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