = OK;
1249
1250 switch ((*ct)->c_type) {
1251 case CT_TEXT: {
1252 /* Nothing to do for text/plain. */
1253 if ((*ct)->c_subtype == TEXT_PLAIN) { return OK; }
1254
1255 if (parent && parent->c_type == CT_MULTIPART &&
1256 parent->c_subtype == MULTI_ALTERNATE) {
1257 int new_subpart_number = 1;
1258 int has_text_plain =
1259 find_textplain_sibling (parent, replacetextplain,
1260 &new_subpart_number);
1261
1262 if (! has_text_plain) {
1263 /* Parent is a multipart/alternative. Insert a new
1264 text/plain subpart. */
1265 const int inserted =
1266 insert_new_text_plain_part (*ct, new_subpart_number,
1267 parent);
1268 if (inserted) {
1269 ++*message_mods;
1270 if (verbosw) {
1271 report (NULL, parent->c_partno, parent->c_file,
1272 "insert text/plain part");
1273 }
1274 } else {
1275 status = NOTOK;
1276 }
1277 }
1278 } else if (parent && parent->c_type == CT_MULTIPART &&
1279 parent->c_subtype == MULTI_RELATED) {
1280 char *type_subtype =
1281 concat ((*ct)->c_ctinfo.ci_type, "/",
1282 (*ct)->c_ctinfo.ci_subtype, NULL);
1283 const char *parent_type =
1284 get_param (parent->c_ctinfo.ci_first_pm, "type", '?', 1);
1285 int new_subpart_number = 1;
1286 int has_text_plain = 0;
1287
1288 /* Have to do string comparison on the subtype because we
1289 don't enumerate all of them in c_subtype values.
1290 parent_type will be NULL if the multipart/related part
1291 doesn't have a type parameter. The type parameter must
1292 be specified according to RFC 2387 Sec. 3.1 but not all
1293 messages comply. */
1294 if (parent_type && strcasecmp (type_subtype, parent_type) == 0) {
1295 /* The type of this part matches the root type of the
1296 parent multipart/related. Look to see if there's
1297 text/plain sibling. */
1298 has_text_plain =
1299 find_textplain_sibling (parent, replacetextplain,
1300 &new_subpart_number);
1301 }
1302
1303 free (type_subtype);
1304
1305 if (! has_text_plain) {
1306 struct multipart *mp = (struct multipart *) parent->c_ctparams;
1307 struct part *part;
1308 int siblings = 0;
1309
1310 for (part = mp->mp_parts; part; part = part->mp_next) {
1311 if (*ct != part->mp_part) {
1312 ++siblings;
1313 }
1314 }
1315
1316 if (siblings) {
1317 /* Parent is a multipart/related. Insert a new
1318 text/plain subpart in a new multipart/alternative. */
1319 if (insert_into_new_mp_alt (ct, message_mods)) {
1320 /* Not an error if text/plain couldn't be added. */
1321 }
1322 } else {
1323 /* There are no siblings, so insert a new text/plain
1324 subpart, and change the parent type from
1325 multipart/related to multipart/alternative. */
1326 const int inserted =
1327 insert_new_text_plain_part (*ct, new_subpart_number,
1328 parent);
1329
1330 if (inserted) {
1331 HF hf;
1332
1333 parent->c_subtype = MULTI_ALTERNATE;
1334 free (parent->c_ctinfo.ci_subtype);
1335 parent->c_ctinfo.ci_subtype = mh_xstrdup("alternative");
1336 if (! replace_substring (&parent->c_ctline, "/related",
1337 "/alternative")) {
1338 inform("did not find multipart/related in %s",
1339 parent->c_ctline);
1340 }
1341
1342 /* Update Content-Type header field. */
1343 for (hf = parent->c_first_hf; hf; hf = hf->next) {
1344 if (! strcasecmp (TYPE_FIELD, hf->name)) {
1345 if (replace_substring (&hf->value, "/related",
1346 "/alternative")) {
1347 ++*message_mods;
1348 if (verbosw) {
1349 report (NULL, parent->c_partno,
1350 parent->c_file,
1351 "insert text/plain part");
1352 }
1353
1354 /* Remove, e.g., type="text/html" from
1355 multipart/alternative. */
1356 remove_parameter (hf->value, "type");
1357 break;
1358 } else {
1359 inform("did not find multipart/"
1360 "related in header %s",
1361 hf->value);
1362 }
1363 }
1364 }
1365 } else {
1366 /* Not an error if text/plain couldn't be inserted. */
1367 }
1368 }
1369 }
1370 } else {
1371 if (insert_into_new_mp_alt (ct, message_mods)) {
1372 status = NOTOK;
1373 }
1374 }
1375 break;
1376 }
1377
1378 case CT_MULTIPART: {
1379 struct multipart *mp = (struct multipart *) (*ct)->c_ctparams;
1380 struct part *part;
1381
1382 for (part = mp->mp_parts; status == OK && part; part = part->mp_next) {
1383 if ((*ct)->c_type == CT_MULTIPART) {
1384 status = ensure_text_plain (&part->mp_part, *ct, message_mods,
1385 replacetextplain);
1386 }
1387 }
1388 break;
1389 }
1390
1391 case CT_MESSAGE:
1392 if ((*ct)->c_subtype == MESSAGE_EXTERNAL) {
1393 struct exbody *e = (struct exbody *) (*ct)->c_ctparams;
1394
1395 status = ensure_text_plain (&e->eb_content, *ct, message_mods,
1396 replacetextplain);
1397 }
1398 break;
1399 }
1400
1401 return status;
1402 }
1403
1404
1405 /*
1406 * See if there is a sibling text/plain, and return its subpart number.
1407 */
1408 static int
1409 find_textplain_sibling (CT parent, int replacetextplain,
1410 int *new_subpart_number) {
1411 struct multipart *mp = (struct multipart *) parent->c_ctparams;
1412 struct part *part, *prev;
1413 int has_text_plain = 0;
1414
1415 for (prev = part = mp->mp_parts; part; part = part->mp_next) {
1416 ++*new_subpart_number;
1417 if (part->mp_part->c_type == CT_TEXT &&
1418 part->mp_part->c_subtype == TEXT_PLAIN) {
1419 if (replacetextplain) {
1420 struct part *old_part;
1421 if (part == mp->mp_parts) {
1422 old_part = mp->mp_parts;
1423 mp->mp_parts = part->mp_next;
1424 } else {
1425 old_part = prev->mp_next;
1426 prev->mp_next = part->mp_next;
1427 }
1428 if (verbosw) {
1429 report (NULL, parent->c_partno, parent->c_file,
1430 "remove text/plain part %s",
1431 old_part->mp_part->c_partno);
1432 }
1433 free_content (old_part->mp_part);
1434 free (old_part);
1435 } else {
1436 has_text_plain = 1;
1437 }
1438 break;
1439 }
1440 prev = part;
1441 }
1442
1443 return has_text_plain;
1444 }
1445
1446
1447 /*
1448 * Insert a new text/plain part.
1449 */
1450 static int
1451 insert_new_text_plain_part (CT ct, int new_subpart_number, CT parent) {
1452 struct multipart *mp = (struct multipart *) parent->c_ctparams;
1453 struct part *new_part;
1454
1455 NEW(new_part);
1456 if ((new_part->mp_part = build_text_plain_part (ct))) {
1457 char buffer[16];
1458 snprintf (buffer, sizeof buffer, "%d", new_subpart_number);
1459
1460 new_part->mp_next = mp->mp_parts;
1461 mp->mp_parts = new_part;
1462 new_part->mp_part->c_partno =
1463 concat (parent->c_partno ? parent->c_partno : "1", ".",
1464 buffer, NULL);
1465
1466 return 1;
1467 }
1468
1469 free_content (new_part->mp_part);
1470 free (new_part);
1471
1472 return 0;
1473 }
1474
1475
1476 /*
1477 * Create a text/plain part to go along with non-plain sibling part.
1478 */
1479 static CT
1480 build_text_plain_part (CT encoded_part) {
1481 CT tp_part = divide_part (encoded_part);
1482 char *tmp_plain_file = NULL;
1483
1484 if (decode_part (tp_part) == OK) {
1485 /* Now, tp_part->c_cefile.ce_file is the name of the tmp file that
1486 contains the decoded contents. And the decoding function, such
1487 as openQuoted, will have set ...->ce_unlink to 1 so that it will
1488 be unlinked by free_content (). */
1489 char *tempfile;
1490
1491 /* This m_mktemp2() call closes the temp file. */
1492 if ((tempfile = m_mktemp2 (NULL, invo_name, NULL, NULL)) == NULL) {
1493 inform("unable to create temporary file in %s",
1494 get_temp_dir());
1495 } else {
1496 tmp_plain_file = mh_xstrdup (tempfile);
1497 if (reformat_part (tp_part, tmp_plain_file,
1498 tp_part->c_ctinfo.ci_type,
1499 tp_part->c_ctinfo.ci_subtype,
1500 tp_part->c_type) == OK) {
1501 return tp_part;
1502 }
1503 }
1504 }
1505
1506 free_content (tp_part);
1507 if (tmp_plain_file) { (void) m_unlink (tmp_plain_file); }
1508 free (tmp_plain_file);
1509
1510 return NULL;
1511 }
1512
1513
1514 /*
1515 * Slip new text/plain part into a new multipart/alternative.
1516 */
1517 static int
1518 insert_into_new_mp_alt (CT *ct, int *message_mods) {
1519 CT tp_part = build_text_plain_part (*ct);
1520 int status = OK;
1521
1522 if (tp_part) {
1523 CT mp_alt = build_multipart_alt (*ct, tp_part, CT_MULTIPART,
1524 MULTI_ALTERNATE);
1525 if (mp_alt) {
1526 struct multipart *mp = (struct multipart *) mp_alt->c_ctparams;
1527
1528 if (mp && mp->mp_parts) {
1529 mp->mp_parts->mp_part = tp_part;
1530 /* Make the new multipart/alternative the parent. */
1531 *ct = mp_alt;
1532
1533 ++*message_mods;
1534 if (verbosw) {
1535 report (NULL, (*ct)->c_partno, (*ct)->c_file,
1536 "insert text/plain part");
1537 }
1538 } else {
1539 free_content (tp_part);
1540 free_content (mp_alt);
1541 status = NOTOK;
1542 }
1543 } else {
1544 status = NOTOK;
1545 }
1546 } else {
1547 /* Not an error if text/plain couldn't be built. */
1548 }
1549
1550 return status;
1551 }
1552
1553
1554 /*
1555 * Clone a MIME part.
1556 */
1557 static CT
1558 divide_part (CT ct) {
1559 CT new_part;
1560
1561 NEW0(new_part);
1562 /* Just copy over what is needed for decoding. c_vrsn and
1563 c_celine aren't necessary. */
1564 new_part->c_file = mh_xstrdup (ct->c_file);
1565 new_part->c_begin = ct->c_begin;
1566 new_part->c_end = ct->c_end;
1567 copy_ctinfo (&new_part->c_ctinfo, &ct->c_ctinfo);
1568 new_part->c_type = ct->c_type;
1569 new_part->c_cefile = ct->c_cefile;
1570 new_part->c_encoding = ct->c_encoding;
1571 new_part->c_ctinitfnx = ct->c_ctinitfnx;
1572 new_part->c_ceopenfnx = ct->c_ceopenfnx;
1573 new_part->c_ceclosefnx = ct->c_ceclosefnx;
1574 new_part->c_cesizefnx = ct->c_cesizefnx;
1575
1576 /* c_ctline is used by reformat__part(), so it can preserve
1577 anything after the type/subtype. */
1578 new_part->c_ctline = mh_xstrdup (ct->c_ctline);
1579
1580 return new_part;
1581 }
1582
1583
1584 /*
1585 * Copy the content info from one part to another.
1586 */
1587 static void
1588 copy_ctinfo (CI dest, CI src) {
1589 PM s_pm, d_pm;
1590
1591 dest->ci_type = src->ci_type ? mh_xstrdup (src->ci_type) : NULL;
1592 dest->ci_subtype = src->ci_subtype ? mh_xstrdup (src->ci_subtype) : NULL;
1593
1594 for (s_pm = src->ci_first_pm; s_pm; s_pm = s_pm->pm_next) {
1595 d_pm = add_param(&dest->ci_first_pm, &dest->ci_last_pm, s_pm->pm_name,
1596 s_pm->pm_value, 0);
1597 if (s_pm->pm_charset) {
1598 d_pm->pm_charset = mh_xstrdup(s_pm->pm_charset);
1599 }
1600 if (s_pm->pm_lang) {
1601 d_pm->pm_lang = mh_xstrdup(s_pm->pm_lang);
1602 }
1603 }
1604
1605 dest->ci_comment = src->ci_comment ? mh_xstrdup (src->ci_comment) : NULL;
1606 dest->ci_magic = src->ci_magic ? mh_xstrdup (src->ci_magic) : NULL;
1607 }
1608
1609
1610 /*
1611 * Decode content.
1612 */
1613 static int
1614 decode_part (CT ct) {
1615 char *tmp_decoded;
1616 int status;
1617 FILE *file;
1618 char *tempfile;
1619
1620 if ((tempfile = m_mktemp2 (NULL, invo_name, NULL, &file)) == NULL) {
1621 adios (NULL, "unable to create temporary file in %s", get_temp_dir());
1622 }
1623 tmp_decoded = mh_xstrdup (tempfile);
1624 /* The following call will load ct->c_cefile.ce_file with the tmp
1625 filename of the decoded content. tmp_decoded will contain the
1626 encoded output, get rid of that. */
1627 status = output_message_fp (ct, file, tmp_decoded);
1628 (void) m_unlink (tmp_decoded);
1629 free (tmp_decoded);
1630 if (fclose (file)) {
1631 inform("unable to close temporary file %s, continuing...", tempfile);
1632 }
1633
1634 return status;
1635 }
1636
1637
1638 /*
1639 * Reformat content as plain text.
1640 * Some of the arguments aren't really needed now, but maybe will
1641 * be in the future for other than text types.
1642 */
1643 static int
1644 reformat_part (CT ct, char *file, char *type, char *subtype, int c_type) {
1645 int output_subtype, output_encoding;
1646 const char *reason = NULL;
1647 char *cp, *cf;
1648 int status;
1649
1650 /* Hacky: this redirects the output from whatever command is used
1651 to show the part to a file. So, the user can't have any output
1652 redirection in that command.
1653 Could show_multi() in mhshowsbr.c avoid this? */
1654
1655 /* Check for invo_name-format-type/subtype. */
1656 if ((cf = context_find_by_type ("format", type, subtype)) == NULL) {
1657 if (verbosw) {
1658 inform("Don't know how to convert %s, there is no "
1659 "%s-format-%s/%s profile entry",
1660 ct->c_file, invo_name, type, subtype);
1661 }
1662 return NOTOK;
1663 }
1664 if (strchr (cf, '>')) {
1665 inform("'>' prohibited in \"%s\",\nplease fix your "
1666 "%s-format-%s/%s profile entry", cf, invo_name, type,
1667 subtype ? subtype : "");
1668
1669 return NOTOK;
1670 }
1671
1672 cp = concat (cf, " >", file, NULL);
1673 status = show_content_aux (ct, 0, cp, NULL, NULL);
1674 free (cp);
1675
1676 /* Unlink decoded content tmp file and free its filename to avoid
1677 leaks. The file stream should already have been closed. */
1678 if (ct->c_cefile.ce_unlink) {
1679 (void) m_unlink (ct->c_cefile.ce_file);
1680 free (ct->c_cefile.ce_file);
1681 ct->c_cefile.ce_file = NULL;
1682 ct->c_cefile.ce_unlink = 0;
1683 }
1684
1685 if (c_type == CT_TEXT) {
1686 output_subtype = TEXT_PLAIN;
1687 } else {
1688 /* Set subtype to 0, which is always an UNKNOWN subtype. */
1689 output_subtype = 0;
1690 }
1691
1692 output_encoding = content_encoding (ct, &reason);
1693 if (status == OK &&
1694 set_ct_type (ct, c_type, output_subtype, output_encoding) == OK) {
1695 ct->c_cefile.ce_file = file;
1696 ct->c_cefile.ce_unlink = 1;
1697 } else {
1698 ct->c_cefile.ce_unlink = 0;
1699 status = NOTOK;
1700 }
1701
1702 return status;
1703 }
1704
1705
1706 /*
1707 * Fill in a multipart/alternative part.
1708 */
1709 static CT
1710 build_multipart_alt (CT first_alt, CT new_part, int type, int subtype) {
1711 char *boundary_prefix = "----=_nmh-multipart";
1712 char *boundary = concat (boundary_prefix, first_alt->c_partno, NULL);
1713 char *boundary_indicator = "; boundary=";
1714 char *typename, *subtypename, *name;
1715 CT ct;
1716 struct part *p;
1717 struct multipart *m;
1718 const struct str2init *ctinit;
1719
1720 NEW0(ct);
1721
1722 /* Set up the multipart/alternative part. These fields of *ct were
1723 initialized to 0 by mh_xcalloc():
1724 c_fp, c_unlink, c_begin, c_end,
1725 c_vrsn, c_ctline, c_celine,
1726 c_id, c_descr, c_dispo, c_partno,
1727 c_ctinfo.ci_comment, c_ctinfo.ci_magic,
1728 c_cefile, c_encoding,
1729 c_digested, c_digest[16], c_ctexbody,
1730 c_ctinitfnx, c_ceopenfnx, c_ceclosefnx, c_cesizefnx,
1731 c_umask, c_rfc934,
1732 c_showproc, c_termproc, c_storeproc, c_storage, c_folder
1733 */
1734
1735 ct->c_file = mh_xstrdup (first_alt->c_file);
1736 ct->c_type = type;
1737 ct->c_subtype = subtype;
1738
1739 ctinit = get_ct_init (ct->c_type);
1740
1741 typename = ct_type_str (type);
1742 subtypename = ct_subtype_str (type, subtype);
1743
1744 {
1745 int serial = 0;
1746 int found_boundary = 1;
1747
1748 while (found_boundary && serial < 1000000) {
1749 found_boundary = 0;
1750
1751 /* Ensure that the boundary doesn't appear in the decoded
1752 content. */
1753 if (new_part->c_cefile.ce_file) {
1754 if ((found_boundary =
1755 boundary_in_content (&new_part->c_cefile.ce_fp,
1756 new_part->c_cefile.ce_file,
1757 boundary)) == NOTOK) {
1758 free_content (ct);
1759 return NULL;
1760 }
1761 }
1762
1763 /* Ensure that the boundary doesn't appear in the encoded
1764 content. */
1765 if (! found_boundary && new_part->c_file) {
1766 if ((found_boundary =
1767 boundary_in_content (&new_part->c_fp,
1768 new_part->c_file,
1769 boundary)) == NOTOK) {
1770 free_content (ct);
1771 return NULL;
1772 }
1773 }
1774
1775 if (found_boundary) {
1776 /* Try a slightly different boundary. */
1777 char buffer2[16];
1778
1779 free (boundary);
1780 ++serial;
1781 snprintf (buffer2, sizeof buffer2, "%d", serial);
1782 boundary =
1783 concat (boundary_prefix,
1784 first_alt->c_partno ? first_alt->c_partno : "",
1785 "-", buffer2, NULL);
1786 }
1787 }
1788
1789 if (found_boundary) {
1790 inform("giving up trying to find a unique boundary");
1791 free_content (ct);
1792 return NULL;
1793 }
1794 }
1795
1796 name = concat (" ", typename, "/", subtypename, boundary_indicator, "\"",
1797 boundary, "\"", NULL);
1798
1799 /* Load c_first_hf and c_last_hf. */
1800 transfer_noncontent_headers (first_alt, ct);
1801 add_header (ct, mh_xstrdup (TYPE_FIELD), concat (name, "\n", NULL));
1802 free (name);
1803
1804 /* Load c_partno. */
1805 if (first_alt->c_partno) {
1806 ct->c_partno = mh_xstrdup (first_alt->c_partno);
1807 free (first_alt->c_partno);
1808 first_alt->c_partno = concat (ct->c_partno, ".1", NULL);
1809 new_part->c_partno = concat (ct->c_partno, ".2", NULL);
1810 } else {
1811 first_alt->c_partno = mh_xstrdup ("1");
1812 new_part->c_partno = mh_xstrdup ("2");
1813 }
1814
1815 if (ctinit) {
1816 ct->c_ctinfo.ci_type = mh_xstrdup (typename);
1817 ct->c_ctinfo.ci_subtype = mh_xstrdup (subtypename);
1818 }
1819
1820 add_param(&ct->c_ctinfo.ci_first_pm, &ct->c_ctinfo.ci_last_pm,
1821 "boundary", boundary, 0);
1822
1823 NEW(p);
1824 NEW(p->mp_next);
1825 p->mp_next->mp_next = NULL;
1826 p->mp_next->mp_part = first_alt;
1827
1828 NEW0(m);
1829 m->mp_start = concat (boundary, "\n", NULL);
1830 m->mp_stop = concat (boundary, "--\n", NULL);
1831 m->mp_parts = p;
1832 ct->c_ctparams = m;
1833
1834 free (boundary);
1835
1836 return ct;
1837 }
1838
1839
1840 /*
1841 * Check that the boundary does not appear in the content.
1842 */
1843 static int
1844 boundary_in_content (FILE **fp, char *file, const char *boundary) {
1845 char buffer[NMH_BUFSIZ];
1846 size_t bytes_read;
1847 int found_boundary = 0;
1848
1849 /* free_content() will close *fp if we fopen it here. */
1850 if (! *fp && (*fp = fopen (file, "r")) == NULL) {
1851 advise (file, "unable to open %s for reading", file);
1852 return NOTOK;
1853 }
1854
1855 fseeko (*fp, 0L, SEEK_SET);
1856 while ((bytes_read = fread (buffer, 1, sizeof buffer, *fp)) > 0) {
1857 if (find_str (buffer, bytes_read, boundary)) {
1858 found_boundary = 1;
1859 break;
1860 }
1861 }
1862
1863 return found_boundary;
1864 }
1865
1866
1867 /*
1868 * Remove all non-Content headers.
1869 */
1870 static void
1871 transfer_noncontent_headers (CT old, CT new) {
1872 HF hp, hp_prev;
1873
1874 hp_prev = hp = old->c_first_hf;
1875 while (hp) {
1876 HF next = hp->next;
1877
1878 if (strncasecmp (XXX_FIELD_PRF, hp->name, strlen (XXX_FIELD_PRF))) {
1879 if (hp == old->c_last_hf) {
1880 if (hp == old->c_first_hf) {
1881 old->c_last_hf = old->c_first_hf = NULL;
1882 } else {
1883 hp_prev->next = NULL;
1884 old->c_last_hf = hp_prev;
1885 }
1886 } else {
1887 if (hp == old->c_first_hf) {
1888 old->c_first_hf = next;
1889 } else {
1890 hp_prev->next = next;
1891 }
1892 }
1893
1894 /* Put node hp in the new CT. */
1895 if (new->c_first_hf == NULL) {
1896 new->c_first_hf = hp;
1897 } else {
1898 new->c_last_hf->next = hp;
1899 }
1900 new->c_last_hf = hp;
1901 } else {
1902 /* A Content- header, leave in old. */
1903 hp_prev = hp;
1904 }
1905
1906 hp = next;
1907 }
1908 }
1909
1910
1911 /*
1912 * Set content type.
1913 */
1914 static int
1915 set_ct_type (CT ct, int type, int subtype, int encoding) {
1916 char *typename = ct_type_str (type);
1917 char *subtypename = ct_subtype_str (type, subtype);
1918 /* E.g, " text/plain" */
1919 char *type_subtypename = concat (" ", typename, "/", subtypename, NULL);
1920 /* E.g, " text/plain\n" */
1921 char *name_plus_nl = concat (type_subtypename, "\n", NULL);
1922 int found_content_type = 0;
1923 HF hf;
1924 const char *cp = NULL;
1925 char *ctline;
1926 int status;
1927
1928 /* Update/add Content-Type header field. */
1929 for (hf = ct->c_first_hf; hf; hf = hf->next) {
1930 if (! strcasecmp (TYPE_FIELD, hf->name)) {
1931 found_content_type = 1;
1932 free (hf->value);
1933 hf->value = (cp = strchr (ct->c_ctline, ';'))
1934 ? concat (type_subtypename, cp, "\n", NULL)
1935 : mh_xstrdup (name_plus_nl);
1936 }
1937 }
1938 if (! found_content_type) {
1939 add_header (ct, mh_xstrdup (TYPE_FIELD),
1940 (cp = strchr (ct->c_ctline, ';'))
1941 ? concat (type_subtypename, cp, "\n", NULL)
1942 : mh_xstrdup (name_plus_nl));
1943 }
1944
1945 /* Some of these might not be used, but set them anyway. */
1946 ctline = cp
1947 ? concat (type_subtypename, cp, NULL)
1948 : concat (type_subtypename, NULL);
1949 free (ct->c_ctline);
1950 ct->c_ctline = ctline;
1951 /* Leave other ctinfo members as they were. */
1952 free (ct->c_ctinfo.ci_type);
1953 ct->c_ctinfo.ci_type = mh_xstrdup (typename);
1954 free (ct->c_ctinfo.ci_subtype);
1955 ct->c_ctinfo.ci_subtype = mh_xstrdup (subtypename);
1956 ct->c_type = type;
1957 ct->c_subtype = subtype;
1958
1959 free (name_plus_nl);
1960 free (type_subtypename);
1961
1962 status = set_ce (ct, encoding);
1963
1964 return status;
1965 }
1966
1967
1968 /*
1969 * It's not necessary to update the charset parameter of a Content-Type
1970 * header for a text part. According to RFC 2045 Sec. 6.4, the body
1971 * (content) was originally in the specified charset, "and will be in
1972 * that character set again after decoding."
1973 */
1974 static int
1975 decode_text_parts (CT ct, int encoding, const char *decodetypes,
1976 int *message_mods) {
1977 int status = OK;
1978 int lf_line_endings = 0;
1979
1980 switch (ct->c_type) {
1981 case CT_MULTIPART: {
1982 struct multipart *m = (struct multipart *) ct->c_ctparams;
1983 struct part *part;
1984
1985 /* Should check to see if the body for this part is encoded?
1986 For now, it gets passed along as-is by InitMultiPart(). */
1987 for (part = m->mp_parts; status == OK && part; part = part->mp_next) {
1988 status = decode_text_parts (part->mp_part, encoding, decodetypes,
1989 message_mods);
1990 }
1991 break;
1992 }
1993
1994 case CT_MESSAGE:
1995 if (ct->c_subtype == MESSAGE_EXTERNAL) {
1996 struct exbody *e = (struct exbody *) ct->c_ctparams;
1997
1998 status = decode_text_parts (e->eb_content, encoding, decodetypes,
1999 message_mods);
2000 }
2001 break;
2002
2003 default:
2004 if (! should_decode(decodetypes, ct->c_ctinfo.ci_type, ct->c_ctinfo.ci_subtype)) {
2005 break;
2006 }
2007
2008 lf_line_endings =
2009 ct->c_ctparams && ((struct text *) ct->c_ctparams)->lf_line_endings;
2010
2011 switch (ct->c_encoding) {
2012 case CE_BASE64:
2013 case CE_QUOTED: {
2014 int ct_encoding;
2015
2016 if (decode_part (ct) == OK && ct->c_cefile.ce_file) {
2017 const char *reason = NULL;
2018
2019 if ((ct_encoding = content_encoding (ct, &reason)) == CE_BINARY
2020 && encoding != CE_BINARY) {
2021 /* The decoding isn't acceptable so discard it.
2022 Leave status as OK to allow other transformations. */
2023 if (verbosw) {
2024 report (NULL, ct->c_partno, ct->c_file,
2025 "will not decode%s because it is binary (%s)",
2026 ct->c_partno ? ""
2027 : ct->c_ctline ? ct->c_ctline
2028 : "",
2029 reason);
2030 }
2031 (void) m_unlink (ct->c_cefile.ce_file);
2032 free (ct->c_cefile.ce_file);
2033 ct->c_cefile.ce_file = NULL;
2034 } else if (ct->c_encoding == CE_QUOTED &&
2035 ct_encoding == CE_8BIT && encoding == CE_7BIT) {
2036 /* The decoding isn't acceptable so discard it.
2037 Leave status as OK to allow other transformations. */
2038 if (verbosw) {
2039 report (NULL, ct->c_partno, ct->c_file,
2040 "will not decode%s because it is 8bit",
2041 ct->c_partno ? ""
2042 : ct->c_ctline ? ct->c_ctline
2043 : "");
2044 }
2045 (void) m_unlink (ct->c_cefile.ce_file);
2046 free (ct->c_cefile.ce_file);
2047 ct->c_cefile.ce_file = NULL;
2048 } else {
2049 int enc;
2050
2051 if (ct_encoding == CE_BINARY) {
2052 enc = CE_BINARY;
2053 } else if (ct_encoding == CE_8BIT && encoding == CE_7BIT) {
2054 enc = CE_QUOTED;
2055 } else {
2056 enc = ct_encoding;
2057 }
2058 if (set_ce (ct, enc) == OK) {
2059 ++*message_mods;
2060 if (verbosw) {
2061 report (NULL, ct->c_partno, ct->c_file, "decode%s",
2062 ct->c_ctline ? ct->c_ctline : "");
2063 }
2064 if (lf_line_endings) {
2065 strip_crs (ct, message_mods);
2066 }
2067 } else {
2068 status = NOTOK;
2069 }
2070 }
2071 } else {
2072 status = NOTOK;
2073 }
2074 break;
2075 }
2076 case CE_8BIT:
2077 case CE_7BIT:
2078 if (lf_line_endings) {
2079 strip_crs (ct, message_mods);
2080 }
2081 break;
2082 default:
2083 break;
2084 }
2085
2086 break;
2087 }
2088
2089 return status;
2090 }
2091
2092
2093 /*
2094 * Determine if the part with type[/subtype] should be decoded, according to
2095 * decodetypes (which came from the -decodetypes switch).
2096 */
2097 static int
2098 should_decode(const char *decodetypes, const char *type, const char *subtype) {
2099 /* Quick search for matching type[/subtype] in decodetypes: bracket
2100 decodetypes with commas, then search for ,type, and ,type/subtype, in
2101 it. */
2102
2103 int found_match = 0;
2104 char *delimited_decodetypes = concat(",", decodetypes, ",", NULL);
2105 char *delimited_type = concat(",", type, ",", NULL);
2106
2107 if (nmh_strcasestr(delimited_decodetypes, delimited_type)) {
2108 found_match = 1;
2109 } else if (subtype != NULL) {
2110 char *delimited_type_subtype =
2111 concat(",", type, "/", subtype, ",", NULL);
2112
2113 if (nmh_strcasestr(delimited_decodetypes, delimited_type_subtype)) {
2114 found_match = 1;
2115 }
2116 free(delimited_type_subtype);
2117 }
2118
2119 free(delimited_type);
2120 free(delimited_decodetypes);
2121
2122 return found_match;
2123 }
2124
2125
2126 /*
2127 * See if the decoded content is 7bit, 8bit, or binary. It's binary
2128 * if it has any NUL characters, a CR not followed by a LF, or lines
2129 * greater than 998 characters in length. If binary, reason is set
2130 * to a string explaining why.
2131 */
2132 static int
2133 content_encoding (CT ct, const char **reason) {
2134 CE ce = &ct->c_cefile;
2135 int encoding = CE_7BIT;
2136
2137 if (ce->ce_file) {
2138 size_t line_len = 0;
2139 char buffer[NMH_BUFSIZ];
2140 size_t inbytes;
2141
2142 if (! ce->ce_fp && (ce->ce_fp = fopen (ce->ce_file, "r")) == NULL) {
2143 advise (ce->ce_file, "unable to open for reading");
2144 return CE_UNKNOWN;
2145 }
2146
2147 fseeko (ce->ce_fp, 0L, SEEK_SET);
2148 while (encoding != CE_BINARY &&
2149 (inbytes = fread (buffer, 1, sizeof buffer, ce->ce_fp)) > 0) {
2150 char *cp;
2151 size_t i;
2152 int last_char_was_cr = 0;
2153
2154 for (i = 0, cp = buffer; i < inbytes; ++i, ++cp) {
2155 if (*cp == '\0' || ++line_len > 998 ||
2156 (*cp != '\n' && last_char_was_cr)) {
2157 encoding = CE_BINARY;
2158 if (*cp == '\0') {
2159 *reason = "null character";
2160 } else if (line_len > 998) {
2161 *reason = "line length > 998";
2162 } else if (*cp != '\n' && last_char_was_cr) {
2163 *reason = "CR not followed by LF";
2164 } else {
2165 /* Should not reach this. */
2166 *reason = "";
2167 }
2168 break;
2169 } else if (*cp == '\n') {
2170 line_len = 0;
2171 } else if (! isascii ((unsigned char) *cp)) {
2172 encoding = CE_8BIT;
2173 }
2174
2175 last_char_was_cr = *cp == '\r' ? 1 : 0;
2176 }
2177 }
2178
2179 fclose (ce->ce_fp);
2180 ce->ce_fp = NULL;
2181 } /* else should never happen */
2182
2183 return encoding;
2184 }
2185
2186
2187 /*
2188 * Strip carriage returns from content.
2189 */
2190 static int
2191 strip_crs (CT ct, int *message_mods) {
2192 char *charset = content_charset (ct);
2193 int status = OK;
2194
2195 /* Only strip carriage returns if content is ASCII or another
2196 charset that has the same readily recognizable CR followed by a
2197 LF. We can include UTF-8 here because if the high-order bit of
2198 a UTF-8 byte is 0, then it must be a single-byte ASCII
2199 character. */
2200 if (! strcasecmp (charset, "US-ASCII") ||
2201 ! strcasecmp (charset, "UTF-8") ||
2202 ! strncasecmp (charset, "ISO-8859-", 9) ||
2203 ! strncasecmp (charset, "WINDOWS-12", 10)) {
2204 char **file = NULL;
2205 FILE **fp = NULL;
2206 size_t begin;
2207 size_t end;
2208 int has_crs = 0;
2209 int opened_input_file = 0;
2210
2211 if (ct->c_cefile.ce_file) {
2212 file = &ct->c_cefile.ce_file;
2213 fp = &ct->c_cefile.ce_fp;
2214 begin = end = 0;
2215 } else if (ct->c_file) {
2216 file = &ct->c_file;
2217 fp = &ct->c_fp;
2218 begin = (size_t) ct->c_begin;
2219 end = (size_t) ct->c_end;
2220 } /* else don't know where the content is */
2221
2222 if (file && *file && fp) {
2223 if (! *fp) {
2224 if ((*fp = fopen (*file, "r")) == NULL) {
2225 advise (*file, "unable to open for reading");
2226 status = NOTOK;
2227 } else {
2228 opened_input_file = 1;
2229 }
2230 }
2231 }
2232
2233 if (fp && *fp) {
2234 char buffer[NMH_BUFSIZ];
2235 size_t bytes_read;
2236 size_t bytes_to_read =
2237 end > 0 && end > begin ? end - begin : sizeof buffer;
2238
2239 fseeko (*fp, begin, SEEK_SET);
2240 while ((bytes_read = fread (buffer, 1,
2241 min (bytes_to_read, sizeof buffer),
2242 *fp)) > 0) {
2243 /* Look for CR followed by a LF. This is supposed to
2244 be text so there should be LF's. If not, don't
2245 modify the content. */
2246 char *cp;
2247 size_t i;
2248 int last_char_was_cr = 0;
2249
2250 if (end > 0) { bytes_to_read -= bytes_read; }
2251
2252 for (i = 0, cp = buffer; i < bytes_read; ++i, ++cp) {
2253 if (*cp == '\n' && last_char_was_cr) {
2254 has_crs = 1;
2255 break;
2256 }
2257
2258 last_char_was_cr = *cp == '\r' ? 1 : 0;
2259 }
2260 }
2261
2262 if (has_crs) {
2263 int fd;
2264 char *stripped_content_file;
2265 char *tempfile = m_mktemp2 (NULL, invo_name, &fd, NULL);
2266
2267 if (tempfile == NULL) {
2268 adios (NULL, "unable to create temporary file in %s",
2269 get_temp_dir());
2270 }
2271 stripped_content_file = mh_xstrdup (tempfile);
2272
2273 /* Strip each CR before a LF from the content. */
2274 fseeko (*fp, begin, SEEK_SET);
2275 while ((bytes_read = fread (buffer, 1, sizeof buffer, *fp)) >
2276 0) {
2277 char *cp;
2278 size_t i;
2279 int last_char_was_cr = 0;
2280
2281 for (i = 0, cp = buffer; i < bytes_read; ++i, ++cp) {
2282 if (*cp == '\r') {
2283 last_char_was_cr = 1;
2284 } else if (last_char_was_cr) {
2285 if (*cp != '\n') {
2286 if (write (fd, "\r", 1) < 0) {
2287 advise (tempfile, "CR write");
2288 }
2289 }
2290 if (write (fd, cp, 1) < 0) {
2291 advise (tempfile, "write");
2292 }
2293 last_char_was_cr = 0;
2294 } else {
2295 if (write (fd, cp, 1) < 0) {
2296 advise (tempfile, "write");
2297 }
2298 last_char_was_cr = 0;
2299 }
2300 }
2301 }
2302
2303 if (close (fd)) {
2304 inform("unable to write temporary file %s, continuing...",
2305 stripped_content_file);
2306 (void) m_unlink (stripped_content_file);
2307 status = NOTOK;
2308 } else {
2309 /* Replace the decoded file with the converted one. */
2310 if (ct->c_cefile.ce_file && ct->c_cefile.ce_unlink)
2311 (void) m_unlink (ct->c_cefile.ce_file);
2312
2313 mh_xfree(ct->c_cefile.ce_file);
2314 ct->c_cefile.ce_file = stripped_content_file;
2315 ct->c_cefile.ce_unlink = 1;
2316
2317 ++*message_mods;
2318 if (verbosw) {
2319 report (NULL, ct->c_partno,
2320 begin == 0 && end == 0 ? "" : *file,
2321 "stripped CRs");
2322 }
2323 }
2324 }
2325
2326 if (opened_input_file) {
2327 fclose (*fp);
2328 *fp = NULL;
2329 }
2330 }
2331 }
2332
2333 free (charset);
2334
2335 return status;
2336 }
2337
2338
2339 /*
2340 * Add/update, if necessary, the message C-T-E, based on the least restrictive
2341 * of the part C-T-E's.
2342 */
2343 static void
2344 update_cte (CT ct) {
2345 const int least_restrictive_enc = least_restrictive_encoding (ct);
2346
2347 if (least_restrictive_enc != CE_UNKNOWN &&
2348 least_restrictive_enc != CE_7BIT) {
2349 char *cte = concat (" ", ce_str (least_restrictive_enc), "\n", NULL);
2350 HF hf;
2351 int found_cte = 0;
2352
2353 /* Update/add Content-Transfer-Encoding header field. */
2354 for (hf = ct->c_first_hf; hf; hf = hf->next) {
2355 if (! strcasecmp (ENCODING_FIELD, hf->name)) {
2356 found_cte = 1;
2357 free (hf->value);
2358 hf->value = cte;
2359 }
2360 }
2361 if (! found_cte) {
2362 add_header (ct, mh_xstrdup (ENCODING_FIELD), cte);
2363 }
2364 }
2365 }
2366
2367
2368 /*
2369 * Find the least restrictive encoding (7bit, 8bit, binary) of the parts
2370 * within a message.
2371 */
2372 static int
2373 least_restrictive_encoding (CT ct) {
2374 int encoding = CE_UNKNOWN;
2375
2376 switch (ct->c_type) {
2377 case CT_MULTIPART: {
2378 struct multipart *m = (struct multipart *) ct->c_ctparams;
2379 struct part *part;
2380
2381 for (part = m->mp_parts; part; part = part->mp_next) {
2382 const int part_encoding =
2383 least_restrictive_encoding (part->mp_part);
2384
2385 if (less_restrictive (encoding, part_encoding)) {
2386 encoding = part_encoding;
2387 }
2388 }
2389 break;
2390 }
2391
2392 case CT_MESSAGE:
2393 if (ct->c_subtype == MESSAGE_EXTERNAL) {
2394 struct exbody *e = (struct exbody *) ct->c_ctparams;
2395 const int part_encoding =
2396 least_restrictive_encoding (e->eb_content);
2397
2398 if (less_restrictive (encoding, part_encoding)) {
2399 encoding = part_encoding;
2400 }
2401 }
2402 break;
2403
2404 default: {
2405 if (less_restrictive (encoding, ct->c_encoding)) {
2406 encoding = ct->c_encoding;
2407 }
2408 }}
2409
2410 return encoding;
2411 }
2412
2413
2414 /*
2415 * Return whether the second encoding is less restrictive than the first, where
2416 * "less restrictive" is in the sense used by RFC 2045 Secs. 6.1 and 6.4. So,
2417 * CE_BINARY is less restrictive than CE_8BIT and
2418 * CE_8BIT is less restrictive than CE_7BIT.
2419 */
2420 static int
2421 less_restrictive (int encoding, int second_encoding) {
2422 switch (second_encoding) {
2423 case CE_BINARY:
2424 return encoding != CE_BINARY;
2425 case CE_8BIT:
2426 return encoding != CE_BINARY && encoding != CE_8BIT;
2427 case CE_7BIT:
2428 return encoding != CE_BINARY && encoding != CE_8BIT &&
2429 encoding != CE_7BIT;
2430 default :
2431 return 0;
2432 }
2433 }
2434
2435
2436 /*
2437 * Convert character set of each part.
2438 */
2439 static int
2440 convert_charsets (CT ct, char *dest_charset, int *message_mods) {
2441 int status = OK;
2442
2443 switch (ct->c_type) {
2444 case CT_TEXT:
2445 if (ct->c_subtype == TEXT_PLAIN) {
2446 status = convert_charset (ct, dest_charset, message_mods);
2447 if (status == OK) {
2448 if (verbosw) {
2449 char *ct_charset = content_charset (ct);
2450
2451 report (NULL, ct->c_partno, ct->c_file,
2452 "convert %s to %s", ct_charset, dest_charset);
2453 free (ct_charset);
2454 }
2455 } else {
2456 char *ct_charset = content_charset (ct);
2457
2458 report ("iconv", ct->c_partno, ct->c_file,
2459 "failed to convert %s to %s", ct_charset, dest_charset);
2460 free (ct_charset);
2461 }
2462 }
2463 break;
2464
2465 case CT_MULTIPART: {
2466 struct multipart *m = (struct multipart *) ct->c_ctparams;
2467 struct part *part;
2468
2469 /* Should check to see if the body for this part is encoded?
2470 For now, it gets passed along as-is by InitMultiPart(). */
2471 for (part = m->mp_parts; status == OK && part; part = part->mp_next) {
2472 status =
2473 convert_charsets (part->mp_part, dest_charset, message_mods);
2474 }
2475 break;
2476 }
2477
2478 case CT_MESSAGE:
2479 if (ct->c_subtype == MESSAGE_EXTERNAL) {
2480 struct exbody *e = (struct exbody *) ct->c_ctparams;
2481
2482 status =
2483 convert_charsets (e->eb_content, dest_charset, message_mods);
2484 }
2485 break;
2486
2487 default:
2488 break;
2489 }
2490
2491 return status;
2492 }
2493
2494
2495 /*
2496 * Fix various problems that aren't handled elsewhere. These
2497 * are fixed unconditionally: there are no switches to disable
2498 * them. Currently, "problems" are these:
2499 * 1) remove extraneous semicolon at the end of a header parameter list
2500 * 2) replace RFC 2047 encoding with RFC 2231 encoding of name and
2501 * filename parameters in Content-Type and Content-Disposition
2502 * headers, respectively.
2503 */
2504 static int
2505 fix_always (CT ct, int *message_mods) {
2506 int status = OK;
2507
2508 switch (ct->c_type) {
2509 case CT_MULTIPART: {
2510 struct multipart *m = (struct multipart *) ct->c_ctparams;
2511 struct part *part;
2512
2513 for (part = m->mp_parts; status == OK && part; part = part->mp_next) {
2514 status = fix_always (part->mp_part, message_mods);
2515 }
2516 break;
2517 }
2518
2519 case CT_MESSAGE:
2520 if (ct->c_subtype == MESSAGE_EXTERNAL) {
2521 struct exbody *e = (struct exbody *) ct->c_ctparams;
2522
2523 status = fix_always (e->eb_content, message_mods);
2524 }
2525 break;
2526
2527 default: {
2528 HF hf;
2529
2530 if (ct->c_first_hf) {
2531 fix_filename_encoding (ct);
2532 }
2533
2534 for (hf = ct->c_first_hf; hf; hf = hf->next) {
2535 size_t len = strlen (hf->value);
2536
2537 if (strcasecmp (hf->name, TYPE_FIELD) != 0 &&
2538 strcasecmp (hf->name, DISPO_FIELD) != 0) {
2539 /* Only do this for Content-Type and
2540 Content-Disposition fields because those are the
2541 only headers that parse_mime() warns about. */
2542 continue;
2543 }
2544
2545 /* whitespace following a trailing ';' will be nuked as well */
2546 if (hf->value[len - 1] == '\n') {
2547 while (isspace((unsigned char)(hf->value[len - 2]))) {
2548 if (len-- == 0) { break; }
2549 }
2550 }
2551
2552 if (hf->value[len - 2] == ';') {
2553 /* Remove trailing ';' from parameter value. */
2554 hf->value[len - 2] = '\n';
2555 hf->value[len - 1] = '\0';
2556
2557 /* Also, if Content-Type parameter, remove trailing ';'
2558 from ct->c_ctline. This probably isn't necessary
2559 but can't hurt. */
2560 if (strcasecmp(hf->name, TYPE_FIELD) == 0 && ct->c_ctline) {
2561 size_t l = strlen(ct->c_ctline) - 1;
2562 while (isspace((unsigned char)(ct->c_ctline[l])) ||
2563 ct->c_ctline[l] == ';') {
2564 ct->c_ctline[l--] = '\0';
2565 if (l == 0) { break; }
2566 }
2567 }
2568
2569 ++*message_mods;
2570 if (verbosw) {
2571 report (NULL, ct->c_partno, ct->c_file,
2572 "remove trailing ; from %s parameter value",
2573 hf->name);
2574 }
2575 }
2576 }
2577 }}
2578
2579 return status;
2580 }
2581
2582
2583 /*
2584 * Factor out common code for loops in fix_filename_encoding().
2585 */
2586 static int
2587 fix_filename_param (char *name, char *value, PM *first_pm, PM *last_pm) {
2588 int fixed = 0;
2589
2590 if (has_prefix(value, "=?") && has_suffix(value, "?=")) {
2591 /* Looks like an RFC 2047 encoded parameter. */
2592 char decoded[PATH_MAX + 1];
2593
2594 if (decode_rfc2047 (value, decoded, sizeof decoded)) {
2595 /* Encode using RFC 2231. */
2596 replace_param (first_pm, last_pm, name, decoded, 0);
2597 fixed = 1;
2598 } else {
2599 inform("failed to decode %s parameter %s", name, value);
2600 }
2601 }
2602
2603 return fixed;
2604 }
2605
2606
2607 /*
2608 * Replace RFC 2047 encoding with RFC 2231 encoding of name and
2609 * filename parameters in Content-Type and Content-Disposition
2610 * headers, respectively.
2611 */
2612 static int
2613 fix_filename_encoding (CT ct) {
2614 PM pm;
2615 HF hf;
2616 int fixed = 0;
2617
2618 for (pm = ct->c_ctinfo.ci_first_pm; pm; pm = pm->pm_next) {
2619 if (pm->pm_name && pm->pm_value &&
2620 strcasecmp (pm->pm_name, "name") == 0) {
2621 fixed = fix_filename_param (pm->pm_name, pm->pm_value,
2622 &ct->c_ctinfo.ci_first_pm,
2623 &ct->c_ctinfo.ci_last_pm);
2624 }
2625 }
2626
2627 for (pm = ct->c_dispo_first; pm; pm = pm->pm_next) {
2628 if (pm->pm_name && pm->pm_value &&
2629 strcasecmp (pm->pm_name, "filename") == 0) {
2630 fixed = fix_filename_param (pm->pm_name, pm->pm_value,
2631 &ct->c_dispo_first,
2632 &ct->c_dispo_last);
2633 }
2634 }
2635
2636 /* Fix hf values to correspond. */
2637 for (hf = ct->c_first_hf; fixed && hf; hf = hf->next) {
2638 enum { OTHER, TYPE_HEADER, DISPO_HEADER } field = OTHER;
2639
2640 if (strcasecmp (hf->name, TYPE_FIELD) == 0) {
2641 field = TYPE_HEADER;
2642 } else if (strcasecmp (hf->name, DISPO_FIELD) == 0) {
2643 field = DISPO_HEADER;
2644 }
2645
2646 if (field != OTHER) {
2647 const char *const semicolon_loc = strchr (hf->value, ';');
2648
2649 if (semicolon_loc) {
2650 const size_t len =
2651 strlen (hf->name) + 1 + semicolon_loc - hf->value;
2652 const char *const params =
2653 output_params (len,
2654 field == TYPE_HEADER
2655 ? ct->c_ctinfo.ci_first_pm
2656 : ct->c_dispo_first,
2657 NULL, 0);
2658 const char *const new_params = concat (params, "\n", NULL);
2659
2660 replace_substring (&hf->value, semicolon_loc, new_params);
2661 free((void *)new_params); /* Cast away const. Sigh. */
2662 free((void *)params);
2663 } else {
2664 inform("did not find semicolon in %s:%s\n",
2665 hf->name, hf->value);
2666 }
2667 }
2668 }
2669
2670 return OK;
2671 }
2672
2673
2674 /*
2675 * Output content in input file to output file.
2676 */
2677 static int
2678 write_content (CT ct, const char *input_filename, char *outfile, FILE *outfp,
2679 int modify_inplace, int message_mods) {
2680 int status = OK;
2681
2682 if (modify_inplace) {
2683 if (message_mods > 0) {
2684 if ((status = output_message_fp (ct, outfp, outfile)) == OK) {
2685 char *infile = input_filename
2686 ? mh_xstrdup (input_filename)
2687 : mh_xstrdup (ct->c_file ? ct->c_file : "-");
2688
2689 if (remove_file (infile) == OK) {
2690 if (rename (outfile, infile)) {
2691 /* Rename didn't work, possibly because of an
2692 attempt to rename across filesystems. Try
2693 brute force copy. */
2694 int old = open (outfile, O_RDONLY);
2695 int new =
2696 open (infile, O_WRONLY | O_CREAT, m_gmprot ());
2697 int i = -1;
2698
2699 if (old != -1 && new != -1) {
2700 char buffer[NMH_BUFSIZ];
2701
2702 while ((i = read (old, buffer, sizeof buffer)) >
2703 0) {
2704 if (write (new, buffer, i) != i) {
2705 i = -1;
2706 break;
2707 }
2708 }
2709 }
2710 if (new != -1) { close (new); }
2711 if (old != -1) { close (old); }
2712 (void) m_unlink (outfile);
2713
2714 if (i < 0) {
2715 /* The -file argument processing used path() to
2716 expand filename to absolute path. */
2717 int file = ct->c_file && ct->c_file[0] == '/';
2718
2719 inform("unable to rename %s %s to %s, continuing...",
2720 file ? "file" : "message", outfile,
2721 infile);
2722 status = NOTOK;
2723 }
2724 }
2725 } else {
2726 inform("unable to remove input file %s, "
2727 "not modifying it, continuing...", infile);
2728 (void) m_unlink (outfile);
2729 status = NOTOK;
2730 }
2731
2732 free (infile);
2733 } else {
2734 status = NOTOK;
2735 }
2736 } else {
2737 /* No modifications and didn't need the tmp outfile. */
2738 (void) m_unlink (outfile);
2739 }
2740 } else {
2741 /* Output is going to some file. Produce it whether or not
2742 there were modifications. */
2743 status = output_message_fp (ct, outfp, outfile);
2744 }
2745
2746 flush_errors ();
2747 return status;
2748 }
2749
2750
2751 /*
2752 * parse_mime() does not set lf_line_endings in struct text, so use this
2753 * function to do it. It touches the parts the decodetypes identifies.
2754 */
2755 static void
2756 set_text_ctparams(CT ct, char *decodetypes, int lf_line_endings) {
2757 switch (ct->c_type) {
2758 case CT_MULTIPART: {
2759 struct multipart *m = (struct multipart *) ct->c_ctparams;
2760 struct part *part;
2761
2762 for (part = m->mp_parts; part; part = part->mp_next) {
2763 set_text_ctparams(part->mp_part, decodetypes, lf_line_endings);
2764 }
2765 break;
2766 }
2767
2768 case CT_MESSAGE:
2769 if (ct->c_subtype == MESSAGE_EXTERNAL) {
2770 struct exbody *e = (struct exbody *) ct->c_ctparams;
2771
2772 set_text_ctparams(e->eb_content, decodetypes, lf_line_endings);
2773 }
2774 break;
2775
2776 default:
2777 if (should_decode(decodetypes, ct->c_ctinfo.ci_type, ct->c_ctinfo.ci_subtype)) {
2778 if (ct->c_ctparams == NULL) {
2779 ct->c_ctparams = mh_xcalloc(1, sizeof (struct text));
2780 }
2781 ((struct text *) ct->c_ctparams)->lf_line_endings = lf_line_endings;
2782 }
2783 }
2784 }
2785
2786
2787 /*
2788 * If "rmmproc" is defined, call that to remove the file. Otherwise,
2789 * use the standard MH backup file.
2790 */
2791 static int
2792 remove_file (const char *file) {
2793 if (rmmproc) {
2794 char *rmm_command = concat (rmmproc, " ", file, NULL);
2795 int status = system (rmm_command);
2796
2797 free (rmm_command);
2798 return WIFEXITED (status) ? WEXITSTATUS (status) : NOTOK;
2799 }
2800 /* This is OK for a non-message file, it still uses the
2801 BACKUP_PREFIX form. The backup file will be in the same
2802 directory as file. */
2803 return rename (file, m_backup (file));
2804 }
2805
2806
2807 /*
2808 * Output formatted message to user.
2809 */
2810 static void
2811 report (char *what, char *partno, char *filename, char *message, ...) {
2812 va_list args;
2813 char *fmt;
2814
2815 if (verbosw) {
2816 va_start (args, message);
2817 fmt = concat (filename, partno ? " part " : ", ",
2818 partno ? partno : "", partno ? ", " : "", message, NULL);
2819
2820 advertise (what, NULL, fmt, args);
2821
2822 free (fmt);
2823 va_end (args);
2824 }
2825 }
2826
2827
2828 static void
2829 pipeser (int i)
2830 {
2831 if (i == SIGQUIT) {
2832 fflush (stdout);
2833 fprintf (stderr, "\n");
2834 fflush (stderr);
2835 }
2836
2837 done (1);
2838 /* NOTREACHED */
2839 }
Fork of https://git.savannah.nongnu.org/cgit/nmh.git