]> diplodocus.org Git - nmh/blob - uip/mhfixmsg.c
projects / nmh / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
cppflags.m4: Don't trample CFLAGS and CPPFLAGS.
[nmh] / uip / mhfixmsg.c
1 /* mhfixmsg.c -- rewrite a message with various transformations
2 *
3 * This code is Copyright (c) 2002 and 2013, by the authors of nmh.
4 * See the COPYRIGHT file in the root directory of the nmh
5 * distribution for complete copyright information.
6 */
7
8 #include <h/mh.h>
9 #include <h/fmt_scan.h>
10 #include <h/mime.h>
11 #include <h/mhparse.h>
12 #include <h/utils.h>
13 #include <h/signals.h>
14 #include "../sbr/m_maildir.h"
15 #include "../sbr/m_mktemp.h"
16 #include "../sbr/mime_type.h"
17 #include "mhfree.h"
18 #include "mhoutsbr.h"
19 #include "mhshowsbr.h"
20 #include <fcntl.h>
21
22 #define MHFIXMSG_SWITCHES \
23 X("decodetext 8bit|7bit|binary", 0, DECODETEXTSW) \
24 X("nodecodetext", 0, NDECODETEXTSW) \
25 X("decodetypes", 0, DECODETYPESW) \
26 X("crlflinebreaks", 0, CRLFLINEBREAKSSW) \
27 X("nocrlflinebreaks", 0, NCRLFLINEBREAKSSW) \
28 X("textcharset", 0, TEXTCHARSETSW) \
29 X("notextcharset", 0, NTEXTCHARSETSW) \
30 X("reformat", 0, REFORMATSW) \
31 X("noreformat", 0, NREFORMATSW) \
32 X("replacetextplain", 0, REPLACETEXTPLAINSW) \
33 X("noreplacetextplain", 0, NREPLACETEXTPLAINSW) \
34 X("fixboundary", 0, FIXBOUNDARYSW) \
35 X("nofixboundary", 0, NFIXBOUNDARYSW) \
36 X("fixcte", 0, FIXCOMPOSITECTESW) \
37 X("nofixcte", 0, NFIXCOMPOSITECTESW) \
38 X("fixtype mimetype", 0, FIXTYPESW) \
39 X("file file", 0, FILESW) \
40 X("outfile file", 0, OUTFILESW) \
41 X("rmmproc program", 0, RPROCSW) \
42 X("normmproc", 0, NRPRCSW) \
43 X("changecur", 0, CHGSW) \
44 X("nochangecur", 0, NCHGSW) \
45 X("verbose", 0, VERBSW) \
46 X("noverbose", 0, NVERBSW) \
47 X("version", 0, VERSIONSW) \
48 X("help", 0, HELPSW) \
49
50 #define X(sw, minchars, id) id,
51 DEFINE_SWITCH_ENUM(MHFIXMSG);
52 #undef X
53
54 #define X(sw, minchars, id) { sw, minchars, id },
55 DEFINE_SWITCH_ARRAY(MHFIXMSG, switches);
56 #undef X
57
58
59 int verbosw;
60 int debugsw; /* Needed by mhparse.c. */
61
62 #define quitser pipeser
63
64 /* mhparse.c */
65 extern int skip_mp_cte_check; /* flag to InitMultiPart */
66 extern int suppress_bogus_mp_content_warning; /* flag to InitMultiPart */
67 extern int bogus_mp_content; /* flag from InitMultiPart */
68 /* flags to/from parse_header_attrs */
69 extern int suppress_extraneous_trailing_semicolon_warning;
70
71 /* mhmisc.c */
72 void flush_errors (void);
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 break;
770 begin -= sizeof buffer;
771 }
772
773 /* Get boundary at beginning of multipart. */
774 if (end_boundary) {
775 fseeko (ct->c_fp, ct->c_begin, SEEK_SET);
776 while ((bytes_read = fread (buffer, 1, sizeof buffer, ct->c_fp)) > 0) {
777 if (bytes_read >= strlen (end_boundary)) {
778 char *cp = find_str (buffer, bytes_read, end_boundary);
779
780 if (cp && cp - buffer >= 2 && *--cp == '-' &&
781 *--cp == '-' && (cp > buffer && *--cp == '\n')) {
782 status = OK;
783 break;
784 }
785 } else {
786 /* The start and end boundaries didn't match, or the
787 start boundary doesn't begin with "\n--" (or "--"
788 if at the beginning of buffer). Keep trying. */
789 status = NOTOK;
790 }
791 }
792 } else {
793 status = NOTOK;
794 }
795
796 if (ct->c_fp) {
797 fclose (ct->c_fp);
798 ct->c_fp = NULL;
799 }
800
801 if (status == OK) {
802 *part_boundary = end_boundary;
803 } else {
804 *part_boundary = NULL;
805 free (end_boundary);
806 }
807
808 return status;
809 }
810
811
812 /*
813 * Open and copy ct->c_file to file, replacing the multipart boundary.
814 */
815 static int
816 replace_boundary (CT ct, char *file, char *boundary) {
817 FILE *fpin, *fpout;
818 int compnum, state;
819 char buf[NMH_BUFSIZ], name[NAMESZ];
820 char *np, *vp;
821 m_getfld_state_t gstate;
822 int status = OK;
823
824 if (ct->c_file == NULL) {
825 inform("missing input filename");
826 return NOTOK;
827 }
828
829 if ((fpin = fopen (ct->c_file, "r")) == NULL) {
830 advise (ct->c_file, "unable to open for reading");
831 return NOTOK;
832 }
833
834 if ((fpout = fopen (file, "w")) == NULL) {
835 fclose (fpin);
836 advise (file, "unable to open for writing");
837 return NOTOK;
838 }
839
840 gstate = m_getfld_state_init(fpin);
841 for (compnum = 1;;) {
842 int bufsz = (int) sizeof buf;
843
844 switch (state = m_getfld2(&gstate, name, buf, &bufsz)) {
845 case FLD:
846 case FLDPLUS:
847 compnum++;
848
849 /* get copies of the buffers */
850 np = mh_xstrdup (name);
851 vp = mh_xstrdup (buf);
852
853 /* if necessary, get rest of field */
854 while (state == FLDPLUS) {
855 bufsz = sizeof buf;
856 state = m_getfld2(&gstate, name, buf, &bufsz);
857 vp = add (buf, vp); /* add to previous value */
858 }
859
860 if (strcasecmp (TYPE_FIELD, np)) {
861 fprintf (fpout, "%s:%s", np, vp);
862 } else {
863 char *new_ctline, *new_params;
864
865 replace_param(&ct->c_ctinfo.ci_first_pm,
866 &ct->c_ctinfo.ci_last_pm, "boundary",
867 boundary, 0);
868
869 new_ctline = concat(" ", ct->c_ctinfo.ci_type, "/",
870 ct->c_ctinfo.ci_subtype, NULL);
871 new_params = output_params(strlen(TYPE_FIELD) +
872 strlen(new_ctline) + 1,
873 ct->c_ctinfo.ci_first_pm, NULL, 0);
874 fprintf (fpout, "%s:%s%s\n", np, new_ctline,
875 FENDNULL(new_params));
876 free(new_ctline);
877 mh_xfree(new_params);
878 }
879
880 free (vp);
881 free (np);
882
883 continue;
884
885 case BODY:
886 putc('\n', fpout);
887 /* buf will have a terminating NULL, skip it. */
888 if ((int) fwrite (buf, 1, bufsz-1, fpout) < bufsz-1) {
889 advise (file, "fwrite");
890 }
891 continue;
892
893 case FILEEOF:
894 break;
895
896 case LENERR:
897 case FMTERR:
898 inform("message format error in component #%d", compnum);
899 status = NOTOK;
900 break;
901
902 default:
903 inform("getfld() returned %d", state);
904 status = NOTOK;
905 break;
906 }
907
908 break;
909 }
910
911 m_getfld_state_destroy (&gstate);
912 fclose (fpout);
913 fclose (fpin);
914
915 return status;
916 }
917
918
919 /*
920 * Fix Content-Type header to reflect the content of its part.
921 */
922 static int
923 fix_types (CT ct, svector_t fixtypes, int *message_mods) {
924 int status = OK;
925
926 switch (ct->c_type) {
927 case CT_MULTIPART: {
928 struct multipart *m = (struct multipart *) ct->c_ctparams;
929 struct part *part;
930
931 for (part = m->mp_parts; status == OK && part; part = part->mp_next) {
932 status = fix_types (part->mp_part, fixtypes, message_mods);
933 }
934 break;
935 }
936
937 case CT_MESSAGE:
938 if (ct->c_subtype == MESSAGE_EXTERNAL) {
939 struct exbody *e = (struct exbody *) ct->c_ctparams;
940
941 status = fix_types (e->eb_content, fixtypes, message_mods);
942 }
943 break;
944
945 default: {
946 char **typep, *type;
947
948 if (ct->c_ctinfo.ci_type && ct->c_ctinfo.ci_subtype) {
949 for (typep = svector_strs (fixtypes);
950 typep && (type = *typep);
951 ++typep) {
952 char *type_subtype =
953 concat (ct->c_ctinfo.ci_type, "/", ct->c_ctinfo.ci_subtype,
954 NULL);
955
956 if (! strcasecmp (type, type_subtype) &&
957 decode_part (ct) == OK &&
958 ct->c_cefile.ce_file != NULL) {
959 char *ct_type_subtype = mime_type (ct->c_cefile.ce_file);
960 char *cp;
961
962 if ((cp = strchr (ct_type_subtype, ';'))) {
963 /* Truncate to remove any parameter list from
964 mime_type () result. */
965 *cp = '\0';
966 }
967
968 if (strcasecmp (type, ct_type_subtype)) {
969 char *ct_type, *ct_subtype;
970 HF hf;
971
972 /* The Content-Type header does not match the
973 content, so update these struct Content
974 fields to match:
975 * c_type, c_subtype
976 * c_ctinfo.ci_type, c_ctinfo.ci_subtype
977 * c_ctline
978 */
979 /* Extract type and subtype from type/subtype. */
980 ct_type = mh_xstrdup(ct_type_subtype);
981 if ((cp = strchr (ct_type, '/'))) {
982 *cp = '\0';
983 ct_subtype = mh_xstrdup(++cp);
984 } else {
985 inform("missing / in MIME type of %s %s",
986 ct->c_file, ct->c_partno);
987 free (ct_type);
988 return NOTOK;
989 }
990
991 ct->c_type = ct_str_type (ct_type);
992 ct->c_subtype = ct_str_subtype (ct->c_type, ct_subtype);
993
994 free (ct->c_ctinfo.ci_type);
995 ct->c_ctinfo.ci_type = ct_type;
996 free (ct->c_ctinfo.ci_subtype);
997 ct->c_ctinfo.ci_subtype = ct_subtype;
998 if (! replace_substring (&ct->c_ctline, type,
999 ct_type_subtype)) {
1000 inform("did not find %s in %s",
1001 type, ct->c_ctline);
1002 }
1003
1004 /* Update Content-Type header field. */
1005 for (hf = ct->c_first_hf; hf; hf = hf->next) {
1006 if (! strcasecmp (TYPE_FIELD, hf->name)) {
1007 if (replace_substring (&hf->value, type,
1008 ct_type_subtype)) {
1009 ++*message_mods;
1010 if (verbosw) {
1011 report (NULL, ct->c_partno, ct->c_file,
1012 "change Content-Type in header "
1013 "from %s to %s",
1014 type, ct_type_subtype);
1015 }
1016 break;
1017 }
1018 inform("did not find %s in %s", type, hf->value);
1019 }
1020 }
1021 }
1022 free (ct_type_subtype);
1023 }
1024 free (type_subtype);
1025 }
1026 }
1027 }}
1028
1029 return status;
1030 }
1031
1032
1033 /*
1034 * Replace a substring, allocating space to hold the new one.
1035 */
1036 char *
1037 replace_substring (char **str, const char *old, const char *new) {
1038 char *cp;
1039
1040 if ((cp = strstr (*str, old))) {
1041 char *remainder = cp + strlen (old);
1042 char *prefix, *new_str;
1043
1044 if (cp - *str) {
1045 prefix = mh_xstrdup(*str);
1046 *(prefix + (cp - *str)) = '\0';
1047 new_str = concat (prefix, new, remainder, NULL);
1048 free (prefix);
1049 } else {
1050 new_str = concat (new, remainder, NULL);
1051 }
1052
1053 free (*str);
1054
1055 return *str = new_str;
1056 }
1057
1058 return NULL;
1059 }
1060
1061
1062 /*
1063 * Remove a name=value parameter, given just its name, from a header value.
1064 */
1065 char *
1066 remove_parameter (char *str, const char *name) {
1067 /* It looks to me, based on the BNF in RFC 2045, than there can't
1068 be whitespace between the parameter name and the "=", or
1069 between the "=" and the parameter value. */
1070 char *param_name = concat (name, "=", NULL);
1071 char *cp;
1072
1073 if ((cp = strstr (str, param_name))) {
1074 char *start, *end;
1075 size_t count = 1;
1076
1077 /* Remove any leading spaces, before the parameter name. */
1078 for (start = cp;
1079 start > str && isspace ((unsigned char) *(start-1));
1080 --start) {
1081 continue;
1082 }
1083 /* Remove a leading semicolon. */
1084 if (start > str && *(start-1) == ';') { --start; }
1085
1086 end = cp + strlen (name) + 1;
1087 if (*end == '"') {
1088 /* Skip past the quoted value, and then the final quote. */
1089 for (++end ; *end && *end != '"'; ++end) { continue; }
1090 ++end;
1091 } else {
1092 /* Skip past the value. */
1093 for (++end ; *end && ! isspace ((unsigned char) *end); ++end) {}
1094 }
1095
1096 /* Count how many characters need to be moved. Include
1097 trailing null, which is accounted for by the
1098 initialization of count to 1. */
1099 for (cp = end; *cp; ++cp) { ++count; }
1100 (void) memmove (start, end, count);
1101 }
1102
1103 free (param_name);
1104
1105 return str;
1106 }
1107
1108
1109 /*
1110 * Fix Content-Transfer-Encoding of composite,, e.g., message or multipart, part.
1111 * According to RFC 2045 Sec. 6.4, it must be 7bit, 8bit, or binary. Set it to
1112 * 8 bit.
1113 */
1114 static int
1115 fix_composite_cte (CT ct, int *message_mods) {
1116 int status = OK;
1117
1118 if (ct->c_type == CT_MESSAGE || ct->c_type == CT_MULTIPART) {
1119 if (ct->c_encoding != CE_7BIT && ct->c_encoding != CE_8BIT &&
1120 ct->c_encoding != CE_BINARY) {
1121 HF hf;
1122
1123 for (hf = ct->c_first_hf; hf; hf = hf->next) {
1124 char *name = hf->name;
1125 for (; *name && isspace ((unsigned char) *name); ++name) {
1126 continue;
1127 }
1128
1129 if (! strncasecmp (name, ENCODING_FIELD,
1130 strlen (ENCODING_FIELD))) {
1131 char *prefix = "Nmh-REPLACED-INVALID-";
1132 HF h;
1133
1134 NEW(h);
1135 h->name = mh_xstrdup (hf->name);
1136 h->hf_encoding = hf->hf_encoding;
1137 h->next = hf->next;
1138 hf->next = h;
1139
1140 /* Retain old header but prefix its name. */
1141 free (hf->name);
1142 hf->name = concat (prefix, h->name, NULL);
1143
1144 ++*message_mods;
1145 if (verbosw) {
1146 char *encoding = cpytrim (hf->value);
1147 report (NULL, ct->c_partno, ct->c_file,
1148 "replace Content-Transfer-Encoding of %s "
1149 "with 8 bit", encoding);
1150 free (encoding);
1151 }
1152
1153 h->value = mh_xstrdup (" 8bit\n");
1154
1155 /* Don't need to warn for multiple C-T-E header
1156 fields, parse_mime() already does that. But
1157 if there are any, fix them all as necessary. */
1158 hf = h;
1159 }
1160 }
1161
1162 set_ce (ct, CE_8BIT);
1163 }
1164
1165 if (ct->c_type == CT_MULTIPART) {
1166 struct multipart *m;
1167 struct part *part;
1168
1169 m = (struct multipart *) ct->c_ctparams;
1170 for (part = m->mp_parts; part; part = part->mp_next) {
1171 if (fix_composite_cte (part->mp_part, message_mods) != OK) {
1172 status = NOTOK;
1173 break;
1174 }
1175 }
1176 }
1177 }
1178
1179 return status;
1180 }
1181
1182
1183 /*
1184 * Set content encoding.
1185 */
1186 static int
1187 set_ce (CT ct, int encoding) {
1188 const char *ce = ce_str (encoding);
1189 const struct str2init *ctinit = get_ce_method (ce);
1190
1191 if (ctinit) {
1192 char *cte = concat (" ", ce, "\n", NULL);
1193 int found_cte = 0;
1194 HF hf;
1195 /* Decoded contents might be in ct->c_cefile.ce_file, if the
1196 caller is decode_text_parts (). Save because we'll
1197 overwrite below. */
1198 struct cefile decoded_content_info = ct->c_cefile;
1199
1200 ct->c_encoding = encoding;
1201
1202 ct->c_ctinitfnx = ctinit->si_init;
1203 /* This will assign ct->c_cefile with an all-0 struct, which
1204 is what we want. */
1205 (*ctinit->si_init) (ct);
1206 /* After returning, the caller should set
1207 ct->c_cefile.ce_file to the name of the file containing
1208 the contents. */
1209
1210 if (ct->c_ceclosefnx) {
1211 (*ct->c_ceclosefnx) (ct);
1212 }
1213
1214 /* Restore the cefile. */
1215 ct->c_cefile = decoded_content_info;
1216
1217 /* Update/add Content-Transfer-Encoding header field. */
1218 for (hf = ct->c_first_hf; hf; hf = hf->next) {
1219 if (! strcasecmp (ENCODING_FIELD, hf->name)) {
1220 found_cte = 1;
1221 free (hf->value);
1222 hf->value = cte;
1223 }
1224 }
1225 if (! found_cte) {
1226 add_header (ct, mh_xstrdup (ENCODING_FIELD), cte);
1227 }
1228
1229 /* Update c_celine. It's used only by mhlist -debug. */
1230 free (ct->c_celine);
1231 ct->c_celine = mh_xstrdup (cte);
1232
1233 return OK;
1234 }
1235
1236 return NOTOK;
1237 }
1238
1239
1240 /*
1241 * Make sure each text part has a corresponding text/plain part.
1242 */
1243 static int
1244 ensure_text_plain (CT *ct, CT parent, int *message_mods, int replacetextplain) {
1245 int status = OK;
1246
1247 switch ((*ct)->c_type) {
1248 case CT_TEXT: {
1249 /* Nothing to do for text/plain. */
1250 if ((*ct)->c_subtype == TEXT_PLAIN) { return OK; }
1251
1252 if (parent && parent->c_type == CT_MULTIPART &&
1253 parent->c_subtype == MULTI_ALTERNATE) {
1254 int new_subpart_number = 1;
1255 int has_text_plain =
1256 find_textplain_sibling (parent, replacetextplain,
1257 &new_subpart_number);
1258
1259 if (! has_text_plain) {
1260 /* Parent is a multipart/alternative. Insert a new
1261 text/plain subpart. */
1262 const int inserted =
1263 insert_new_text_plain_part (*ct, new_subpart_number,
1264 parent);
1265 if (inserted) {
1266 ++*message_mods;
1267 if (verbosw) {
1268 report (NULL, parent->c_partno, parent->c_file,
1269 "insert text/plain part");
1270 }
1271 } else {
1272 status = NOTOK;
1273 }
1274 }
1275 } else if (parent && parent->c_type == CT_MULTIPART &&
1276 parent->c_subtype == MULTI_RELATED) {
1277 char *type_subtype =
1278 concat ((*ct)->c_ctinfo.ci_type, "/",
1279 (*ct)->c_ctinfo.ci_subtype, NULL);
1280 const char *parent_type =
1281 get_param (parent->c_ctinfo.ci_first_pm, "type", '?', 1);
1282 int new_subpart_number = 1;
1283 int has_text_plain = 0;
1284
1285 /* Have to do string comparison on the subtype because we
1286 don't enumerate all of them in c_subtype values.
1287 parent_type will be NULL if the multipart/related part
1288 doesn't have a type parameter. The type parameter must
1289 be specified according to RFC 2387 Sec. 3.1 but not all
1290 messages comply. */
1291 if (parent_type && strcasecmp (type_subtype, parent_type) == 0) {
1292 /* The type of this part matches the root type of the
1293 parent multipart/related. Look to see if there's
1294 text/plain sibling. */
1295 has_text_plain =
1296 find_textplain_sibling (parent, replacetextplain,
1297 &new_subpart_number);
1298 }
1299
1300 free (type_subtype);
1301
1302 if (! has_text_plain) {
1303 struct multipart *mp = (struct multipart *) parent->c_ctparams;
1304 struct part *part;
1305 int siblings = 0;
1306
1307 for (part = mp->mp_parts; part; part = part->mp_next) {
1308 if (*ct != part->mp_part) {
1309 ++siblings;
1310 }
1311 }
1312
1313 if (siblings) {
1314 /* Parent is a multipart/related. Insert a new
1315 text/plain subpart in a new multipart/alternative. */
1316 if (insert_into_new_mp_alt (ct, message_mods)) {
1317 /* Not an error if text/plain couldn't be added. */
1318 }
1319 } else {
1320 /* There are no siblings, so insert a new text/plain
1321 subpart, and change the parent type from
1322 multipart/related to multipart/alternative. */
1323 const int inserted =
1324 insert_new_text_plain_part (*ct, new_subpart_number,
1325 parent);
1326
1327 if (inserted) {
1328 HF hf;
1329
1330 parent->c_subtype = MULTI_ALTERNATE;
1331 free (parent->c_ctinfo.ci_subtype);
1332 parent->c_ctinfo.ci_subtype = mh_xstrdup("alternative");
1333 if (! replace_substring (&parent->c_ctline, "/related",
1334 "/alternative")) {
1335 inform("did not find multipart/related in %s",
1336 parent->c_ctline);
1337 }
1338
1339 /* Update Content-Type header field. */
1340 for (hf = parent->c_first_hf; hf; hf = hf->next) {
1341 if (! strcasecmp (TYPE_FIELD, hf->name)) {
1342 if (replace_substring (&hf->value, "/related",
1343 "/alternative")) {
1344 ++*message_mods;
1345 if (verbosw) {
1346 report (NULL, parent->c_partno,
1347 parent->c_file,
1348 "insert text/plain part");
1349 }
1350
1351 /* Remove, e.g., type="text/html" from
1352 multipart/alternative. */
1353 remove_parameter (hf->value, "type");
1354 break;
1355 }
1356 inform("did not find multipart/"
1357 "related in header %s", hf->value);
1358 }
1359 }
1360 } else {
1361 /* Not an error if text/plain couldn't be inserted. */
1362 }
1363 }
1364 }
1365 } else {
1366 if (insert_into_new_mp_alt (ct, message_mods)) {
1367 status = NOTOK;
1368 }
1369 }
1370 break;
1371 }
1372
1373 case CT_MULTIPART: {
1374 struct multipart *mp = (struct multipart *) (*ct)->c_ctparams;
1375 struct part *part;
1376
1377 for (part = mp->mp_parts; status == OK && part; part = part->mp_next) {
1378 if ((*ct)->c_type == CT_MULTIPART) {
1379 status = ensure_text_plain (&part->mp_part, *ct, message_mods,
1380 replacetextplain);
1381 }
1382 }
1383 break;
1384 }
1385
1386 case CT_MESSAGE:
1387 if ((*ct)->c_subtype == MESSAGE_EXTERNAL) {
1388 struct exbody *e = (struct exbody *) (*ct)->c_ctparams;
1389
1390 status = ensure_text_plain (&e->eb_content, *ct, message_mods,
1391 replacetextplain);
1392 }
1393 break;
1394 }
1395
1396 return status;
1397 }
1398
1399
1400 /*
1401 * See if there is a sibling text/plain, and return its subpart number.
1402 */
1403 static int
1404 find_textplain_sibling (CT parent, int replacetextplain,
1405 int *new_subpart_number) {
1406 struct multipart *mp = (struct multipart *) parent->c_ctparams;
1407 struct part *part, *prev;
1408 int has_text_plain = 0;
1409
1410 for (prev = part = mp->mp_parts; part; part = part->mp_next) {
1411 ++*new_subpart_number;
1412 if (part->mp_part->c_type == CT_TEXT &&
1413 part->mp_part->c_subtype == TEXT_PLAIN) {
1414 if (replacetextplain) {
1415 struct part *old_part;
1416 if (part == mp->mp_parts) {
1417 old_part = mp->mp_parts;
1418 mp->mp_parts = part->mp_next;
1419 } else {
1420 old_part = prev->mp_next;
1421 prev->mp_next = part->mp_next;
1422 }
1423 if (verbosw) {
1424 report (NULL, parent->c_partno, parent->c_file,
1425 "remove text/plain part %s",
1426 old_part->mp_part->c_partno);
1427 }
1428 free_content (old_part->mp_part);
1429 free (old_part);
1430 } else {
1431 has_text_plain = 1;
1432 }
1433 break;
1434 }
1435 prev = part;
1436 }
1437
1438 return has_text_plain;
1439 }
1440
1441
1442 /*
1443 * Insert a new text/plain part.
1444 */
1445 static int
1446 insert_new_text_plain_part (CT ct, int new_subpart_number, CT parent) {
1447 struct multipart *mp = (struct multipart *) parent->c_ctparams;
1448 struct part *new_part;
1449
1450 NEW(new_part);
1451 if ((new_part->mp_part = build_text_plain_part (ct))) {
1452 char buffer[16];
1453 snprintf (buffer, sizeof buffer, "%d", new_subpart_number);
1454
1455 new_part->mp_next = mp->mp_parts;
1456 mp->mp_parts = new_part;
1457 new_part->mp_part->c_partno =
1458 concat (parent->c_partno ? parent->c_partno : "1", ".",
1459 buffer, NULL);
1460
1461 return 1;
1462 }
1463
1464 free_content (new_part->mp_part);
1465 free (new_part);
1466
1467 return 0;
1468 }
1469
1470
1471 /*
1472 * Create a text/plain part to go along with non-plain sibling part.
1473 */
1474 static CT
1475 build_text_plain_part (CT encoded_part) {
1476 CT tp_part = divide_part (encoded_part);
1477 char *tmp_plain_file = NULL;
1478
1479 if (decode_part (tp_part) == OK) {
1480 /* Now, tp_part->c_cefile.ce_file is the name of the tmp file that
1481 contains the decoded contents. And the decoding function, such
1482 as openQuoted, will have set ...->ce_unlink to 1 so that it will
1483 be unlinked by free_content (). */
1484 char *tempfile;
1485
1486 /* This m_mktemp2() call closes the temp file. */
1487 if ((tempfile = m_mktemp2 (NULL, invo_name, NULL, NULL)) == NULL) {
1488 inform("unable to create temporary file in %s",
1489 get_temp_dir());
1490 } else {
1491 tmp_plain_file = mh_xstrdup (tempfile);
1492 if (reformat_part (tp_part, tmp_plain_file,
1493 tp_part->c_ctinfo.ci_type,
1494 tp_part->c_ctinfo.ci_subtype,
1495 tp_part->c_type) == OK) {
1496 return tp_part;
1497 }
1498 }
1499 }
1500
1501 free_content (tp_part);
1502 if (tmp_plain_file) { (void) m_unlink (tmp_plain_file); }
1503 free (tmp_plain_file);
1504
1505 return NULL;
1506 }
1507
1508
1509 /*
1510 * Slip new text/plain part into a new multipart/alternative.
1511 */
1512 static int
1513 insert_into_new_mp_alt (CT *ct, int *message_mods) {
1514 CT tp_part = build_text_plain_part (*ct);
1515 int status = OK;
1516
1517 if (tp_part) {
1518 CT mp_alt = build_multipart_alt (*ct, tp_part, CT_MULTIPART,
1519 MULTI_ALTERNATE);
1520 if (mp_alt) {
1521 struct multipart *mp = (struct multipart *) mp_alt->c_ctparams;
1522
1523 if (mp && mp->mp_parts) {
1524 mp->mp_parts->mp_part = tp_part;
1525 /* Make the new multipart/alternative the parent. */
1526 *ct = mp_alt;
1527
1528 ++*message_mods;
1529 if (verbosw) {
1530 report (NULL, (*ct)->c_partno, (*ct)->c_file,
1531 "insert text/plain part");
1532 }
1533 } else {
1534 free_content (tp_part);
1535 free_content (mp_alt);
1536 status = NOTOK;
1537 }
1538 } else {
1539 status = NOTOK;
1540 }
1541 } else {
1542 /* Not an error if text/plain couldn't be built. */
1543 }
1544
1545 return status;
1546 }
1547
1548
1549 /*
1550 * Clone a MIME part.
1551 */
1552 static CT
1553 divide_part (CT ct) {
1554 CT new_part;
1555
1556 NEW0(new_part);
1557 /* Just copy over what is needed for decoding. c_vrsn and
1558 c_celine aren't necessary. */
1559 new_part->c_file = mh_xstrdup (ct->c_file);
1560 new_part->c_begin = ct->c_begin;
1561 new_part->c_end = ct->c_end;
1562 copy_ctinfo (&new_part->c_ctinfo, &ct->c_ctinfo);
1563 new_part->c_type = ct->c_type;
1564 new_part->c_cefile = ct->c_cefile;
1565 new_part->c_encoding = ct->c_encoding;
1566 new_part->c_ctinitfnx = ct->c_ctinitfnx;
1567 new_part->c_ceopenfnx = ct->c_ceopenfnx;
1568 new_part->c_ceclosefnx = ct->c_ceclosefnx;
1569 new_part->c_cesizefnx = ct->c_cesizefnx;
1570
1571 /* c_ctline is used by reformat__part(), so it can preserve
1572 anything after the type/subtype. */
1573 new_part->c_ctline = mh_xstrdup (ct->c_ctline);
1574
1575 return new_part;
1576 }
1577
1578
1579 /*
1580 * Copy the content info from one part to another.
1581 */
1582 static void
1583 copy_ctinfo (CI dest, CI src) {
1584 PM s_pm, d_pm;
1585
1586 dest->ci_type = src->ci_type ? mh_xstrdup (src->ci_type) : NULL;
1587 dest->ci_subtype = src->ci_subtype ? mh_xstrdup (src->ci_subtype) : NULL;
1588
1589 for (s_pm = src->ci_first_pm; s_pm; s_pm = s_pm->pm_next) {
1590 d_pm = add_param(&dest->ci_first_pm, &dest->ci_last_pm, s_pm->pm_name,
1591 s_pm->pm_value, 0);
1592 if (s_pm->pm_charset) {
1593 d_pm->pm_charset = mh_xstrdup(s_pm->pm_charset);
1594 }
1595 if (s_pm->pm_lang) {
1596 d_pm->pm_lang = mh_xstrdup(s_pm->pm_lang);
1597 }
1598 }
1599
1600 dest->ci_comment = src->ci_comment ? mh_xstrdup (src->ci_comment) : NULL;
1601 dest->ci_magic = src->ci_magic ? mh_xstrdup (src->ci_magic) : NULL;
1602 }
1603
1604
1605 /*
1606 * Decode content.
1607 */
1608 static int
1609 decode_part (CT ct) {
1610 char *tmp_decoded;
1611 int status;
1612 FILE *file;
1613 char *tempfile;
1614
1615 if ((tempfile = m_mktemp2 (NULL, invo_name, NULL, &file)) == NULL) {
1616 adios (NULL, "unable to create temporary file in %s", get_temp_dir());
1617 }
1618 tmp_decoded = mh_xstrdup (tempfile);
1619 /* The following call will load ct->c_cefile.ce_file with the tmp
1620 filename of the decoded content. tmp_decoded will contain the
1621 encoded output, get rid of that. */
1622 status = output_message_fp (ct, file, tmp_decoded);
1623 (void) m_unlink (tmp_decoded);
1624 free (tmp_decoded);
1625 if (fclose (file)) {
1626 inform("unable to close temporary file %s, continuing...", tempfile);
1627 }
1628
1629 return status;
1630 }
1631
1632
1633 /*
1634 * Reformat content as plain text.
1635 * Some of the arguments aren't really needed now, but maybe will
1636 * be in the future for other than text types.
1637 */
1638 static int
1639 reformat_part (CT ct, char *file, char *type, char *subtype, int c_type) {
1640 int output_subtype, output_encoding;
1641 const char *reason = NULL;
1642 char *cp, *cf;
1643 int status;
1644
1645 /* Hacky: this redirects the output from whatever command is used
1646 to show the part to a file. So, the user can't have any output
1647 redirection in that command.
1648 Could show_multi() in mhshowsbr.c avoid this? */
1649
1650 /* Check for invo_name-format-type/subtype. */
1651 if ((cf = context_find_by_type ("format", type, subtype)) == NULL) {
1652 if (verbosw) {
1653 inform("Don't know how to convert %s, there is no "
1654 "%s-format-%s/%s profile entry",
1655 ct->c_file, invo_name, type, subtype);
1656 }
1657 return NOTOK;
1658 }
1659 if (strchr (cf, '>')) {
1660 inform("'>' prohibited in \"%s\",\nplease fix your "
1661 "%s-format-%s/%s profile entry", cf, invo_name, type,
1662 FENDNULL(subtype));
1663
1664 return NOTOK;
1665 }
1666
1667 cp = concat (cf, " >", file, NULL);
1668 status = show_content_aux (ct, 0, cp, NULL, NULL);
1669 free (cp);
1670
1671 /* Unlink decoded content tmp file and free its filename to avoid
1672 leaks. The file stream should already have been closed. */
1673 if (ct->c_cefile.ce_unlink) {
1674 (void) m_unlink (ct->c_cefile.ce_file);
1675 free (ct->c_cefile.ce_file);
1676 ct->c_cefile.ce_file = NULL;
1677 ct->c_cefile.ce_unlink = 0;
1678 }
1679
1680 if (c_type == CT_TEXT) {
1681 output_subtype = TEXT_PLAIN;
1682 } else {
1683 /* Set subtype to 0, which is always an UNKNOWN subtype. */
1684 output_subtype = 0;
1685 }
1686
1687 output_encoding = content_encoding (ct, &reason);
1688 if (status == OK &&
1689 set_ct_type (ct, c_type, output_subtype, output_encoding) == OK) {
1690 ct->c_cefile.ce_file = file;
1691 ct->c_cefile.ce_unlink = 1;
1692 } else {
1693 ct->c_cefile.ce_unlink = 0;
1694 status = NOTOK;
1695 }
1696
1697 return status;
1698 }
1699
1700
1701 /*
1702 * Fill in a multipart/alternative part.
1703 */
1704 static CT
1705 build_multipart_alt (CT first_alt, CT new_part, int type, int subtype) {
1706 char *boundary_prefix = "----=_nmh-multipart";
1707 char *boundary = concat (boundary_prefix, first_alt->c_partno, NULL);
1708 char *boundary_indicator = "; boundary=";
1709 char *typename, *subtypename, *name;
1710 CT ct;
1711 struct part *p;
1712 struct multipart *m;
1713 const struct str2init *ctinit;
1714
1715 NEW0(ct);
1716
1717 /* Set up the multipart/alternative part. These fields of *ct were
1718 initialized to 0 by mh_xcalloc():
1719 c_fp, c_unlink, c_begin, c_end,
1720 c_vrsn, c_ctline, c_celine,
1721 c_id, c_descr, c_dispo, c_partno,
1722 c_ctinfo.ci_comment, c_ctinfo.ci_magic,
1723 c_cefile, c_encoding,
1724 c_digested, c_digest[16], c_ctexbody,
1725 c_ctinitfnx, c_ceopenfnx, c_ceclosefnx, c_cesizefnx,
1726 c_umask, c_rfc934,
1727 c_showproc, c_termproc, c_storeproc, c_storage, c_folder
1728 */
1729
1730 ct->c_file = mh_xstrdup (first_alt->c_file);
1731 ct->c_type = type;
1732 ct->c_subtype = subtype;
1733
1734 ctinit = get_ct_init (ct->c_type);
1735
1736 typename = ct_type_str (type);
1737 subtypename = ct_subtype_str (type, subtype);
1738
1739 {
1740 int serial = 0;
1741 int found_boundary = 1;
1742
1743 while (found_boundary && serial < 1000000) {
1744 found_boundary = 0;
1745
1746 /* Ensure that the boundary doesn't appear in the decoded
1747 content. */
1748 if (new_part->c_cefile.ce_file) {
1749 if ((found_boundary =
1750 boundary_in_content (&new_part->c_cefile.ce_fp,
1751 new_part->c_cefile.ce_file,
1752 boundary)) == NOTOK) {
1753 free_content (ct);
1754 return NULL;
1755 }
1756 }
1757
1758 /* Ensure that the boundary doesn't appear in the encoded
1759 content. */
1760 if (! found_boundary && new_part->c_file) {
1761 if ((found_boundary =
1762 boundary_in_content (&new_part->c_fp,
1763 new_part->c_file,
1764 boundary)) == NOTOK) {
1765 free_content (ct);
1766 return NULL;
1767 }
1768 }
1769
1770 if (found_boundary) {
1771 /* Try a slightly different boundary. */
1772 char buffer2[16];
1773
1774 free (boundary);
1775 ++serial;
1776 snprintf (buffer2, sizeof buffer2, "%d", serial);
1777 boundary =
1778 concat (boundary_prefix,
1779 FENDNULL(first_alt->c_partno),
1780 "-", buffer2, NULL);
1781 }
1782 }
1783
1784 if (found_boundary) {
1785 inform("giving up trying to find a unique boundary");
1786 free_content (ct);
1787 return NULL;
1788 }
1789 }
1790
1791 name = concat (" ", typename, "/", subtypename, boundary_indicator, "\"",
1792 boundary, "\"", NULL);
1793
1794 /* Load c_first_hf and c_last_hf. */
1795 transfer_noncontent_headers (first_alt, ct);
1796 add_header (ct, mh_xstrdup (TYPE_FIELD), concat (name, "\n", NULL));
1797 free (name);
1798
1799 /* Load c_partno. */
1800 if (first_alt->c_partno) {
1801 ct->c_partno = mh_xstrdup (first_alt->c_partno);
1802 free (first_alt->c_partno);
1803 first_alt->c_partno = concat (ct->c_partno, ".1", NULL);
1804 new_part->c_partno = concat (ct->c_partno, ".2", NULL);
1805 } else {
1806 first_alt->c_partno = mh_xstrdup ("1");
1807 new_part->c_partno = mh_xstrdup ("2");
1808 }
1809
1810 if (ctinit) {
1811 ct->c_ctinfo.ci_type = mh_xstrdup (typename);
1812 ct->c_ctinfo.ci_subtype = mh_xstrdup (subtypename);
1813 }
1814
1815 add_param(&ct->c_ctinfo.ci_first_pm, &ct->c_ctinfo.ci_last_pm,
1816 "boundary", boundary, 0);
1817
1818 NEW(p);
1819 NEW(p->mp_next);
1820 p->mp_next->mp_next = NULL;
1821 p->mp_next->mp_part = first_alt;
1822
1823 NEW0(m);
1824 m->mp_start = concat (boundary, "\n", NULL);
1825 m->mp_stop = concat (boundary, "--\n", NULL);
1826 m->mp_parts = p;
1827 ct->c_ctparams = m;
1828
1829 free (boundary);
1830
1831 return ct;
1832 }
1833
1834
1835 /*
1836 * Check that the boundary does not appear in the content.
1837 */
1838 static int
1839 boundary_in_content (FILE **fp, char *file, const char *boundary) {
1840 char buffer[NMH_BUFSIZ];
1841 size_t bytes_read;
1842 int found_boundary = 0;
1843
1844 /* free_content() will close *fp if we fopen it here. */
1845 if (! *fp && (*fp = fopen (file, "r")) == NULL) {
1846 advise (file, "unable to open %s for reading", file);
1847 return NOTOK;
1848 }
1849
1850 fseeko (*fp, 0L, SEEK_SET);
1851 while ((bytes_read = fread (buffer, 1, sizeof buffer, *fp)) > 0) {
1852 if (find_str (buffer, bytes_read, boundary)) {
1853 found_boundary = 1;
1854 break;
1855 }
1856 }
1857
1858 return found_boundary;
1859 }
1860
1861
1862 /*
1863 * Remove all non-Content headers.
1864 */
1865 static void
1866 transfer_noncontent_headers (CT old, CT new) {
1867 HF hp, hp_prev;
1868
1869 hp_prev = hp = old->c_first_hf;
1870 while (hp) {
1871 HF next = hp->next;
1872
1873 if (strncasecmp (XXX_FIELD_PRF, hp->name, strlen (XXX_FIELD_PRF))) {
1874 if (hp == old->c_last_hf) {
1875 if (hp == old->c_first_hf) {
1876 old->c_last_hf = old->c_first_hf = NULL;
1877 } else {
1878 hp_prev->next = NULL;
1879 old->c_last_hf = hp_prev;
1880 }
1881 } else {
1882 if (hp == old->c_first_hf) {
1883 old->c_first_hf = next;
1884 } else {
1885 hp_prev->next = next;
1886 }
1887 }
1888
1889 /* Put node hp in the new CT. */
1890 if (new->c_first_hf == NULL) {
1891 new->c_first_hf = hp;
1892 } else {
1893 new->c_last_hf->next = hp;
1894 }
1895 new->c_last_hf = hp;
1896 } else {
1897 /* A Content- header, leave in old. */
1898 hp_prev = hp;
1899 }
1900
1901 hp = next;
1902 }
1903 }
1904
1905
1906 /*
1907 * Set content type.
1908 */
1909 static int
1910 set_ct_type (CT ct, int type, int subtype, int encoding) {
1911 char *typename = ct_type_str (type);
1912 char *subtypename = ct_subtype_str (type, subtype);
1913 /* E.g, " text/plain" */
1914 char *type_subtypename = concat (" ", typename, "/", subtypename, NULL);
1915 /* E.g, " text/plain\n" */
1916 char *name_plus_nl = concat (type_subtypename, "\n", NULL);
1917 int found_content_type = 0;
1918 HF hf;
1919 const char *cp = NULL;
1920 char *ctline;
1921 int status;
1922
1923 /* Update/add Content-Type header field. */
1924 for (hf = ct->c_first_hf; hf; hf = hf->next) {
1925 if (! strcasecmp (TYPE_FIELD, hf->name)) {
1926 found_content_type = 1;
1927 free (hf->value);
1928 hf->value = (cp = strchr (ct->c_ctline, ';'))
1929 ? concat (type_subtypename, cp, "\n", NULL)
1930 : mh_xstrdup (name_plus_nl);
1931 }
1932 }
1933 if (! found_content_type) {
1934 add_header (ct, mh_xstrdup (TYPE_FIELD),
1935 (cp = strchr (ct->c_ctline, ';'))
1936 ? concat (type_subtypename, cp, "\n", NULL)
1937 : mh_xstrdup (name_plus_nl));
1938 }
1939
1940 /* Some of these might not be used, but set them anyway. */
1941 ctline = cp
1942 ? concat (type_subtypename, cp, NULL)
1943 : concat (type_subtypename, NULL);
1944 free (ct->c_ctline);
1945 ct->c_ctline = ctline;
1946 /* Leave other ctinfo members as they were. */
1947 free (ct->c_ctinfo.ci_type);
1948 ct->c_ctinfo.ci_type = mh_xstrdup (typename);
1949 free (ct->c_ctinfo.ci_subtype);
1950 ct->c_ctinfo.ci_subtype = mh_xstrdup (subtypename);
1951 ct->c_type = type;
1952 ct->c_subtype = subtype;
1953
1954 free (name_plus_nl);
1955 free (type_subtypename);
1956
1957 status = set_ce (ct, encoding);
1958
1959 return status;
1960 }
1961
1962
1963 /*
1964 * It's not necessary to update the charset parameter of a Content-Type
1965 * header for a text part. According to RFC 2045 Sec. 6.4, the body
1966 * (content) was originally in the specified charset, "and will be in
1967 * that character set again after decoding."
1968 */
1969 static int
1970 decode_text_parts (CT ct, int encoding, const char *decodetypes,
1971 int *message_mods) {
1972 int status = OK;
1973 int lf_line_endings = 0;
1974
1975 switch (ct->c_type) {
1976 case CT_MULTIPART: {
1977 struct multipart *m = (struct multipart *) ct->c_ctparams;
1978 struct part *part;
1979
1980 /* Should check to see if the body for this part is encoded?
1981 For now, it gets passed along as-is by InitMultiPart(). */
1982 for (part = m->mp_parts; status == OK && part; part = part->mp_next) {
1983 status = decode_text_parts (part->mp_part, encoding, decodetypes,
1984 message_mods);
1985 }
1986 break;
1987 }
1988
1989 case CT_MESSAGE:
1990 if (ct->c_subtype == MESSAGE_EXTERNAL) {
1991 struct exbody *e = (struct exbody *) ct->c_ctparams;
1992
1993 status = decode_text_parts (e->eb_content, encoding, decodetypes,
1994 message_mods);
1995 }
1996 break;
1997
1998 default:
1999 if (! should_decode(decodetypes, ct->c_ctinfo.ci_type, ct->c_ctinfo.ci_subtype)) {
2000 break;
2001 }
2002
2003 lf_line_endings =
2004 ct->c_ctparams && ((struct text *) ct->c_ctparams)->lf_line_endings;
2005
2006 switch (ct->c_encoding) {
2007 case CE_BASE64:
2008 case CE_QUOTED: {
2009 int ct_encoding;
2010
2011 if (decode_part (ct) == OK && ct->c_cefile.ce_file) {
2012 const char *reason = NULL;
2013
2014 if ((ct_encoding = content_encoding (ct, &reason)) == CE_BINARY
2015 && encoding != CE_BINARY) {
2016 /* The decoding isn't acceptable so discard it.
2017 Leave status as OK to allow other transformations. */
2018 if (verbosw) {
2019 report (NULL, ct->c_partno, ct->c_file,
2020 "will not decode%s because it is binary (%s)",
2021 ct->c_partno ? ""
2022 : (FENDNULL(ct->c_ctline)),
2023 reason);
2024 }
2025 (void) m_unlink (ct->c_cefile.ce_file);
2026 free (ct->c_cefile.ce_file);
2027 ct->c_cefile.ce_file = NULL;
2028 } else if (ct->c_encoding == CE_QUOTED &&
2029 ct_encoding == CE_8BIT && encoding == CE_7BIT) {
2030 /* The decoding isn't acceptable so discard it.
2031 Leave status as OK to allow other transformations. */
2032 if (verbosw) {
2033 report (NULL, ct->c_partno, ct->c_file,
2034 "will not decode%s because it is 8bit",
2035 ct->c_partno ? ""
2036 : (FENDNULL(ct->c_ctline)));
2037 }
2038 (void) m_unlink (ct->c_cefile.ce_file);
2039 free (ct->c_cefile.ce_file);
2040 ct->c_cefile.ce_file = NULL;
2041 } else {
2042 int enc;
2043
2044 if (ct_encoding == CE_BINARY) {
2045 enc = CE_BINARY;
2046 } else if (ct_encoding == CE_8BIT && encoding == CE_7BIT) {
2047 enc = CE_QUOTED;
2048 } else {
2049 enc = ct_encoding;
2050 }
2051 if (set_ce (ct, enc) == OK) {
2052 ++*message_mods;
2053 if (verbosw) {
2054 report (NULL, ct->c_partno, ct->c_file, "decode%s",
2055 FENDNULL(ct->c_ctline));
2056 }
2057 if (lf_line_endings) {
2058 strip_crs (ct, message_mods);
2059 }
2060 } else {
2061 status = NOTOK;
2062 }
2063 }
2064 } else {
2065 status = NOTOK;
2066 }
2067 break;
2068 }
2069 case CE_8BIT:
2070 case CE_7BIT:
2071 if (lf_line_endings) {
2072 strip_crs (ct, message_mods);
2073 }
2074 break;
2075 default:
2076 break;
2077 }
2078
2079 break;
2080 }
2081
2082 return status;
2083 }
2084
2085
2086 /*
2087 * Determine if the part with type[/subtype] should be decoded, according to
2088 * decodetypes (which came from the -decodetypes switch).
2089 */
2090 static int
2091 should_decode(const char *decodetypes, const char *type, const char *subtype) {
2092 /* Quick search for matching type[/subtype] in decodetypes: bracket
2093 decodetypes with commas, then search for ,type, and ,type/subtype, in
2094 it. */
2095
2096 int found_match = 0;
2097 char *delimited_decodetypes = concat(",", decodetypes, ",", NULL);
2098 char *delimited_type = concat(",", type, ",", NULL);
2099
2100 if (nmh_strcasestr(delimited_decodetypes, delimited_type)) {
2101 found_match = 1;
2102 } else if (subtype != NULL) {
2103 char *delimited_type_subtype =
2104 concat(",", type, "/", subtype, ",", NULL);
2105
2106 if (nmh_strcasestr(delimited_decodetypes, delimited_type_subtype)) {
2107 found_match = 1;
2108 }
2109 free(delimited_type_subtype);
2110 }
2111
2112 free(delimited_type);
2113 free(delimited_decodetypes);
2114
2115 return found_match;
2116 }
2117
2118
2119 /*
2120 * See if the decoded content is 7bit, 8bit, or binary. It's binary
2121 * if it has any NUL characters, a CR not followed by a LF, or lines
2122 * greater than 998 characters in length. If binary, reason is set
2123 * to a string explaining why.
2124 */
2125 static int
2126 content_encoding (CT ct, const char **reason) {
2127 CE ce = &ct->c_cefile;
2128 int encoding = CE_7BIT;
2129
2130 if (ce->ce_file) {
2131 size_t line_len = 0;
2132 char buffer[NMH_BUFSIZ];
2133 size_t inbytes;
2134
2135 if (! ce->ce_fp && (ce->ce_fp = fopen (ce->ce_file, "r")) == NULL) {
2136 advise (ce->ce_file, "unable to open for reading");
2137 return CE_UNKNOWN;
2138 }
2139
2140 fseeko (ce->ce_fp, 0L, SEEK_SET);
2141 while (encoding != CE_BINARY &&
2142 (inbytes = fread (buffer, 1, sizeof buffer, ce->ce_fp)) > 0) {
2143 char *cp;
2144 size_t i;
2145 int last_char_was_cr = 0;
2146
2147 for (i = 0, cp = buffer; i < inbytes; ++i, ++cp) {
2148 if (*cp == '\0' || ++line_len > 998 ||
2149 (*cp != '\n' && last_char_was_cr)) {
2150 encoding = CE_BINARY;
2151 if (*cp == '\0') {
2152 *reason = "null character";
2153 } else if (line_len > 998) {
2154 *reason = "line length > 998";
2155 } else if (*cp != '\n' && last_char_was_cr) {
2156 *reason = "CR not followed by LF";
2157 } else {
2158 /* Should not reach this. */
2159 *reason = "";
2160 }
2161 break;
2162 }
2163 if (*cp == '\n') {
2164 line_len = 0;
2165 } else if (! isascii ((unsigned char) *cp)) {
2166 encoding = CE_8BIT;
2167 }
2168
2169 last_char_was_cr = *cp == '\r';
2170 }
2171 }
2172
2173 fclose (ce->ce_fp);
2174 ce->ce_fp = NULL;
2175 } /* else should never happen */
2176
2177 return encoding;
2178 }
2179
2180
2181 /*
2182 * Strip carriage returns from content.
2183 */
2184 static int
2185 strip_crs (CT ct, int *message_mods) {
2186 char *charset = content_charset (ct);
2187 int status = OK;
2188
2189 /* Only strip carriage returns if content is ASCII or another
2190 charset that has the same readily recognizable CR followed by a
2191 LF. We can include UTF-8 here because if the high-order bit of
2192 a UTF-8 byte is 0, then it must be a single-byte ASCII
2193 character. */
2194 if (! strcasecmp (charset, "US-ASCII") ||
2195 ! strcasecmp (charset, "UTF-8") ||
2196 ! strncasecmp (charset, "ISO-8859-", 9) ||
2197 ! strncasecmp (charset, "WINDOWS-12", 10)) {
2198 char **file = NULL;
2199 FILE **fp = NULL;
2200 size_t begin;
2201 size_t end;
2202 int has_crs = 0;
2203 int opened_input_file = 0;
2204
2205 if (ct->c_cefile.ce_file) {
2206 file = &ct->c_cefile.ce_file;
2207 fp = &ct->c_cefile.ce_fp;
2208 begin = end = 0;
2209 } else if (ct->c_file) {
2210 file = &ct->c_file;
2211 fp = &ct->c_fp;
2212 begin = (size_t) ct->c_begin;
2213 end = (size_t) ct->c_end;
2214 } /* else don't know where the content is */
2215
2216 if (file && *file && fp) {
2217 if (! *fp) {
2218 if ((*fp = fopen (*file, "r")) == NULL) {
2219 advise (*file, "unable to open for reading");
2220 status = NOTOK;
2221 } else {
2222 opened_input_file = 1;
2223 }
2224 }
2225 }
2226
2227 if (fp && *fp) {
2228 char buffer[NMH_BUFSIZ];
2229 size_t bytes_read;
2230 size_t bytes_to_read =
2231 end > 0 && end > begin ? end - begin : sizeof buffer;
2232
2233 fseeko (*fp, begin, SEEK_SET);
2234 while ((bytes_read = fread (buffer, 1,
2235 min (bytes_to_read, sizeof buffer),
2236 *fp)) > 0) {
2237 /* Look for CR followed by a LF. This is supposed to
2238 be text so there should be LF's. If not, don't
2239 modify the content. */
2240 char *cp;
2241 size_t i;
2242 int last_char_was_cr = 0;
2243
2244 if (end > 0) { bytes_to_read -= bytes_read; }
2245
2246 for (i = 0, cp = buffer; i < bytes_read; ++i, ++cp) {
2247 if (*cp == '\n' && last_char_was_cr) {
2248 has_crs = 1;
2249 break;
2250 }
2251
2252 last_char_was_cr = *cp == '\r';
2253 }
2254 }
2255
2256 if (has_crs) {
2257 int fd;
2258 char *stripped_content_file;
2259 char *tempfile = m_mktemp2 (NULL, invo_name, &fd, NULL);
2260
2261 if (tempfile == NULL) {
2262 adios (NULL, "unable to create temporary file in %s",
2263 get_temp_dir());
2264 }
2265 stripped_content_file = mh_xstrdup (tempfile);
2266
2267 /* Strip each CR before a LF from the content. */
2268 fseeko (*fp, begin, SEEK_SET);
2269 while ((bytes_read = fread (buffer, 1, sizeof buffer, *fp)) >
2270 0) {
2271 char *cp;
2272 size_t i;
2273 int last_char_was_cr = 0;
2274
2275 for (i = 0, cp = buffer; i < bytes_read; ++i, ++cp) {
2276 if (*cp == '\r') {
2277 last_char_was_cr = 1;
2278 } else if (last_char_was_cr) {
2279 if (*cp != '\n') {
2280 if (write (fd, "\r", 1) < 0) {
2281 advise (tempfile, "CR write");
2282 }
2283 }
2284 if (write (fd, cp, 1) < 0) {
2285 advise (tempfile, "write");
2286 }
2287 last_char_was_cr = 0;
2288 } else {
2289 if (write (fd, cp, 1) < 0) {
2290 advise (tempfile, "write");
2291 }
2292 last_char_was_cr = 0;
2293 }
2294 }
2295 }
2296
2297 if (close (fd)) {
2298 inform("unable to write temporary file %s, continuing...",
2299 stripped_content_file);
2300 (void) m_unlink (stripped_content_file);
2301 status = NOTOK;
2302 } else {
2303 /* Replace the decoded file with the converted one. */
2304 if (ct->c_cefile.ce_file && ct->c_cefile.ce_unlink)
2305 (void) m_unlink (ct->c_cefile.ce_file);
2306
2307 mh_xfree(ct->c_cefile.ce_file);
2308 ct->c_cefile.ce_file = stripped_content_file;
2309 ct->c_cefile.ce_unlink = 1;
2310
2311 ++*message_mods;
2312 if (verbosw) {
2313 report (NULL, ct->c_partno,
2314 begin == 0 && end == 0 ? "" : *file,
2315 "stripped CRs");
2316 }
2317 }
2318 }
2319
2320 if (opened_input_file) {
2321 fclose (*fp);
2322 *fp = NULL;
2323 }
2324 }
2325 }
2326
2327 free (charset);
2328
2329 return status;
2330 }
2331
2332
2333 /*
2334 * Add/update, if necessary, the message C-T-E, based on the least restrictive
2335 * of the part C-T-E's.
2336 */
2337 static void
2338 update_cte (CT ct) {
2339 const int least_restrictive_enc = least_restrictive_encoding (ct);
2340
2341 if (least_restrictive_enc != CE_UNKNOWN &&
2342 least_restrictive_enc != CE_7BIT) {
2343 char *cte = concat (" ", ce_str (least_restrictive_enc), "\n", NULL);
2344 HF hf;
2345 int found_cte = 0;
2346
2347 /* Update/add Content-Transfer-Encoding header field. */
2348 for (hf = ct->c_first_hf; hf; hf = hf->next) {
2349 if (! strcasecmp (ENCODING_FIELD, hf->name)) {
2350 found_cte = 1;
2351 free (hf->value);
2352 hf->value = cte;
2353 }
2354 }
2355 if (! found_cte) {
2356 add_header (ct, mh_xstrdup (ENCODING_FIELD), cte);
2357 }
2358 }
2359 }
2360
2361
2362 /*
2363 * Find the least restrictive encoding (7bit, 8bit, binary) of the parts
2364 * within a message.
2365 */
2366 static int
2367 least_restrictive_encoding (CT ct) {
2368 int encoding = CE_UNKNOWN;
2369
2370 switch (ct->c_type) {
2371 case CT_MULTIPART: {
2372 struct multipart *m = (struct multipart *) ct->c_ctparams;
2373 struct part *part;
2374
2375 for (part = m->mp_parts; part; part = part->mp_next) {
2376 const int part_encoding =
2377 least_restrictive_encoding (part->mp_part);
2378
2379 if (less_restrictive (encoding, part_encoding)) {
2380 encoding = part_encoding;
2381 }
2382 }
2383 break;
2384 }
2385
2386 case CT_MESSAGE:
2387 if (ct->c_subtype == MESSAGE_EXTERNAL) {
2388 struct exbody *e = (struct exbody *) ct->c_ctparams;
2389 const int part_encoding =
2390 least_restrictive_encoding (e->eb_content);
2391
2392 if (less_restrictive (encoding, part_encoding)) {
2393 encoding = part_encoding;
2394 }
2395 }
2396 break;
2397
2398 default: {
2399 if (less_restrictive (encoding, ct->c_encoding)) {
2400 encoding = ct->c_encoding;
2401 }
2402 }}
2403
2404 return encoding;
2405 }
2406
2407
2408 /*
2409 * Return whether the second encoding is less restrictive than the first, where
2410 * "less restrictive" is in the sense used by RFC 2045 Secs. 6.1 and 6.4. So,
2411 * CE_BINARY is less restrictive than CE_8BIT and
2412 * CE_8BIT is less restrictive than CE_7BIT.
2413 */
2414 static int
2415 less_restrictive (int encoding, int second_encoding) {
2416 switch (second_encoding) {
2417 case CE_BINARY:
2418 return encoding != CE_BINARY;
2419 case CE_8BIT:
2420 return encoding != CE_BINARY && encoding != CE_8BIT;
2421 case CE_7BIT:
2422 return encoding != CE_BINARY && encoding != CE_8BIT &&
2423 encoding != CE_7BIT;
2424 default :
2425 return 0;
2426 }
2427 }
2428
2429
2430 /*
2431 * Convert character set of each part.
2432 */
2433 static int
2434 convert_charsets (CT ct, char *dest_charset, int *message_mods) {
2435 int status = OK;
2436
2437 switch (ct->c_type) {
2438 case CT_TEXT:
2439 if (ct->c_subtype == TEXT_PLAIN) {
2440 status = convert_charset (ct, dest_charset, message_mods);
2441 if (status == OK) {
2442 if (verbosw) {
2443 char *ct_charset = content_charset (ct);
2444
2445 report (NULL, ct->c_partno, ct->c_file,
2446 "convert %s to %s", ct_charset, dest_charset);
2447 free (ct_charset);
2448 }
2449 } else {
2450 char *ct_charset = content_charset (ct);
2451
2452 report ("iconv", ct->c_partno, ct->c_file,
2453 "failed to convert %s to %s", ct_charset, dest_charset);
2454 free (ct_charset);
2455 }
2456 }
2457 break;
2458
2459 case CT_MULTIPART: {
2460 struct multipart *m = (struct multipart *) ct->c_ctparams;
2461 struct part *part;
2462
2463 /* Should check to see if the body for this part is encoded?
2464 For now, it gets passed along as-is by InitMultiPart(). */
2465 for (part = m->mp_parts; status == OK && part; part = part->mp_next) {
2466 status =
2467 convert_charsets (part->mp_part, dest_charset, message_mods);
2468 }
2469 break;
2470 }
2471
2472 case CT_MESSAGE:
2473 if (ct->c_subtype == MESSAGE_EXTERNAL) {
2474 struct exbody *e = (struct exbody *) ct->c_ctparams;
2475
2476 status =
2477 convert_charsets (e->eb_content, dest_charset, message_mods);
2478 }
2479 break;
2480
2481 default:
2482 break;
2483 }
2484
2485 return status;
2486 }
2487
2488
2489 /*
2490 * Fix various problems that aren't handled elsewhere. These
2491 * are fixed unconditionally: there are no switches to disable
2492 * them. Currently, "problems" are these:
2493 * 1) remove extraneous semicolon at the end of a header parameter list
2494 * 2) replace RFC 2047 encoding with RFC 2231 encoding of name and
2495 * filename parameters in Content-Type and Content-Disposition
2496 * headers, respectively.
2497 */
2498 static int
2499 fix_always (CT ct, int *message_mods) {
2500 int status = OK;
2501
2502 switch (ct->c_type) {
2503 case CT_MULTIPART: {
2504 struct multipart *m = (struct multipart *) ct->c_ctparams;
2505 struct part *part;
2506
2507 for (part = m->mp_parts; status == OK && part; part = part->mp_next) {
2508 status = fix_always (part->mp_part, message_mods);
2509 }
2510 break;
2511 }
2512
2513 case CT_MESSAGE:
2514 if (ct->c_subtype == MESSAGE_EXTERNAL) {
2515 struct exbody *e = (struct exbody *) ct->c_ctparams;
2516
2517 status = fix_always (e->eb_content, message_mods);
2518 }
2519 break;
2520
2521 default: {
2522 HF hf;
2523
2524 if (ct->c_first_hf) {
2525 fix_filename_encoding (ct);
2526 }
2527
2528 for (hf = ct->c_first_hf; hf; hf = hf->next) {
2529 size_t len = strlen (hf->value);
2530
2531 if (strcasecmp (hf->name, TYPE_FIELD) != 0 &&
2532 strcasecmp (hf->name, DISPO_FIELD) != 0) {
2533 /* Only do this for Content-Type and
2534 Content-Disposition fields because those are the
2535 only headers that parse_mime() warns about. */
2536 continue;
2537 }
2538
2539 /* whitespace following a trailing ';' will be nuked as well */
2540 if (hf->value[len - 1] == '\n') {
2541 while (isspace((unsigned char)(hf->value[len - 2]))) {
2542 if (len-- == 0) { break; }
2543 }
2544 }
2545
2546 if (hf->value[len - 2] == ';') {
2547 /* Remove trailing ';' from parameter value. */
2548 hf->value[len - 2] = '\n';
2549 hf->value[len - 1] = '\0';
2550
2551 /* Also, if Content-Type parameter, remove trailing ';'
2552 from ct->c_ctline. This probably isn't necessary
2553 but can't hurt. */
2554 if (strcasecmp(hf->name, TYPE_FIELD) == 0 && ct->c_ctline) {
2555 size_t l = strlen(ct->c_ctline) - 1;
2556 while (isspace((unsigned char)(ct->c_ctline[l])) ||
2557 ct->c_ctline[l] == ';') {
2558 ct->c_ctline[l--] = '\0';
2559 if (l == 0) { break; }
2560 }
2561 }
2562
2563 ++*message_mods;
2564 if (verbosw) {
2565 report (NULL, ct->c_partno, ct->c_file,
2566 "remove trailing ; from %s parameter value",
2567 hf->name);
2568 }
2569 }
2570 }
2571 }}
2572
2573 return status;
2574 }
2575
2576
2577 /*
2578 * Factor out common code for loops in fix_filename_encoding().
2579 */
2580 static int
2581 fix_filename_param (char *name, char *value, PM *first_pm, PM *last_pm) {
2582 int fixed = 0;
2583
2584 if (has_prefix(value, "=?") && has_suffix(value, "?=")) {
2585 /* Looks like an RFC 2047 encoded parameter. */
2586 char decoded[PATH_MAX + 1];
2587
2588 if (decode_rfc2047 (value, decoded, sizeof decoded)) {
2589 /* Encode using RFC 2231. */
2590 replace_param (first_pm, last_pm, name, decoded, 0);
2591 fixed = 1;
2592 } else {
2593 inform("failed to decode %s parameter %s", name, value);
2594 }
2595 }
2596
2597 return fixed;
2598 }
2599
2600
2601 /*
2602 * Replace RFC 2047 encoding with RFC 2231 encoding of name and
2603 * filename parameters in Content-Type and Content-Disposition
2604 * headers, respectively.
2605 */
2606 static int
2607 fix_filename_encoding (CT ct) {
2608 PM pm;
2609 HF hf;
2610 int fixed = 0;
2611
2612 for (pm = ct->c_ctinfo.ci_first_pm; pm; pm = pm->pm_next) {
2613 if (pm->pm_name && pm->pm_value &&
2614 strcasecmp (pm->pm_name, "name") == 0) {
2615 fixed = fix_filename_param (pm->pm_name, pm->pm_value,
2616 &ct->c_ctinfo.ci_first_pm,
2617 &ct->c_ctinfo.ci_last_pm);
2618 }
2619 }
2620
2621 for (pm = ct->c_dispo_first; pm; pm = pm->pm_next) {
2622 if (pm->pm_name && pm->pm_value &&
2623 strcasecmp (pm->pm_name, "filename") == 0) {
2624 fixed = fix_filename_param (pm->pm_name, pm->pm_value,
2625 &ct->c_dispo_first,
2626 &ct->c_dispo_last);
2627 }
2628 }
2629
2630 /* Fix hf values to correspond. */
2631 for (hf = ct->c_first_hf; fixed && hf; hf = hf->next) {
2632 enum { OTHER, TYPE_HEADER, DISPO_HEADER } field = OTHER;
2633
2634 if (strcasecmp (hf->name, TYPE_FIELD) == 0) {
2635 field = TYPE_HEADER;
2636 } else if (strcasecmp (hf->name, DISPO_FIELD) == 0) {
2637 field = DISPO_HEADER;
2638 }
2639
2640 if (field != OTHER) {
2641 const char *const semicolon_loc = strchr (hf->value, ';');
2642
2643 if (semicolon_loc) {
2644 const size_t len =
2645 strlen (hf->name) + 1 + semicolon_loc - hf->value;
2646 const char *const params =
2647 output_params (len,
2648 field == TYPE_HEADER
2649 ? ct->c_ctinfo.ci_first_pm
2650 : ct->c_dispo_first,
2651 NULL, 0);
2652 const char *const new_params = concat (params, "\n", NULL);
2653
2654 replace_substring (&hf->value, semicolon_loc, new_params);
2655 free((void *)new_params); /* Cast away const. Sigh. */
2656 free((void *)params);
2657 } else {
2658 inform("did not find semicolon in %s:%s\n",
2659 hf->name, hf->value);
2660 }
2661 }
2662 }
2663
2664 return OK;
2665 }
2666
2667
2668 /*
2669 * Output content in input file to output file.
2670 */
2671 static int
2672 write_content (CT ct, const char *input_filename, char *outfile, FILE *outfp,
2673 int modify_inplace, int message_mods) {
2674 int status = OK;
2675
2676 if (modify_inplace) {
2677 if (message_mods > 0) {
2678 if ((status = output_message_fp (ct, outfp, outfile)) == OK) {
2679 char *infile = input_filename
2680 ? mh_xstrdup (input_filename)
2681 : mh_xstrdup (ct->c_file ? ct->c_file : "-");
2682
2683 if (remove_file (infile) == OK) {
2684 if (rename (outfile, infile)) {
2685 /* Rename didn't work, possibly because of an
2686 attempt to rename across filesystems. Try
2687 brute force copy. */
2688 int old = open (outfile, O_RDONLY);
2689 int new =
2690 open (infile, O_WRONLY | O_CREAT, m_gmprot ());
2691 int i = -1;
2692
2693 if (old != -1 && new != -1) {
2694 char buffer[NMH_BUFSIZ];
2695
2696 while ((i = read (old, buffer, sizeof buffer)) >
2697 0) {
2698 if (write (new, buffer, i) != i) {
2699 i = -1;
2700 break;
2701 }
2702 }
2703 }
2704 if (new != -1) { close (new); }
2705 if (old != -1) { close (old); }
2706 (void) m_unlink (outfile);
2707
2708 if (i < 0) {
2709 /* The -file argument processing used path() to
2710 expand filename to absolute path. */
2711 int file = ct->c_file && ct->c_file[0] == '/';
2712
2713 inform("unable to rename %s %s to %s, continuing...",
2714 file ? "file" : "message", outfile,
2715 infile);
2716 status = NOTOK;
2717 }
2718 }
2719 } else {
2720 inform("unable to remove input file %s, "
2721 "not modifying it, continuing...", infile);
2722 (void) m_unlink (outfile);
2723 status = NOTOK;
2724 }
2725
2726 free (infile);
2727 } else {
2728 status = NOTOK;
2729 }
2730 } else {
2731 /* No modifications and didn't need the tmp outfile. */
2732 (void) m_unlink (outfile);
2733 }
2734 } else {
2735 /* Output is going to some file. Produce it whether or not
2736 there were modifications. */
2737 status = output_message_fp (ct, outfp, outfile);
2738 }
2739
2740 flush_errors ();
2741 return status;
2742 }
2743
2744
2745 /*
2746 * parse_mime() does not set lf_line_endings in struct text, so use this
2747 * function to do it. It touches the parts the decodetypes identifies.
2748 */
2749 static void
2750 set_text_ctparams(CT ct, char *decodetypes, int lf_line_endings) {
2751 switch (ct->c_type) {
2752 case CT_MULTIPART: {
2753 struct multipart *m = (struct multipart *) ct->c_ctparams;
2754 struct part *part;
2755
2756 for (part = m->mp_parts; part; part = part->mp_next) {
2757 set_text_ctparams(part->mp_part, decodetypes, lf_line_endings);
2758 }
2759 break;
2760 }
2761
2762 case CT_MESSAGE:
2763 if (ct->c_subtype == MESSAGE_EXTERNAL) {
2764 struct exbody *e = (struct exbody *) ct->c_ctparams;
2765
2766 set_text_ctparams(e->eb_content, decodetypes, lf_line_endings);
2767 }
2768 break;
2769
2770 default:
2771 if (should_decode(decodetypes, ct->c_ctinfo.ci_type, ct->c_ctinfo.ci_subtype)) {
2772 if (ct->c_ctparams == NULL) {
2773 ct->c_ctparams = mh_xcalloc(1, sizeof (struct text));
2774 }
2775 ((struct text *) ct->c_ctparams)->lf_line_endings = lf_line_endings;
2776 }
2777 }
2778 }
2779
2780
2781 /*
2782 * If "rmmproc" is defined, call that to remove the file. Otherwise,
2783 * use the standard MH backup file.
2784 */
2785 static int
2786 remove_file (const char *file) {
2787 if (rmmproc) {
2788 char *rmm_command = concat (rmmproc, " ", file, NULL);
2789 int status = system (rmm_command);
2790
2791 free (rmm_command);
2792 return WIFEXITED (status) ? WEXITSTATUS (status) : NOTOK;
2793 }
2794 /* This is OK for a non-message file, it still uses the
2795 BACKUP_PREFIX form. The backup file will be in the same
2796 directory as file. */
2797 return rename (file, m_backup (file));
2798 }
2799
2800
2801 /*
2802 * Output formatted message to user.
2803 */
2804 static void
2805 report (char *what, char *partno, char *filename, char *message, ...) {
2806 va_list args;
2807 char *fmt;
2808
2809 if (verbosw) {
2810 va_start (args, message);
2811 fmt = concat (filename, partno ? " part " : ", ",
2812 FENDNULL(partno), partno ? ", " : "", message, NULL);
2813
2814 advertise (what, NULL, fmt, args);
2815
2816 free (fmt);
2817 va_end (args);
2818 }
2819 }
2820
2821
2822 static void
2823 pipeser (int i)
2824 {
2825 if (i == SIGQUIT) {
2826 fflush (stdout);
2827 fprintf (stderr, "\n");
2828 fflush (stderr);
2829 }
2830
2831 done (1);
2832 /* NOTREACHED */
2833 }
Fork of https://git.savannah.nongnu.org/cgit/nmh.git