TODO updates.
[sgt/agedu] / agedu.c
1 /*
2 * Main program for agedu.
3 */
4
5 #include "agedu.h"
6
7 #include "du.h"
8 #include "trie.h"
9 #include "index.h"
10 #include "alloc.h"
11 #include "html.h"
12 #include "httpd.h"
13 #include "fgetline.h"
14
15 /*
16 * Path separator. This global variable affects the behaviour of
17 * various parts of the code when they need to deal with path
18 * separators. The path separator appropriate to a particular data
19 * set is encoded in the index file storing that data set; data
20 * sets generated on Unix will of course have the default '/', but
21 * foreign data sets are conceivable and must be handled correctly.
22 */
23 char pathsep = '/';
24
25 void fatal(const char *fmt, ...)
26 {
27 va_list ap;
28 fprintf(stderr, "%s: ", PNAME);
29 va_start(ap, fmt);
30 vfprintf(stderr, fmt, ap);
31 va_end(ap);
32 fprintf(stderr, "\n");
33 exit(1);
34 }
35
36 struct inclusion_exclusion {
37 int type;
38 const char *wildcard;
39 int path;
40 };
41
42 struct ctx {
43 triebuild *tb;
44 dev_t datafile_dev, filesystem_dev;
45 ino_t datafile_ino;
46 time_t last_output_update;
47 int progress, progwidth;
48 int straight_to_dump;
49 struct inclusion_exclusion *inex;
50 int ninex;
51 int crossfs;
52 int usemtime;
53 int fakeatimes;
54 };
55
56 static void dump_line(const char *pathname, const struct trie_file *tf)
57 {
58 const char *p;
59 if (printf("%llu %llu ", tf->size, tf->atime) < 0) goto error;
60 for (p = pathname; *p; p++) {
61 if (*p >= ' ' && *p < 127 && *p != '%') {
62 if (putchar(*p) == EOF) goto error;
63 } else {
64 if (printf("%%%02x", (unsigned char)*p) < 0) goto error;
65 }
66 }
67 if (putchar('\n') == EOF) goto error;
68 return;
69 error:
70 fatal("standard output: %s", strerror(errno));
71 }
72
73 static int gotdata(void *vctx, const char *pathname, const STRUCT_STAT *st)
74 {
75 struct ctx *ctx = (struct ctx *)vctx;
76 struct trie_file file;
77 time_t t;
78 int i, include;
79 const char *filename;
80
81 /*
82 * Filter out our own data file.
83 */
84 if (st->st_dev == ctx->datafile_dev && st->st_ino == ctx->datafile_ino)
85 return 0;
86
87 /*
88 * Don't cross the streams^W^Wany file system boundary.
89 */
90 if (!ctx->crossfs && st->st_dev != ctx->filesystem_dev)
91 return 0;
92
93 file.size = (unsigned long long)512 * st->st_blocks;
94 if (ctx->usemtime || (ctx->fakeatimes && S_ISDIR(st->st_mode)))
95 file.atime = st->st_mtime;
96 else
97 file.atime = max(st->st_mtime, st->st_atime);
98
99 /*
100 * Filter based on wildcards.
101 */
102 include = 1;
103 filename = strrchr(pathname, pathsep);
104 if (!filename)
105 filename = pathname;
106 else
107 filename++;
108 for (i = 0; i < ctx->ninex; i++) {
109 if (fnmatch(ctx->inex[i].wildcard,
110 ctx->inex[i].path ? pathname : filename, 0) == 0)
111 include = ctx->inex[i].type;
112 }
113 if (include == -1)
114 return 0; /* ignore this entry and any subdirs */
115 if (include == 0) {
116 /*
117 * Here we are supposed to be filtering an entry out, but
118 * still recursing into it if it's a directory. However,
119 * we can't actually leave out any directory whose
120 * subdirectories we then look at. So we cheat, in that
121 * case, by setting the size to zero.
122 */
123 if (!S_ISDIR(st->st_mode))
124 return 0; /* just ignore */
125 else
126 file.size = 0;
127 }
128
129 if (ctx->straight_to_dump)
130 dump_line(pathname, &file);
131 else
132 triebuild_add(ctx->tb, pathname, &file);
133
134 if (ctx->progress) {
135 t = time(NULL);
136 if (t != ctx->last_output_update) {
137 fprintf(stderr, "%-*.*s\r", ctx->progwidth, ctx->progwidth,
138 pathname);
139 fflush(stderr);
140 ctx->last_output_update = t;
141 }
142 }
143
144 return 1;
145 }
146
147 static void scan_error(void *vctx, const char *fmt, ...)
148 {
149 struct ctx *ctx = (struct ctx *)vctx;
150 va_list ap;
151
152 if (ctx->progress) {
153 fprintf(stderr, "%-*s\r", ctx->progwidth, "");
154 fflush(stderr);
155 }
156
157 fprintf(stderr, "%s: ", PNAME);
158 va_start(ap, fmt);
159 vfprintf(stderr, fmt, ap);
160 va_end(ap);
161
162 ctx->last_output_update--; /* force a progress report next time */
163 }
164
165 static void text_query(const void *mappedfile, const char *querydir,
166 time_t t, int showfiles, int depth, FILE *fp)
167 {
168 size_t maxpathlen;
169 char *pathbuf;
170 unsigned long xi1, xi2;
171 unsigned long long size;
172
173 maxpathlen = trie_maxpathlen(mappedfile);
174 pathbuf = snewn(maxpathlen + 1, char);
175
176 /*
177 * We want to query everything between the supplied filename
178 * (inclusive) and that filename with a ^A on the end
179 * (exclusive). So find the x indices for each.
180 */
181 strcpy(pathbuf, querydir);
182 make_successor(pathbuf);
183 xi1 = trie_before(mappedfile, querydir);
184 xi2 = trie_before(mappedfile, pathbuf);
185
186 if (!showfiles && xi2 - xi1 == 1)
187 return; /* file, or empty dir => no display */
188
189 /*
190 * Now do the lookups in the age index.
191 */
192 if (xi2 - xi1 == 1) {
193 /*
194 * We are querying an individual file, so we should not
195 * depend on the index entries either side of the node,
196 * since they almost certainly don't both exist. Instead,
197 * just look up the file's size and atime in the main trie.
198 */
199 const struct trie_file *f = trie_getfile(mappedfile, xi1);
200 if (f->atime < t)
201 size = f->size;
202 else
203 size = 0;
204 } else {
205 unsigned long long s1, s2;
206 s1 = index_query(mappedfile, xi1, t);
207 s2 = index_query(mappedfile, xi2, t);
208 size = s2 - s1;
209 }
210
211 if (size == 0)
212 return; /* no space taken up => no display */
213
214 if (depth != 0) {
215 /*
216 * Now scan for first-level subdirectories and report
217 * those too.
218 */
219 int newdepth = (depth > 0 ? depth - 1 : depth);
220 xi1++;
221 while (xi1 < xi2) {
222 trie_getpath(mappedfile, xi1, pathbuf);
223 text_query(mappedfile, pathbuf, t, showfiles, newdepth, fp);
224 make_successor(pathbuf);
225 xi1 = trie_before(mappedfile, pathbuf);
226 }
227 }
228
229 /* Display in units of 1Kb */
230 fprintf(fp, "%-11llu %s\n", (size) / 1024, querydir);
231 }
232
233 /*
234 * Largely frivolous way to define all my command-line options. I
235 * present here a parametric macro which declares a series of
236 * _logical_ option identifiers, and for each one declares zero or
237 * more short option characters and zero or more long option
238 * words. Then I repeatedly invoke that macro with its arguments
239 * defined to be various other macros, which allows me to
240 * variously:
241 *
242 * - define an enum allocating a distinct integer value to each
243 * logical option id
244 * - define a string consisting of precisely all the short option
245 * characters
246 * - define a string array consisting of all the long option
247 * strings
248 * - define (with help from auxiliary enums) integer arrays
249 * parallel to both of the above giving the logical option id
250 * for each physical short and long option
251 * - define an array indexed by logical option id indicating
252 * whether the option in question takes a value
253 * - define a function which prints out brief online help for all
254 * the options.
255 *
256 * It's not at all clear to me that this trickery is actually
257 * particularly _efficient_ - it still, after all, requires going
258 * linearly through the option list at run time and doing a
259 * strcmp, whereas in an ideal world I'd have liked the lists of
260 * long and short options to be pre-sorted so that a binary search
261 * or some other more efficient lookup was possible. (Not that
262 * asymptotic algorithmic complexity is remotely vital in option
263 * parsing, but if I were doing this in, say, Lisp or something
264 * with an equivalently powerful preprocessor then once I'd had
265 * the idea of preparing the option-parsing data structures at
266 * compile time I would probably have made the effort to prepare
267 * them _properly_. I could have Perl generate me a source file
268 * from some sort of description, I suppose, but that would seem
269 * like overkill. And in any case, it's more of a challenge to
270 * achieve as much as possible by cunning use of cpp and enum than
271 * to just write some sensible and logical code in a Turing-
272 * complete language. I said it was largely frivolous :-)
273 *
274 * This approach does have the virtue that it brings together the
275 * option ids, option spellings and help text into a single
276 * combined list and defines them all in exactly one place. If I
277 * want to add a new option, or a new spelling for an option, I
278 * only have to modify the main OPTHELP macro below and then add
279 * code to process the new logical id.
280 *
281 * (Though, really, even that isn't ideal, since it still involves
282 * modifying the source file in more than one place. In a
283 * _properly_ ideal world, I'd be able to interleave the option
284 * definitions with the code fragments that process them. And then
285 * not bother defining logical identifiers for them at all - those
286 * would be automatically generated, since I wouldn't have any
287 * need to specify them manually in another part of the code.)
288 *
289 * One other helpful consequence of the enum-based structure here
290 * is that it causes a compiler error if I accidentally try to
291 * define the same option (short or long) twice.
292 */
293
294 #define OPTHELP(NOVAL, VAL, SHORT, LONG, HELPPFX, HELPARG, HELPLINE, HELPOPT) \
295 HELPPFX("usage") HELPLINE(PNAME " [options] action [action...]") \
296 HELPPFX("actions") \
297 VAL(SCAN) SHORT(s) LONG(scan) \
298 HELPARG("directory") HELPOPT("scan and index a directory") \
299 NOVAL(HTTPD) SHORT(w) LONG(web) LONG(server) LONG(httpd) \
300 HELPOPT("serve HTML reports from a temporary web server") \
301 VAL(TEXT) SHORT(t) LONG(text) \
302 HELPARG("subdir") HELPOPT("print a plain text report on a subdirectory") \
303 NOVAL(REMOVE) SHORT(R) LONG(remove) LONG(delete) LONG(unlink) \
304 HELPOPT("remove the index file") \
305 NOVAL(DUMP) SHORT(D) LONG(dump) HELPOPT("dump the index file on stdout") \
306 NOVAL(LOAD) SHORT(L) LONG(load) \
307 HELPOPT("load and index a dump file") \
308 VAL(SCANDUMP) SHORT(S) LONG(scan_dump) \
309 HELPARG("directory") HELPOPT("scan only, generating a dump") \
310 VAL(HTML) SHORT(H) LONG(html) \
311 HELPARG("subdir") HELPOPT("print an HTML report on a subdirectory") \
312 NOVAL(CGI) LONG(cgi) \
313 HELPOPT("do the right thing when run from a CGI script") \
314 HELPPFX("options") \
315 VAL(DATAFILE) SHORT(f) LONG(file) \
316 HELPARG("filename") HELPOPT("[most modes] specify index file") \
317 NOVAL(CROSSFS) LONG(cross_fs) \
318 HELPOPT("[--scan] cross filesystem boundaries") \
319 NOVAL(NOCROSSFS) LONG(no_cross_fs) \
320 HELPOPT("[--scan] stick to one filesystem") \
321 VAL(PRUNE) LONG(prune) \
322 HELPARG("wildcard") HELPOPT("[--scan] prune files matching pattern") \
323 VAL(PRUNEPATH) LONG(prune_path) \
324 HELPARG("wildcard") HELPOPT("[--scan] prune pathnames matching pattern") \
325 VAL(EXCLUDE) LONG(exclude) \
326 HELPARG("wildcard") HELPOPT("[--scan] exclude files matching pattern") \
327 VAL(EXCLUDEPATH) LONG(exclude_path) \
328 HELPARG("wildcard") HELPOPT("[--scan] exclude pathnames matching pattern") \
329 VAL(INCLUDE) LONG(include) \
330 HELPARG("wildcard") HELPOPT("[--scan] include files matching pattern") \
331 VAL(INCLUDEPATH) LONG(include_path) \
332 HELPARG("wildcard") HELPOPT("[--scan] include pathnames matching pattern") \
333 NOVAL(PROGRESS) LONG(progress) LONG(scan_progress) \
334 HELPOPT("[--scan] report progress on stderr") \
335 NOVAL(NOPROGRESS) LONG(no_progress) LONG(no_scan_progress) \
336 HELPOPT("[--scan] do not report progress") \
337 NOVAL(TTYPROGRESS) LONG(tty_progress) LONG(tty_scan_progress) \
338 LONG(progress_tty) LONG(scan_progress_tty) \
339 HELPOPT("[--scan] report progress if stderr is a tty") \
340 NOVAL(DIRATIME) LONG(dir_atime) LONG(dir_atimes) \
341 HELPOPT("[--scan,--load] keep real atimes on directories") \
342 NOVAL(NODIRATIME) LONG(no_dir_atime) LONG(no_dir_atimes) \
343 HELPOPT("[--scan,--load] fake atimes on directories") \
344 NOVAL(MTIME) LONG(mtime) \
345 HELPOPT("[--scan] use mtime instead of atime") \
346 NOVAL(SHOWFILES) LONG(files) \
347 HELPOPT("[--web,--html,--text] list individual files") \
348 VAL(AGERANGE) SHORT(r) LONG(age_range) LONG(range) LONG(ages) \
349 HELPARG("age[-age]") HELPOPT("[--web,--html] set limits of colour coding") \
350 VAL(OUTFILE) SHORT(o) LONG(output) \
351 HELPARG("filename") HELPOPT("[--html] specify output file or directory name") \
352 VAL(SERVERADDR) LONG(address) LONG(addr) LONG(server_address) \
353 LONG(server_addr) \
354 HELPARG("addr[:port]") HELPOPT("[--web] specify HTTP server address") \
355 VAL(AUTH) LONG(auth) LONG(http_auth) LONG(httpd_auth) \
356 LONG(server_auth) LONG(web_auth) \
357 HELPARG("type") HELPOPT("[--web] specify HTTP authentication method") \
358 VAL(AUTHFILE) LONG(auth_file) \
359 HELPARG("filename") HELPOPT("[--web] read HTTP Basic user/pass from file") \
360 VAL(AUTHFD) LONG(auth_fd) \
361 HELPARG("fd") HELPOPT("[--web] read HTTP Basic user/pass from fd") \
362 VAL(DEPTH) SHORT(d) LONG(depth) LONG(max_depth) LONG(maximum_depth) \
363 HELPARG("levels") HELPOPT("[--text,--html] recurse to this many levels") \
364 VAL(MINAGE) SHORT(a) LONG(age) LONG(min_age) LONG(minimum_age) \
365 HELPARG("age") HELPOPT("[--text] include only files older than this") \
366 HELPPFX("also") \
367 NOVAL(HELP) SHORT(h) LONG(help) HELPOPT("display this help text") \
368 NOVAL(VERSION) SHORT(V) LONG(version) HELPOPT("report version number") \
369 NOVAL(LICENCE) LONG(licence) LONG(license) \
370 HELPOPT("display (MIT) licence text") \
371
372 #define IGNORE(x)
373 #define DEFENUM(x) OPT_ ## x,
374 #define ZERO(x) 0,
375 #define ONE(x) 1,
376 #define STRING(x) #x ,
377 #define STRINGNOCOMMA(x) #x
378 #define SHORTNEWOPT(x) SHORTtmp_ ## x = OPT_ ## x,
379 #define SHORTTHISOPT(x) SHORTtmp2_ ## x, SHORTVAL_ ## x = SHORTtmp2_ ## x - 1,
380 #define SHORTOPTVAL(x) SHORTVAL_ ## x,
381 #define SHORTTMP(x) SHORTtmp3_ ## x,
382 #define LONGNEWOPT(x) LONGtmp_ ## x = OPT_ ## x,
383 #define LONGTHISOPT(x) LONGtmp2_ ## x, LONGVAL_ ## x = LONGtmp2_ ## x - 1,
384 #define LONGOPTVAL(x) LONGVAL_ ## x,
385 #define LONGTMP(x) SHORTtmp3_ ## x,
386
387 #define OPTIONS(NOVAL, VAL, SHORT, LONG) \
388 OPTHELP(NOVAL, VAL, SHORT, LONG, IGNORE, IGNORE, IGNORE, IGNORE)
389
390 enum { OPTIONS(DEFENUM,DEFENUM,IGNORE,IGNORE) NOPTIONS };
391 enum { OPTIONS(IGNORE,IGNORE,SHORTTMP,IGNORE) NSHORTOPTS };
392 enum { OPTIONS(IGNORE,IGNORE,IGNORE,LONGTMP) NLONGOPTS };
393 static const int opthasval[NOPTIONS] = {OPTIONS(ZERO,ONE,IGNORE,IGNORE)};
394 static const char shortopts[] = {OPTIONS(IGNORE,IGNORE,STRINGNOCOMMA,IGNORE)};
395 static const char *const longopts[] = {OPTIONS(IGNORE,IGNORE,IGNORE,STRING)};
396 enum { OPTIONS(SHORTNEWOPT,SHORTNEWOPT,SHORTTHISOPT,IGNORE) };
397 enum { OPTIONS(LONGNEWOPT,LONGNEWOPT,IGNORE,LONGTHISOPT) };
398 static const int shortvals[] = {OPTIONS(IGNORE,IGNORE,SHORTOPTVAL,IGNORE)};
399 static const int longvals[] = {OPTIONS(IGNORE,IGNORE,IGNORE,LONGOPTVAL)};
400
401 static void usage(FILE *fp)
402 {
403 char longbuf[80];
404 const char *prefix, *shortopt, *longopt, *optarg;
405 int i, optex;
406
407 #define HELPRESET prefix = shortopt = longopt = optarg = NULL, optex = -1
408 #define HELPNOVAL(s) optex = 0;
409 #define HELPVAL(s) optex = 1;
410 #define HELPSHORT(s) if (!shortopt) shortopt = "-" #s;
411 #define HELPLONG(s) if (!longopt) { \
412 strcpy(longbuf, "--" #s); longopt = longbuf; \
413 for (i = 0; longbuf[i]; i++) if (longbuf[i] == '_') longbuf[i] = '-'; }
414 #define HELPPFX(s) prefix = s;
415 #define HELPARG(s) optarg = s;
416 #define HELPLINE(s) assert(optex == -1); \
417 fprintf(fp, "%7s%c %s\n", prefix?prefix:"", prefix?':':' ', s); \
418 HELPRESET;
419 #define HELPOPT(s) assert((optex == 1 && optarg) || (optex == 0 && !optarg)); \
420 assert(shortopt || longopt); \
421 i = fprintf(fp, "%7s%c %s%s%s%s%s", prefix?prefix:"", prefix?':':' ', \
422 shortopt?shortopt:"", shortopt&&longopt?", ":"", longopt?longopt:"", \
423 optarg?" ":"", optarg?optarg:""); \
424 fprintf(fp, "%*s %s\n", i<32?32-i:0,"",s); HELPRESET;
425
426 HELPRESET;
427 OPTHELP(HELPNOVAL, HELPVAL, HELPSHORT, HELPLONG,
428 HELPPFX, HELPARG, HELPLINE, HELPOPT);
429
430 #undef HELPRESET
431 #undef HELPNOVAL
432 #undef HELPVAL
433 #undef HELPSHORT
434 #undef HELPLONG
435 #undef HELPPFX
436 #undef HELPARG
437 #undef HELPLINE
438 #undef HELPOPT
439 }
440
441 static time_t parse_age(time_t now, const char *agestr)
442 {
443 time_t t;
444 struct tm tm;
445 int nunits;
446 char unit[2];
447
448 t = now;
449
450 if (2 != sscanf(agestr, "%d%1[DdWwMmYy]", &nunits, unit)) {
451 fprintf(stderr, "%s: age specification should be a number followed by"
452 " one of d,w,m,y\n", PNAME);
453 exit(1);
454 }
455
456 if (unit[0] == 'd') {
457 t -= 86400 * nunits;
458 } else if (unit[0] == 'w') {
459 t -= 86400 * 7 * nunits;
460 } else {
461 int ym;
462
463 tm = *localtime(&t);
464 ym = tm.tm_year * 12 + tm.tm_mon;
465
466 if (unit[0] == 'm')
467 ym -= nunits;
468 else
469 ym -= 12 * nunits;
470
471 tm.tm_year = ym / 12;
472 tm.tm_mon = ym % 12;
473
474 t = mktime(&tm);
475 }
476
477 return t;
478 }
479
480 int main(int argc, char **argv)
481 {
482 int fd, count;
483 struct ctx actx, *ctx = &actx;
484 struct stat st;
485 off_t totalsize, realsize;
486 void *mappedfile;
487 triewalk *tw;
488 indexbuild *ib;
489 const struct trie_file *tf, *prevtf;
490 char *filename = PNAME ".dat";
491 int doing_opts = 1;
492 enum { TEXT, HTML, SCAN, DUMP, SCANDUMP, LOAD, HTTPD, REMOVE };
493 struct action {
494 int mode;
495 char *arg;
496 } *actions = NULL;
497 int nactions = 0, actionsize = 0, action;
498 time_t now = time(NULL);
499 time_t textcutoff = now, htmlnewest = now, htmloldest = now;
500 int htmlautoagerange = 1;
501 const char *httpserveraddr = NULL;
502 int httpserverport = 0;
503 const char *httpauthdata = NULL;
504 const char *outfile = NULL;
505 int auth = HTTPD_AUTH_MAGIC | HTTPD_AUTH_BASIC;
506 int progress = 1;
507 struct inclusion_exclusion *inex = NULL;
508 int ninex = 0, inexsize = 0;
509 int crossfs = 0;
510 int depth = -1, gotdepth = 0;
511 int fakediratimes = 1;
512 int mtime = 0;
513 int showfiles = 0;
514
515 #ifdef DEBUG_MAD_OPTION_PARSING_MACROS
516 {
517 static const char *const optnames[NOPTIONS] = {
518 OPTIONS(STRING,STRING,IGNORE,IGNORE)
519 };
520 int i;
521 for (i = 0; i < NSHORTOPTS; i++)
522 printf("-%c == %s [%s]\n", shortopts[i], optnames[shortvals[i]],
523 opthasval[shortvals[i]] ? "value" : "no value");
524 for (i = 0; i < NLONGOPTS; i++)
525 printf("--%s == %s [%s]\n", longopts[i], optnames[longvals[i]],
526 opthasval[longvals[i]] ? "value" : "no value");
527 }
528 #endif
529
530 while (--argc > 0) {
531 char *p = *++argv;
532
533 if (doing_opts && *p == '-') {
534 int wordstart = 1;
535
536 if (!strcmp(p, "--")) {
537 doing_opts = 0;
538 continue;
539 }
540
541 p++;
542 while (*p) {
543 int optid = -1;
544 int i;
545 char *optval;
546
547 if (wordstart && *p == '-') {
548 /*
549 * GNU-style long option.
550 */
551 p++;
552 optval = strchr(p, '=');
553 if (optval)
554 *optval++ = '\0';
555
556 for (i = 0; i < NLONGOPTS; i++) {
557 const char *opt = longopts[i], *s = p;
558 int match = 1;
559 /*
560 * The underscores in the option names
561 * defined above may be given by the user
562 * as underscores or dashes, or omitted
563 * entirely.
564 */
565 while (*opt) {
566 if (*opt == '_') {
567 if (*s == '-' || *s == '_')
568 s++;
569 } else {
570 if (*opt != *s) {
571 match = 0;
572 break;
573 }
574 s++;
575 }
576 opt++;
577 }
578 if (match && !*s) {
579 optid = longvals[i];
580 break;
581 }
582 }
583
584 if (optid < 0) {
585 fprintf(stderr, "%s: unrecognised option '--%s'\n",
586 PNAME, p);
587 return 1;
588 }
589
590 if (!opthasval[optid]) {
591 if (optval) {
592 fprintf(stderr, "%s: unexpected argument to option"
593 " '--%s'\n", PNAME, p);
594 return 1;
595 }
596 } else {
597 if (!optval) {
598 if (--argc > 0) {
599 optval = *++argv;
600 } else {
601 fprintf(stderr, "%s: option '--%s' expects"
602 " an argument\n", PNAME, p);
603 return 1;
604 }
605 }
606 }
607
608 p += strlen(p); /* finished with this argument word */
609 } else {
610 /*
611 * Short option.
612 */
613 char c = *p++;
614
615 for (i = 0; i < NSHORTOPTS; i++)
616 if (c == shortopts[i]) {
617 optid = shortvals[i];
618 break;
619 }
620
621 if (optid < 0) {
622 fprintf(stderr, "%s: unrecognised option '-%c'\n",
623 PNAME, c);
624 return 1;
625 }
626
627 if (opthasval[optid]) {
628 if (*p) {
629 optval = p;
630 p += strlen(p);
631 } else if (--argc > 0) {
632 optval = *++argv;
633 } else {
634 fprintf(stderr, "%s: option '-%c' expects"
635 " an argument\n", PNAME, c);
636 return 1;
637 }
638 } else {
639 optval = NULL;
640 }
641 }
642
643 wordstart = 0;
644
645 /*
646 * Now actually process the option.
647 */
648 switch (optid) {
649 case OPT_HELP:
650 usage(stdout);
651 return 0;
652 case OPT_VERSION:
653 #ifdef PACKAGE_VERSION
654 printf("%s, revision %s\n", PNAME, PACKAGE_VERSION);
655 #else
656 printf("%s: version number not available when not built"
657 " via automake\n", PNAME);
658 #endif
659 return 0;
660 case OPT_LICENCE:
661 {
662 extern const char *const licence[];
663 int i;
664
665 for (i = 0; licence[i]; i++)
666 fputs(licence[i], stdout);
667
668 return 0;
669 }
670 return 0;
671 case OPT_SCAN:
672 if (nactions >= actionsize) {
673 actionsize = nactions * 3 / 2 + 16;
674 actions = sresize(actions, actionsize, struct action);
675 }
676 actions[nactions].mode = SCAN;
677 actions[nactions].arg = optval;
678 nactions++;
679 break;
680 case OPT_SCANDUMP:
681 if (nactions >= actionsize) {
682 actionsize = nactions * 3 / 2 + 16;
683 actions = sresize(actions, actionsize, struct action);
684 }
685 actions[nactions].mode = SCANDUMP;
686 actions[nactions].arg = optval;
687 nactions++;
688 break;
689 case OPT_DUMP:
690 if (nactions >= actionsize) {
691 actionsize = nactions * 3 / 2 + 16;
692 actions = sresize(actions, actionsize, struct action);
693 }
694 actions[nactions].mode = DUMP;
695 actions[nactions].arg = NULL;
696 nactions++;
697 break;
698 case OPT_LOAD:
699 if (nactions >= actionsize) {
700 actionsize = nactions * 3 / 2 + 16;
701 actions = sresize(actions, actionsize, struct action);
702 }
703 actions[nactions].mode = LOAD;
704 actions[nactions].arg = NULL;
705 nactions++;
706 break;
707 case OPT_TEXT:
708 if (nactions >= actionsize) {
709 actionsize = nactions * 3 / 2 + 16;
710 actions = sresize(actions, actionsize, struct action);
711 }
712 actions[nactions].mode = TEXT;
713 actions[nactions].arg = optval;
714 nactions++;
715 break;
716 case OPT_HTML:
717 case OPT_CGI:
718 if (nactions >= actionsize) {
719 actionsize = nactions * 3 / 2 + 16;
720 actions = sresize(actions, actionsize, struct action);
721 }
722 actions[nactions].mode = HTML;
723 actions[nactions].arg = (optid == OPT_HTML ? optval :
724 NULL);
725 nactions++;
726 break;
727 case OPT_HTTPD:
728 if (nactions >= actionsize) {
729 actionsize = nactions * 3 / 2 + 16;
730 actions = sresize(actions, actionsize, struct action);
731 }
732 actions[nactions].mode = HTTPD;
733 actions[nactions].arg = NULL;
734 nactions++;
735 break;
736 case OPT_REMOVE:
737 if (nactions >= actionsize) {
738 actionsize = nactions * 3 / 2 + 16;
739 actions = sresize(actions, actionsize, struct action);
740 }
741 actions[nactions].mode = REMOVE;
742 actions[nactions].arg = NULL;
743 nactions++;
744 break;
745 case OPT_PROGRESS:
746 progress = 2;
747 break;
748 case OPT_NOPROGRESS:
749 progress = 0;
750 break;
751 case OPT_TTYPROGRESS:
752 progress = 1;
753 break;
754 case OPT_CROSSFS:
755 crossfs = 1;
756 break;
757 case OPT_NOCROSSFS:
758 crossfs = 0;
759 break;
760 case OPT_DIRATIME:
761 fakediratimes = 0;
762 break;
763 case OPT_NODIRATIME:
764 fakediratimes = 1;
765 break;
766 case OPT_SHOWFILES:
767 showfiles = 1;
768 break;
769 case OPT_MTIME:
770 mtime = 1;
771 break;
772 case OPT_DATAFILE:
773 filename = optval;
774 break;
775 case OPT_DEPTH:
776 if (!strcasecmp(optval, "unlimited") ||
777 !strcasecmp(optval, "infinity") ||
778 !strcasecmp(optval, "infinite") ||
779 !strcasecmp(optval, "inf") ||
780 !strcasecmp(optval, "maximum") ||
781 !strcasecmp(optval, "max"))
782 depth = -1;
783 else
784 depth = atoi(optval);
785 gotdepth = 1;
786 break;
787 case OPT_OUTFILE:
788 outfile = optval;
789 break;
790 case OPT_MINAGE:
791 textcutoff = parse_age(now, optval);
792 break;
793 case OPT_AGERANGE:
794 if (!strcmp(optval, "auto")) {
795 htmlautoagerange = 1;
796 } else {
797 char *q = optval + strcspn(optval, "-:");
798 if (*q)
799 *q++ = '\0';
800 htmloldest = parse_age(now, optval);
801 htmlnewest = *q ? parse_age(now, q) : now;
802 htmlautoagerange = 0;
803 }
804 break;
805 case OPT_SERVERADDR:
806 {
807 char *port;
808 if (optval[0] == '[' &&
809 (port = strchr(optval, ']')) != NULL)
810 port++;
811 else
812 port = optval;
813 port += strcspn(port, ":");
814 if (port)
815 *port++ = '\0';
816 httpserveraddr = optval;
817 httpserverport = atoi(port);
818 }
819 break;
820 case OPT_AUTH:
821 if (!strcmp(optval, "magic"))
822 auth = HTTPD_AUTH_MAGIC;
823 else if (!strcmp(optval, "basic"))
824 auth = HTTPD_AUTH_BASIC;
825 else if (!strcmp(optval, "none"))
826 auth = HTTPD_AUTH_NONE;
827 else if (!strcmp(optval, "default"))
828 auth = HTTPD_AUTH_MAGIC | HTTPD_AUTH_BASIC;
829 else if (!strcmp(optval, "help") ||
830 !strcmp(optval, "list")) {
831 printf(PNAME ": supported HTTP authentication types"
832 " are:\n"
833 " magic use Linux /proc/net/tcp to"
834 " determine owner of peer socket\n"
835 " basic HTTP Basic username and"
836 " password authentication\n"
837 " default use 'magic' if possible, "
838 " otherwise fall back to 'basic'\n"
839 " none unauthenticated HTTP (if"
840 " the data file is non-confidential)\n");
841 return 0;
842 } else {
843 fprintf(stderr, "%s: unrecognised authentication"
844 " type '%s'\n%*s options are 'magic',"
845 " 'basic', 'none', 'default'\n",
846 PNAME, optval, (int)strlen(PNAME), "");
847 return 1;
848 }
849 break;
850 case OPT_AUTHFILE:
851 case OPT_AUTHFD:
852 {
853 int fd;
854 char namebuf[40];
855 const char *name;
856 char *authbuf;
857 int authlen, authsize;
858 int ret;
859
860 if (optid == OPT_AUTHFILE) {
861 fd = open(optval, O_RDONLY);
862 if (fd < 0) {
863 fprintf(stderr, "%s: %s: open: %s\n", PNAME,
864 optval, strerror(errno));
865 return 1;
866 }
867 name = optval;
868 } else {
869 fd = atoi(optval);
870 name = namebuf;
871 sprintf(namebuf, "fd %d", fd);
872 }
873
874 authlen = 0;
875 authsize = 256;
876 authbuf = snewn(authsize, char);
877 while ((ret = read(fd, authbuf+authlen,
878 authsize-authlen)) > 0) {
879 authlen += ret;
880 if ((authsize - authlen) < (authsize / 16)) {
881 authsize = authlen * 3 / 2 + 4096;
882 authbuf = sresize(authbuf, authsize, char);
883 }
884 }
885 if (ret < 0) {
886 fprintf(stderr, "%s: %s: read: %s\n", PNAME,
887 name, strerror(errno));
888 return 1;
889 }
890 if (optid == OPT_AUTHFILE)
891 close(fd);
892 httpauthdata = authbuf;
893 }
894 break;
895 case OPT_INCLUDE:
896 case OPT_INCLUDEPATH:
897 case OPT_EXCLUDE:
898 case OPT_EXCLUDEPATH:
899 case OPT_PRUNE:
900 case OPT_PRUNEPATH:
901 if (ninex >= inexsize) {
902 inexsize = ninex * 3 / 2 + 16;
903 inex = sresize(inex, inexsize,
904 struct inclusion_exclusion);
905 }
906 inex[ninex].path = (optid == OPT_INCLUDEPATH ||
907 optid == OPT_EXCLUDEPATH ||
908 optid == OPT_PRUNEPATH);
909 inex[ninex].type = (optid == OPT_INCLUDE ? 1 :
910 optid == OPT_INCLUDEPATH ? 1 :
911 optid == OPT_EXCLUDE ? 0 :
912 optid == OPT_EXCLUDEPATH ? 0 :
913 optid == OPT_PRUNE ? -1 :
914 /* optid == OPT_PRUNEPATH ? */ -1);
915 inex[ninex].wildcard = optval;
916 ninex++;
917 break;
918 }
919 }
920 } else {
921 fprintf(stderr, "%s: unexpected argument '%s'\n", PNAME, p);
922 return 1;
923 }
924 }
925
926 if (nactions == 0) {
927 usage(stderr);
928 return 1;
929 }
930
931 for (action = 0; action < nactions; action++) {
932 int mode = actions[action].mode;
933
934 if (mode == SCAN || mode == SCANDUMP || mode == LOAD) {
935 const char *scandir = actions[action].arg;
936
937 if (mode == LOAD) {
938 char *buf = fgetline(stdin);
939 unsigned newpathsep;
940 buf[strcspn(buf, "\r\n")] = '\0';
941 if (1 != sscanf(buf, DUMPHDR "%x",
942 &newpathsep)) {
943 fprintf(stderr, "%s: header in dump file not recognised\n",
944 PNAME);
945 return 1;
946 }
947 pathsep = (char)newpathsep;
948 sfree(buf);
949 }
950
951 if (mode == SCAN || mode == LOAD) {
952 /*
953 * Prepare to write out the index file.
954 */
955 fd = open(filename, O_RDWR | O_TRUNC | O_CREAT,
956 S_IRUSR | S_IWUSR);
957 if (fd < 0) {
958 fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename,
959 strerror(errno));
960 return 1;
961 }
962 if (fstat(fd, &st) < 0) {
963 perror(PNAME ": fstat");
964 return 1;
965 }
966 ctx->datafile_dev = st.st_dev;
967 ctx->datafile_ino = st.st_ino;
968 ctx->straight_to_dump = 0;
969 } else {
970 ctx->datafile_dev = -1;
971 ctx->datafile_ino = -1;
972 ctx->straight_to_dump = 1;
973 }
974
975 if (mode == SCAN || mode == SCANDUMP) {
976 if (stat(scandir, &st) < 0) {
977 fprintf(stderr, "%s: %s: stat: %s\n", PNAME, scandir,
978 strerror(errno));
979 return 1;
980 }
981 ctx->filesystem_dev = crossfs ? 0 : st.st_dev;
982 }
983
984 ctx->inex = inex;
985 ctx->ninex = ninex;
986 ctx->crossfs = crossfs;
987 ctx->fakeatimes = fakediratimes;
988 ctx->usemtime = mtime;
989
990 ctx->last_output_update = time(NULL);
991
992 /* progress==1 means report progress only if stderr is a tty */
993 if (progress == 1)
994 progress = isatty(2) ? 2 : 0;
995 ctx->progress = progress;
996 {
997 struct winsize ws;
998 if (progress &&
999 ioctl(2, TIOCGWINSZ, &ws) == 0 &&
1000 ws.ws_col > 0)
1001 ctx->progwidth = ws.ws_col - 1;
1002 else
1003 ctx->progwidth = 79;
1004 }
1005
1006 if (mode == SCANDUMP)
1007 printf(DUMPHDR "%02x\n", (unsigned char)pathsep);
1008
1009 /*
1010 * Scan the directory tree, and write out the trie component
1011 * of the data file.
1012 */
1013 if (mode != SCANDUMP) {
1014 ctx->tb = triebuild_new(fd);
1015 }
1016 if (mode == LOAD) {
1017 char *buf;
1018 int line = 2;
1019 while ((buf = fgetline(stdin)) != NULL) {
1020 struct trie_file tf;
1021 char *p, *q;
1022
1023 buf[strcspn(buf, "\r\n")] = '\0';
1024
1025 p = buf;
1026 q = p;
1027 while (*p && *p != ' ') p++;
1028 if (!*p) {
1029 fprintf(stderr, "%s: dump file line %d: expected at least"
1030 " three fields\n", PNAME, line);
1031 return 1;
1032 }
1033 *p++ = '\0';
1034 tf.size = strtoull(q, NULL, 10);
1035 q = p;
1036 while (*p && *p != ' ') p++;
1037 if (!*p) {
1038 fprintf(stderr, "%s: dump file line %d: expected at least"
1039 " three fields\n", PNAME, line);
1040 return 1;
1041 }
1042 *p++ = '\0';
1043 tf.atime = strtoull(q, NULL, 10);
1044 q = buf;
1045 while (*p) {
1046 int c = *p;
1047 if (*p == '%') {
1048 int i;
1049 p++;
1050 c = 0;
1051 for (i = 0; i < 2; i++) {
1052 c *= 16;
1053 if (*p >= '0' && *p <= '9')
1054 c += *p - '0';
1055 else if (*p >= 'A' && *p <= 'F')
1056 c += *p - ('A' - 10);
1057 else if (*p >= 'a' && *p <= 'f')
1058 c += *p - ('a' - 10);
1059 else {
1060 fprintf(stderr, "%s: dump file line %d: unable"
1061 " to parse hex escape\n", PNAME, line);
1062 }
1063 p++;
1064 }
1065 } else {
1066 p++;
1067 }
1068 *q++ = c;
1069 }
1070 *q = '\0';
1071 triebuild_add(ctx->tb, buf, &tf);
1072 sfree(buf);
1073 line++;
1074 }
1075 } else {
1076 du(scandir, gotdata, scan_error, ctx);
1077 }
1078 if (mode != SCANDUMP) {
1079 size_t maxpathlen;
1080 size_t delta;
1081 char *buf, *prevbuf;
1082
1083 count = triebuild_finish(ctx->tb);
1084 triebuild_free(ctx->tb);
1085
1086 if (ctx->progress) {
1087 fprintf(stderr, "%-*s\r", ctx->progwidth, "");
1088 fflush(stderr);
1089 }
1090
1091 /*
1092 * Work out how much space the cumulative index trees
1093 * will take; enlarge the file, and memory-map it.
1094 */
1095 if (fstat(fd, &st) < 0) {
1096 perror(PNAME ": fstat");
1097 return 1;
1098 }
1099
1100 printf("Built pathname index, %d entries,"
1101 " %llu bytes of index\n", count,
1102 (unsigned long long)st.st_size);
1103
1104 totalsize = index_initial_size(st.st_size, count);
1105 totalsize += totalsize / 10;
1106
1107 if (lseek(fd, totalsize-1, SEEK_SET) < 0) {
1108 perror(PNAME ": lseek");
1109 return 1;
1110 }
1111 if (write(fd, "\0", 1) < 1) {
1112 perror(PNAME ": write");
1113 return 1;
1114 }
1115
1116 mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0);
1117 if (!mappedfile) {
1118 perror(PNAME ": mmap");
1119 return 1;
1120 }
1121
1122 if (fakediratimes) {
1123 printf("Faking directory atimes\n");
1124 trie_fake_dir_atimes(mappedfile);
1125 }
1126
1127 printf("Building index\n");
1128 ib = indexbuild_new(mappedfile, st.st_size, count, &delta);
1129 maxpathlen = trie_maxpathlen(mappedfile);
1130 buf = snewn(maxpathlen, char);
1131 prevbuf = snewn(maxpathlen, char);
1132 tw = triewalk_new(mappedfile);
1133 prevbuf[0] = '\0';
1134 tf = triewalk_next(tw, buf);
1135 assert(tf);
1136 while (1) {
1137 int i;
1138
1139 if (totalsize - indexbuild_realsize(ib) < delta) {
1140 const void *oldfile = mappedfile;
1141 ptrdiff_t diff;
1142
1143 /*
1144 * Unmap the file, grow it, and remap it.
1145 */
1146 munmap(mappedfile, totalsize);
1147
1148 totalsize += delta;
1149 totalsize += totalsize / 10;
1150
1151 if (lseek(fd, totalsize-1, SEEK_SET) < 0) {
1152 perror(PNAME ": lseek");
1153 return 1;
1154 }
1155 if (write(fd, "\0", 1) < 1) {
1156 perror(PNAME ": write");
1157 return 1;
1158 }
1159
1160 mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0);
1161 if (!mappedfile) {
1162 perror(PNAME ": mmap");
1163 return 1;
1164 }
1165
1166 indexbuild_rebase(ib, mappedfile);
1167 triewalk_rebase(tw, mappedfile);
1168 diff = (const unsigned char *)mappedfile -
1169 (const unsigned char *)oldfile;
1170 if (prevtf)
1171 prevtf = (const struct trie_file *)
1172 (((const unsigned char *)prevtf) + diff);
1173 if (tf)
1174 tf = (const struct trie_file *)
1175 (((const unsigned char *)tf) + diff);
1176 }
1177
1178 /*
1179 * Get the next file from the index. So we are
1180 * currently holding, and have not yet
1181 * indexed, prevtf (with pathname prevbuf) and
1182 * tf (with pathname buf).
1183 */
1184 prevtf = tf;
1185 memcpy(prevbuf, buf, maxpathlen);
1186 tf = triewalk_next(tw, buf);
1187
1188 if (!tf)
1189 buf[0] = '\0';
1190
1191 /*
1192 * Find the first differing character position
1193 * between our two pathnames.
1194 */
1195 for (i = 0; prevbuf[i] && prevbuf[i] == buf[i]; i++);
1196
1197 /*
1198 * If prevbuf was a directory name and buf is
1199 * something inside that directory, then
1200 * trie_before() will be called on prevbuf
1201 * itself. Hence we must drop a tag before it,
1202 * so that the resulting index is usable.
1203 */
1204 if ((!prevbuf[i] && (buf[i] == pathsep ||
1205 (i > 0 && buf[i-1] == pathsep))))
1206 indexbuild_tag(ib);
1207
1208 /*
1209 * Add prevtf to the index.
1210 */
1211 indexbuild_add(ib, prevtf);
1212
1213 if (!tf) {
1214 /*
1215 * Drop an unconditional final tag, and
1216 * get out of this loop.
1217 */
1218 indexbuild_tag(ib);
1219 break;
1220 }
1221
1222 /*
1223 * If prevbuf was a filename inside some
1224 * directory which buf is outside, then
1225 * trie_before() will be called on some
1226 * pathname either equal to buf or epsilon
1227 * less than it. Either way, we're going to
1228 * need to drop a tag after prevtf.
1229 */
1230 if (strchr(prevbuf+i, pathsep) || !tf)
1231 indexbuild_tag(ib);
1232 }
1233
1234 triewalk_free(tw);
1235 realsize = indexbuild_realsize(ib);
1236 indexbuild_free(ib);
1237
1238 munmap(mappedfile, totalsize);
1239 ftruncate(fd, realsize);
1240 close(fd);
1241 printf("Final index file size = %llu bytes\n",
1242 (unsigned long long)realsize);
1243 }
1244 } else if (mode == TEXT) {
1245 char *querydir = actions[action].arg;
1246 size_t pathlen;
1247
1248 fd = open(filename, O_RDONLY);
1249 if (fd < 0) {
1250 fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename,
1251 strerror(errno));
1252 return 1;
1253 }
1254 if (fstat(fd, &st) < 0) {
1255 perror(PNAME ": fstat");
1256 return 1;
1257 }
1258 totalsize = st.st_size;
1259 mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0);
1260 if (!mappedfile) {
1261 perror(PNAME ": mmap");
1262 return 1;
1263 }
1264 pathsep = trie_pathsep(mappedfile);
1265
1266 /*
1267 * Trim trailing slash, just in case.
1268 */
1269 pathlen = strlen(querydir);
1270 if (pathlen > 0 && querydir[pathlen-1] == pathsep)
1271 querydir[--pathlen] = '\0';
1272
1273 if (!gotdepth)
1274 depth = 1; /* default for text mode */
1275 if (outfile != NULL) {
1276 FILE *fp = fopen(outfile, "w");
1277 if (!fp) {
1278 fprintf(stderr, "%s: %s: open: %s\n", PNAME,
1279 outfile, strerror(errno));
1280 return 1;
1281 }
1282 text_query(mappedfile, querydir, textcutoff, showfiles,
1283 depth, fp);
1284 fclose(fp);
1285 } else {
1286 text_query(mappedfile, querydir, textcutoff, showfiles,
1287 depth, stdout);
1288 }
1289
1290 munmap(mappedfile, totalsize);
1291 } else if (mode == HTML) {
1292 char *querydir = actions[action].arg;
1293 size_t pathlen, maxpathlen;
1294 char *pathbuf;
1295 struct html_config cfg;
1296 unsigned long xi;
1297 char *html;
1298
1299 fd = open(filename, O_RDONLY);
1300 if (fd < 0) {
1301 fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename,
1302 strerror(errno));
1303 if (!querydir) {
1304 printf("Status: 500\nContent-type: text/html\n\n"
1305 "<html><head>"
1306 "<title>500 Internal Server Error</title>"
1307 "</head><body>"
1308 "<h1>500 Internal Server Error</h1>"
1309 "<p><code>agedu</code> suffered an internal error."
1310 "</body></html>\n");
1311 return 0;
1312 }
1313 return 1;
1314 }
1315 if (fstat(fd, &st) < 0) {
1316 fprintf(stderr, "%s: %s: fstat: %s\n", PNAME, filename,
1317 strerror(errno));
1318 if (!querydir) {
1319 printf("Status: 500\nContent-type: text/html\n\n"
1320 "<html><head>"
1321 "<title>500 Internal Server Error</title>"
1322 "</head><body>"
1323 "<h1>500 Internal Server Error</h1>"
1324 "<p><code>agedu</code> suffered an internal error."
1325 "</body></html>\n");
1326 return 0;
1327 }
1328 return 1;
1329 }
1330 totalsize = st.st_size;
1331 mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0);
1332 if (!mappedfile) {
1333 fprintf(stderr, "%s: %s: mmap: %s\n", PNAME, filename,
1334 strerror(errno));
1335 if (!querydir) {
1336 printf("Status: 500\nContent-type: text/html\n\n"
1337 "<html><head>"
1338 "<title>500 Internal Server Error</title>"
1339 "</head><body>"
1340 "<h1>500 Internal Server Error</h1>"
1341 "<p><code>agedu</code> suffered an internal error."
1342 "</body></html>\n");
1343 return 0;
1344 }
1345 return 1;
1346 }
1347 pathsep = trie_pathsep(mappedfile);
1348
1349 maxpathlen = trie_maxpathlen(mappedfile);
1350 pathbuf = snewn(maxpathlen, char);
1351
1352 if (!querydir) {
1353 /*
1354 * If we're run in --cgi mode, read PATH_INFO to get
1355 * a numeric pathname index.
1356 */
1357 char *path_info = getenv("PATH_INFO");
1358
1359 if (!path_info)
1360 path_info = "";
1361
1362 /*
1363 * Because we need relative links to go to the
1364 * right place, it's important that our
1365 * PATH_INFO should contain a slash right at the
1366 * start, and no slashes anywhere else.
1367 */
1368 if (path_info[0] != '/') {
1369 char *servername = getenv("SERVER_NAME");
1370 char *scriptname = getenv("SCRIPT_NAME");
1371 if (!servername || !scriptname) {
1372 if (servername)
1373 fprintf(stderr, "%s: SCRIPT_NAME unset\n", PNAME);
1374 else if (scriptname)
1375 fprintf(stderr, "%s: SCRIPT_NAME unset\n", PNAME);
1376 else
1377 fprintf(stderr, "%s: SERVER_NAME and "
1378 "SCRIPT_NAME both unset\n", PNAME);
1379 printf("Status: 500\nContent-type: text/html\n\n"
1380 "<html><head>"
1381 "<title>500 Internal Server Error</title>"
1382 "</head><body>"
1383 "<h1>500 Internal Server Error</h1>"
1384 "<p><code>agedu</code> suffered an internal "
1385 "error."
1386 "</body></html>\n");
1387 return 0;
1388 }
1389 printf("Status: 301\n"
1390 "Location: http://%s/%s/\n"
1391 "Content-type: text/html\n\n"
1392 "<html><head>"
1393 "<title>301 Moved</title>"
1394 "</head><body>"
1395 "<h1>301 Moved</h1>"
1396 "<p>Moved."
1397 "</body></html>\n",
1398 servername, scriptname);
1399 return 0;
1400 } else if (strchr(path_info+1, '/')) {
1401 printf("Status: 404\nContent-type: text/html\n\n"
1402 "<html><head>"
1403 "<title>404 Not Found</title>"
1404 "</head><body>"
1405 "<h1>400 Not Found</h1>"
1406 "<p>Invalid <code>agedu</code> pathname."
1407 "</body></html>\n");
1408 return 0;
1409 }
1410 xi = atoi(path_info + 1);
1411
1412 if (xi >= trie_count(mappedfile)) {
1413 printf("Status: 404\nContent-type: text/html\n\n"
1414 "<html><head>"
1415 "<title>404 Not Found</title>"
1416 "</head><body>"
1417 "<h1>400 Not Found</h1>"
1418 "<p>This is not a valid pathname index."
1419 "</body></html>\n");
1420 return 0;
1421 } else if (!index_has_root(mappedfile, xi)) {
1422 printf("Status: 404\nContent-type: text/html\n\n"
1423 "<html><head>"
1424 "<title>404 Not Found</title>"
1425 "</head><body>"
1426 "<h1>404 Not Found</h1>"
1427 "<p>Pathname index out of range."
1428 "</body></html>\n");
1429 return 0;
1430 }
1431 } else {
1432 /*
1433 * In ordinary --html mode, process a query
1434 * directory passed in on the command line.
1435 */
1436
1437 /*
1438 * Trim trailing slash, just in case.
1439 */
1440 pathlen = strlen(querydir);
1441 if (pathlen > 0 && querydir[pathlen-1] == pathsep)
1442 querydir[--pathlen] = '\0';
1443
1444 xi = trie_before(mappedfile, querydir);
1445 if (xi >= trie_count(mappedfile) ||
1446 (trie_getpath(mappedfile, xi, pathbuf),
1447 strcmp(pathbuf, querydir))) {
1448 fprintf(stderr, "%s: pathname '%s' does not exist in index\n"
1449 "%*s(check it is spelled exactly as it is in the "
1450 "index, including\n%*sany leading './')\n",
1451 PNAME, querydir,
1452 (int)(1+sizeof(PNAME)), "",
1453 (int)(1+sizeof(PNAME)), "");
1454 return 1;
1455 } else if (!index_has_root(mappedfile, xi)) {
1456 fprintf(stderr, "%s: pathname '%s' is"
1457 " a file, not a directory\n", PNAME, querydir);
1458 return 1;
1459 }
1460 }
1461
1462 if (!querydir || !gotdepth) {
1463 /*
1464 * Single output file.
1465 */
1466 if (!querydir) {
1467 cfg.format = "%.0lu"; /* use crosslinks in --cgi mode */
1468 } else {
1469 cfg.format = NULL;
1470 }
1471 cfg.rootpage = NULL;
1472 cfg.autoage = htmlautoagerange;
1473 cfg.oldest = htmloldest;
1474 cfg.newest = htmlnewest;
1475 cfg.showfiles = showfiles;
1476 html = html_query(mappedfile, xi, &cfg, 1);
1477 if (querydir && outfile != NULL) {
1478 FILE *fp = fopen(outfile, "w");
1479 if (!fp) {
1480 fprintf(stderr, "%s: %s: open: %s\n", PNAME,
1481 outfile, strerror(errno));
1482 return 1;
1483 } else if (fputs(html, fp) < 0) {
1484 fprintf(stderr, "%s: %s: write: %s\n", PNAME,
1485 outfile, strerror(errno));
1486 fclose(fp);
1487 return 1;
1488 } else if (fclose(fp) < 0) {
1489 fprintf(stderr, "%s: %s: fclose: %s\n", PNAME,
1490 outfile, strerror(errno));
1491 return 1;
1492 }
1493 } else {
1494 if (!querydir) {
1495 printf("Content-type: text/html\n\n");
1496 }
1497 fputs(html, stdout);
1498 }
1499 } else {
1500 /*
1501 * Multiple output files.
1502 */
1503 int dirlen = outfile ? 2+strlen(outfile) : 3;
1504 char prefix[dirlen];
1505 if (outfile) {
1506 if (mkdir(outfile, 0777) < 0 && errno != EEXIST) {
1507 fprintf(stderr, "%s: %s: mkdir: %s\n", PNAME,
1508 outfile, strerror(errno));
1509 return 1;
1510 }
1511 snprintf(prefix, dirlen, "%s/", outfile);
1512 } else
1513 snprintf(prefix, dirlen, "./");
1514
1515 unsigned long xi2;
1516 /*
1517 * pathbuf is only set up in the plain-HTML case and
1518 * not in the CGI case; but that's OK, because the
1519 * CGI case can't come to this branch of the if
1520 * anyway.
1521 */
1522 make_successor(pathbuf);
1523 xi2 = trie_before(mappedfile, pathbuf);
1524
1525 cfg.format = "%lu.html";
1526 cfg.rootpage = "index.html";
1527 cfg.autoage = htmlautoagerange;
1528 cfg.oldest = htmloldest;
1529 cfg.newest = htmlnewest;
1530 cfg.showfiles = showfiles;
1531 if (html_dump(mappedfile, xi, xi2, depth, &cfg, prefix))
1532 return 1;
1533 }
1534
1535 munmap(mappedfile, totalsize);
1536 sfree(pathbuf);
1537 } else if (mode == DUMP) {
1538 size_t maxpathlen;
1539 char *buf;
1540
1541 fd = open(filename, O_RDONLY);
1542 if (fd < 0) {
1543 fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename,
1544 strerror(errno));
1545 return 1;
1546 }
1547 if (fstat(fd, &st) < 0) {
1548 perror(PNAME ": fstat");
1549 return 1;
1550 }
1551 totalsize = st.st_size;
1552 mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0);
1553 if (!mappedfile) {
1554 perror(PNAME ": mmap");
1555 return 1;
1556 }
1557 pathsep = trie_pathsep(mappedfile);
1558
1559 maxpathlen = trie_maxpathlen(mappedfile);
1560 buf = snewn(maxpathlen, char);
1561
1562 printf(DUMPHDR "%02x\n", (unsigned char)pathsep);
1563 tw = triewalk_new(mappedfile);
1564 while ((tf = triewalk_next(tw, buf)) != NULL)
1565 dump_line(buf, tf);
1566 triewalk_free(tw);
1567
1568 munmap(mappedfile, totalsize);
1569 } else if (mode == HTTPD) {
1570 struct html_config pcfg;
1571 struct httpd_config dcfg;
1572
1573 fd = open(filename, O_RDONLY);
1574 if (fd < 0) {
1575 fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename,
1576 strerror(errno));
1577 return 1;
1578 }
1579 if (fstat(fd, &st) < 0) {
1580 perror(PNAME ": fstat");
1581 return 1;
1582 }
1583 totalsize = st.st_size;
1584 mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0);
1585 if (!mappedfile) {
1586 perror(PNAME ": mmap");
1587 return 1;
1588 }
1589 pathsep = trie_pathsep(mappedfile);
1590
1591 dcfg.address = httpserveraddr;
1592 dcfg.port = httpserverport;
1593 dcfg.basicauthdata = httpauthdata;
1594 pcfg.format = NULL;
1595 pcfg.rootpage = NULL;
1596 pcfg.autoage = htmlautoagerange;
1597 pcfg.oldest = htmloldest;
1598 pcfg.newest = htmlnewest;
1599 pcfg.showfiles = showfiles;
1600 run_httpd(mappedfile, auth, &dcfg, &pcfg);
1601 munmap(mappedfile, totalsize);
1602 } else if (mode == REMOVE) {
1603 if (remove(filename) < 0) {
1604 fprintf(stderr, "%s: %s: remove: %s\n", PNAME, filename,
1605 strerror(errno));
1606 return 1;
1607 }
1608 }
1609 }
1610
1611 return 0;
1612 }