packv4: nth_packed_object_sha1() packv4 support
[git/packv4.git] / builtin-mailinfo.c
blobb558754142b26399df78472b0bb124144dc194da
1 /*
2 * Another stupid program, this one parsing the headers of an
3 * email to figure out authorship and subject
4 */
5 #include "cache.h"
6 #include "builtin.h"
7 #include "utf8.h"
9 static FILE *cmitmsg, *patchfile, *fin, *fout;
11 static int keep_subject;
12 static const char *metainfo_charset;
13 static char line[1000];
14 static char name[1000];
15 static char email[1000];
17 static enum {
18 TE_DONTCARE, TE_QP, TE_BASE64,
19 } transfer_encoding;
20 static enum {
21 TYPE_TEXT, TYPE_OTHER,
22 } message_type;
24 static char charset[256];
25 static int patch_lines;
26 static char **p_hdr_data, **s_hdr_data;
28 #define MAX_HDR_PARSED 10
29 #define MAX_BOUNDARIES 5
31 static char *sanity_check(char *name, char *email)
33 int len = strlen(name);
34 if (len < 3 || len > 60)
35 return email;
36 if (strchr(name, '@') || strchr(name, '<') || strchr(name, '>'))
37 return email;
38 return name;
41 static int bogus_from(char *line)
43 /* John Doe <johndoe> */
44 char *bra, *ket, *dst, *cp;
46 /* This is fallback, so do not bother if we already have an
47 * e-mail address.
49 if (*email)
50 return 0;
52 bra = strchr(line, '<');
53 if (!bra)
54 return 0;
55 ket = strchr(bra, '>');
56 if (!ket)
57 return 0;
59 for (dst = email, cp = bra+1; cp < ket; )
60 *dst++ = *cp++;
61 *dst = 0;
62 for (cp = line; isspace(*cp); cp++)
64 for (bra--; isspace(*bra); bra--)
65 *bra = 0;
66 cp = sanity_check(cp, email);
67 strcpy(name, cp);
68 return 1;
71 static int handle_from(char *in_line)
73 char line[1000];
74 char *at;
75 char *dst;
77 strcpy(line, in_line);
78 at = strchr(line, '@');
79 if (!at)
80 return bogus_from(line);
83 * If we already have one email, don't take any confusing lines
85 if (*email && strchr(at+1, '@'))
86 return 0;
88 /* Pick up the string around '@', possibly delimited with <>
89 * pair; that is the email part. White them out while copying.
91 while (at > line) {
92 char c = at[-1];
93 if (isspace(c))
94 break;
95 if (c == '<') {
96 at[-1] = ' ';
97 break;
99 at--;
101 dst = email;
102 for (;;) {
103 unsigned char c = *at;
104 if (!c || c == '>' || isspace(c)) {
105 if (c == '>')
106 *at = ' ';
107 break;
109 *at++ = ' ';
110 *dst++ = c;
112 *dst++ = 0;
114 /* The remainder is name. It could be "John Doe <john.doe@xz>"
115 * or "john.doe@xz (John Doe)", but we have whited out the
116 * email part, so trim from both ends, possibly removing
117 * the () pair at the end.
119 at = line + strlen(line);
120 while (at > line) {
121 unsigned char c = *--at;
122 if (!isspace(c)) {
123 at[(c == ')') ? 0 : 1] = 0;
124 break;
128 at = line;
129 for (;;) {
130 unsigned char c = *at;
131 if (!c || !isspace(c)) {
132 if (c == '(')
133 at++;
134 break;
136 at++;
138 at = sanity_check(at, email);
139 strcpy(name, at);
140 return 1;
143 static int handle_header(char *line, char *data, int ofs)
145 if (!line || !data)
146 return 1;
148 strcpy(data, line+ofs);
150 return 0;
153 /* NOTE NOTE NOTE. We do not claim we do full MIME. We just attempt
154 * to have enough heuristics to grok MIME encoded patches often found
155 * on our mailing lists. For example, we do not even treat header lines
156 * case insensitively.
159 static int slurp_attr(const char *line, const char *name, char *attr)
161 const char *ends, *ap = strcasestr(line, name);
162 size_t sz;
164 if (!ap) {
165 *attr = 0;
166 return 0;
168 ap += strlen(name);
169 if (*ap == '"') {
170 ap++;
171 ends = "\"";
173 else
174 ends = "; \t";
175 sz = strcspn(ap, ends);
176 memcpy(attr, ap, sz);
177 attr[sz] = 0;
178 return 1;
181 struct content_type {
182 char *boundary;
183 int boundary_len;
186 static struct content_type content[MAX_BOUNDARIES];
188 static struct content_type *content_top = content;
190 static int handle_content_type(char *line)
192 char boundary[256];
194 if (strcasestr(line, "text/") == NULL)
195 message_type = TYPE_OTHER;
196 if (slurp_attr(line, "boundary=", boundary + 2)) {
197 memcpy(boundary, "--", 2);
198 if (content_top++ >= &content[MAX_BOUNDARIES]) {
199 fprintf(stderr, "Too many boundaries to handle\n");
200 exit(1);
202 content_top->boundary_len = strlen(boundary);
203 content_top->boundary = xmalloc(content_top->boundary_len+1);
204 strcpy(content_top->boundary, boundary);
206 if (slurp_attr(line, "charset=", charset)) {
207 int i, c;
208 for (i = 0; (c = charset[i]) != 0; i++)
209 charset[i] = tolower(c);
211 return 0;
214 static int handle_content_transfer_encoding(char *line)
216 if (strcasestr(line, "base64"))
217 transfer_encoding = TE_BASE64;
218 else if (strcasestr(line, "quoted-printable"))
219 transfer_encoding = TE_QP;
220 else
221 transfer_encoding = TE_DONTCARE;
222 return 0;
225 static int is_multipart_boundary(const char *line)
227 return (!memcmp(line, content_top->boundary, content_top->boundary_len));
230 static int eatspace(char *line)
232 int len = strlen(line);
233 while (len > 0 && isspace(line[len-1]))
234 line[--len] = 0;
235 return len;
238 static char *cleanup_subject(char *subject)
240 for (;;) {
241 char *p;
242 int len, remove;
243 switch (*subject) {
244 case 'r': case 'R':
245 if (!memcmp("e:", subject+1, 2)) {
246 subject += 3;
247 continue;
249 break;
250 case ' ': case '\t': case ':':
251 subject++;
252 continue;
254 case '[':
255 p = strchr(subject, ']');
256 if (!p) {
257 subject++;
258 continue;
260 len = strlen(p);
261 remove = p - subject;
262 if (remove <= len *2) {
263 subject = p+1;
264 continue;
266 break;
268 eatspace(subject);
269 return subject;
273 static void cleanup_space(char *buf)
275 unsigned char c;
276 while ((c = *buf) != 0) {
277 buf++;
278 if (isspace(c)) {
279 buf[-1] = ' ';
280 c = *buf;
281 while (isspace(c)) {
282 int len = strlen(buf);
283 memmove(buf, buf+1, len);
284 c = *buf;
290 static void decode_header(char *it);
291 static char *header[MAX_HDR_PARSED] = {
292 "From","Subject","Date",
295 static int check_header(char *line, char **hdr_data, int overwrite)
297 int i;
299 /* search for the interesting parts */
300 for (i = 0; header[i]; i++) {
301 int len = strlen(header[i]);
302 if ((!hdr_data[i] || overwrite) &&
303 !strncasecmp(line, header[i], len) &&
304 line[len] == ':' && isspace(line[len + 1])) {
305 /* Unwrap inline B and Q encoding, and optionally
306 * normalize the meta information to utf8.
308 decode_header(line + len + 2);
309 hdr_data[i] = xmalloc(1000 * sizeof(char));
310 if (! handle_header(line, hdr_data[i], len + 2)) {
311 return 1;
316 /* Content stuff */
317 if (!strncasecmp(line, "Content-Type", 12) &&
318 line[12] == ':' && isspace(line[12 + 1])) {
319 decode_header(line + 12 + 2);
320 if (! handle_content_type(line)) {
321 return 1;
324 if (!strncasecmp(line, "Content-Transfer-Encoding", 25) &&
325 line[25] == ':' && isspace(line[25 + 1])) {
326 decode_header(line + 25 + 2);
327 if (! handle_content_transfer_encoding(line)) {
328 return 1;
332 /* for inbody stuff */
333 if (!memcmp(">From", line, 5) && isspace(line[5]))
334 return 1;
335 if (!memcmp("[PATCH]", line, 7) && isspace(line[7])) {
336 for (i = 0; header[i]; i++) {
337 if (!memcmp("Subject: ", header[i], 9)) {
338 if (! handle_header(line, hdr_data[i], 0)) {
339 return 1;
345 /* no match */
346 return 0;
349 static int is_rfc2822_header(char *line)
352 * The section that defines the loosest possible
353 * field name is "3.6.8 Optional fields".
355 * optional-field = field-name ":" unstructured CRLF
356 * field-name = 1*ftext
357 * ftext = %d33-57 / %59-126
359 int ch;
360 char *cp = line;
362 /* Count mbox From headers as headers */
363 if (!memcmp(line, "From ", 5) || !memcmp(line, ">From ", 6))
364 return 1;
366 while ((ch = *cp++)) {
367 if (ch == ':')
368 return cp != line;
369 if ((33 <= ch && ch <= 57) ||
370 (59 <= ch && ch <= 126))
371 continue;
372 break;
374 return 0;
378 * sz is size of 'line' buffer in bytes. Must be reasonably
379 * long enough to hold one physical real-world e-mail line.
381 static int read_one_header_line(char *line, int sz, FILE *in)
383 int len;
386 * We will read at most (sz-1) bytes and then potentially
387 * re-add NUL after it. Accessing line[sz] after this is safe
388 * and we can allow len to grow up to and including sz.
390 sz--;
392 /* Get the first part of the line. */
393 if (!fgets(line, sz, in))
394 return 0;
397 * Is it an empty line or not a valid rfc2822 header?
398 * If so, stop here, and return false ("not a header")
400 len = eatspace(line);
401 if (!len || !is_rfc2822_header(line)) {
402 /* Re-add the newline */
403 line[len] = '\n';
404 line[len + 1] = '\0';
405 return 0;
409 * Now we need to eat all the continuation lines..
410 * Yuck, 2822 header "folding"
412 for (;;) {
413 int peek, addlen;
414 static char continuation[1000];
416 peek = fgetc(in); ungetc(peek, in);
417 if (peek != ' ' && peek != '\t')
418 break;
419 if (!fgets(continuation, sizeof(continuation), in))
420 break;
421 addlen = eatspace(continuation);
422 if (len < sz - 1) {
423 if (addlen >= sz - len)
424 addlen = sz - len - 1;
425 memcpy(line + len, continuation, addlen);
426 line[len] = '\n';
427 len += addlen;
430 line[len] = 0;
432 return 1;
435 static int decode_q_segment(char *in, char *ot, char *ep, int rfc2047)
437 int c;
438 while ((c = *in++) != 0 && (in <= ep)) {
439 if (c == '=') {
440 int d = *in++;
441 if (d == '\n' || !d)
442 break; /* drop trailing newline */
443 *ot++ = ((hexval(d) << 4) | hexval(*in++));
444 continue;
446 if (rfc2047 && c == '_') /* rfc2047 4.2 (2) */
447 c = 0x20;
448 *ot++ = c;
450 *ot = 0;
451 return 0;
454 static int decode_b_segment(char *in, char *ot, char *ep)
456 /* Decode in..ep, possibly in-place to ot */
457 int c, pos = 0, acc = 0;
459 while ((c = *in++) != 0 && (in <= ep)) {
460 if (c == '+')
461 c = 62;
462 else if (c == '/')
463 c = 63;
464 else if ('A' <= c && c <= 'Z')
465 c -= 'A';
466 else if ('a' <= c && c <= 'z')
467 c -= 'a' - 26;
468 else if ('0' <= c && c <= '9')
469 c -= '0' - 52;
470 else if (c == '=') {
471 /* padding is almost like (c == 0), except we do
472 * not output NUL resulting only from it;
473 * for now we just trust the data.
475 c = 0;
477 else
478 continue; /* garbage */
479 switch (pos++) {
480 case 0:
481 acc = (c << 2);
482 break;
483 case 1:
484 *ot++ = (acc | (c >> 4));
485 acc = (c & 15) << 4;
486 break;
487 case 2:
488 *ot++ = (acc | (c >> 2));
489 acc = (c & 3) << 6;
490 break;
491 case 3:
492 *ot++ = (acc | c);
493 acc = pos = 0;
494 break;
497 *ot = 0;
498 return 0;
502 * When there is no known charset, guess.
504 * Right now we assume that if the target is UTF-8 (the default),
505 * and it already looks like UTF-8 (which includes US-ASCII as its
506 * subset, of course) then that is what it is and there is nothing
507 * to do.
509 * Otherwise, we default to assuming it is Latin1 for historical
510 * reasons.
512 static const char *guess_charset(const char *line, const char *target_charset)
514 if (is_encoding_utf8(target_charset)) {
515 if (is_utf8(line))
516 return NULL;
518 return "latin1";
521 static void convert_to_utf8(char *line, const char *charset)
523 char *out;
525 if (!charset || !*charset) {
526 charset = guess_charset(line, metainfo_charset);
527 if (!charset)
528 return;
531 if (!strcmp(metainfo_charset, charset))
532 return;
533 out = reencode_string(line, metainfo_charset, charset);
534 if (!out)
535 die("cannot convert from %s to %s\n",
536 charset, metainfo_charset);
537 strcpy(line, out);
538 free(out);
541 static int decode_header_bq(char *it)
543 char *in, *out, *ep, *cp, *sp;
544 char outbuf[1000];
545 int rfc2047 = 0;
547 in = it;
548 out = outbuf;
549 while ((ep = strstr(in, "=?")) != NULL) {
550 int sz, encoding;
551 char charset_q[256], piecebuf[256];
552 rfc2047 = 1;
554 if (in != ep) {
555 sz = ep - in;
556 memcpy(out, in, sz);
557 out += sz;
558 in += sz;
560 /* E.g.
561 * ep : "=?iso-2022-jp?B?GyR...?= foo"
562 * ep : "=?ISO-8859-1?Q?Foo=FCbar?= baz"
564 ep += 2;
565 cp = strchr(ep, '?');
566 if (!cp)
567 return rfc2047; /* no munging */
568 for (sp = ep; sp < cp; sp++)
569 charset_q[sp - ep] = tolower(*sp);
570 charset_q[cp - ep] = 0;
571 encoding = cp[1];
572 if (!encoding || cp[2] != '?')
573 return rfc2047; /* no munging */
574 ep = strstr(cp + 3, "?=");
575 if (!ep)
576 return rfc2047; /* no munging */
577 switch (tolower(encoding)) {
578 default:
579 return rfc2047; /* no munging */
580 case 'b':
581 sz = decode_b_segment(cp + 3, piecebuf, ep);
582 break;
583 case 'q':
584 sz = decode_q_segment(cp + 3, piecebuf, ep, 1);
585 break;
587 if (sz < 0)
588 return rfc2047;
589 if (metainfo_charset)
590 convert_to_utf8(piecebuf, charset_q);
591 strcpy(out, piecebuf);
592 out += strlen(out);
593 in = ep + 2;
595 strcpy(out, in);
596 strcpy(it, outbuf);
597 return rfc2047;
600 static void decode_header(char *it)
603 if (decode_header_bq(it))
604 return;
605 /* otherwise "it" is a straight copy of the input.
606 * This can be binary guck but there is no charset specified.
608 if (metainfo_charset)
609 convert_to_utf8(it, "");
612 static void decode_transfer_encoding(char *line)
614 char *ep;
616 switch (transfer_encoding) {
617 case TE_QP:
618 ep = line + strlen(line);
619 decode_q_segment(line, line, ep, 0);
620 break;
621 case TE_BASE64:
622 ep = line + strlen(line);
623 decode_b_segment(line, line, ep);
624 break;
625 case TE_DONTCARE:
626 break;
630 static int handle_filter(char *line);
632 static int find_boundary(void)
634 while(fgets(line, sizeof(line), fin) != NULL) {
635 if (is_multipart_boundary(line))
636 return 1;
638 return 0;
641 static int handle_boundary(void)
643 char newline[]="\n";
644 again:
645 if (!memcmp(line+content_top->boundary_len, "--", 2)) {
646 /* we hit an end boundary */
647 /* pop the current boundary off the stack */
648 free(content_top->boundary);
650 /* technically won't happen as is_multipart_boundary()
651 will fail first. But just in case..
653 if (content_top-- < content) {
654 fprintf(stderr, "Detected mismatched boundaries, "
655 "can't recover\n");
656 exit(1);
658 handle_filter(newline);
660 /* skip to the next boundary */
661 if (!find_boundary())
662 return 0;
663 goto again;
666 /* set some defaults */
667 transfer_encoding = TE_DONTCARE;
668 charset[0] = 0;
669 message_type = TYPE_TEXT;
671 /* slurp in this section's info */
672 while (read_one_header_line(line, sizeof(line), fin))
673 check_header(line, p_hdr_data, 0);
675 /* eat the blank line after section info */
676 return (fgets(line, sizeof(line), fin) != NULL);
679 static inline int patchbreak(const char *line)
681 /* Beginning of a "diff -" header? */
682 if (!memcmp("diff -", line, 6))
683 return 1;
685 /* CVS "Index: " line? */
686 if (!memcmp("Index: ", line, 7))
687 return 1;
690 * "--- <filename>" starts patches without headers
691 * "---<sp>*" is a manual separator
693 if (!memcmp("---", line, 3)) {
694 line += 3;
695 /* space followed by a filename? */
696 if (line[0] == ' ' && !isspace(line[1]))
697 return 1;
698 /* Just whitespace? */
699 for (;;) {
700 unsigned char c = *line++;
701 if (c == '\n')
702 return 1;
703 if (!isspace(c))
704 break;
706 return 0;
708 return 0;
712 static int handle_commit_msg(char *line)
714 static int still_looking = 1;
716 if (!cmitmsg)
717 return 0;
719 if (still_looking) {
720 char *cp = line;
721 if (isspace(*line)) {
722 for (cp = line + 1; *cp; cp++) {
723 if (!isspace(*cp))
724 break;
726 if (!*cp)
727 return 0;
729 if ((still_looking = check_header(cp, s_hdr_data, 0)) != 0)
730 return 0;
733 /* normalize the log message to UTF-8. */
734 if (metainfo_charset)
735 convert_to_utf8(line, charset);
737 if (patchbreak(line)) {
738 fclose(cmitmsg);
739 cmitmsg = NULL;
740 return 1;
743 fputs(line, cmitmsg);
744 return 0;
747 static int handle_patch(char *line)
749 fputs(line, patchfile);
750 patch_lines++;
751 return 0;
754 static int handle_filter(char *line)
756 static int filter = 0;
758 /* filter tells us which part we left off on
759 * a non-zero return indicates we hit a filter point
761 switch (filter) {
762 case 0:
763 if (!handle_commit_msg(line))
764 break;
765 filter++;
766 case 1:
767 if (!handle_patch(line))
768 break;
769 filter++;
770 default:
771 return 1;
774 return 0;
777 static void handle_body(void)
779 int rc = 0;
780 static char newline[2000];
781 static char *np = newline;
783 /* Skip up to the first boundary */
784 if (content_top->boundary) {
785 if (!find_boundary())
786 return;
789 do {
790 /* process any boundary lines */
791 if (content_top->boundary && is_multipart_boundary(line)) {
792 /* flush any leftover */
793 if ((transfer_encoding == TE_BASE64) &&
794 (np != newline)) {
795 handle_filter(newline);
797 if (!handle_boundary())
798 return;
801 /* Unwrap transfer encoding */
802 decode_transfer_encoding(line);
804 switch (transfer_encoding) {
805 case TE_BASE64:
807 char *op = line;
809 /* binary data most likely doesn't have newlines */
810 if (message_type != TYPE_TEXT) {
811 rc = handle_filter(line);
812 break;
815 /* this is a decoded line that may contain
816 * multiple new lines. Pass only one chunk
817 * at a time to handle_filter()
820 do {
821 while (*op != '\n' && *op != 0)
822 *np++ = *op++;
823 *np = *op;
824 if (*np != 0) {
825 /* should be sitting on a new line */
826 *(++np) = 0;
827 op++;
828 rc = handle_filter(newline);
829 np = newline;
831 } while (*op != 0);
832 /* the partial chunk is saved in newline and
833 * will be appended by the next iteration of fgets
835 break;
837 default:
838 rc = handle_filter(line);
840 if (rc)
841 /* nothing left to filter */
842 break;
843 } while (fgets(line, sizeof(line), fin));
845 return;
848 static void output_header_lines(FILE *fout, const char *hdr, char *data)
850 while (1) {
851 char *ep = strchr(data, '\n');
852 int len;
853 if (!ep)
854 len = strlen(data);
855 else
856 len = ep - data;
857 fprintf(fout, "%s: %.*s\n", hdr, len, data);
858 if (!ep)
859 break;
860 data = ep + 1;
864 static void handle_info(void)
866 char *sub;
867 char *hdr;
868 int i;
870 for (i = 0; header[i]; i++) {
872 /* only print inbody headers if we output a patch file */
873 if (patch_lines && s_hdr_data[i])
874 hdr = s_hdr_data[i];
875 else if (p_hdr_data[i])
876 hdr = p_hdr_data[i];
877 else
878 continue;
880 if (!memcmp(header[i], "Subject", 7)) {
881 if (keep_subject)
882 sub = hdr;
883 else {
884 sub = cleanup_subject(hdr);
885 cleanup_space(sub);
887 output_header_lines(fout, "Subject", sub);
888 } else if (!memcmp(header[i], "From", 4)) {
889 handle_from(hdr);
890 fprintf(fout, "Author: %s\n", name);
891 fprintf(fout, "Email: %s\n", email);
892 } else {
893 cleanup_space(hdr);
894 fprintf(fout, "%s: %s\n", header[i], hdr);
897 fprintf(fout, "\n");
900 static int mailinfo(FILE *in, FILE *out, int ks, const char *encoding,
901 const char *msg, const char *patch)
903 keep_subject = ks;
904 metainfo_charset = encoding;
905 fin = in;
906 fout = out;
908 cmitmsg = fopen(msg, "w");
909 if (!cmitmsg) {
910 perror(msg);
911 return -1;
913 patchfile = fopen(patch, "w");
914 if (!patchfile) {
915 perror(patch);
916 fclose(cmitmsg);
917 return -1;
920 p_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(char *));
921 s_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(char *));
923 /* process the email header */
924 while (read_one_header_line(line, sizeof(line), fin))
925 check_header(line, p_hdr_data, 1);
927 handle_body();
928 handle_info();
930 return 0;
933 static const char mailinfo_usage[] =
934 "git-mailinfo [-k] [-u | --encoding=<encoding>] msg patch <mail >info";
936 int cmd_mailinfo(int argc, const char **argv, const char *prefix)
938 const char *def_charset;
940 /* NEEDSWORK: might want to do the optional .git/ directory
941 * discovery
943 git_config(git_default_config);
945 def_charset = (git_commit_encoding ? git_commit_encoding : "utf-8");
946 metainfo_charset = def_charset;
948 while (1 < argc && argv[1][0] == '-') {
949 if (!strcmp(argv[1], "-k"))
950 keep_subject = 1;
951 else if (!strcmp(argv[1], "-u"))
952 metainfo_charset = def_charset;
953 else if (!strcmp(argv[1], "-n"))
954 metainfo_charset = NULL;
955 else if (!prefixcmp(argv[1], "--encoding="))
956 metainfo_charset = argv[1] + 11;
957 else
958 usage(mailinfo_usage);
959 argc--; argv++;
962 if (argc != 3)
963 usage(mailinfo_usage);
965 return !!mailinfo(stdin, stdout, keep_subject, metainfo_charset, argv[1], argv[2]);