~mdw / sgt / agedu / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Tweak the options list a bit: change around some short options, and
[sgt/agedu] / agedu.c
CommitLineData
70322ae3 1/*
2 * Main program for agedu.
3 */
4
5#define _GNU_SOURCE
6#include <stdio.h>
7#include <errno.h>
8#include <stdarg.h>
9#include <stdlib.h>
10#include <stdint.h>
11#include <string.h>
12#include <time.h>
e9e7a1bf 13#include <assert.h>
70322ae3 14
15#include <unistd.h>
16#include <sys/types.h>
17#include <fcntl.h>
18#include <sys/mman.h>
8b1f55d6 19#include <termios.h>
20#include <sys/ioctl.h>
9d0b9596 21#include <fnmatch.h>
70322ae3 22
353bc75d 23#include "agedu.h"
70322ae3 24#include "du.h"
25#include "trie.h"
26#include "index.h"
27#include "malloc.h"
28#include "html.h"
29#include "httpd.h"
84849cbd 30#include "fgetline.h"
70322ae3 31
373a02e5 32/*
33 * Path separator. This global variable affects the behaviour of
34 * various parts of the code when they need to deal with path
35 * separators. The path separator appropriate to a particular data
36 * set is encoded in the index file storing that data set; data
37 * sets generated on Unix will of course have the default '/', but
38 * foreign data sets are conceivable and must be handled correctly.
39 */
40char pathsep = '/';
41
70322ae3 42void fatal(const char *fmt, ...)
43{
44 va_list ap;
45 fprintf(stderr, "%s: ", PNAME);
46 va_start(ap, fmt);
47 vfprintf(stderr, fmt, ap);
48 va_end(ap);
49 fprintf(stderr, "\n");
50 exit(1);
51}
52
9d0b9596 53struct inclusion_exclusion {
0ba55302 54 int type;
9d0b9596 55 const char *wildcard;
56 int path;
57};
58
70322ae3 59struct ctx {
60 triebuild *tb;
61 dev_t datafile_dev, filesystem_dev;
62 ino_t datafile_ino;
63 time_t last_output_update;
8b1f55d6 64 int progress, progwidth;
84849cbd 65 int straight_to_dump;
9d0b9596 66 struct inclusion_exclusion *inex;
67 int ninex;
68 int crossfs;
05b0f827 69 int fakeatimes;
70322ae3 70};
71
84849cbd 72static void dump_line(const char *pathname, const struct trie_file *tf)
73{
74 const char *p;
75 printf("%llu %llu ", tf->size, tf->atime);
76 for (p = pathname; *p; p++) {
77 if (*p >= ' ' && *p < 127 && *p != '%')
78 putchar(*p);
79 else
80 printf("%%%02x", (unsigned char)*p);
81 }
82 putchar('\n');
83}
84
70322ae3 85static int gotdata(void *vctx, const char *pathname, const struct stat64 *st)
86{
87 struct ctx *ctx = (struct ctx *)vctx;
88 struct trie_file file;
89 time_t t;
9d0b9596 90 int i, include;
91 const char *filename;
70322ae3 92
93 /*
94 * Filter out our own data file.
95 */
96 if (st->st_dev == ctx->datafile_dev && st->st_ino == ctx->datafile_ino)
97 return 0;
98
99 /*
100 * Don't cross the streams^W^Wany file system boundary.
70322ae3 101 */
9d0b9596 102 if (!ctx->crossfs && st->st_dev != ctx->filesystem_dev)
70322ae3 103 return 0;
104
84849cbd 105 file.size = (unsigned long long)512 * st->st_blocks;
05b0f827 106 if (ctx->fakeatimes && S_ISDIR(st->st_mode))
107 file.atime = st->st_mtime;
108 else
109 file.atime = st->st_atime;
0ba55302 110
70322ae3 111 /*
9d0b9596 112 * Filter based on wildcards.
70322ae3 113 */
9d0b9596 114 include = 1;
373a02e5 115 filename = strrchr(pathname, pathsep);
9d0b9596 116 if (!filename)
117 filename = pathname;
118 else
119 filename++;
120 for (i = 0; i < ctx->ninex; i++) {
121 if (fnmatch(ctx->inex[i].wildcard,
0ba55302 122 ctx->inex[i].path ? pathname : filename, 0) == 0)
123 include = ctx->inex[i].type;
124 }
125 if (include == -1)
126 return 0; /* ignore this entry and any subdirs */
127 if (include == 0) {
128 /*
129 * Here we are supposed to be filtering an entry out, but
130 * still recursing into it if it's a directory. However,
131 * we can't actually leave out any directory whose
132 * subdirectories we then look at. So we cheat, in that
133 * case, by setting the size to zero.
134 */
135 if (!S_ISDIR(st->st_mode))
136 return 0; /* just ignore */
137 else
84849cbd 138 file.size = 0;
9d0b9596 139 }
70322ae3 140
84849cbd 141 if (ctx->straight_to_dump)
142 dump_line(pathname, &file);
143 else
144 triebuild_add(ctx->tb, pathname, &file);
70322ae3 145
84849cbd 146 if (ctx->progress) {
147 t = time(NULL);
148 if (t != ctx->last_output_update) {
8b1f55d6 149 fprintf(stderr, "%-*.*s\r", ctx->progwidth, ctx->progwidth,
150 pathname);
151 fflush(stderr);
84849cbd 152 ctx->last_output_update = t;
8b1f55d6 153 }
70322ae3 154 }
155
156 return 1;
157}
158
e9e7a1bf 159static void text_query(const void *mappedfile, const char *querydir,
7cf11b75 160 time_t t, int depth)
70322ae3 161{
162 size_t maxpathlen;
163 char *pathbuf;
164 unsigned long xi1, xi2;
165 unsigned long long s1, s2;
166
167 maxpathlen = trie_maxpathlen(mappedfile);
168 pathbuf = snewn(maxpathlen + 1, char);
169
170 /*
171 * We want to query everything between the supplied filename
172 * (inclusive) and that filename with a ^A on the end
173 * (exclusive). So find the x indices for each.
174 */
256c29a2 175 strcpy(pathbuf, querydir);
176 make_successor(pathbuf);
e9e7a1bf 177 xi1 = trie_before(mappedfile, querydir);
70322ae3 178 xi2 = trie_before(mappedfile, pathbuf);
179
0313b788 180 if (xi2 - xi1 == 1)
181 return; /* file, or empty dir => no display */
182
70322ae3 183 /*
184 * Now do the lookups in the age index.
185 */
186 s1 = index_query(mappedfile, xi1, t);
187 s2 = index_query(mappedfile, xi2, t);
188
010dd2a2 189 if (s1 == s2)
190 return; /* no space taken up => no display */
191
70322ae3 192 if (depth > 0) {
193 /*
194 * Now scan for first-level subdirectories and report
195 * those too.
196 */
197 xi1++;
198 while (xi1 < xi2) {
199 trie_getpath(mappedfile, xi1, pathbuf);
7cf11b75 200 text_query(mappedfile, pathbuf, t, depth-1);
256c29a2 201 make_successor(pathbuf);
70322ae3 202 xi1 = trie_before(mappedfile, pathbuf);
203 }
204 }
16e591d6 205
206 /* Display in units of 1Kb */
207 printf("%-11llu %s\n", (s2 - s1) / 1024, querydir);
70322ae3 208}
209
56fa1896 210/*
211 * Largely frivolous way to define all my command-line options. I
212 * present here a parametric macro which declares a series of
213 * _logical_ option identifiers, and for each one declares zero or
214 * more short option characters and zero or more long option
215 * words. Then I repeatedly invoke that macro with its arguments
216 * defined to be various other macros, which allows me to
217 * variously:
218 *
219 * - define an enum allocating a distinct integer value to each
220 * logical option id
221 * - define a string consisting of precisely all the short option
222 * characters
223 * - define a string array consisting of all the long option
224 * strings
225 * - define (with help from auxiliary enums) integer arrays
226 * parallel to both of the above giving the logical option id
227 * for each physical short and long option
228 * - define an array indexed by logical option id indicating
e9e7a1bf 229 * whether the option in question takes a value
230 * - define a function which prints out brief online help for all
231 * the options.
56fa1896 232 *
233 * It's not at all clear to me that this trickery is actually
234 * particularly _efficient_ - it still, after all, requires going
235 * linearly through the option list at run time and doing a
236 * strcmp, whereas in an ideal world I'd have liked the lists of
237 * long and short options to be pre-sorted so that a binary search
238 * or some other more efficient lookup was possible. (Not that
239 * asymptotic algorithmic complexity is remotely vital in option
240 * parsing, but if I were doing this in, say, Lisp or something
241 * with an equivalently powerful preprocessor then once I'd had
242 * the idea of preparing the option-parsing data structures at
243 * compile time I would probably have made the effort to prepare
244 * them _properly_. I could have Perl generate me a source file
245 * from some sort of description, I suppose, but that would seem
246 * like overkill. And in any case, it's more of a challenge to
247 * achieve as much as possible by cunning use of cpp and enum than
248 * to just write some sensible and logical code in a Turing-
249 * complete language. I said it was largely frivolous :-)
250 *
251 * This approach does have the virtue that it brings together the
e9e7a1bf 252 * option ids, option spellings and help text into a single
253 * combined list and defines them all in exactly one place. If I
254 * want to add a new option, or a new spelling for an option, I
255 * only have to modify the main OPTHELP macro below and then add
256 * code to process the new logical id.
56fa1896 257 *
258 * (Though, really, even that isn't ideal, since it still involves
259 * modifying the source file in more than one place. In a
260 * _properly_ ideal world, I'd be able to interleave the option
261 * definitions with the code fragments that process them. And then
262 * not bother defining logical identifiers for them at all - those
263 * would be automatically generated, since I wouldn't have any
264 * need to specify them manually in another part of the code.)
c5c3510f 265 *
266 * One other helpful consequence of the enum-based structure here
267 * is that it causes a compiler error if I accidentally try to
268 * define the same option (short or long) twice.
56fa1896 269 */
270
e9e7a1bf 271#define OPTHELP(NOVAL, VAL, SHORT, LONG, HELPPFX, HELPARG, HELPLINE, HELPOPT) \
bf53e756 272 HELPPFX("usage") HELPLINE(PNAME " [options] action [action...]") \
e9e7a1bf 273 HELPPFX("actions") \
274 VAL(SCAN) SHORT(s) LONG(scan) \
275 HELPARG("directory") HELPOPT("scan and index a directory") \
c5c3510f 276 NOVAL(DUMP) SHORT(D) LONG(dump) HELPOPT("dump the index file on stdout") \
84849cbd 277 VAL(SCANDUMP) SHORT(S) LONG(scan_dump) \
278 HELPARG("directory") HELPOPT("scan only, generating a dump") \
c5c3510f 279 NOVAL(LOAD) SHORT(L) LONG(load) \
84849cbd 280 HELPOPT("load and index a dump file") \
e9e7a1bf 281 VAL(TEXT) SHORT(t) LONG(text) \
282 HELPARG("subdir") HELPOPT("print a plain text report on a subdirectory") \
283 VAL(HTML) SHORT(H) LONG(html) \
284 HELPARG("subdir") HELPOPT("print an HTML report on a subdirectory") \
56fa1896 285 NOVAL(HTTPD) SHORT(w) LONG(web) LONG(server) LONG(httpd) \
84849cbd 286 HELPOPT("serve HTML reports from a temporary web server") \
e9e7a1bf 287 HELPPFX("options") \
288 VAL(DATAFILE) SHORT(f) LONG(file) \
c5c3510f 289 HELPARG("filename") HELPOPT("[most modes] specify index file") \
56fa1896 290 NOVAL(PROGRESS) LONG(progress) LONG(scan_progress) \
e9e7a1bf 291 HELPOPT("[--scan] report progress on stderr") \
56fa1896 292 NOVAL(NOPROGRESS) LONG(no_progress) LONG(no_scan_progress) \
e9e7a1bf 293 HELPOPT("[--scan] do not report progress") \
56fa1896 294 NOVAL(TTYPROGRESS) LONG(tty_progress) LONG(tty_scan_progress) \
295 LONG(progress_tty) LONG(scan_progress_tty) \
e9e7a1bf 296 HELPOPT("[--scan] report progress if stderr is a tty") \
56fa1896 297 NOVAL(CROSSFS) LONG(cross_fs) \
e9e7a1bf 298 HELPOPT("[--scan] cross filesystem boundaries") \
56fa1896 299 NOVAL(NOCROSSFS) LONG(no_cross_fs) \
e9e7a1bf 300 HELPOPT("[--scan] stick to one filesystem") \
56fa1896 301 VAL(INCLUDE) LONG(include) \
e9e7a1bf 302 HELPARG("wildcard") HELPOPT("[--scan] include files matching pattern") \
56fa1896 303 VAL(INCLUDEPATH) LONG(include_path) \
e9e7a1bf 304 HELPARG("wildcard") HELPOPT("[--scan] include pathnames matching pattern") \
56fa1896 305 VAL(EXCLUDE) LONG(exclude) \
e9e7a1bf 306 HELPARG("wildcard") HELPOPT("[--scan] exclude files matching pattern") \
307 VAL(EXCLUDEPATH) LONG(exclude_path) \
308 HELPARG("wildcard") HELPOPT("[--scan] exclude pathnames matching pattern") \
0ba55302 309 VAL(PRUNE) LONG(prune) \
310 HELPARG("wildcard") HELPOPT("[--scan] prune files matching pattern") \
311 VAL(PRUNEPATH) LONG(prune_path) \
312 HELPARG("wildcard") HELPOPT("[--scan] prune pathnames matching pattern") \
05b0f827 313 NOVAL(DIRATIME) LONG(dir_atime) LONG(dir_atimes) \
314 HELPOPT("[--scan] keep real atimes on directories") \
315 NOVAL(NODIRATIME) LONG(no_dir_atime) LONG(no_dir_atimes) \
316 HELPOPT("[--scan] fake atimes on directories") \
c5c3510f 317 VAL(TQDEPTH) SHORT(d) LONG(depth) LONG(max_depth) LONG(maximum_depth) \
16e591d6 318 HELPARG("levels") HELPOPT("[--text] recurse to this many levels") \
e9e7a1bf 319 VAL(MINAGE) SHORT(a) LONG(age) LONG(min_age) LONG(minimum_age) \
320 HELPARG("age") HELPOPT("[--text] include only files older than this") \
f2e52893 321 VAL(AGERANGE) SHORT(r) LONG(age_range) LONG(range) LONG(ages) \
322 HELPARG("age[-age]") HELPOPT("[--html,--web] set limits of colour coding") \
1e8d78b9 323 VAL(SERVERADDR) LONG(address) LONG(addr) LONG(server_address) \
324 LONG(server_addr) \
325 HELPARG("addr[:port]") HELPOPT("[--web] specify HTTP server address") \
e9e7a1bf 326 VAL(AUTH) LONG(auth) LONG(http_auth) LONG(httpd_auth) \
327 LONG(server_auth) LONG(web_auth) \
328 HELPARG("type") HELPOPT("[--web] specify HTTP authentication method") \
1e8d78b9 329 VAL(AUTHFILE) LONG(auth_file) \
330 HELPARG("filename") HELPOPT("[--web] read HTTP Basic user/pass from file") \
331 VAL(AUTHFD) LONG(auth_fd) \
332 HELPARG("fd") HELPOPT("[--web] read HTTP Basic user/pass from fd") \
e9e7a1bf 333 HELPPFX("also") \
334 NOVAL(HELP) SHORT(h) LONG(help) HELPOPT("display this help text") \
335 NOVAL(VERSION) SHORT(V) LONG(version) HELPOPT("report version number") \
336 NOVAL(LICENCE) LONG(licence) LONG(license) \
337 HELPOPT("display (MIT) licence text") \
56fa1896 338
339#define IGNORE(x)
340#define DEFENUM(x) OPT_ ## x,
341#define ZERO(x) 0,
342#define ONE(x) 1,
343#define STRING(x) #x ,
344#define STRINGNOCOMMA(x) #x
345#define SHORTNEWOPT(x) SHORTtmp_ ## x = OPT_ ## x,
346#define SHORTTHISOPT(x) SHORTtmp2_ ## x, SHORTVAL_ ## x = SHORTtmp2_ ## x - 1,
347#define SHORTOPTVAL(x) SHORTVAL_ ## x,
348#define SHORTTMP(x) SHORTtmp3_ ## x,
349#define LONGNEWOPT(x) LONGtmp_ ## x = OPT_ ## x,
350#define LONGTHISOPT(x) LONGtmp2_ ## x, LONGVAL_ ## x = LONGtmp2_ ## x - 1,
351#define LONGOPTVAL(x) LONGVAL_ ## x,
352#define LONGTMP(x) SHORTtmp3_ ## x,
353
e9e7a1bf 354#define OPTIONS(NOVAL, VAL, SHORT, LONG) \
355 OPTHELP(NOVAL, VAL, SHORT, LONG, IGNORE, IGNORE, IGNORE, IGNORE)
356
56fa1896 357enum { OPTIONS(DEFENUM,DEFENUM,IGNORE,IGNORE) NOPTIONS };
358enum { OPTIONS(IGNORE,IGNORE,SHORTTMP,IGNORE) NSHORTOPTS };
359enum { OPTIONS(IGNORE,IGNORE,IGNORE,LONGTMP) NLONGOPTS };
360static const int opthasval[NOPTIONS] = {OPTIONS(ZERO,ONE,IGNORE,IGNORE)};
361static const char shortopts[] = {OPTIONS(IGNORE,IGNORE,STRINGNOCOMMA,IGNORE)};
362static const char *const longopts[] = {OPTIONS(IGNORE,IGNORE,IGNORE,STRING)};
363enum { OPTIONS(SHORTNEWOPT,SHORTNEWOPT,SHORTTHISOPT,IGNORE) };
364enum { OPTIONS(LONGNEWOPT,LONGNEWOPT,IGNORE,LONGTHISOPT) };
365static const int shortvals[] = {OPTIONS(IGNORE,IGNORE,SHORTOPTVAL,IGNORE)};
366static const int longvals[] = {OPTIONS(IGNORE,IGNORE,IGNORE,LONGOPTVAL)};
367
e9e7a1bf 368static void usage(FILE *fp)
369{
370 char longbuf[80];
371 const char *prefix, *shortopt, *longopt, *optarg;
372 int i, optex;
373
374#define HELPRESET prefix = shortopt = longopt = optarg = NULL, optex = -1
375#define HELPNOVAL(s) optex = 0;
376#define HELPVAL(s) optex = 1;
377#define HELPSHORT(s) if (!shortopt) shortopt = "-" #s;
378#define HELPLONG(s) if (!longopt) { \
379 strcpy(longbuf, "--" #s); longopt = longbuf; \
380 for (i = 0; longbuf[i]; i++) if (longbuf[i] == '_') longbuf[i] = '-'; }
381#define HELPPFX(s) prefix = s;
382#define HELPARG(s) optarg = s;
383#define HELPLINE(s) assert(optex == -1); \
384 fprintf(fp, "%7s%c %s\n", prefix?prefix:"", prefix?':':' ', s); \
385 HELPRESET;
386#define HELPOPT(s) assert((optex == 1 && optarg) || (optex == 0 && !optarg)); \
387 assert(shortopt || longopt); \
388 i = fprintf(fp, "%7s%c %s%s%s%s%s", prefix?prefix:"", prefix?':':' ', \
389 shortopt?shortopt:"", shortopt&&longopt?", ":"", longopt?longopt:"", \
390 optarg?" ":"", optarg?optarg:""); \
391 fprintf(fp, "%*s %s\n", i<32?32-i:0,"",s); HELPRESET;
392
393 HELPRESET;
394 OPTHELP(HELPNOVAL, HELPVAL, HELPSHORT, HELPLONG,
395 HELPPFX, HELPARG, HELPLINE, HELPOPT);
396
397#undef HELPRESET
398#undef HELPNOVAL
399#undef HELPVAL
400#undef HELPSHORT
401#undef HELPLONG
402#undef HELPPFX
403#undef HELPARG
404#undef HELPLINE
405#undef HELPOPT
406}
407
f2e52893 408static time_t parse_age(time_t now, const char *agestr)
409{
410 time_t t;
411 struct tm tm;
412 int nunits;
413 char unit[2];
414
415 t = now;
416
417 if (2 != sscanf(agestr, "%d%1[DdWwMmYy]", &nunits, unit)) {
418 fprintf(stderr, "%s: age specification should be a number followed by"
419 " one of d,w,m,y\n", PNAME);
420 exit(1);
421 }
422
423 if (unit[0] == 'd') {
424 t -= 86400 * nunits;
425 } else if (unit[0] == 'w') {
426 t -= 86400 * 7 * nunits;
427 } else {
428 int ym;
429
430 tm = *localtime(&t);
431 ym = tm.tm_year * 12 + tm.tm_mon;
432
433 if (unit[0] == 'm')
434 ym -= nunits;
435 else
436 ym -= 12 * nunits;
437
438 tm.tm_year = ym / 12;
439 tm.tm_mon = ym % 12;
440
441 t = mktime(&tm);
442 }
443
444 return t;
445}
446
70322ae3 447int main(int argc, char **argv)
448{
449 int fd, count;
450 struct ctx actx, *ctx = &actx;
451 struct stat st;
452 off_t totalsize, realsize;
453 void *mappedfile;
454 triewalk *tw;
455 indexbuild *ib;
456 const struct trie_file *tf;
bf53e756 457 char *filename = PNAME ".dat";
70322ae3 458 int doing_opts = 1;
444c684c 459 enum { TEXT, HTML, SCAN, DUMP, SCANDUMP, LOAD, HTTPD };
460 struct action {
461 int mode;
462 char *arg;
463 } *actions = NULL;
464 int nactions = 0, actionsize = 0, action;
f2e52893 465 time_t now = time(NULL);
466 time_t textcutoff = now, htmlnewest = now, htmloldest = now;
467 int htmlautoagerange = 1;
1e8d78b9 468 const char *httpserveraddr = NULL;
469 int httpserverport = 0;
470 const char *httpauthdata = NULL;
812e4bf2 471 int auth = HTTPD_AUTH_MAGIC | HTTPD_AUTH_BASIC;
8b1f55d6 472 int progress = 1;
9d0b9596 473 struct inclusion_exclusion *inex = NULL;
474 int ninex = 0, inexsize = 0;
475 int crossfs = 0;
16e591d6 476 int tqdepth = 1;
05b0f827 477 int fakediratimes = 1;
70322ae3 478
56fa1896 479#ifdef DEBUG_MAD_OPTION_PARSING_MACROS
480 {
481 static const char *const optnames[NOPTIONS] = {
482 OPTIONS(STRING,STRING,IGNORE,IGNORE)
483 };
484 int i;
485 for (i = 0; i < NSHORTOPTS; i++)
486 printf("-%c == %s [%s]\n", shortopts[i], optnames[shortvals[i]],
487 opthasval[shortvals[i]] ? "value" : "no value");
488 for (i = 0; i < NLONGOPTS; i++)
489 printf("--%s == %s [%s]\n", longopts[i], optnames[longvals[i]],
490 opthasval[longvals[i]] ? "value" : "no value");
491 }
492#endif
493
70322ae3 494 while (--argc > 0) {
495 char *p = *++argv;
70322ae3 496
497 if (doing_opts && *p == '-') {
56fa1896 498 int wordstart = 1;
499
70322ae3 500 if (!strcmp(p, "--")) {
501 doing_opts = 0;
56fa1896 502 continue;
503 }
504
505 p++;
506 while (*p) {
507 int optid = -1;
508 int i;
509 char *optval;
510
511 if (wordstart && *p == '-') {
70322ae3 512 /*
56fa1896 513 * GNU-style long option.
70322ae3 514 */
56fa1896 515 p++;
516 optval = strchr(p, '=');
517 if (optval)
518 *optval++ = '\0';
519
520 for (i = 0; i < NLONGOPTS; i++) {
521 const char *opt = longopts[i], *s = p;
522 int match = 1;
523 /*
524 * The underscores in the option names
525 * defined above may be given by the user
526 * as underscores or dashes, or omitted
527 * entirely.
528 */
529 while (*opt) {
530 if (*opt == '_') {
531 if (*s == '-' || *s == '_')
532 s++;
533 } else {
534 if (*opt != *s) {
535 match = 0;
536 break;
537 }
538 s++;
539 }
540 opt++;
541 }
542 if (match && !*s) {
543 optid = longvals[i];
544 break;
70322ae3 545 }
546 }
56fa1896 547
548 if (optid < 0) {
549 fprintf(stderr, "%s: unrecognised option '--%s'\n",
550 PNAME, p);
551 return 1;
552 }
553
554 if (!opthasval[optid]) {
555 if (optval) {
556 fprintf(stderr, "%s: unexpected argument to option"
557 " '--%s'\n", PNAME, p);
812e4bf2 558 return 1;
559 }
56fa1896 560 } else {
561 if (!optval) {
562 if (--argc > 0) {
563 optval = *++argv;
564 } else {
565 fprintf(stderr, "%s: option '--%s' expects"
566 " an argument\n", PNAME, p);
567 return 1;
568 }
9d0b9596 569 }
70322ae3 570 }
56fa1896 571
572 p += strlen(p); /* finished with this argument word */
70322ae3 573 } else {
56fa1896 574 /*
575 * Short option.
576 */
70322ae3 577 char c = *p++;
578
56fa1896 579 for (i = 0; i < NSHORTOPTS; i++)
580 if (c == shortopts[i]) {
581 optid = shortvals[i];
582 break;
583 }
584
585 if (optid < 0) {
586 fprintf(stderr, "%s: unrecognised option '-%c'\n",
587 PNAME, c);
588 return 1;
589 }
590
591 if (opthasval[optid]) {
70322ae3 592 if (*p) {
593 optval = p;
594 p += strlen(p);
595 } else if (--argc > 0) {
596 optval = *++argv;
597 } else {
56fa1896 598 fprintf(stderr, "%s: option '-%c' expects"
70322ae3 599 " an argument\n", PNAME, c);
600 return 1;
601 }
56fa1896 602 } else {
603 optval = NULL;
604 }
605 }
606
607 wordstart = 0;
608
609 /*
610 * Now actually process the option.
611 */
612 switch (optid) {
613 case OPT_HELP:
e9e7a1bf 614 usage(stdout);
56fa1896 615 return 0;
616 case OPT_VERSION:
617 printf("FIXME: version();\n");
618 return 0;
619 case OPT_LICENCE:
5a29503d 620 {
621 extern const char *const licence[];
622 int i;
623
624 for (i = 0; licence[i]; i++)
625 fputs(licence[i], stdout);
626
627 return 0;
628 }
56fa1896 629 return 0;
630 case OPT_SCAN:
444c684c 631 if (nactions >= actionsize) {
632 actionsize = nactions * 3 / 2 + 16;
633 actions = sresize(actions, actionsize, struct action);
634 }
635 actions[nactions].mode = SCAN;
636 actions[nactions].arg = optval;
637 nactions++;
56fa1896 638 break;
84849cbd 639 case OPT_SCANDUMP:
444c684c 640 if (nactions >= actionsize) {
641 actionsize = nactions * 3 / 2 + 16;
642 actions = sresize(actions, actionsize, struct action);
643 }
644 actions[nactions].mode = SCANDUMP;
645 actions[nactions].arg = optval;
646 nactions++;
84849cbd 647 break;
56fa1896 648 case OPT_DUMP:
444c684c 649 if (nactions >= actionsize) {
650 actionsize = nactions * 3 / 2 + 16;
651 actions = sresize(actions, actionsize, struct action);
652 }
653 actions[nactions].mode = DUMP;
654 actions[nactions].arg = NULL;
655 nactions++;
56fa1896 656 break;
84849cbd 657 case OPT_LOAD:
444c684c 658 if (nactions >= actionsize) {
659 actionsize = nactions * 3 / 2 + 16;
660 actions = sresize(actions, actionsize, struct action);
661 }
662 actions[nactions].mode = LOAD;
663 actions[nactions].arg = NULL;
664 nactions++;
84849cbd 665 break;
56fa1896 666 case OPT_TEXT:
444c684c 667 if (nactions >= actionsize) {
668 actionsize = nactions * 3 / 2 + 16;
669 actions = sresize(actions, actionsize, struct action);
670 }
671 actions[nactions].mode = TEXT;
672 actions[nactions].arg = optval;
673 nactions++;
56fa1896 674 break;
675 case OPT_HTML:
444c684c 676 if (nactions >= actionsize) {
677 actionsize = nactions * 3 / 2 + 16;
678 actions = sresize(actions, actionsize, struct action);
679 }
680 actions[nactions].mode = HTML;
681 actions[nactions].arg = optval;
682 nactions++;
56fa1896 683 break;
684 case OPT_HTTPD:
444c684c 685 if (nactions >= actionsize) {
686 actionsize = nactions * 3 / 2 + 16;
687 actions = sresize(actions, actionsize, struct action);
688 }
689 actions[nactions].mode = HTTPD;
690 actions[nactions].arg = NULL;
691 nactions++;
56fa1896 692 break;
693 case OPT_PROGRESS:
694 progress = 2;
695 break;
696 case OPT_NOPROGRESS:
697 progress = 0;
698 break;
699 case OPT_TTYPROGRESS:
700 progress = 1;
701 break;
702 case OPT_CROSSFS:
703 crossfs = 1;
704 break;
705 case OPT_NOCROSSFS:
706 crossfs = 0;
707 break;
05b0f827 708 case OPT_DIRATIME:
709 fakediratimes = 0;
710 break;
711 case OPT_NODIRATIME:
712 fakediratimes = 1;
713 break;
56fa1896 714 case OPT_DATAFILE:
715 filename = optval;
716 break;
16e591d6 717 case OPT_TQDEPTH:
718 tqdepth = atoi(optval);
719 break;
56fa1896 720 case OPT_MINAGE:
f2e52893 721 textcutoff = parse_age(now, optval);
722 break;
723 case OPT_AGERANGE:
724 if (!strcmp(optval, "auto")) {
725 htmlautoagerange = 1;
726 } else {
727 char *q = optval + strcspn(optval, "-:");
728 if (*q)
729 *q++ = '\0';
730 htmloldest = parse_age(now, optval);
731 htmlnewest = *q ? parse_age(now, q) : now;
732 htmlautoagerange = 0;
733 }
56fa1896 734 break;
1e8d78b9 735 case OPT_SERVERADDR:
736 {
737 char *port;
738 if (optval[0] == '[' &&
739 (port = strchr(optval, ']')) != NULL)
740 port++;
741 else
742 port = optval;
743 port += strcspn(port, ":");
744 if (port)
745 *port++ = '\0';
746 httpserveraddr = optval;
747 httpserverport = atoi(port);
748 }
749 break;
56fa1896 750 case OPT_AUTH:
751 if (!strcmp(optval, "magic"))
752 auth = HTTPD_AUTH_MAGIC;
753 else if (!strcmp(optval, "basic"))
754 auth = HTTPD_AUTH_BASIC;
755 else if (!strcmp(optval, "none"))
756 auth = HTTPD_AUTH_NONE;
757 else if (!strcmp(optval, "default"))
758 auth = HTTPD_AUTH_MAGIC | HTTPD_AUTH_BASIC;
f2e52893 759 else if (!strcmp(optval, "help") ||
760 !strcmp(optval, "list")) {
bf53e756 761 printf(PNAME ": supported HTTP authentication types"
f2e52893 762 " are:\n"
763 " magic use Linux /proc/net/tcp to"
764 " determine owner of peer socket\n"
765 " basic HTTP Basic username and"
766 " password authentication\n"
767 " default use 'magic' if possible, "
768 " otherwise fall back to 'basic'\n"
769 " none unauthenticated HTTP (if"
770 " the data file is non-confidential)\n");
771 return 0;
772 } else {
56fa1896 773 fprintf(stderr, "%s: unrecognised authentication"
774 " type '%s'\n%*s options are 'magic',"
775 " 'basic', 'none', 'default'\n",
776 PNAME, optval, (int)strlen(PNAME), "");
777 return 1;
778 }
779 break;
1e8d78b9 780 case OPT_AUTHFILE:
781 case OPT_AUTHFD:
782 {
783 int fd;
784 char namebuf[40];
785 const char *name;
786 char *authbuf;
787 int authlen, authsize;
788 int ret;
789
790 if (optid == OPT_AUTHFILE) {
791 fd = open(optval, O_RDONLY);
792 if (fd < 0) {
793 fprintf(stderr, "%s: %s: open: %s\n", PNAME,
794 optval, strerror(errno));
795 return 1;
796 }
797 name = optval;
798 } else {
799 fd = atoi(optval);
800 name = namebuf;
801 sprintf(namebuf, "fd %d", fd);
802 }
803
804 authlen = 0;
805 authsize = 256;
806 authbuf = snewn(authsize, char);
807 while ((ret = read(fd, authbuf+authlen,
808 authsize-authlen)) > 0) {
809 authlen += ret;
810 if ((authsize - authlen) < (authsize / 16)) {
811 authsize = authlen * 3 / 2 + 4096;
812 authbuf = sresize(authbuf, authsize, char);
813 }
814 }
815 if (ret < 0) {
816 fprintf(stderr, "%s: %s: read: %s\n", PNAME,
817 name, strerror(errno));
818 return 1;
819 }
820 if (optid == OPT_AUTHFILE)
821 close(fd);
822 httpauthdata = authbuf;
823 }
824 break;
56fa1896 825 case OPT_INCLUDE:
826 case OPT_INCLUDEPATH:
827 case OPT_EXCLUDE:
828 case OPT_EXCLUDEPATH:
0ba55302 829 case OPT_PRUNE:
830 case OPT_PRUNEPATH:
56fa1896 831 if (ninex >= inexsize) {
832 inexsize = ninex * 3 / 2 + 16;
833 inex = sresize(inex, inexsize,
834 struct inclusion_exclusion);
835 }
836 inex[ninex].path = (optid == OPT_INCLUDEPATH ||
0ba55302 837 optid == OPT_EXCLUDEPATH ||
838 optid == OPT_PRUNEPATH);
839 inex[ninex].type = (optid == OPT_INCLUDE ? 1 :
840 optid == OPT_INCLUDEPATH ? 1 :
841 optid == OPT_EXCLUDE ? 0 :
842 optid == OPT_EXCLUDEPATH ? 0 :
843 optid == OPT_PRUNE ? -1 :
844 /* optid == OPT_PRUNEPATH ? */ -1);
56fa1896 845 inex[ninex].wildcard = optval;
846 ninex++;
847 break;
848 }
849 }
70322ae3 850 } else {
e9e7a1bf 851 fprintf(stderr, "%s: unexpected argument '%s'\n", PNAME, p);
852 return 1;
70322ae3 853 }
854 }
855
444c684c 856 if (nactions == 0) {
e9e7a1bf 857 usage(stderr);
858 return 1;
444c684c 859 }
860
861 for (action = 0; action < nactions; action++) {
862 int mode = actions[action].mode;
863
864 if (mode == SCAN || mode == SCANDUMP || mode == LOAD) {
865 const char *scandir = actions[action].arg;
866 if (mode == LOAD) {
867 char *buf = fgetline(stdin);
868 unsigned newpathsep;
869 buf[strcspn(buf, "\r\n")] = '\0';
bf53e756 870 if (1 != sscanf(buf, DUMPHDR "%x",
444c684c 871 &newpathsep)) {
872 fprintf(stderr, "%s: header in dump file not recognised\n",
873 PNAME);
874 return 1;
875 }
876 pathsep = (char)newpathsep;
877 sfree(buf);
84849cbd 878 }
70322ae3 879
444c684c 880 if (mode == SCAN || mode == LOAD) {
881 /*
882 * Prepare to write out the index file.
883 */
884 fd = open(filename, O_RDWR | O_TRUNC | O_CREAT, S_IRWXU);
885 if (fd < 0) {
886 fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename,
887 strerror(errno));
888 return 1;
889 }
890 if (fstat(fd, &st) < 0) {
bf53e756 891 perror(PNAME ": fstat");
444c684c 892 return 1;
893 }
894 ctx->datafile_dev = st.st_dev;
895 ctx->datafile_ino = st.st_ino;
896 ctx->straight_to_dump = 0;
897 } else {
898 ctx->datafile_dev = -1;
899 ctx->datafile_ino = -1;
900 ctx->straight_to_dump = 1;
84849cbd 901 }
444c684c 902
903 if (mode == SCAN || mode == SCANDUMP) {
904 if (stat(scandir, &st) < 0) {
905 fprintf(stderr, "%s: %s: stat: %s\n", PNAME, scandir,
906 strerror(errno));
907 return 1;
908 }
909 ctx->filesystem_dev = crossfs ? 0 : st.st_dev;
84849cbd 910 }
70322ae3 911
444c684c 912 ctx->inex = inex;
913 ctx->ninex = ninex;
914 ctx->crossfs = crossfs;
05b0f827 915 ctx->fakeatimes = fakediratimes;
444c684c 916
917 ctx->last_output_update = time(NULL);
918
919 /* progress==1 means report progress only if stderr is a tty */
920 if (progress == 1)
921 progress = isatty(2) ? 2 : 0;
922 ctx->progress = progress;
923 {
924 struct winsize ws;
925 if (progress && ioctl(2, TIOCGWINSZ, &ws) == 0)
926 ctx->progwidth = ws.ws_col - 1;
927 else
928 ctx->progwidth = 79;
84849cbd 929 }
84849cbd 930
444c684c 931 if (mode == SCANDUMP)
bf53e756 932 printf(DUMPHDR "%02x\n", (unsigned char)pathsep);
8b1f55d6 933
444c684c 934 /*
935 * Scan the directory tree, and write out the trie component
936 * of the data file.
937 */
938 if (mode != SCANDUMP) {
939 ctx->tb = triebuild_new(fd);
940 }
941 if (mode == LOAD) {
942 char *buf;
943 int line = 2;
944 while ((buf = fgetline(stdin)) != NULL) {
945 struct trie_file tf;
946 char *p, *q;
947
948 buf[strcspn(buf, "\r\n")] = '\0';
949
950 p = buf;
951 q = p;
952 while (*p && *p != ' ') p++;
953 if (!*p) {
954 fprintf(stderr, "%s: dump file line %d: expected at least"
955 " three fields\n", PNAME, line);
956 return 1;
957 }
958 *p++ = '\0';
959 tf.size = strtoull(q, NULL, 10);
960 q = p;
961 while (*p && *p != ' ') p++;
962 if (!*p) {
963 fprintf(stderr, "%s: dump file line %d: expected at least"
964 " three fields\n", PNAME, line);
965 return 1;
966 }
967 *p++ = '\0';
968 tf.atime = strtoull(q, NULL, 10);
969 q = buf;
970 while (*p) {
971 int c = *p;
972 if (*p == '%') {
973 int i;
974 p++;
975 c = 0;
976 for (i = 0; i < 2; i++) {
de693987 977 c *= 16;
444c684c 978 if (*p >= '0' && *p <= '9')
979 c += *p - '0';
980 else if (*p >= 'A' && *p <= 'F')
981 c += *p - ('A' - 10);
982 else if (*p >= 'a' && *p <= 'f')
983 c += *p - ('a' - 10);
984 else {
985 fprintf(stderr, "%s: dump file line %d: unable"
986 " to parse hex escape\n", PNAME, line);
987 }
988 p++;
989 }
990 }
991 *q++ = c;
992 p++;
993 }
994 *q = '\0';
995 triebuild_add(ctx->tb, buf, &tf);
996 sfree(buf);
de693987 997 line++;
444c684c 998 }
999 } else {
1000 du(scandir, gotdata, ctx);
1001 }
1002 if (mode != SCANDUMP) {
1003 count = triebuild_finish(ctx->tb);
1004 triebuild_free(ctx->tb);
84849cbd 1005
444c684c 1006 if (ctx->progress) {
1007 fprintf(stderr, "%-*s\r", ctx->progwidth, "");
1008 fflush(stderr);
1009 }
84849cbd 1010
444c684c 1011 /*
1012 * Work out how much space the cumulative index trees
1013 * will take; enlarge the file, and memory-map it.
1014 */
1015 if (fstat(fd, &st) < 0) {
bf53e756 1016 perror(PNAME ": fstat");
444c684c 1017 return 1;
1018 }
84849cbd 1019
444c684c 1020 printf("Built pathname index, %d entries, %ju bytes\n", count,
1021 (intmax_t)st.st_size);
1022
1023 totalsize = index_compute_size(st.st_size, count);
1024
1025 if (lseek(fd, totalsize-1, SEEK_SET) < 0) {
bf53e756 1026 perror(PNAME ": lseek");
84849cbd 1027 return 1;
1028 }
444c684c 1029 if (write(fd, "\0", 1) < 1) {
bf53e756 1030 perror(PNAME ": write");
84849cbd 1031 return 1;
1032 }
444c684c 1033
1034 printf("Upper bound on index file size = %ju bytes\n",
1035 (intmax_t)totalsize);
1036
1037 mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0);
1038 if (!mappedfile) {
bf53e756 1039 perror(PNAME ": mmap");
444c684c 1040 return 1;
84849cbd 1041 }
444c684c 1042
05b0f827 1043 if (fakediratimes) {
1044 printf("Faking directory atimes\n");
1045 trie_fake_dir_atimes(mappedfile);
1046 }
1047
1048 printf("Building index\n");
444c684c 1049 ib = indexbuild_new(mappedfile, st.st_size, count);
1050 tw = triewalk_new(mappedfile);
1051 while ((tf = triewalk_next(tw, NULL)) != NULL)
1052 indexbuild_add(ib, tf);
1053 triewalk_free(tw);
1054 realsize = indexbuild_realsize(ib);
1055 indexbuild_free(ib);
1056
1057 munmap(mappedfile, totalsize);
1058 ftruncate(fd, realsize);
1059 close(fd);
1060 printf("Actual index file size = %ju bytes\n", (intmax_t)realsize);
84849cbd 1061 }
444c684c 1062 } else if (mode == TEXT) {
1063 char *querydir = actions[action].arg;
1064 size_t pathlen;
70322ae3 1065
444c684c 1066 fd = open(filename, O_RDONLY);
1067 if (fd < 0) {
1068 fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename,
1069 strerror(errno));
1070 return 1;
1071 }
1072 if (fstat(fd, &st) < 0) {
bf53e756 1073 perror(PNAME ": fstat");
444c684c 1074 return 1;
1075 }
1076 totalsize = st.st_size;
1077 mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0);
1078 if (!mappedfile) {
bf53e756 1079 perror(PNAME ": mmap");
444c684c 1080 return 1;
84849cbd 1081 }
444c684c 1082 pathsep = trie_pathsep(mappedfile);
70322ae3 1083
84849cbd 1084 /*
444c684c 1085 * Trim trailing slash, just in case.
84849cbd 1086 */
444c684c 1087 pathlen = strlen(querydir);
1088 if (pathlen > 0 && querydir[pathlen-1] == pathsep)
1089 querydir[--pathlen] = '\0';
1090
16e591d6 1091 text_query(mappedfile, querydir, textcutoff, tqdepth);
444c684c 1092 } else if (mode == HTML) {
1093 char *querydir = actions[action].arg;
1094 size_t pathlen;
1095 struct html_config cfg;
1096 unsigned long xi;
1097 char *html;
1098
1099 fd = open(filename, O_RDONLY);
1100 if (fd < 0) {
1101 fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename,
1102 strerror(errno));
1103 return 1;
1104 }
84849cbd 1105 if (fstat(fd, &st) < 0) {
bf53e756 1106 perror(PNAME ": fstat");
84849cbd 1107 return 1;
1108 }
444c684c 1109 totalsize = st.st_size;
1110 mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0);
1111 if (!mappedfile) {
bf53e756 1112 perror(PNAME ": mmap");
444c684c 1113 return 1;
1114 }
1115 pathsep = trie_pathsep(mappedfile);
70322ae3 1116
444c684c 1117 /*
1118 * Trim trailing slash, just in case.
1119 */
1120 pathlen = strlen(querydir);
1121 if (pathlen > 0 && querydir[pathlen-1] == pathsep)
1122 querydir[--pathlen] = '\0';
1123
1124 xi = trie_before(mappedfile, querydir);
1125 cfg.format = NULL;
1126 cfg.autoage = htmlautoagerange;
1127 cfg.oldest = htmloldest;
1128 cfg.newest = htmlnewest;
1129 html = html_query(mappedfile, xi, &cfg);
1130 fputs(html, stdout);
1131 } else if (mode == DUMP) {
1132 size_t maxpathlen;
1133 char *buf;
70322ae3 1134
444c684c 1135 fd = open(filename, O_RDONLY);
1136 if (fd < 0) {
1137 fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename,
1138 strerror(errno));
84849cbd 1139 return 1;
1140 }
444c684c 1141 if (fstat(fd, &st) < 0) {
bf53e756 1142 perror(PNAME ": fstat");
84849cbd 1143 return 1;
1144 }
444c684c 1145 totalsize = st.st_size;
1146 mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0);
84849cbd 1147 if (!mappedfile) {
bf53e756 1148 perror(PNAME ": mmap");
84849cbd 1149 return 1;
1150 }
444c684c 1151 pathsep = trie_pathsep(mappedfile);
1152
1153 maxpathlen = trie_maxpathlen(mappedfile);
1154 buf = snewn(maxpathlen, char);
84849cbd 1155
bf53e756 1156 printf(DUMPHDR "%02x\n", (unsigned char)pathsep);
84849cbd 1157 tw = triewalk_new(mappedfile);
444c684c 1158 while ((tf = triewalk_next(tw, buf)) != NULL)
1159 dump_line(buf, tf);
84849cbd 1160 triewalk_free(tw);
444c684c 1161 } else if (mode == HTTPD) {
1162 struct html_config pcfg;
1163 struct httpd_config dcfg;
70322ae3 1164
444c684c 1165 fd = open(filename, O_RDONLY);
1166 if (fd < 0) {
1167 fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename,
1168 strerror(errno));
1169 return 1;
1170 }
1171 if (fstat(fd, &st) < 0) {
bf53e756 1172 perror(PNAME ": fstat");
444c684c 1173 return 1;
1174 }
1175 totalsize = st.st_size;
1176 mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0);
1177 if (!mappedfile) {
bf53e756 1178 perror(PNAME ": mmap");
444c684c 1179 return 1;
1180 }
1181 pathsep = trie_pathsep(mappedfile);
1182
1183 dcfg.address = httpserveraddr;
1184 dcfg.port = httpserverport;
1185 dcfg.basicauthdata = httpauthdata;
1186 pcfg.format = NULL;
1187 pcfg.autoage = htmlautoagerange;
1188 pcfg.oldest = htmloldest;
1189 pcfg.newest = htmlnewest;
1190 run_httpd(mappedfile, auth, &dcfg, &pcfg);
70322ae3 1191 }
70322ae3 1192 }
1193
1194 return 0;
1195}
Simon Tatham's `agedu' utility; import of svn://svn.tartarus.org/sgt/agedu