]> git.armaanb.net Git - stagit.git/blob - stagit.c
escape HTML in paths
[stagit.git] / stagit.c
1 #include <sys/stat.h>
2 #include <sys/types.h>
3
4 #include <err.h>
5 #include <errno.h>
6 #include <inttypes.h>
7 #include <libgen.h>
8 #include <limits.h>
9 #include <stdio.h>
10 #include <stdlib.h>
11 #include <string.h>
12 #include <unistd.h>
13
14 #include <git2.h>
15
16 #include "compat.h"
17
18 struct deltainfo {
19         git_patch *patch;
20
21         size_t addcount;
22         size_t delcount;
23 };
24
25 struct commitinfo {
26         const git_oid *id;
27
28         char oid[GIT_OID_HEXSZ + 1];
29         char parentoid[GIT_OID_HEXSZ + 1];
30
31         const git_signature *author;
32         const git_signature *committer;
33         const char          *summary;
34         const char          *msg;
35
36         git_diff   *diff;
37         git_commit *commit;
38         git_commit *parent;
39         git_tree   *commit_tree;
40         git_tree   *parent_tree;
41
42         size_t addcount;
43         size_t delcount;
44         size_t filecount;
45
46         struct deltainfo **deltas;
47         size_t ndeltas;
48 };
49
50 static git_repository *repo;
51
52 static const char *relpath = "";
53 static const char *repodir;
54
55 static char *name = "";
56 static char *strippedname = "";
57 static char description[255];
58 static char cloneurl[1024];
59 static char *submodules;
60 static char *licensefiles[] = { "HEAD:LICENSE", "HEAD:LICENSE.md", "HEAD:COPYING" };
61 static char *license;
62 static char *readmefiles[] = { "HEAD:README", "HEAD:README.md" };
63 static char *readme;
64 static long long nlogcommits = -1; /* < 0 indicates not used */
65
66 /* cache */
67 static git_oid lastoid;
68 static char lastoidstr[GIT_OID_HEXSZ + 2]; /* id + newline + NUL byte */
69 static FILE *rcachefp, *wcachefp;
70 static const char *cachefile;
71
72 void
73 joinpath(char *buf, size_t bufsiz, const char *path, const char *path2)
74 {
75         int r;
76
77         r = snprintf(buf, bufsiz, "%s%s%s",
78                 path, path[0] && path[strlen(path) - 1] != '/' ? "/" : "", path2);
79         if (r < 0 || (size_t)r >= bufsiz)
80                 errx(1, "path truncated: '%s%s%s'",
81                         path, path[0] && path[strlen(path) - 1] != '/' ? "/" : "", path2);
82 }
83
84 void
85 deltainfo_free(struct deltainfo *di)
86 {
87         if (!di)
88                 return;
89         git_patch_free(di->patch);
90         memset(di, 0, sizeof(*di));
91         free(di);
92 }
93
94 int
95 commitinfo_getstats(struct commitinfo *ci)
96 {
97         struct deltainfo *di;
98         git_diff_options opts;
99         git_diff_find_options fopts;
100         const git_diff_delta *delta;
101         const git_diff_hunk *hunk;
102         const git_diff_line *line;
103         git_patch *patch = NULL;
104         size_t ndeltas, nhunks, nhunklines;
105         size_t i, j, k;
106
107         if (git_tree_lookup(&(ci->commit_tree), repo, git_commit_tree_id(ci->commit)))
108                 goto err;
109         if (!git_commit_parent(&(ci->parent), ci->commit, 0)) {
110                 if (git_tree_lookup(&(ci->parent_tree), repo, git_commit_tree_id(ci->parent))) {
111                         ci->parent = NULL;
112                         ci->parent_tree = NULL;
113                 }
114         }
115
116         git_diff_init_options(&opts, GIT_DIFF_OPTIONS_VERSION);
117         opts.flags |= GIT_DIFF_DISABLE_PATHSPEC_MATCH |
118                       GIT_DIFF_IGNORE_SUBMODULES |
119                       GIT_DIFF_INCLUDE_TYPECHANGE;
120         if (git_diff_tree_to_tree(&(ci->diff), repo, ci->parent_tree, ci->commit_tree, &opts))
121                 goto err;
122
123         if (git_diff_find_init_options(&fopts, GIT_DIFF_FIND_OPTIONS_VERSION))
124                 goto err;
125         /* find renames and copies, exact matches (no heuristic) for renames. */
126         fopts.flags |= GIT_DIFF_FIND_RENAMES | GIT_DIFF_FIND_COPIES |
127                        GIT_DIFF_FIND_EXACT_MATCH_ONLY;
128         if (git_diff_find_similar(ci->diff, &fopts))
129                 goto err;
130
131         ndeltas = git_diff_num_deltas(ci->diff);
132         if (ndeltas && !(ci->deltas = calloc(ndeltas, sizeof(struct deltainfo *))))
133                 err(1, "calloc");
134
135         for (i = 0; i < ndeltas; i++) {
136                 if (git_patch_from_diff(&patch, ci->diff, i))
137                         goto err;
138
139                 if (!(di = calloc(1, sizeof(struct deltainfo))))
140                         err(1, "calloc");
141                 di->patch = patch;
142                 ci->deltas[i] = di;
143
144                 delta = git_patch_get_delta(patch);
145
146                 /* skip stats for binary data */
147                 if (delta->flags & GIT_DIFF_FLAG_BINARY)
148                         continue;
149
150                 nhunks = git_patch_num_hunks(patch);
151                 for (j = 0; j < nhunks; j++) {
152                         if (git_patch_get_hunk(&hunk, &nhunklines, patch, j))
153                                 break;
154                         for (k = 0; ; k++) {
155                                 if (git_patch_get_line_in_hunk(&line, patch, j, k))
156                                         break;
157                                 if (line->old_lineno == -1) {
158                                         di->addcount++;
159                                         ci->addcount++;
160                                 } else if (line->new_lineno == -1) {
161                                         di->delcount++;
162                                         ci->delcount++;
163                                 }
164                         }
165                 }
166         }
167         ci->ndeltas = i;
168         ci->filecount = i;
169
170         return 0;
171
172 err:
173         git_diff_free(ci->diff);
174         ci->diff = NULL;
175         git_tree_free(ci->commit_tree);
176         ci->commit_tree = NULL;
177         git_tree_free(ci->parent_tree);
178         ci->parent_tree = NULL;
179         git_commit_free(ci->parent);
180         ci->parent = NULL;
181
182         if (ci->deltas)
183                 for (i = 0; i < ci->ndeltas; i++)
184                         deltainfo_free(ci->deltas[i]);
185         free(ci->deltas);
186         ci->deltas = NULL;
187         ci->ndeltas = 0;
188         ci->addcount = 0;
189         ci->delcount = 0;
190         ci->filecount = 0;
191
192         return -1;
193 }
194
195 void
196 commitinfo_free(struct commitinfo *ci)
197 {
198         size_t i;
199
200         if (!ci)
201                 return;
202         if (ci->deltas)
203                 for (i = 0; i < ci->ndeltas; i++)
204                         deltainfo_free(ci->deltas[i]);
205
206         free(ci->deltas);
207         git_diff_free(ci->diff);
208         git_tree_free(ci->commit_tree);
209         git_tree_free(ci->parent_tree);
210         git_commit_free(ci->commit);
211         git_commit_free(ci->parent);
212         memset(ci, 0, sizeof(*ci));
213         free(ci);
214 }
215
216 struct commitinfo *
217 commitinfo_getbyoid(const git_oid *id)
218 {
219         struct commitinfo *ci;
220
221         if (!(ci = calloc(1, sizeof(struct commitinfo))))
222                 err(1, "calloc");
223
224         if (git_commit_lookup(&(ci->commit), repo, id))
225                 goto err;
226         ci->id = id;
227
228         git_oid_tostr(ci->oid, sizeof(ci->oid), git_commit_id(ci->commit));
229         git_oid_tostr(ci->parentoid, sizeof(ci->parentoid), git_commit_parent_id(ci->commit, 0));
230
231         ci->author = git_commit_author(ci->commit);
232         ci->committer = git_commit_committer(ci->commit);
233         ci->summary = git_commit_summary(ci->commit);
234         ci->msg = git_commit_message(ci->commit);
235
236         return ci;
237
238 err:
239         commitinfo_free(ci);
240
241         return NULL;
242 }
243
244 FILE *
245 efopen(const char *name, const char *flags)
246 {
247         FILE *fp;
248
249         if (!(fp = fopen(name, flags)))
250                 err(1, "fopen: '%s'", name);
251
252         return fp;
253 }
254
255 /* Escape characters below as HTML 2.0 / XML 1.0. */
256 void
257 xmlencode(FILE *fp, const char *s, size_t len)
258 {
259         size_t i;
260
261         for (i = 0; *s && i < len; s++, i++) {
262                 switch(*s) {
263                 case '<':  fputs("&lt;",   fp); break;
264                 case '>':  fputs("&gt;",   fp); break;
265                 case '\'': fputs("&#39;",  fp); break;
266                 case '&':  fputs("&amp;",  fp); break;
267                 case '"':  fputs("&quot;", fp); break;
268                 default:   fputc(*s, fp);
269                 }
270         }
271 }
272
273 int
274 mkdirp(const char *path)
275 {
276         char tmp[PATH_MAX], *p;
277
278         if (strlcpy(tmp, path, sizeof(tmp)) >= sizeof(tmp))
279                 errx(1, "path truncated: '%s'", path);
280         for (p = tmp + (tmp[0] == '/'); *p; p++) {
281                 if (*p != '/')
282                         continue;
283                 *p = '\0';
284                 if (mkdir(tmp, S_IRWXU | S_IRWXG | S_IRWXO) < 0 && errno != EEXIST)
285                         return -1;
286                 *p = '/';
287         }
288         if (mkdir(tmp, S_IRWXU | S_IRWXG | S_IRWXO) < 0 && errno != EEXIST)
289                 return -1;
290         return 0;
291 }
292
293 void
294 printtimez(FILE *fp, const git_time *intime)
295 {
296         struct tm *intm;
297         time_t t;
298         char out[32];
299
300         t = (time_t)intime->time;
301         if (!(intm = gmtime(&t)))
302                 return;
303         strftime(out, sizeof(out), "%Y-%m-%dT%H:%M:%SZ", intm);
304         fputs(out, fp);
305 }
306
307 void
308 printtime(FILE *fp, const git_time *intime)
309 {
310         struct tm *intm;
311         time_t t;
312         char out[32];
313
314         t = (time_t)intime->time + (intime->offset * 60);
315         if (!(intm = gmtime(&t)))
316                 return;
317         strftime(out, sizeof(out), "%a, %e %b %Y %H:%M:%S", intm);
318         if (intime->offset < 0)
319                 fprintf(fp, "%s -%02d%02d", out,
320                             -(intime->offset) / 60, -(intime->offset) % 60);
321         else
322                 fprintf(fp, "%s +%02d%02d", out,
323                             intime->offset / 60, intime->offset % 60);
324 }
325
326 void
327 printtimeshort(FILE *fp, const git_time *intime)
328 {
329         struct tm *intm;
330         time_t t;
331         char out[32];
332
333         t = (time_t)intime->time;
334         if (!(intm = gmtime(&t)))
335                 return;
336         strftime(out, sizeof(out), "%Y-%m-%d %H:%M", intm);
337         fputs(out, fp);
338 }
339
340 void
341 writeheader(FILE *fp, const char *title)
342 {
343         fputs("<!DOCTYPE html>\n"
344                 "<html>\n<head>\n"
345                 "<meta http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\" />\n"
346                 "<title>", fp);
347         xmlencode(fp, title, strlen(title));
348         if (title[0] && strippedname[0])
349                 fputs(" - ", fp);
350         xmlencode(fp, strippedname, strlen(strippedname));
351         if (description[0])
352                 fputs(" - ", fp);
353         xmlencode(fp, description, strlen(description));
354         fprintf(fp, "</title>\n<link rel=\"icon\" type=\"image/png\" href=\"%sfavicon.png\" />\n", relpath);
355         fprintf(fp, "<link rel=\"alternate\" type=\"application/atom+xml\" title=\"%s Atom Feed\" href=\"%satom.xml\" />\n",
356                 name, relpath);
357         fprintf(fp, "<link rel=\"stylesheet\" type=\"text/css\" href=\"%sstyle.css\" />\n", relpath);
358         fputs("</head>\n<body>\n<table><tr><td>", fp);
359         fprintf(fp, "<a href=\"../%s\"><img src=\"%slogo.png\" alt=\"\" width=\"32\" height=\"32\" /></a>",
360                 relpath, relpath);
361         fputs("</td><td><h1>", fp);
362         xmlencode(fp, strippedname, strlen(strippedname));
363         fputs("</h1><span class=\"desc\">", fp);
364         xmlencode(fp, description, strlen(description));
365         fputs("</span></td></tr>", fp);
366         if (cloneurl[0]) {
367                 fputs("<tr class=\"url\"><td></td><td>git clone <a href=\"", fp);
368                 xmlencode(fp, cloneurl, strlen(cloneurl));
369                 fputs("\">", fp);
370                 xmlencode(fp, cloneurl, strlen(cloneurl));
371                 fputs("</a></td></tr>", fp);
372         }
373         fputs("<tr><td></td><td>\n", fp);
374         fprintf(fp, "<a href=\"%slog.html\">Log</a> | ", relpath);
375         fprintf(fp, "<a href=\"%sfiles.html\">Files</a> | ", relpath);
376         fprintf(fp, "<a href=\"%srefs.html\">Refs</a>", relpath);
377         if (submodules)
378                 fprintf(fp, " | <a href=\"%sfile/%s.html\">Submodules</a>",
379                         relpath, submodules);
380         if (readme)
381                 fprintf(fp, " | <a href=\"%sfile/%s.html\">README</a>",
382                         relpath, readme);
383         if (license)
384                 fprintf(fp, " | <a href=\"%sfile/%s.html\">LICENSE</a>",
385                         relpath, license);
386         fputs("</td></tr></table>\n<hr/>\n<div id=\"content\">\n", fp);
387 }
388
389 void
390 writefooter(FILE *fp)
391 {
392         fputs("</div>\n</body>\n</html>\n", fp);
393 }
394
395 int
396 writeblobhtml(FILE *fp, const git_blob *blob)
397 {
398         size_t n = 0, i, prev;
399         const char *nfmt = "<a href=\"#l%d\" class=\"line\" id=\"l%d\">%7d</a> ";
400         const char *s = git_blob_rawcontent(blob);
401         git_off_t len = git_blob_rawsize(blob);
402
403         fputs("<pre id=\"blob\">\n", fp);
404
405         if (len > 0) {
406                 for (i = 0, prev = 0; i < (size_t)len; i++) {
407                         if (s[i] != '\n')
408                                 continue;
409                         n++;
410                         fprintf(fp, nfmt, n, n, n);
411                         xmlencode(fp, &s[prev], i - prev + 1);
412                         prev = i + 1;
413                 }
414                 /* trailing data */
415                 if ((len - prev) > 0) {
416                         n++;
417                         fprintf(fp, nfmt, n, n, n);
418                         xmlencode(fp, &s[prev], len - prev);
419                 }
420         }
421
422         fputs("</pre>\n", fp);
423
424         return n;
425 }
426
427 void
428 printcommit(FILE *fp, struct commitinfo *ci)
429 {
430         fprintf(fp, "<b>commit</b> <a href=\"%scommit/%s.html\">%s</a>\n",
431                 relpath, ci->oid, ci->oid);
432
433         if (ci->parentoid[0])
434                 fprintf(fp, "<b>parent</b> <a href=\"%scommit/%s.html\">%s</a>\n",
435                         relpath, ci->parentoid, ci->parentoid);
436
437         if (ci->author) {
438                 fputs("<b>Author:</b> ", fp);
439                 xmlencode(fp, ci->author->name, strlen(ci->author->name));
440                 fputs(" &lt;<a href=\"mailto:", fp);
441                 xmlencode(fp, ci->author->email, strlen(ci->author->email));
442                 fputs("\">", fp);
443                 xmlencode(fp, ci->author->email, strlen(ci->author->email));
444                 fputs("</a>&gt;\n<b>Date:</b>   ", fp);
445                 printtime(fp, &(ci->author->when));
446                 fputc('\n', fp);
447         }
448         if (ci->msg) {
449                 fputc('\n', fp);
450                 xmlencode(fp, ci->msg, strlen(ci->msg));
451                 fputc('\n', fp);
452         }
453 }
454
455 void
456 printshowfile(FILE *fp, struct commitinfo *ci)
457 {
458         const git_diff_delta *delta;
459         const git_diff_hunk *hunk;
460         const git_diff_line *line;
461         git_patch *patch;
462         size_t nhunks, nhunklines, changed, add, del, total, i, j, k;
463         char linestr[80];
464         int c;
465
466         printcommit(fp, ci);
467
468         if (!ci->deltas)
469                 return;
470
471         if (ci->filecount > 1000   ||
472             ci->ndeltas   > 1000   ||
473             ci->addcount  > 100000 ||
474             ci->delcount  > 100000) {
475                 fputs("Diff is too large, output suppressed.\n", fp);
476                 return;
477         }
478
479         /* diff stat */
480         fputs("<b>Diffstat:</b>\n<table>", fp);
481         for (i = 0; i < ci->ndeltas; i++) {
482                 delta = git_patch_get_delta(ci->deltas[i]->patch);
483
484                 switch (delta->status) {
485                 case GIT_DELTA_ADDED:      c = 'A'; break;
486                 case GIT_DELTA_COPIED:     c = 'C'; break;
487                 case GIT_DELTA_DELETED:    c = 'D'; break;
488                 case GIT_DELTA_MODIFIED:   c = 'M'; break;
489                 case GIT_DELTA_RENAMED:    c = 'R'; break;
490                 case GIT_DELTA_TYPECHANGE: c = 'T'; break;
491                 default:                   c = ' '; break;
492                 }
493                 if (c == ' ')
494                         fprintf(fp, "<tr><td>%c", c);
495                 else
496                         fprintf(fp, "<tr><td class=\"%c\">%c", c, c);
497
498                 fprintf(fp, "</td><td><a href=\"#h%zu\">", i);
499                 xmlencode(fp, delta->old_file.path, strlen(delta->old_file.path));
500                 if (strcmp(delta->old_file.path, delta->new_file.path)) {
501                         fputs(" -&gt; ", fp);
502                         xmlencode(fp, delta->new_file.path, strlen(delta->new_file.path));
503                 }
504
505                 add = ci->deltas[i]->addcount;
506                 del = ci->deltas[i]->delcount;
507                 changed = add + del;
508                 total = sizeof(linestr) - 2;
509                 if (changed > total) {
510                         if (add)
511                                 add = ((float)total / changed * add) + 1;
512                         if (del)
513                                 del = ((float)total / changed * del) + 1;
514                 }
515                 memset(&linestr, '+', add);
516                 memset(&linestr[add], '-', del);
517
518                 fprintf(fp, "</a></td><td> | </td><td class=\"num\">%zu</td><td><span class=\"i\">",
519                         ci->deltas[i]->addcount + ci->deltas[i]->delcount);
520                 fwrite(&linestr, 1, add, fp);
521                 fputs("</span><span class=\"d\">", fp);
522                 fwrite(&linestr[add], 1, del, fp);
523                 fputs("</span></td></tr>\n", fp);
524         }
525         fprintf(fp, "</table></pre><pre>%zu file%s changed, %zu insertion%s(+), %zu deletion%s(-)\n",
526                 ci->filecount, ci->filecount == 1 ? "" : "s",
527                 ci->addcount,  ci->addcount  == 1 ? "" : "s",
528                 ci->delcount,  ci->delcount  == 1 ? "" : "s");
529
530         fputs("<hr/>", fp);
531
532         for (i = 0; i < ci->ndeltas; i++) {
533                 patch = ci->deltas[i]->patch;
534                 delta = git_patch_get_delta(patch);
535                 fprintf(fp, "<b>diff --git a/<a id=\"h%zu\" href=\"%sfile/", i, relpath);
536                 xmlencode(fp, delta->old_file.path, strlen(delta->old_file.path));
537                 fputs(".html\">", fp);
538                 xmlencode(fp, delta->old_file.path, strlen(delta->old_file.path));
539                 fprintf(fp, "</a> b/<a href=\"%sfile/", relpath);
540                 xmlencode(fp, delta->new_file.path, strlen(delta->new_file.path));
541                 fprintf(fp, ".html\">");
542                 xmlencode(fp, delta->new_file.path, strlen(delta->new_file.path));
543                 fprintf(fp, "</a></b>\n");
544
545                 /* check binary data */
546                 if (delta->flags & GIT_DIFF_FLAG_BINARY) {
547                         fputs("Binary files differ.\n", fp);
548                         continue;
549                 }
550
551                 nhunks = git_patch_num_hunks(patch);
552                 for (j = 0; j < nhunks; j++) {
553                         if (git_patch_get_hunk(&hunk, &nhunklines, patch, j))
554                                 break;
555
556                         fprintf(fp, "<a href=\"#h%zu-%zu\" id=\"h%zu-%zu\" class=\"h\">", i, j, i, j);
557                         xmlencode(fp, hunk->header, hunk->header_len);
558                         fputs("</a>", fp);
559
560                         for (k = 0; ; k++) {
561                                 if (git_patch_get_line_in_hunk(&line, patch, j, k))
562                                         break;
563                                 if (line->old_lineno == -1)
564                                         fprintf(fp, "<a href=\"#h%zu-%zu-%zu\" id=\"h%zu-%zu-%zu\" class=\"i\">+",
565                                                 i, j, k, i, j, k);
566                                 else if (line->new_lineno == -1)
567                                         fprintf(fp, "<a href=\"#h%zu-%zu-%zu\" id=\"h%zu-%zu-%zu\" class=\"d\">-",
568                                                 i, j, k, i, j, k);
569                                 else
570                                         fputc(' ', fp);
571                                 xmlencode(fp, line->content, line->content_len);
572                                 if (line->old_lineno == -1 || line->new_lineno == -1)
573                                         fputs("</a>", fp);
574                         }
575                 }
576         }
577 }
578
579 void
580 writelogline(FILE *fp, struct commitinfo *ci)
581 {
582         fputs("<tr><td>", fp);
583         if (ci->author)
584                 printtimeshort(fp, &(ci->author->when));
585         fputs("</td><td>", fp);
586         if (ci->summary) {
587                 fprintf(fp, "<a href=\"%scommit/%s.html\">", relpath, ci->oid);
588                 xmlencode(fp, ci->summary, strlen(ci->summary));
589                 fputs("</a>", fp);
590         }
591         fputs("</td><td>", fp);
592         if (ci->author)
593                 xmlencode(fp, ci->author->name, strlen(ci->author->name));
594         fputs("</td><td class=\"num\" align=\"right\">", fp);
595         fprintf(fp, "%zu", ci->filecount);
596         fputs("</td><td class=\"num\" align=\"right\">", fp);
597         fprintf(fp, "+%zu", ci->addcount);
598         fputs("</td><td class=\"num\" align=\"right\">", fp);
599         fprintf(fp, "-%zu", ci->delcount);
600         fputs("</td></tr>\n", fp);
601 }
602
603 int
604 writelog(FILE *fp, const git_oid *oid)
605 {
606         struct commitinfo *ci;
607         git_revwalk *w = NULL;
608         git_oid id;
609         char path[PATH_MAX], oidstr[GIT_OID_HEXSZ + 1];
610         FILE *fpfile;
611         int r;
612
613         git_revwalk_new(&w, repo);
614         git_revwalk_push(w, oid);
615         git_revwalk_simplify_first_parent(w);
616
617         while (!git_revwalk_next(&id, w)) {
618                 relpath = "";
619
620                 if (cachefile && !memcmp(&id, &lastoid, sizeof(id)))
621                         break;
622
623                 git_oid_tostr(oidstr, sizeof(oidstr), &id);
624                 r = snprintf(path, sizeof(path), "commit/%s.html", oidstr);
625                 if (r < 0 || (size_t)r >= sizeof(path))
626                         errx(1, "path truncated: 'commit/%s.html'", oidstr);
627                 r = access(path, F_OK);
628
629                 /* optimization: if there are no log lines to write and
630                    the commit file already exists: skip the diffstat */
631                 if (!nlogcommits && !r)
632                         continue;
633
634                 if (!(ci = commitinfo_getbyoid(&id)))
635                         break;
636                 /* diffstat: for stagit HTML required for the log.html line */
637                 if (commitinfo_getstats(ci) == -1)
638                         goto err;
639
640                 if (nlogcommits < 0) {
641                         writelogline(fp, ci);
642                 } else if (nlogcommits > 0) {
643                         writelogline(fp, ci);
644                         nlogcommits--;
645                         if (!nlogcommits && ci->parentoid[0])
646                                 fputs("<tr><td></td><td colspan=\"5\">"
647                                       "More commits remaining [...]</td>"
648                                       "</tr>\n", fp);
649                 }
650
651                 if (cachefile)
652                         writelogline(wcachefp, ci);
653
654                 /* check if file exists if so skip it */
655                 if (r) {
656                         relpath = "../";
657                         fpfile = efopen(path, "w");
658                         writeheader(fpfile, ci->summary);
659                         fputs("<pre>", fpfile);
660                         printshowfile(fpfile, ci);
661                         fputs("</pre>\n", fpfile);
662                         writefooter(fpfile);
663                         fclose(fpfile);
664                 }
665 err:
666                 commitinfo_free(ci);
667         }
668         git_revwalk_free(w);
669
670         relpath = "";
671
672         return 0;
673 }
674
675 void
676 printcommitatom(FILE *fp, struct commitinfo *ci)
677 {
678         fputs("<entry>\n", fp);
679
680         fprintf(fp, "<id>%s</id>\n", ci->oid);
681         if (ci->author) {
682                 fputs("<published>", fp);
683                 printtimez(fp, &(ci->author->when));
684                 fputs("</published>\n", fp);
685         }
686         if (ci->committer) {
687                 fputs("<updated>", fp);
688                 printtimez(fp, &(ci->committer->when));
689                 fputs("</updated>\n", fp);
690         }
691         if (ci->summary) {
692                 fputs("<title type=\"text\">", fp);
693                 xmlencode(fp, ci->summary, strlen(ci->summary));
694                 fputs("</title>\n", fp);
695         }
696         fprintf(fp, "<link rel=\"alternate\" type=\"text/html\" href=\"commit/%s.html\" />",
697                 ci->oid);
698
699         if (ci->author) {
700                 fputs("<author><name>", fp);
701                 xmlencode(fp, ci->author->name, strlen(ci->author->name));
702                 fputs("</name>\n<email>", fp);
703                 xmlencode(fp, ci->author->email, strlen(ci->author->email));
704                 fputs("</email>\n</author>\n", fp);
705         }
706
707         fputs("<content type=\"text\">", fp);
708         fprintf(fp, "commit %s\n", ci->oid);
709         if (ci->parentoid[0])
710                 fprintf(fp, "parent %s\n", ci->parentoid);
711         if (ci->author) {
712                 fputs("Author: ", fp);
713                 xmlencode(fp, ci->author->name, strlen(ci->author->name));
714                 fputs(" &lt;", fp);
715                 xmlencode(fp, ci->author->email, strlen(ci->author->email));
716                 fputs("&gt;\nDate:   ", fp);
717                 printtime(fp, &(ci->author->when));
718                 fputc('\n', fp);
719         }
720         if (ci->msg) {
721                 fputc('\n', fp);
722                 xmlencode(fp, ci->msg, strlen(ci->msg));
723         }
724         fputs("\n</content>\n</entry>\n", fp);
725 }
726
727 int
728 writeatom(FILE *fp)
729 {
730         struct commitinfo *ci;
731         git_revwalk *w = NULL;
732         git_oid id;
733         size_t i, m = 100; /* last 'm' commits */
734
735         fputs("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
736               "<feed xmlns=\"http://www.w3.org/2005/Atom\">\n<title>", fp);
737         xmlencode(fp, strippedname, strlen(strippedname));
738         fputs(", branch HEAD</title>\n<subtitle>", fp);
739         xmlencode(fp, description, strlen(description));
740         fputs("</subtitle>\n", fp);
741
742         git_revwalk_new(&w, repo);
743         git_revwalk_push_head(w);
744         git_revwalk_simplify_first_parent(w);
745
746         for (i = 0; i < m && !git_revwalk_next(&id, w); i++) {
747                 if (!(ci = commitinfo_getbyoid(&id)))
748                         break;
749                 printcommitatom(fp, ci);
750                 commitinfo_free(ci);
751         }
752         git_revwalk_free(w);
753
754         fputs("</feed>\n", fp);
755
756         return 0;
757 }
758
759 int
760 writeblob(git_object *obj, const char *fpath, const char *filename, git_off_t filesize)
761 {
762         char tmp[PATH_MAX] = "", *d;
763         const char *p;
764         int lc = 0;
765         FILE *fp;
766
767         if (strlcpy(tmp, fpath, sizeof(tmp)) >= sizeof(tmp))
768                 errx(1, "path truncated: '%s'", fpath);
769         if (!(d = dirname(tmp)))
770                 err(1, "dirname");
771         if (mkdirp(d))
772                 return -1;
773
774         for (p = fpath, tmp[0] = '\0'; *p; p++) {
775                 if (*p == '/' && strlcat(tmp, "../", sizeof(tmp)) >= sizeof(tmp))
776                         errx(1, "path truncated: '../%s'", tmp);
777         }
778         relpath = tmp;
779
780         fp = efopen(fpath, "w");
781         writeheader(fp, filename);
782         fputs("<p> ", fp);
783         xmlencode(fp, filename, strlen(filename));
784         fprintf(fp, " (%juB)", (uintmax_t)filesize);
785         fputs("</p><hr/>", fp);
786
787         if (git_blob_is_binary((git_blob *)obj)) {
788                 fputs("<p>Binary file.</p>\n", fp);
789         } else {
790                 lc = writeblobhtml(fp, (git_blob *)obj);
791                 if (ferror(fp))
792                         err(1, "fwrite");
793         }
794         writefooter(fp);
795         fclose(fp);
796
797         relpath = "";
798
799         return lc;
800 }
801
802 const char *
803 filemode(git_filemode_t m)
804 {
805         static char mode[11];
806
807         memset(mode, '-', sizeof(mode) - 1);
808         mode[10] = '\0';
809
810         if (S_ISREG(m))
811                 mode[0] = '-';
812         else if (S_ISBLK(m))
813                 mode[0] = 'b';
814         else if (S_ISCHR(m))
815                 mode[0] = 'c';
816         else if (S_ISDIR(m))
817                 mode[0] = 'd';
818         else if (S_ISFIFO(m))
819                 mode[0] = 'p';
820         else if (S_ISLNK(m))
821                 mode[0] = 'l';
822         else if (S_ISSOCK(m))
823                 mode[0] = 's';
824         else
825                 mode[0] = '?';
826
827         if (m & S_IRUSR) mode[1] = 'r';
828         if (m & S_IWUSR) mode[2] = 'w';
829         if (m & S_IXUSR) mode[3] = 'x';
830         if (m & S_IRGRP) mode[4] = 'r';
831         if (m & S_IWGRP) mode[5] = 'w';
832         if (m & S_IXGRP) mode[6] = 'x';
833         if (m & S_IROTH) mode[7] = 'r';
834         if (m & S_IWOTH) mode[8] = 'w';
835         if (m & S_IXOTH) mode[9] = 'x';
836
837         if (m & S_ISUID) mode[3] = (mode[3] == 'x') ? 's' : 'S';
838         if (m & S_ISGID) mode[6] = (mode[6] == 'x') ? 's' : 'S';
839         if (m & S_ISVTX) mode[9] = (mode[9] == 'x') ? 't' : 'T';
840
841         return mode;
842 }
843
844 int
845 writefilestree(FILE *fp, git_tree *tree, const char *path)
846 {
847         const git_tree_entry *entry = NULL;
848         git_submodule *module = NULL;
849         git_object *obj = NULL;
850         git_off_t filesize;
851         const char *entryname;
852         char filepath[PATH_MAX], entrypath[PATH_MAX];
853         size_t count, i;
854         int lc, r, ret;
855
856         count = git_tree_entrycount(tree);
857         for (i = 0; i < count; i++) {
858                 if (!(entry = git_tree_entry_byindex(tree, i)) ||
859                     !(entryname = git_tree_entry_name(entry)))
860                         return -1;
861                 joinpath(entrypath, sizeof(entrypath), path, entryname);
862
863                 r = snprintf(filepath, sizeof(filepath), "file/%s.html",
864                          entrypath);
865                 if (r < 0 || (size_t)r >= sizeof(filepath))
866                         errx(1, "path truncated: 'file/%s.html'", entrypath);
867
868                 if (!git_tree_entry_to_object(&obj, repo, entry)) {
869                         switch (git_object_type(obj)) {
870                         case GIT_OBJ_BLOB:
871                                 break;
872                         case GIT_OBJ_TREE:
873                                 /* NOTE: recurses */
874                                 ret = writefilestree(fp, (git_tree *)obj,
875                                                      entrypath);
876                                 git_object_free(obj);
877                                 if (ret)
878                                         return ret;
879                                 continue;
880                         default:
881                                 git_object_free(obj);
882                                 continue;
883                         }
884
885                         filesize = git_blob_rawsize((git_blob *)obj);
886                         lc = writeblob(obj, filepath, entryname, filesize);
887
888                         fputs("<tr><td>", fp);
889                         fputs(filemode(git_tree_entry_filemode(entry)), fp);
890                         fprintf(fp, "</td><td><a href=\"%s", relpath);
891                         xmlencode(fp, filepath, strlen(filepath));
892                         fputs("\">", fp);
893                         xmlencode(fp, entrypath, strlen(entrypath));
894                         fputs("</a></td><td class=\"num\" align=\"right\">", fp);
895                         if (lc > 0)
896                                 fprintf(fp, "%dL", lc);
897                         else
898                                 fprintf(fp, "%juB", (uintmax_t)filesize);
899                         fputs("</td></tr>\n", fp);
900                         git_object_free(obj);
901                 } else if (!git_submodule_lookup(&module, repo, entryname)) {
902                         fprintf(fp, "<tr><td>m---------</td><td><a href=\"%sfile/.gitmodules.html\">",
903                                 relpath);
904                         xmlencode(fp, entrypath, strlen(entrypath));
905                         git_submodule_free(module);
906                         fputs("</a></td><td class=\"num\" align=\"right\"></td></tr>\n", fp);
907                 }
908         }
909
910         return 0;
911 }
912
913 int
914 writefiles(FILE *fp, const git_oid *id)
915 {
916         git_tree *tree = NULL;
917         git_commit *commit = NULL;
918         int ret = -1;
919
920         fputs("<table id=\"files\"><thead>\n<tr>"
921               "<td><b>Mode</b></td><td><b>Name</b></td>"
922               "<td class=\"num\" align=\"right\"><b>Size</b></td>"
923               "</tr>\n</thead><tbody>\n", fp);
924
925         if (!git_commit_lookup(&commit, repo, id) &&
926             !git_commit_tree(&tree, commit))
927                 ret = writefilestree(fp, tree, "");
928
929         fputs("</tbody></table>", fp);
930
931         git_commit_free(commit);
932         git_tree_free(tree);
933
934         return ret;
935 }
936
937 int
938 refs_cmp(const void *v1, const void *v2)
939 {
940         git_reference *r1 = (*(git_reference **)v1);
941         git_reference *r2 = (*(git_reference **)v2);
942         int r;
943
944         if ((r = git_reference_is_branch(r1) - git_reference_is_branch(r2)))
945                 return r;
946
947         return strcmp(git_reference_shorthand(r1),
948                       git_reference_shorthand(r2));
949 }
950
951 int
952 writerefs(FILE *fp)
953 {
954         struct commitinfo *ci;
955         const git_oid *id = NULL;
956         git_object *obj = NULL;
957         git_reference *dref = NULL, *r, *ref = NULL;
958         git_reference_iterator *it = NULL;
959         git_reference **refs = NULL;
960         size_t count, i, j, refcount;
961         const char *titles[] = { "Branches", "Tags" };
962         const char *ids[] = { "branches", "tags" };
963         const char *name;
964
965         if (git_reference_iterator_new(&it, repo))
966                 return -1;
967
968         for (refcount = 0; !git_reference_next(&ref, it); refcount++) {
969                 if (!(refs = reallocarray(refs, refcount + 1, sizeof(git_reference *))))
970                         err(1, "realloc");
971                 refs[refcount] = ref;
972         }
973         git_reference_iterator_free(it);
974
975         /* sort by type then shorthand name */
976         qsort(refs, refcount, sizeof(git_reference *), refs_cmp);
977
978         for (j = 0; j < 2; j++) {
979                 for (i = 0, count = 0; i < refcount; i++) {
980                         if (!(git_reference_is_branch(refs[i]) && j == 0) &&
981                             !(git_reference_is_tag(refs[i]) && j == 1))
982                                 continue;
983
984                         switch (git_reference_type(refs[i])) {
985                         case GIT_REF_SYMBOLIC:
986                                 if (git_reference_resolve(&dref, refs[i]))
987                                         goto err;
988                                 r = dref;
989                                 break;
990                         case GIT_REF_OID:
991                                 r = refs[i];
992                                 break;
993                         default:
994                                 continue;
995                         }
996                         if (!git_reference_target(r) ||
997                             git_reference_peel(&obj, r, GIT_OBJ_ANY))
998                                 goto err;
999                         if (!(id = git_object_id(obj)))
1000                                 goto err;
1001                         if (!(ci = commitinfo_getbyoid(id)))
1002                                 break;
1003
1004                         /* print header if it has an entry (first). */
1005                         if (++count == 1) {
1006                                 fprintf(fp, "<h2>%s</h2><table id=\"%s\">"
1007                                         "<thead>\n<tr><td><b>Name</b></td>"
1008                                         "<td><b>Last commit date</b></td>"
1009                                         "<td><b>Author</b></td>\n</tr>\n"
1010                                         "</thead><tbody>\n",
1011                                          titles[j], ids[j]);
1012                         }
1013
1014                         relpath = "";
1015                         name = git_reference_shorthand(r);
1016
1017                         fputs("<tr><td>", fp);
1018                         xmlencode(fp, name, strlen(name));
1019                         fputs("</td><td>", fp);
1020                         if (ci->author)
1021                                 printtimeshort(fp, &(ci->author->when));
1022                         fputs("</td><td>", fp);
1023                         if (ci->author)
1024                                 xmlencode(fp, ci->author->name, strlen(ci->author->name));
1025                         fputs("</td></tr>\n", fp);
1026
1027                         relpath = "../";
1028
1029                         commitinfo_free(ci);
1030                         git_object_free(obj);
1031                         obj = NULL;
1032                         git_reference_free(dref);
1033                         dref = NULL;
1034                 }
1035                 /* table footer */
1036                 if (count)
1037                         fputs("</tbody></table><br/>", fp);
1038         }
1039
1040 err:
1041         git_object_free(obj);
1042         git_reference_free(dref);
1043
1044         for (i = 0; i < refcount; i++)
1045                 git_reference_free(refs[i]);
1046         free(refs);
1047
1048         return 0;
1049 }
1050
1051 void
1052 usage(char *argv0)
1053 {
1054         fprintf(stderr, "%s [-c cachefile | -l commits] repodir\n", argv0);
1055         exit(1);
1056 }
1057
1058 int
1059 main(int argc, char *argv[])
1060 {
1061         git_object *obj = NULL;
1062         const git_oid *head = NULL;
1063         mode_t mask;
1064         FILE *fp, *fpread;
1065         char path[PATH_MAX], repodirabs[PATH_MAX + 1], *p;
1066         char tmppath[64] = "cache.XXXXXXXXXXXX", buf[BUFSIZ];
1067         size_t n;
1068         int i, fd;
1069
1070         for (i = 1; i < argc; i++) {
1071                 if (argv[i][0] != '-') {
1072                         if (repodir)
1073                                 usage(argv[0]);
1074                         repodir = argv[i];
1075                 } else if (argv[i][1] == 'c') {
1076                         if (nlogcommits > 0 || i + 1 >= argc)
1077                                 usage(argv[0]);
1078                         cachefile = argv[++i];
1079                 } else if (argv[i][1] == 'l') {
1080                         if (cachefile || i + 1 >= argc)
1081                                 usage(argv[0]);
1082                         errno = 0;
1083                         nlogcommits = strtoll(argv[++i], &p, 10);
1084                         if (argv[i][0] == '\0' || *p != '\0' ||
1085                             nlogcommits <= 0 || errno)
1086                                 usage(argv[0]);
1087                 }
1088         }
1089         if (!repodir)
1090                 usage(argv[0]);
1091
1092         if (!realpath(repodir, repodirabs))
1093                 err(1, "realpath");
1094
1095         git_libgit2_init();
1096
1097 #ifdef __OpenBSD__
1098         if (cachefile) {
1099                 if (pledge("stdio rpath wpath cpath fattr", NULL) == -1)
1100                         err(1, "pledge");
1101         } else {
1102                 if (pledge("stdio rpath wpath cpath", NULL) == -1)
1103                         err(1, "pledge");
1104         }
1105 #endif
1106
1107         if (git_repository_open_ext(&repo, repodir,
1108                 GIT_REPOSITORY_OPEN_NO_SEARCH, NULL) < 0) {
1109                 fprintf(stderr, "%s: cannot open repository\n", argv[0]);
1110                 return 1;
1111         }
1112
1113         /* find HEAD */
1114         if (!git_revparse_single(&obj, repo, "HEAD"))
1115                 head = git_object_id(obj);
1116         git_object_free(obj);
1117
1118         /* use directory name as name */
1119         if ((name = strrchr(repodirabs, '/')))
1120                 name++;
1121         else
1122                 name = "";
1123
1124         /* strip .git suffix */
1125         if (!(strippedname = strdup(name)))
1126                 err(1, "strdup");
1127         if ((p = strrchr(strippedname, '.')))
1128                 if (!strcmp(p, ".git"))
1129                         *p = '\0';
1130
1131         /* read description or .git/description */
1132         joinpath(path, sizeof(path), repodir, "description");
1133         if (!(fpread = fopen(path, "r"))) {
1134                 joinpath(path, sizeof(path), repodir, ".git/description");
1135                 fpread = fopen(path, "r");
1136         }
1137         if (fpread) {
1138                 if (!fgets(description, sizeof(description), fpread))
1139                         description[0] = '\0';
1140                 fclose(fpread);
1141         }
1142
1143         /* read url or .git/url */
1144         joinpath(path, sizeof(path), repodir, "url");
1145         if (!(fpread = fopen(path, "r"))) {
1146                 joinpath(path, sizeof(path), repodir, ".git/url");
1147                 fpread = fopen(path, "r");
1148         }
1149         if (fpread) {
1150                 if (!fgets(cloneurl, sizeof(cloneurl), fpread))
1151                         cloneurl[0] = '\0';
1152                 cloneurl[strcspn(cloneurl, "\n")] = '\0';
1153                 fclose(fpread);
1154         }
1155
1156         /* check LICENSE */
1157         for (i = 0; i < sizeof(licensefiles) / sizeof(*licensefiles) && !license; i++) {
1158                 if (!git_revparse_single(&obj, repo, licensefiles[i]) &&
1159                     git_object_type(obj) == GIT_OBJ_BLOB)
1160                         license = licensefiles[i] + strlen("HEAD:");
1161                 git_object_free(obj);
1162         }
1163
1164         /* check README */
1165         for (i = 0; i < sizeof(readmefiles) / sizeof(*readmefiles) && !readme; i++) {
1166                 if (!git_revparse_single(&obj, repo, readmefiles[i]) &&
1167                     git_object_type(obj) == GIT_OBJ_BLOB)
1168                         readme = readmefiles[i] + strlen("HEAD:");
1169                 git_object_free(obj);
1170         }
1171
1172         if (!git_revparse_single(&obj, repo, "HEAD:.gitmodules") &&
1173             git_object_type(obj) == GIT_OBJ_BLOB)
1174                 submodules = ".gitmodules";
1175         git_object_free(obj);
1176
1177         /* log for HEAD */
1178         fp = efopen("log.html", "w");
1179         relpath = "";
1180         mkdir("commit", S_IRWXU | S_IRWXG | S_IRWXO);
1181         writeheader(fp, "Log");
1182         fputs("<table id=\"log\"><thead>\n<tr><td><b>Date</b></td>"
1183               "<td><b>Commit message</b></td>"
1184               "<td><b>Author</b></td><td class=\"num\" align=\"right\"><b>Files</b></td>"
1185               "<td class=\"num\" align=\"right\"><b>+</b></td>"
1186               "<td class=\"num\" align=\"right\"><b>-</b></td></tr>\n</thead><tbody>\n", fp);
1187
1188         if (cachefile && head) {
1189                 /* read from cache file (does not need to exist) */
1190                 if ((rcachefp = fopen(cachefile, "r"))) {
1191                         if (!fgets(lastoidstr, sizeof(lastoidstr), rcachefp))
1192                                 errx(1, "%s: no object id", cachefile);
1193                         if (git_oid_fromstr(&lastoid, lastoidstr))
1194                                 errx(1, "%s: invalid object id", cachefile);
1195                 }
1196
1197                 /* write log to (temporary) cache */
1198                 if ((fd = mkstemp(tmppath)) == -1)
1199                         err(1, "mkstemp");
1200                 if (!(wcachefp = fdopen(fd, "w")))
1201                         err(1, "fdopen: '%s'", tmppath);
1202                 /* write last commit id (HEAD) */
1203                 git_oid_tostr(buf, sizeof(buf), head);
1204                 fprintf(wcachefp, "%s\n", buf);
1205
1206                 writelog(fp, head);
1207
1208                 if (rcachefp) {
1209                         /* append previous log to log.html and the new cache */
1210                         while (!feof(rcachefp)) {
1211                                 n = fread(buf, 1, sizeof(buf), rcachefp);
1212                                 if (ferror(rcachefp))
1213                                         err(1, "fread");
1214                                 if (fwrite(buf, 1, n, fp) != n ||
1215                                     fwrite(buf, 1, n, wcachefp) != n)
1216                                         err(1, "fwrite");
1217                         }
1218                         fclose(rcachefp);
1219                 }
1220                 fclose(wcachefp);
1221         } else {
1222                 if (head)
1223                         writelog(fp, head);
1224         }
1225
1226         fputs("</tbody></table>", fp);
1227         writefooter(fp);
1228         fclose(fp);
1229
1230         /* files for HEAD */
1231         fp = efopen("files.html", "w");
1232         writeheader(fp, "Files");
1233         if (head)
1234                 writefiles(fp, head);
1235         writefooter(fp);
1236         fclose(fp);
1237
1238         /* summary page with branches and tags */
1239         fp = efopen("refs.html", "w");
1240         writeheader(fp, "Refs");
1241         writerefs(fp);
1242         writefooter(fp);
1243         fclose(fp);
1244
1245         /* Atom feed */
1246         fp = efopen("atom.xml", "w");
1247         writeatom(fp);
1248         fclose(fp);
1249
1250         /* rename new cache file on success */
1251         if (cachefile && head) {
1252                 if (rename(tmppath, cachefile))
1253                         err(1, "rename: '%s' to '%s'", tmppath, cachefile);
1254                 umask((mask = umask(0)));
1255                 if (chmod(cachefile,
1256                     (S_IRUSR|S_IWUSR|S_IRGRP|S_IWGRP|S_IROTH|S_IWOTH) & ~mask))
1257                         err(1, "chmod: '%s'", cachefile);
1258         }
1259
1260         /* cleanup */
1261         git_repository_free(repo);
1262         git_libgit2_shutdown();
1263
1264         return 0;
1265 }