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 | |
23 | #include "du.h" |
24 | #include "trie.h" |
25 | #include "index.h" |
26 | #include "malloc.h" |
27 | #include "html.h" |
28 | #include "httpd.h" |
29 | |
30 | #define PNAME "agedu" |
31 | |
56fa1896 |
32 | #define lenof(x) (sizeof((x))/sizeof(*(x))) |
33 | |
70322ae3 |
34 | void fatal(const char *fmt, ...) |
35 | { |
36 | va_list ap; |
37 | fprintf(stderr, "%s: ", PNAME); |
38 | va_start(ap, fmt); |
39 | vfprintf(stderr, fmt, ap); |
40 | va_end(ap); |
41 | fprintf(stderr, "\n"); |
42 | exit(1); |
43 | } |
44 | |
9d0b9596 |
45 | struct inclusion_exclusion { |
0ba55302 |
46 | int type; |
9d0b9596 |
47 | const char *wildcard; |
48 | int path; |
49 | }; |
50 | |
70322ae3 |
51 | struct ctx { |
52 | triebuild *tb; |
53 | dev_t datafile_dev, filesystem_dev; |
54 | ino_t datafile_ino; |
55 | time_t last_output_update; |
8b1f55d6 |
56 | int progress, progwidth; |
9d0b9596 |
57 | struct inclusion_exclusion *inex; |
58 | int ninex; |
59 | int crossfs; |
70322ae3 |
60 | }; |
61 | |
62 | static int gotdata(void *vctx, const char *pathname, const struct stat64 *st) |
63 | { |
64 | struct ctx *ctx = (struct ctx *)vctx; |
65 | struct trie_file file; |
66 | time_t t; |
9d0b9596 |
67 | int i, include; |
68 | const char *filename; |
70322ae3 |
69 | |
70 | /* |
71 | * Filter out our own data file. |
72 | */ |
73 | if (st->st_dev == ctx->datafile_dev && st->st_ino == ctx->datafile_ino) |
74 | return 0; |
75 | |
76 | /* |
77 | * Don't cross the streams^W^Wany file system boundary. |
70322ae3 |
78 | */ |
9d0b9596 |
79 | if (!ctx->crossfs && st->st_dev != ctx->filesystem_dev) |
70322ae3 |
80 | return 0; |
81 | |
0ba55302 |
82 | file.blocks = st->st_blocks; |
83 | file.atime = st->st_atime; |
84 | |
70322ae3 |
85 | /* |
9d0b9596 |
86 | * Filter based on wildcards. |
70322ae3 |
87 | */ |
9d0b9596 |
88 | include = 1; |
89 | filename = strrchr(pathname, '/'); |
90 | if (!filename) |
91 | filename = pathname; |
92 | else |
93 | filename++; |
94 | for (i = 0; i < ctx->ninex; i++) { |
95 | if (fnmatch(ctx->inex[i].wildcard, |
0ba55302 |
96 | ctx->inex[i].path ? pathname : filename, 0) == 0) |
97 | include = ctx->inex[i].type; |
98 | } |
99 | if (include == -1) |
100 | return 0; /* ignore this entry and any subdirs */ |
101 | if (include == 0) { |
102 | /* |
103 | * Here we are supposed to be filtering an entry out, but |
104 | * still recursing into it if it's a directory. However, |
105 | * we can't actually leave out any directory whose |
106 | * subdirectories we then look at. So we cheat, in that |
107 | * case, by setting the size to zero. |
108 | */ |
109 | if (!S_ISDIR(st->st_mode)) |
110 | return 0; /* just ignore */ |
111 | else |
112 | file.blocks = 0; |
9d0b9596 |
113 | } |
70322ae3 |
114 | |
70322ae3 |
115 | triebuild_add(ctx->tb, pathname, &file); |
116 | |
117 | t = time(NULL); |
118 | if (t != ctx->last_output_update) { |
8b1f55d6 |
119 | if (ctx->progress) { |
120 | fprintf(stderr, "%-*.*s\r", ctx->progwidth, ctx->progwidth, |
121 | pathname); |
122 | fflush(stderr); |
123 | } |
70322ae3 |
124 | ctx->last_output_update = t; |
125 | } |
126 | |
127 | return 1; |
128 | } |
129 | |
e9e7a1bf |
130 | static void text_query(const void *mappedfile, const char *querydir, |
7cf11b75 |
131 | time_t t, int depth) |
70322ae3 |
132 | { |
133 | size_t maxpathlen; |
134 | char *pathbuf; |
135 | unsigned long xi1, xi2; |
136 | unsigned long long s1, s2; |
137 | |
138 | maxpathlen = trie_maxpathlen(mappedfile); |
139 | pathbuf = snewn(maxpathlen + 1, char); |
140 | |
141 | /* |
142 | * We want to query everything between the supplied filename |
143 | * (inclusive) and that filename with a ^A on the end |
144 | * (exclusive). So find the x indices for each. |
145 | */ |
e9e7a1bf |
146 | sprintf(pathbuf, "%s\001", querydir); |
147 | xi1 = trie_before(mappedfile, querydir); |
70322ae3 |
148 | xi2 = trie_before(mappedfile, pathbuf); |
149 | |
150 | /* |
151 | * Now do the lookups in the age index. |
152 | */ |
153 | s1 = index_query(mappedfile, xi1, t); |
154 | s2 = index_query(mappedfile, xi2, t); |
155 | |
010dd2a2 |
156 | if (s1 == s2) |
157 | return; /* no space taken up => no display */ |
158 | |
70322ae3 |
159 | /* Display in units of 2 512-byte blocks = 1Kb */ |
e9e7a1bf |
160 | printf("%-11llu %s\n", (s2 - s1) / 2, querydir); |
70322ae3 |
161 | |
162 | if (depth > 0) { |
163 | /* |
164 | * Now scan for first-level subdirectories and report |
165 | * those too. |
166 | */ |
167 | xi1++; |
168 | while (xi1 < xi2) { |
169 | trie_getpath(mappedfile, xi1, pathbuf); |
7cf11b75 |
170 | text_query(mappedfile, pathbuf, t, depth-1); |
70322ae3 |
171 | strcat(pathbuf, "\001"); |
172 | xi1 = trie_before(mappedfile, pathbuf); |
173 | } |
174 | } |
175 | } |
176 | |
56fa1896 |
177 | /* |
178 | * Largely frivolous way to define all my command-line options. I |
179 | * present here a parametric macro which declares a series of |
180 | * _logical_ option identifiers, and for each one declares zero or |
181 | * more short option characters and zero or more long option |
182 | * words. Then I repeatedly invoke that macro with its arguments |
183 | * defined to be various other macros, which allows me to |
184 | * variously: |
185 | * |
186 | * - define an enum allocating a distinct integer value to each |
187 | * logical option id |
188 | * - define a string consisting of precisely all the short option |
189 | * characters |
190 | * - define a string array consisting of all the long option |
191 | * strings |
192 | * - define (with help from auxiliary enums) integer arrays |
193 | * parallel to both of the above giving the logical option id |
194 | * for each physical short and long option |
195 | * - define an array indexed by logical option id indicating |
e9e7a1bf |
196 | * whether the option in question takes a value |
197 | * - define a function which prints out brief online help for all |
198 | * the options. |
56fa1896 |
199 | * |
200 | * It's not at all clear to me that this trickery is actually |
201 | * particularly _efficient_ - it still, after all, requires going |
202 | * linearly through the option list at run time and doing a |
203 | * strcmp, whereas in an ideal world I'd have liked the lists of |
204 | * long and short options to be pre-sorted so that a binary search |
205 | * or some other more efficient lookup was possible. (Not that |
206 | * asymptotic algorithmic complexity is remotely vital in option |
207 | * parsing, but if I were doing this in, say, Lisp or something |
208 | * with an equivalently powerful preprocessor then once I'd had |
209 | * the idea of preparing the option-parsing data structures at |
210 | * compile time I would probably have made the effort to prepare |
211 | * them _properly_. I could have Perl generate me a source file |
212 | * from some sort of description, I suppose, but that would seem |
213 | * like overkill. And in any case, it's more of a challenge to |
214 | * achieve as much as possible by cunning use of cpp and enum than |
215 | * to just write some sensible and logical code in a Turing- |
216 | * complete language. I said it was largely frivolous :-) |
217 | * |
218 | * This approach does have the virtue that it brings together the |
e9e7a1bf |
219 | * option ids, option spellings and help text into a single |
220 | * combined list and defines them all in exactly one place. If I |
221 | * want to add a new option, or a new spelling for an option, I |
222 | * only have to modify the main OPTHELP macro below and then add |
223 | * code to process the new logical id. |
56fa1896 |
224 | * |
225 | * (Though, really, even that isn't ideal, since it still involves |
226 | * modifying the source file in more than one place. In a |
227 | * _properly_ ideal world, I'd be able to interleave the option |
228 | * definitions with the code fragments that process them. And then |
229 | * not bother defining logical identifiers for them at all - those |
230 | * would be automatically generated, since I wouldn't have any |
231 | * need to specify them manually in another part of the code.) |
232 | */ |
233 | |
e9e7a1bf |
234 | #define OPTHELP(NOVAL, VAL, SHORT, LONG, HELPPFX, HELPARG, HELPLINE, HELPOPT) \ |
235 | HELPPFX("usage") HELPLINE("agedu [options] action") \ |
236 | HELPPFX("actions") \ |
237 | VAL(SCAN) SHORT(s) LONG(scan) \ |
238 | HELPARG("directory") HELPOPT("scan and index a directory") \ |
239 | NOVAL(DUMP) SHORT(d) LONG(dump) HELPOPT("dump the index file") \ |
240 | VAL(TEXT) SHORT(t) LONG(text) \ |
241 | HELPARG("subdir") HELPOPT("print a plain text report on a subdirectory") \ |
242 | VAL(HTML) SHORT(H) LONG(html) \ |
243 | HELPARG("subdir") HELPOPT("print an HTML report on a subdirectory") \ |
56fa1896 |
244 | NOVAL(HTTPD) SHORT(w) LONG(web) LONG(server) LONG(httpd) \ |
e9e7a1bf |
245 | HELPOPT("serve reports from a temporary web server") \ |
246 | HELPPFX("options") \ |
247 | VAL(DATAFILE) SHORT(f) LONG(file) \ |
248 | HELPARG("filename") HELPOPT("[all modes] specify index file") \ |
56fa1896 |
249 | NOVAL(PROGRESS) LONG(progress) LONG(scan_progress) \ |
e9e7a1bf |
250 | HELPOPT("[--scan] report progress on stderr") \ |
56fa1896 |
251 | NOVAL(NOPROGRESS) LONG(no_progress) LONG(no_scan_progress) \ |
e9e7a1bf |
252 | HELPOPT("[--scan] do not report progress") \ |
56fa1896 |
253 | NOVAL(TTYPROGRESS) LONG(tty_progress) LONG(tty_scan_progress) \ |
254 | LONG(progress_tty) LONG(scan_progress_tty) \ |
e9e7a1bf |
255 | HELPOPT("[--scan] report progress if stderr is a tty") \ |
56fa1896 |
256 | NOVAL(CROSSFS) LONG(cross_fs) \ |
e9e7a1bf |
257 | HELPOPT("[--scan] cross filesystem boundaries") \ |
56fa1896 |
258 | NOVAL(NOCROSSFS) LONG(no_cross_fs) \ |
e9e7a1bf |
259 | HELPOPT("[--scan] stick to one filesystem") \ |
56fa1896 |
260 | VAL(INCLUDE) LONG(include) \ |
e9e7a1bf |
261 | HELPARG("wildcard") HELPOPT("[--scan] include files matching pattern") \ |
56fa1896 |
262 | VAL(INCLUDEPATH) LONG(include_path) \ |
e9e7a1bf |
263 | HELPARG("wildcard") HELPOPT("[--scan] include pathnames matching pattern") \ |
56fa1896 |
264 | VAL(EXCLUDE) LONG(exclude) \ |
e9e7a1bf |
265 | HELPARG("wildcard") HELPOPT("[--scan] exclude files matching pattern") \ |
266 | VAL(EXCLUDEPATH) LONG(exclude_path) \ |
267 | HELPARG("wildcard") HELPOPT("[--scan] exclude pathnames matching pattern") \ |
0ba55302 |
268 | VAL(PRUNE) LONG(prune) \ |
269 | HELPARG("wildcard") HELPOPT("[--scan] prune files matching pattern") \ |
270 | VAL(PRUNEPATH) LONG(prune_path) \ |
271 | HELPARG("wildcard") HELPOPT("[--scan] prune pathnames matching pattern") \ |
e9e7a1bf |
272 | VAL(MINAGE) SHORT(a) LONG(age) LONG(min_age) LONG(minimum_age) \ |
273 | HELPARG("age") HELPOPT("[--text] include only files older than this") \ |
f2e52893 |
274 | VAL(AGERANGE) SHORT(r) LONG(age_range) LONG(range) LONG(ages) \ |
275 | HELPARG("age[-age]") HELPOPT("[--html,--web] set limits of colour coding") \ |
e9e7a1bf |
276 | VAL(AUTH) LONG(auth) LONG(http_auth) LONG(httpd_auth) \ |
277 | LONG(server_auth) LONG(web_auth) \ |
278 | HELPARG("type") HELPOPT("[--web] specify HTTP authentication method") \ |
279 | HELPPFX("also") \ |
280 | NOVAL(HELP) SHORT(h) LONG(help) HELPOPT("display this help text") \ |
281 | NOVAL(VERSION) SHORT(V) LONG(version) HELPOPT("report version number") \ |
282 | NOVAL(LICENCE) LONG(licence) LONG(license) \ |
283 | HELPOPT("display (MIT) licence text") \ |
56fa1896 |
284 | |
285 | #define IGNORE(x) |
286 | #define DEFENUM(x) OPT_ ## x, |
287 | #define ZERO(x) 0, |
288 | #define ONE(x) 1, |
289 | #define STRING(x) #x , |
290 | #define STRINGNOCOMMA(x) #x |
291 | #define SHORTNEWOPT(x) SHORTtmp_ ## x = OPT_ ## x, |
292 | #define SHORTTHISOPT(x) SHORTtmp2_ ## x, SHORTVAL_ ## x = SHORTtmp2_ ## x - 1, |
293 | #define SHORTOPTVAL(x) SHORTVAL_ ## x, |
294 | #define SHORTTMP(x) SHORTtmp3_ ## x, |
295 | #define LONGNEWOPT(x) LONGtmp_ ## x = OPT_ ## x, |
296 | #define LONGTHISOPT(x) LONGtmp2_ ## x, LONGVAL_ ## x = LONGtmp2_ ## x - 1, |
297 | #define LONGOPTVAL(x) LONGVAL_ ## x, |
298 | #define LONGTMP(x) SHORTtmp3_ ## x, |
299 | |
e9e7a1bf |
300 | #define OPTIONS(NOVAL, VAL, SHORT, LONG) \ |
301 | OPTHELP(NOVAL, VAL, SHORT, LONG, IGNORE, IGNORE, IGNORE, IGNORE) |
302 | |
56fa1896 |
303 | enum { OPTIONS(DEFENUM,DEFENUM,IGNORE,IGNORE) NOPTIONS }; |
304 | enum { OPTIONS(IGNORE,IGNORE,SHORTTMP,IGNORE) NSHORTOPTS }; |
305 | enum { OPTIONS(IGNORE,IGNORE,IGNORE,LONGTMP) NLONGOPTS }; |
306 | static const int opthasval[NOPTIONS] = {OPTIONS(ZERO,ONE,IGNORE,IGNORE)}; |
307 | static const char shortopts[] = {OPTIONS(IGNORE,IGNORE,STRINGNOCOMMA,IGNORE)}; |
308 | static const char *const longopts[] = {OPTIONS(IGNORE,IGNORE,IGNORE,STRING)}; |
309 | enum { OPTIONS(SHORTNEWOPT,SHORTNEWOPT,SHORTTHISOPT,IGNORE) }; |
310 | enum { OPTIONS(LONGNEWOPT,LONGNEWOPT,IGNORE,LONGTHISOPT) }; |
311 | static const int shortvals[] = {OPTIONS(IGNORE,IGNORE,SHORTOPTVAL,IGNORE)}; |
312 | static const int longvals[] = {OPTIONS(IGNORE,IGNORE,IGNORE,LONGOPTVAL)}; |
313 | |
e9e7a1bf |
314 | static void usage(FILE *fp) |
315 | { |
316 | char longbuf[80]; |
317 | const char *prefix, *shortopt, *longopt, *optarg; |
318 | int i, optex; |
319 | |
320 | #define HELPRESET prefix = shortopt = longopt = optarg = NULL, optex = -1 |
321 | #define HELPNOVAL(s) optex = 0; |
322 | #define HELPVAL(s) optex = 1; |
323 | #define HELPSHORT(s) if (!shortopt) shortopt = "-" #s; |
324 | #define HELPLONG(s) if (!longopt) { \ |
325 | strcpy(longbuf, "--" #s); longopt = longbuf; \ |
326 | for (i = 0; longbuf[i]; i++) if (longbuf[i] == '_') longbuf[i] = '-'; } |
327 | #define HELPPFX(s) prefix = s; |
328 | #define HELPARG(s) optarg = s; |
329 | #define HELPLINE(s) assert(optex == -1); \ |
330 | fprintf(fp, "%7s%c %s\n", prefix?prefix:"", prefix?':':' ', s); \ |
331 | HELPRESET; |
332 | #define HELPOPT(s) assert((optex == 1 && optarg) || (optex == 0 && !optarg)); \ |
333 | assert(shortopt || longopt); \ |
334 | i = fprintf(fp, "%7s%c %s%s%s%s%s", prefix?prefix:"", prefix?':':' ', \ |
335 | shortopt?shortopt:"", shortopt&&longopt?", ":"", longopt?longopt:"", \ |
336 | optarg?" ":"", optarg?optarg:""); \ |
337 | fprintf(fp, "%*s %s\n", i<32?32-i:0,"",s); HELPRESET; |
338 | |
339 | HELPRESET; |
340 | OPTHELP(HELPNOVAL, HELPVAL, HELPSHORT, HELPLONG, |
341 | HELPPFX, HELPARG, HELPLINE, HELPOPT); |
342 | |
343 | #undef HELPRESET |
344 | #undef HELPNOVAL |
345 | #undef HELPVAL |
346 | #undef HELPSHORT |
347 | #undef HELPLONG |
348 | #undef HELPPFX |
349 | #undef HELPARG |
350 | #undef HELPLINE |
351 | #undef HELPOPT |
352 | } |
353 | |
f2e52893 |
354 | static time_t parse_age(time_t now, const char *agestr) |
355 | { |
356 | time_t t; |
357 | struct tm tm; |
358 | int nunits; |
359 | char unit[2]; |
360 | |
361 | t = now; |
362 | |
363 | if (2 != sscanf(agestr, "%d%1[DdWwMmYy]", &nunits, unit)) { |
364 | fprintf(stderr, "%s: age specification should be a number followed by" |
365 | " one of d,w,m,y\n", PNAME); |
366 | exit(1); |
367 | } |
368 | |
369 | if (unit[0] == 'd') { |
370 | t -= 86400 * nunits; |
371 | } else if (unit[0] == 'w') { |
372 | t -= 86400 * 7 * nunits; |
373 | } else { |
374 | int ym; |
375 | |
376 | tm = *localtime(&t); |
377 | ym = tm.tm_year * 12 + tm.tm_mon; |
378 | |
379 | if (unit[0] == 'm') |
380 | ym -= nunits; |
381 | else |
382 | ym -= 12 * nunits; |
383 | |
384 | tm.tm_year = ym / 12; |
385 | tm.tm_mon = ym % 12; |
386 | |
387 | t = mktime(&tm); |
388 | } |
389 | |
390 | return t; |
391 | } |
392 | |
70322ae3 |
393 | int main(int argc, char **argv) |
394 | { |
395 | int fd, count; |
396 | struct ctx actx, *ctx = &actx; |
397 | struct stat st; |
398 | off_t totalsize, realsize; |
399 | void *mappedfile; |
400 | triewalk *tw; |
401 | indexbuild *ib; |
402 | const struct trie_file *tf; |
403 | char *filename = "agedu.dat"; |
e9e7a1bf |
404 | char *scandir = NULL; |
405 | char *querydir = NULL; |
70322ae3 |
406 | int doing_opts = 1; |
7cf11b75 |
407 | enum { USAGE, TEXT, HTML, SCAN, DUMP, HTTPD } mode = USAGE; |
f2e52893 |
408 | time_t now = time(NULL); |
409 | time_t textcutoff = now, htmlnewest = now, htmloldest = now; |
410 | int htmlautoagerange = 1; |
812e4bf2 |
411 | int auth = HTTPD_AUTH_MAGIC | HTTPD_AUTH_BASIC; |
8b1f55d6 |
412 | int progress = 1; |
9d0b9596 |
413 | struct inclusion_exclusion *inex = NULL; |
414 | int ninex = 0, inexsize = 0; |
415 | int crossfs = 0; |
70322ae3 |
416 | |
56fa1896 |
417 | #ifdef DEBUG_MAD_OPTION_PARSING_MACROS |
418 | { |
419 | static const char *const optnames[NOPTIONS] = { |
420 | OPTIONS(STRING,STRING,IGNORE,IGNORE) |
421 | }; |
422 | int i; |
423 | for (i = 0; i < NSHORTOPTS; i++) |
424 | printf("-%c == %s [%s]\n", shortopts[i], optnames[shortvals[i]], |
425 | opthasval[shortvals[i]] ? "value" : "no value"); |
426 | for (i = 0; i < NLONGOPTS; i++) |
427 | printf("--%s == %s [%s]\n", longopts[i], optnames[longvals[i]], |
428 | opthasval[longvals[i]] ? "value" : "no value"); |
429 | } |
430 | #endif |
431 | |
70322ae3 |
432 | while (--argc > 0) { |
433 | char *p = *++argv; |
70322ae3 |
434 | |
435 | if (doing_opts && *p == '-') { |
56fa1896 |
436 | int wordstart = 1; |
437 | |
70322ae3 |
438 | if (!strcmp(p, "--")) { |
439 | doing_opts = 0; |
56fa1896 |
440 | continue; |
441 | } |
442 | |
443 | p++; |
444 | while (*p) { |
445 | int optid = -1; |
446 | int i; |
447 | char *optval; |
448 | |
449 | if (wordstart && *p == '-') { |
70322ae3 |
450 | /* |
56fa1896 |
451 | * GNU-style long option. |
70322ae3 |
452 | */ |
56fa1896 |
453 | p++; |
454 | optval = strchr(p, '='); |
455 | if (optval) |
456 | *optval++ = '\0'; |
457 | |
458 | for (i = 0; i < NLONGOPTS; i++) { |
459 | const char *opt = longopts[i], *s = p; |
460 | int match = 1; |
461 | /* |
462 | * The underscores in the option names |
463 | * defined above may be given by the user |
464 | * as underscores or dashes, or omitted |
465 | * entirely. |
466 | */ |
467 | while (*opt) { |
468 | if (*opt == '_') { |
469 | if (*s == '-' || *s == '_') |
470 | s++; |
471 | } else { |
472 | if (*opt != *s) { |
473 | match = 0; |
474 | break; |
475 | } |
476 | s++; |
477 | } |
478 | opt++; |
479 | } |
480 | if (match && !*s) { |
481 | optid = longvals[i]; |
482 | break; |
70322ae3 |
483 | } |
484 | } |
56fa1896 |
485 | |
486 | if (optid < 0) { |
487 | fprintf(stderr, "%s: unrecognised option '--%s'\n", |
488 | PNAME, p); |
489 | return 1; |
490 | } |
491 | |
492 | if (!opthasval[optid]) { |
493 | if (optval) { |
494 | fprintf(stderr, "%s: unexpected argument to option" |
495 | " '--%s'\n", PNAME, p); |
812e4bf2 |
496 | return 1; |
497 | } |
56fa1896 |
498 | } else { |
499 | if (!optval) { |
500 | if (--argc > 0) { |
501 | optval = *++argv; |
502 | } else { |
503 | fprintf(stderr, "%s: option '--%s' expects" |
504 | " an argument\n", PNAME, p); |
505 | return 1; |
506 | } |
9d0b9596 |
507 | } |
70322ae3 |
508 | } |
56fa1896 |
509 | |
510 | p += strlen(p); /* finished with this argument word */ |
70322ae3 |
511 | } else { |
56fa1896 |
512 | /* |
513 | * Short option. |
514 | */ |
70322ae3 |
515 | char c = *p++; |
516 | |
56fa1896 |
517 | for (i = 0; i < NSHORTOPTS; i++) |
518 | if (c == shortopts[i]) { |
519 | optid = shortvals[i]; |
520 | break; |
521 | } |
522 | |
523 | if (optid < 0) { |
524 | fprintf(stderr, "%s: unrecognised option '-%c'\n", |
525 | PNAME, c); |
526 | return 1; |
527 | } |
528 | |
529 | if (opthasval[optid]) { |
70322ae3 |
530 | if (*p) { |
531 | optval = p; |
532 | p += strlen(p); |
533 | } else if (--argc > 0) { |
534 | optval = *++argv; |
535 | } else { |
56fa1896 |
536 | fprintf(stderr, "%s: option '-%c' expects" |
70322ae3 |
537 | " an argument\n", PNAME, c); |
538 | return 1; |
539 | } |
56fa1896 |
540 | } else { |
541 | optval = NULL; |
542 | } |
543 | } |
544 | |
545 | wordstart = 0; |
546 | |
547 | /* |
548 | * Now actually process the option. |
549 | */ |
550 | switch (optid) { |
551 | case OPT_HELP: |
e9e7a1bf |
552 | usage(stdout); |
56fa1896 |
553 | return 0; |
554 | case OPT_VERSION: |
555 | printf("FIXME: version();\n"); |
556 | return 0; |
557 | case OPT_LICENCE: |
558 | printf("FIXME: licence();\n"); |
559 | return 0; |
560 | case OPT_SCAN: |
561 | mode = SCAN; |
e9e7a1bf |
562 | scandir = optval; |
56fa1896 |
563 | break; |
564 | case OPT_DUMP: |
565 | mode = DUMP; |
566 | break; |
567 | case OPT_TEXT: |
e9e7a1bf |
568 | querydir = optval; |
56fa1896 |
569 | mode = TEXT; |
570 | break; |
571 | case OPT_HTML: |
572 | mode = HTML; |
e9e7a1bf |
573 | querydir = optval; |
56fa1896 |
574 | break; |
575 | case OPT_HTTPD: |
576 | mode = HTTPD; |
577 | break; |
578 | case OPT_PROGRESS: |
579 | progress = 2; |
580 | break; |
581 | case OPT_NOPROGRESS: |
582 | progress = 0; |
583 | break; |
584 | case OPT_TTYPROGRESS: |
585 | progress = 1; |
586 | break; |
587 | case OPT_CROSSFS: |
588 | crossfs = 1; |
589 | break; |
590 | case OPT_NOCROSSFS: |
591 | crossfs = 0; |
592 | break; |
593 | case OPT_DATAFILE: |
594 | filename = optval; |
595 | break; |
596 | case OPT_MINAGE: |
f2e52893 |
597 | textcutoff = parse_age(now, optval); |
598 | break; |
599 | case OPT_AGERANGE: |
600 | if (!strcmp(optval, "auto")) { |
601 | htmlautoagerange = 1; |
602 | } else { |
603 | char *q = optval + strcspn(optval, "-:"); |
604 | if (*q) |
605 | *q++ = '\0'; |
606 | htmloldest = parse_age(now, optval); |
607 | htmlnewest = *q ? parse_age(now, q) : now; |
608 | htmlautoagerange = 0; |
609 | } |
56fa1896 |
610 | break; |
611 | case OPT_AUTH: |
612 | if (!strcmp(optval, "magic")) |
613 | auth = HTTPD_AUTH_MAGIC; |
614 | else if (!strcmp(optval, "basic")) |
615 | auth = HTTPD_AUTH_BASIC; |
616 | else if (!strcmp(optval, "none")) |
617 | auth = HTTPD_AUTH_NONE; |
618 | else if (!strcmp(optval, "default")) |
619 | auth = HTTPD_AUTH_MAGIC | HTTPD_AUTH_BASIC; |
f2e52893 |
620 | else if (!strcmp(optval, "help") || |
621 | !strcmp(optval, "list")) { |
622 | printf("agedu: supported HTTP authentication types" |
623 | " are:\n" |
624 | " magic use Linux /proc/net/tcp to" |
625 | " determine owner of peer socket\n" |
626 | " basic HTTP Basic username and" |
627 | " password authentication\n" |
628 | " default use 'magic' if possible, " |
629 | " otherwise fall back to 'basic'\n" |
630 | " none unauthenticated HTTP (if" |
631 | " the data file is non-confidential)\n"); |
632 | return 0; |
633 | } else { |
56fa1896 |
634 | fprintf(stderr, "%s: unrecognised authentication" |
635 | " type '%s'\n%*s options are 'magic'," |
636 | " 'basic', 'none', 'default'\n", |
637 | PNAME, optval, (int)strlen(PNAME), ""); |
638 | return 1; |
639 | } |
640 | break; |
641 | case OPT_INCLUDE: |
642 | case OPT_INCLUDEPATH: |
643 | case OPT_EXCLUDE: |
644 | case OPT_EXCLUDEPATH: |
0ba55302 |
645 | case OPT_PRUNE: |
646 | case OPT_PRUNEPATH: |
56fa1896 |
647 | if (ninex >= inexsize) { |
648 | inexsize = ninex * 3 / 2 + 16; |
649 | inex = sresize(inex, inexsize, |
650 | struct inclusion_exclusion); |
651 | } |
652 | inex[ninex].path = (optid == OPT_INCLUDEPATH || |
0ba55302 |
653 | optid == OPT_EXCLUDEPATH || |
654 | optid == OPT_PRUNEPATH); |
655 | inex[ninex].type = (optid == OPT_INCLUDE ? 1 : |
656 | optid == OPT_INCLUDEPATH ? 1 : |
657 | optid == OPT_EXCLUDE ? 0 : |
658 | optid == OPT_EXCLUDEPATH ? 0 : |
659 | optid == OPT_PRUNE ? -1 : |
660 | /* optid == OPT_PRUNEPATH ? */ -1); |
56fa1896 |
661 | inex[ninex].wildcard = optval; |
662 | ninex++; |
663 | break; |
664 | } |
665 | } |
70322ae3 |
666 | } else { |
e9e7a1bf |
667 | fprintf(stderr, "%s: unexpected argument '%s'\n", PNAME, p); |
668 | return 1; |
70322ae3 |
669 | } |
670 | } |
671 | |
7cf11b75 |
672 | if (mode == USAGE) { |
e9e7a1bf |
673 | usage(stderr); |
674 | return 1; |
7cf11b75 |
675 | } else if (mode == SCAN) { |
70322ae3 |
676 | |
677 | fd = open(filename, O_RDWR | O_TRUNC | O_CREAT, S_IRWXU); |
678 | if (fd < 0) { |
679 | fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, |
680 | strerror(errno)); |
681 | return 1; |
682 | } |
683 | |
e9e7a1bf |
684 | if (stat(scandir, &st) < 0) { |
685 | fprintf(stderr, "%s: %s: stat: %s\n", PNAME, scandir, |
70322ae3 |
686 | strerror(errno)); |
687 | return 1; |
688 | } |
9d0b9596 |
689 | ctx->filesystem_dev = crossfs ? 0 : st.st_dev; |
70322ae3 |
690 | |
691 | if (fstat(fd, &st) < 0) { |
692 | perror("agedu: fstat"); |
693 | return 1; |
694 | } |
695 | ctx->datafile_dev = st.st_dev; |
696 | ctx->datafile_ino = st.st_ino; |
9d0b9596 |
697 | ctx->inex = inex; |
698 | ctx->ninex = ninex; |
699 | ctx->crossfs = crossfs; |
70322ae3 |
700 | |
701 | ctx->last_output_update = time(NULL); |
702 | |
8b1f55d6 |
703 | /* progress==1 means report progress only if stderr is a tty */ |
704 | if (progress == 1) |
705 | progress = isatty(2) ? 2 : 0; |
706 | ctx->progress = progress; |
707 | { |
708 | struct winsize ws; |
709 | if (progress && ioctl(2, TIOCGWINSZ, &ws) == 0) |
710 | ctx->progwidth = ws.ws_col - 1; |
711 | else |
712 | ctx->progwidth = 79; |
713 | } |
714 | |
70322ae3 |
715 | /* |
716 | * Scan the directory tree, and write out the trie component |
717 | * of the data file. |
718 | */ |
719 | ctx->tb = triebuild_new(fd); |
e9e7a1bf |
720 | du(scandir, gotdata, ctx); |
70322ae3 |
721 | count = triebuild_finish(ctx->tb); |
722 | triebuild_free(ctx->tb); |
723 | |
8b1f55d6 |
724 | if (ctx->progress) { |
725 | fprintf(stderr, "%-*s\r", ctx->progwidth, ""); |
726 | fflush(stderr); |
727 | } |
70322ae3 |
728 | |
729 | /* |
730 | * Work out how much space the cumulative index trees will |
731 | * take; enlarge the file, and memory-map it. |
732 | */ |
733 | if (fstat(fd, &st) < 0) { |
734 | perror("agedu: fstat"); |
735 | return 1; |
736 | } |
737 | |
738 | printf("Built pathname index, %d entries, %ju bytes\n", count, |
739 | (intmax_t)st.st_size); |
740 | |
741 | totalsize = index_compute_size(st.st_size, count); |
742 | |
743 | if (lseek(fd, totalsize-1, SEEK_SET) < 0) { |
744 | perror("agedu: lseek"); |
745 | return 1; |
746 | } |
747 | if (write(fd, "\0", 1) < 1) { |
748 | perror("agedu: write"); |
749 | return 1; |
750 | } |
751 | |
752 | printf("Upper bound on index file size = %ju bytes\n", |
753 | (intmax_t)totalsize); |
754 | |
755 | mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0); |
756 | if (!mappedfile) { |
757 | perror("agedu: mmap"); |
758 | return 1; |
759 | } |
760 | |
761 | ib = indexbuild_new(mappedfile, st.st_size, count); |
762 | tw = triewalk_new(mappedfile); |
763 | while ((tf = triewalk_next(tw, NULL)) != NULL) |
764 | indexbuild_add(ib, tf); |
765 | triewalk_free(tw); |
766 | realsize = indexbuild_realsize(ib); |
767 | indexbuild_free(ib); |
768 | |
769 | munmap(mappedfile, totalsize); |
770 | ftruncate(fd, realsize); |
771 | close(fd); |
772 | printf("Actual index file size = %ju bytes\n", (intmax_t)realsize); |
7cf11b75 |
773 | } else if (mode == TEXT) { |
70322ae3 |
774 | size_t pathlen; |
775 | |
70322ae3 |
776 | fd = open(filename, O_RDONLY); |
777 | if (fd < 0) { |
778 | fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, |
779 | strerror(errno)); |
780 | return 1; |
781 | } |
782 | if (fstat(fd, &st) < 0) { |
783 | perror("agedu: fstat"); |
784 | return 1; |
785 | } |
786 | totalsize = st.st_size; |
787 | mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); |
788 | if (!mappedfile) { |
789 | perror("agedu: mmap"); |
790 | return 1; |
791 | } |
792 | |
793 | /* |
794 | * Trim trailing slash, just in case. |
795 | */ |
e9e7a1bf |
796 | pathlen = strlen(querydir); |
797 | if (pathlen > 0 && querydir[pathlen-1] == '/') |
798 | querydir[--pathlen] = '\0'; |
70322ae3 |
799 | |
f2e52893 |
800 | text_query(mappedfile, querydir, textcutoff, 1); |
70322ae3 |
801 | } else if (mode == HTML) { |
802 | size_t pathlen; |
f2e52893 |
803 | struct html_config cfg; |
70322ae3 |
804 | unsigned long xi; |
805 | char *html; |
806 | |
807 | fd = open(filename, O_RDONLY); |
808 | if (fd < 0) { |
809 | fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, |
810 | strerror(errno)); |
811 | return 1; |
812 | } |
813 | if (fstat(fd, &st) < 0) { |
814 | perror("agedu: fstat"); |
815 | return 1; |
816 | } |
817 | totalsize = st.st_size; |
818 | mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); |
819 | if (!mappedfile) { |
820 | perror("agedu: mmap"); |
821 | return 1; |
822 | } |
823 | |
824 | /* |
825 | * Trim trailing slash, just in case. |
826 | */ |
e9e7a1bf |
827 | pathlen = strlen(querydir); |
828 | if (pathlen > 0 && querydir[pathlen-1] == '/') |
829 | querydir[--pathlen] = '\0'; |
70322ae3 |
830 | |
e9e7a1bf |
831 | xi = trie_before(mappedfile, querydir); |
f2e52893 |
832 | cfg.format = NULL; |
833 | cfg.autoage = htmlautoagerange; |
834 | cfg.oldest = htmloldest; |
835 | cfg.newest = htmlnewest; |
836 | html = html_query(mappedfile, xi, &cfg); |
70322ae3 |
837 | fputs(html, stdout); |
838 | } else if (mode == DUMP) { |
839 | size_t maxpathlen; |
840 | char *buf; |
841 | |
842 | fd = open(filename, O_RDONLY); |
843 | if (fd < 0) { |
844 | fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, |
845 | strerror(errno)); |
846 | return 1; |
847 | } |
848 | if (fstat(fd, &st) < 0) { |
849 | perror("agedu: fstat"); |
850 | return 1; |
851 | } |
852 | totalsize = st.st_size; |
853 | mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); |
854 | if (!mappedfile) { |
855 | perror("agedu: mmap"); |
856 | return 1; |
857 | } |
858 | |
859 | maxpathlen = trie_maxpathlen(mappedfile); |
860 | buf = snewn(maxpathlen, char); |
861 | |
862 | tw = triewalk_new(mappedfile); |
863 | while ((tf = triewalk_next(tw, buf)) != NULL) { |
864 | printf("%s: %llu %llu\n", buf, tf->blocks, tf->atime); |
865 | } |
866 | triewalk_free(tw); |
867 | } else if (mode == HTTPD) { |
f2e52893 |
868 | struct html_config cfg; |
869 | |
70322ae3 |
870 | fd = open(filename, O_RDONLY); |
871 | if (fd < 0) { |
872 | fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, |
873 | strerror(errno)); |
874 | return 1; |
875 | } |
876 | if (fstat(fd, &st) < 0) { |
877 | perror("agedu: fstat"); |
878 | return 1; |
879 | } |
880 | totalsize = st.st_size; |
881 | mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); |
882 | if (!mappedfile) { |
883 | perror("agedu: mmap"); |
884 | return 1; |
885 | } |
886 | |
f2e52893 |
887 | cfg.format = NULL; |
888 | cfg.autoage = htmlautoagerange; |
889 | cfg.oldest = htmloldest; |
890 | cfg.newest = htmlnewest; |
891 | run_httpd(mappedfile, auth, &cfg); |
70322ae3 |
892 | } |
893 | |
894 | return 0; |
895 | } |