70322ae3 |
1 | /* |
2 | * Main program for agedu. |
3 | */ |
4 | |
353bc75d |
5 | #include "agedu.h" |
50e82fdc |
6 | |
70322ae3 |
7 | #include "du.h" |
8 | #include "trie.h" |
9 | #include "index.h" |
995db599 |
10 | #include "alloc.h" |
70322ae3 |
11 | #include "html.h" |
12 | #include "httpd.h" |
84849cbd |
13 | #include "fgetline.h" |
70322ae3 |
14 | |
373a02e5 |
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 | |
70322ae3 |
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 | |
9d0b9596 |
36 | struct inclusion_exclusion { |
0ba55302 |
37 | int type; |
9d0b9596 |
38 | const char *wildcard; |
39 | int path; |
40 | }; |
41 | |
70322ae3 |
42 | struct ctx { |
43 | triebuild *tb; |
44 | dev_t datafile_dev, filesystem_dev; |
45 | ino_t datafile_ino; |
46 | time_t last_output_update; |
8b1f55d6 |
47 | int progress, progwidth; |
84849cbd |
48 | int straight_to_dump; |
9d0b9596 |
49 | struct inclusion_exclusion *inex; |
50 | int ninex; |
51 | int crossfs; |
f59a5d34 |
52 | int usemtime; |
05b0f827 |
53 | int fakeatimes; |
70322ae3 |
54 | }; |
55 | |
84849cbd |
56 | static void dump_line(const char *pathname, const struct trie_file *tf) |
57 | { |
58 | const char *p; |
59 | printf("%llu %llu ", tf->size, tf->atime); |
60 | for (p = pathname; *p; p++) { |
61 | if (*p >= ' ' && *p < 127 && *p != '%') |
62 | putchar(*p); |
63 | else |
64 | printf("%%%02x", (unsigned char)*p); |
65 | } |
66 | putchar('\n'); |
67 | } |
68 | |
9c6e61f2 |
69 | static int gotdata(void *vctx, const char *pathname, const STRUCT_STAT *st) |
70322ae3 |
70 | { |
71 | struct ctx *ctx = (struct ctx *)vctx; |
72 | struct trie_file file; |
73 | time_t t; |
9d0b9596 |
74 | int i, include; |
75 | const char *filename; |
70322ae3 |
76 | |
77 | /* |
78 | * Filter out our own data file. |
79 | */ |
80 | if (st->st_dev == ctx->datafile_dev && st->st_ino == ctx->datafile_ino) |
81 | return 0; |
82 | |
83 | /* |
84 | * Don't cross the streams^W^Wany file system boundary. |
70322ae3 |
85 | */ |
9d0b9596 |
86 | if (!ctx->crossfs && st->st_dev != ctx->filesystem_dev) |
70322ae3 |
87 | return 0; |
88 | |
84849cbd |
89 | file.size = (unsigned long long)512 * st->st_blocks; |
f59a5d34 |
90 | if (ctx->usemtime || (ctx->fakeatimes && S_ISDIR(st->st_mode))) |
05b0f827 |
91 | file.atime = st->st_mtime; |
92 | else |
93 | file.atime = st->st_atime; |
0ba55302 |
94 | |
70322ae3 |
95 | /* |
9d0b9596 |
96 | * Filter based on wildcards. |
70322ae3 |
97 | */ |
9d0b9596 |
98 | include = 1; |
373a02e5 |
99 | filename = strrchr(pathname, pathsep); |
9d0b9596 |
100 | if (!filename) |
101 | filename = pathname; |
102 | else |
103 | filename++; |
104 | for (i = 0; i < ctx->ninex; i++) { |
105 | if (fnmatch(ctx->inex[i].wildcard, |
0ba55302 |
106 | ctx->inex[i].path ? pathname : filename, 0) == 0) |
107 | include = ctx->inex[i].type; |
108 | } |
109 | if (include == -1) |
110 | return 0; /* ignore this entry and any subdirs */ |
111 | if (include == 0) { |
112 | /* |
113 | * Here we are supposed to be filtering an entry out, but |
114 | * still recursing into it if it's a directory. However, |
115 | * we can't actually leave out any directory whose |
116 | * subdirectories we then look at. So we cheat, in that |
117 | * case, by setting the size to zero. |
118 | */ |
119 | if (!S_ISDIR(st->st_mode)) |
120 | return 0; /* just ignore */ |
121 | else |
84849cbd |
122 | file.size = 0; |
9d0b9596 |
123 | } |
70322ae3 |
124 | |
84849cbd |
125 | if (ctx->straight_to_dump) |
126 | dump_line(pathname, &file); |
127 | else |
128 | triebuild_add(ctx->tb, pathname, &file); |
70322ae3 |
129 | |
84849cbd |
130 | if (ctx->progress) { |
131 | t = time(NULL); |
132 | if (t != ctx->last_output_update) { |
8b1f55d6 |
133 | fprintf(stderr, "%-*.*s\r", ctx->progwidth, ctx->progwidth, |
134 | pathname); |
135 | fflush(stderr); |
84849cbd |
136 | ctx->last_output_update = t; |
8b1f55d6 |
137 | } |
70322ae3 |
138 | } |
139 | |
140 | return 1; |
141 | } |
142 | |
09fd7619 |
143 | static void scan_error(void *vctx, const char *fmt, ...) |
144 | { |
145 | struct ctx *ctx = (struct ctx *)vctx; |
146 | va_list ap; |
147 | |
148 | if (ctx->progress) { |
149 | fprintf(stderr, "%-*s\r", ctx->progwidth, ""); |
150 | fflush(stderr); |
151 | } |
152 | |
153 | fprintf(stderr, "%s: ", PNAME); |
154 | va_start(ap, fmt); |
155 | vfprintf(stderr, fmt, ap); |
156 | va_end(ap); |
157 | |
158 | ctx->last_output_update--; /* force a progress report next time */ |
159 | } |
160 | |
e9e7a1bf |
161 | static void text_query(const void *mappedfile, const char *querydir, |
7cf11b75 |
162 | time_t t, int depth) |
70322ae3 |
163 | { |
164 | size_t maxpathlen; |
165 | char *pathbuf; |
166 | unsigned long xi1, xi2; |
167 | unsigned long long s1, s2; |
168 | |
169 | maxpathlen = trie_maxpathlen(mappedfile); |
170 | pathbuf = snewn(maxpathlen + 1, char); |
171 | |
172 | /* |
173 | * We want to query everything between the supplied filename |
174 | * (inclusive) and that filename with a ^A on the end |
175 | * (exclusive). So find the x indices for each. |
176 | */ |
256c29a2 |
177 | strcpy(pathbuf, querydir); |
178 | make_successor(pathbuf); |
e9e7a1bf |
179 | xi1 = trie_before(mappedfile, querydir); |
70322ae3 |
180 | xi2 = trie_before(mappedfile, pathbuf); |
181 | |
0313b788 |
182 | if (xi2 - xi1 == 1) |
183 | return; /* file, or empty dir => no display */ |
184 | |
70322ae3 |
185 | /* |
186 | * Now do the lookups in the age index. |
187 | */ |
188 | s1 = index_query(mappedfile, xi1, t); |
189 | s2 = index_query(mappedfile, xi2, t); |
190 | |
010dd2a2 |
191 | if (s1 == s2) |
192 | return; /* no space taken up => no display */ |
193 | |
70322ae3 |
194 | if (depth > 0) { |
195 | /* |
196 | * Now scan for first-level subdirectories and report |
197 | * those too. |
198 | */ |
199 | xi1++; |
200 | while (xi1 < xi2) { |
201 | trie_getpath(mappedfile, xi1, pathbuf); |
7cf11b75 |
202 | text_query(mappedfile, pathbuf, t, depth-1); |
256c29a2 |
203 | make_successor(pathbuf); |
70322ae3 |
204 | xi1 = trie_before(mappedfile, pathbuf); |
205 | } |
206 | } |
16e591d6 |
207 | |
208 | /* Display in units of 1Kb */ |
209 | printf("%-11llu %s\n", (s2 - s1) / 1024, querydir); |
70322ae3 |
210 | } |
211 | |
56fa1896 |
212 | /* |
213 | * Largely frivolous way to define all my command-line options. I |
214 | * present here a parametric macro which declares a series of |
215 | * _logical_ option identifiers, and for each one declares zero or |
216 | * more short option characters and zero or more long option |
217 | * words. Then I repeatedly invoke that macro with its arguments |
218 | * defined to be various other macros, which allows me to |
219 | * variously: |
220 | * |
221 | * - define an enum allocating a distinct integer value to each |
222 | * logical option id |
223 | * - define a string consisting of precisely all the short option |
224 | * characters |
225 | * - define a string array consisting of all the long option |
226 | * strings |
227 | * - define (with help from auxiliary enums) integer arrays |
228 | * parallel to both of the above giving the logical option id |
229 | * for each physical short and long option |
230 | * - define an array indexed by logical option id indicating |
e9e7a1bf |
231 | * whether the option in question takes a value |
232 | * - define a function which prints out brief online help for all |
233 | * the options. |
56fa1896 |
234 | * |
235 | * It's not at all clear to me that this trickery is actually |
236 | * particularly _efficient_ - it still, after all, requires going |
237 | * linearly through the option list at run time and doing a |
238 | * strcmp, whereas in an ideal world I'd have liked the lists of |
239 | * long and short options to be pre-sorted so that a binary search |
240 | * or some other more efficient lookup was possible. (Not that |
241 | * asymptotic algorithmic complexity is remotely vital in option |
242 | * parsing, but if I were doing this in, say, Lisp or something |
243 | * with an equivalently powerful preprocessor then once I'd had |
244 | * the idea of preparing the option-parsing data structures at |
245 | * compile time I would probably have made the effort to prepare |
246 | * them _properly_. I could have Perl generate me a source file |
247 | * from some sort of description, I suppose, but that would seem |
248 | * like overkill. And in any case, it's more of a challenge to |
249 | * achieve as much as possible by cunning use of cpp and enum than |
250 | * to just write some sensible and logical code in a Turing- |
251 | * complete language. I said it was largely frivolous :-) |
252 | * |
253 | * This approach does have the virtue that it brings together the |
e9e7a1bf |
254 | * option ids, option spellings and help text into a single |
255 | * combined list and defines them all in exactly one place. If I |
256 | * want to add a new option, or a new spelling for an option, I |
257 | * only have to modify the main OPTHELP macro below and then add |
258 | * code to process the new logical id. |
56fa1896 |
259 | * |
260 | * (Though, really, even that isn't ideal, since it still involves |
261 | * modifying the source file in more than one place. In a |
262 | * _properly_ ideal world, I'd be able to interleave the option |
263 | * definitions with the code fragments that process them. And then |
264 | * not bother defining logical identifiers for them at all - those |
265 | * would be automatically generated, since I wouldn't have any |
266 | * need to specify them manually in another part of the code.) |
c5c3510f |
267 | * |
268 | * One other helpful consequence of the enum-based structure here |
269 | * is that it causes a compiler error if I accidentally try to |
270 | * define the same option (short or long) twice. |
56fa1896 |
271 | */ |
272 | |
e9e7a1bf |
273 | #define OPTHELP(NOVAL, VAL, SHORT, LONG, HELPPFX, HELPARG, HELPLINE, HELPOPT) \ |
bf53e756 |
274 | HELPPFX("usage") HELPLINE(PNAME " [options] action [action...]") \ |
e9e7a1bf |
275 | HELPPFX("actions") \ |
276 | VAL(SCAN) SHORT(s) LONG(scan) \ |
277 | HELPARG("directory") HELPOPT("scan and index a directory") \ |
67159944 |
278 | NOVAL(HTTPD) SHORT(w) LONG(web) LONG(server) LONG(httpd) \ |
279 | HELPOPT("serve HTML reports from a temporary web server") \ |
280 | VAL(TEXT) SHORT(t) LONG(text) \ |
281 | HELPARG("subdir") HELPOPT("print a plain text report on a subdirectory") \ |
282 | NOVAL(REMOVE) SHORT(R) LONG(remove) LONG(delete) LONG(unlink) \ |
283 | HELPOPT("remove the index file") \ |
c5c3510f |
284 | NOVAL(DUMP) SHORT(D) LONG(dump) HELPOPT("dump the index file on stdout") \ |
c5c3510f |
285 | NOVAL(LOAD) SHORT(L) LONG(load) \ |
84849cbd |
286 | HELPOPT("load and index a dump file") \ |
67159944 |
287 | VAL(SCANDUMP) SHORT(S) LONG(scan_dump) \ |
288 | HELPARG("directory") HELPOPT("scan only, generating a dump") \ |
e9e7a1bf |
289 | VAL(HTML) SHORT(H) LONG(html) \ |
290 | HELPARG("subdir") HELPOPT("print an HTML report on a subdirectory") \ |
e9e7a1bf |
291 | HELPPFX("options") \ |
292 | VAL(DATAFILE) SHORT(f) LONG(file) \ |
c5c3510f |
293 | HELPARG("filename") HELPOPT("[most modes] specify index file") \ |
56fa1896 |
294 | NOVAL(CROSSFS) LONG(cross_fs) \ |
e9e7a1bf |
295 | HELPOPT("[--scan] cross filesystem boundaries") \ |
56fa1896 |
296 | NOVAL(NOCROSSFS) LONG(no_cross_fs) \ |
e9e7a1bf |
297 | HELPOPT("[--scan] stick to one filesystem") \ |
0ba55302 |
298 | VAL(PRUNE) LONG(prune) \ |
299 | HELPARG("wildcard") HELPOPT("[--scan] prune files matching pattern") \ |
300 | VAL(PRUNEPATH) LONG(prune_path) \ |
301 | HELPARG("wildcard") HELPOPT("[--scan] prune pathnames matching pattern") \ |
67159944 |
302 | VAL(EXCLUDE) LONG(exclude) \ |
303 | HELPARG("wildcard") HELPOPT("[--scan] exclude files matching pattern") \ |
304 | VAL(EXCLUDEPATH) LONG(exclude_path) \ |
305 | HELPARG("wildcard") HELPOPT("[--scan] exclude pathnames matching pattern") \ |
306 | VAL(INCLUDE) LONG(include) \ |
307 | HELPARG("wildcard") HELPOPT("[--scan] include files matching pattern") \ |
308 | VAL(INCLUDEPATH) LONG(include_path) \ |
309 | HELPARG("wildcard") HELPOPT("[--scan] include pathnames matching pattern") \ |
310 | NOVAL(PROGRESS) LONG(progress) LONG(scan_progress) \ |
311 | HELPOPT("[--scan] report progress on stderr") \ |
312 | NOVAL(NOPROGRESS) LONG(no_progress) LONG(no_scan_progress) \ |
313 | HELPOPT("[--scan] do not report progress") \ |
314 | NOVAL(TTYPROGRESS) LONG(tty_progress) LONG(tty_scan_progress) \ |
315 | LONG(progress_tty) LONG(scan_progress_tty) \ |
316 | HELPOPT("[--scan] report progress if stderr is a tty") \ |
05b0f827 |
317 | NOVAL(DIRATIME) LONG(dir_atime) LONG(dir_atimes) \ |
67159944 |
318 | HELPOPT("[--scan,--load] keep real atimes on directories") \ |
05b0f827 |
319 | NOVAL(NODIRATIME) LONG(no_dir_atime) LONG(no_dir_atimes) \ |
67159944 |
320 | HELPOPT("[--scan,--load] fake atimes on directories") \ |
f59a5d34 |
321 | NOVAL(MTIME) LONG(mtime) \ |
322 | HELPOPT("[--scan] use mtime instead of atime") \ |
f2e52893 |
323 | VAL(AGERANGE) SHORT(r) LONG(age_range) LONG(range) LONG(ages) \ |
67159944 |
324 | HELPARG("age[-age]") HELPOPT("[--web,--html] set limits of colour coding") \ |
1e8d78b9 |
325 | VAL(SERVERADDR) LONG(address) LONG(addr) LONG(server_address) \ |
326 | LONG(server_addr) \ |
327 | HELPARG("addr[:port]") HELPOPT("[--web] specify HTTP server address") \ |
e9e7a1bf |
328 | VAL(AUTH) LONG(auth) LONG(http_auth) LONG(httpd_auth) \ |
329 | LONG(server_auth) LONG(web_auth) \ |
330 | HELPARG("type") HELPOPT("[--web] specify HTTP authentication method") \ |
1e8d78b9 |
331 | VAL(AUTHFILE) LONG(auth_file) \ |
332 | HELPARG("filename") HELPOPT("[--web] read HTTP Basic user/pass from file") \ |
333 | VAL(AUTHFD) LONG(auth_fd) \ |
334 | HELPARG("fd") HELPOPT("[--web] read HTTP Basic user/pass from fd") \ |
67159944 |
335 | VAL(TQDEPTH) SHORT(d) LONG(depth) LONG(max_depth) LONG(maximum_depth) \ |
336 | HELPARG("levels") HELPOPT("[--text] recurse to this many levels") \ |
337 | VAL(MINAGE) SHORT(a) LONG(age) LONG(min_age) LONG(minimum_age) \ |
338 | HELPARG("age") HELPOPT("[--text] include only files older than this") \ |
e9e7a1bf |
339 | HELPPFX("also") \ |
340 | NOVAL(HELP) SHORT(h) LONG(help) HELPOPT("display this help text") \ |
341 | NOVAL(VERSION) SHORT(V) LONG(version) HELPOPT("report version number") \ |
342 | NOVAL(LICENCE) LONG(licence) LONG(license) \ |
343 | HELPOPT("display (MIT) licence text") \ |
56fa1896 |
344 | |
345 | #define IGNORE(x) |
346 | #define DEFENUM(x) OPT_ ## x, |
347 | #define ZERO(x) 0, |
348 | #define ONE(x) 1, |
349 | #define STRING(x) #x , |
350 | #define STRINGNOCOMMA(x) #x |
351 | #define SHORTNEWOPT(x) SHORTtmp_ ## x = OPT_ ## x, |
352 | #define SHORTTHISOPT(x) SHORTtmp2_ ## x, SHORTVAL_ ## x = SHORTtmp2_ ## x - 1, |
353 | #define SHORTOPTVAL(x) SHORTVAL_ ## x, |
354 | #define SHORTTMP(x) SHORTtmp3_ ## x, |
355 | #define LONGNEWOPT(x) LONGtmp_ ## x = OPT_ ## x, |
356 | #define LONGTHISOPT(x) LONGtmp2_ ## x, LONGVAL_ ## x = LONGtmp2_ ## x - 1, |
357 | #define LONGOPTVAL(x) LONGVAL_ ## x, |
358 | #define LONGTMP(x) SHORTtmp3_ ## x, |
359 | |
e9e7a1bf |
360 | #define OPTIONS(NOVAL, VAL, SHORT, LONG) \ |
361 | OPTHELP(NOVAL, VAL, SHORT, LONG, IGNORE, IGNORE, IGNORE, IGNORE) |
362 | |
56fa1896 |
363 | enum { OPTIONS(DEFENUM,DEFENUM,IGNORE,IGNORE) NOPTIONS }; |
364 | enum { OPTIONS(IGNORE,IGNORE,SHORTTMP,IGNORE) NSHORTOPTS }; |
365 | enum { OPTIONS(IGNORE,IGNORE,IGNORE,LONGTMP) NLONGOPTS }; |
366 | static const int opthasval[NOPTIONS] = {OPTIONS(ZERO,ONE,IGNORE,IGNORE)}; |
367 | static const char shortopts[] = {OPTIONS(IGNORE,IGNORE,STRINGNOCOMMA,IGNORE)}; |
368 | static const char *const longopts[] = {OPTIONS(IGNORE,IGNORE,IGNORE,STRING)}; |
369 | enum { OPTIONS(SHORTNEWOPT,SHORTNEWOPT,SHORTTHISOPT,IGNORE) }; |
370 | enum { OPTIONS(LONGNEWOPT,LONGNEWOPT,IGNORE,LONGTHISOPT) }; |
371 | static const int shortvals[] = {OPTIONS(IGNORE,IGNORE,SHORTOPTVAL,IGNORE)}; |
372 | static const int longvals[] = {OPTIONS(IGNORE,IGNORE,IGNORE,LONGOPTVAL)}; |
373 | |
e9e7a1bf |
374 | static void usage(FILE *fp) |
375 | { |
376 | char longbuf[80]; |
377 | const char *prefix, *shortopt, *longopt, *optarg; |
378 | int i, optex; |
379 | |
380 | #define HELPRESET prefix = shortopt = longopt = optarg = NULL, optex = -1 |
381 | #define HELPNOVAL(s) optex = 0; |
382 | #define HELPVAL(s) optex = 1; |
383 | #define HELPSHORT(s) if (!shortopt) shortopt = "-" #s; |
384 | #define HELPLONG(s) if (!longopt) { \ |
385 | strcpy(longbuf, "--" #s); longopt = longbuf; \ |
386 | for (i = 0; longbuf[i]; i++) if (longbuf[i] == '_') longbuf[i] = '-'; } |
387 | #define HELPPFX(s) prefix = s; |
388 | #define HELPARG(s) optarg = s; |
389 | #define HELPLINE(s) assert(optex == -1); \ |
390 | fprintf(fp, "%7s%c %s\n", prefix?prefix:"", prefix?':':' ', s); \ |
391 | HELPRESET; |
392 | #define HELPOPT(s) assert((optex == 1 && optarg) || (optex == 0 && !optarg)); \ |
393 | assert(shortopt || longopt); \ |
394 | i = fprintf(fp, "%7s%c %s%s%s%s%s", prefix?prefix:"", prefix?':':' ', \ |
395 | shortopt?shortopt:"", shortopt&&longopt?", ":"", longopt?longopt:"", \ |
396 | optarg?" ":"", optarg?optarg:""); \ |
397 | fprintf(fp, "%*s %s\n", i<32?32-i:0,"",s); HELPRESET; |
398 | |
399 | HELPRESET; |
400 | OPTHELP(HELPNOVAL, HELPVAL, HELPSHORT, HELPLONG, |
401 | HELPPFX, HELPARG, HELPLINE, HELPOPT); |
402 | |
403 | #undef HELPRESET |
404 | #undef HELPNOVAL |
405 | #undef HELPVAL |
406 | #undef HELPSHORT |
407 | #undef HELPLONG |
408 | #undef HELPPFX |
409 | #undef HELPARG |
410 | #undef HELPLINE |
411 | #undef HELPOPT |
412 | } |
413 | |
f2e52893 |
414 | static time_t parse_age(time_t now, const char *agestr) |
415 | { |
416 | time_t t; |
417 | struct tm tm; |
418 | int nunits; |
419 | char unit[2]; |
420 | |
421 | t = now; |
422 | |
423 | if (2 != sscanf(agestr, "%d%1[DdWwMmYy]", &nunits, unit)) { |
424 | fprintf(stderr, "%s: age specification should be a number followed by" |
425 | " one of d,w,m,y\n", PNAME); |
426 | exit(1); |
427 | } |
428 | |
429 | if (unit[0] == 'd') { |
430 | t -= 86400 * nunits; |
431 | } else if (unit[0] == 'w') { |
432 | t -= 86400 * 7 * nunits; |
433 | } else { |
434 | int ym; |
435 | |
436 | tm = *localtime(&t); |
437 | ym = tm.tm_year * 12 + tm.tm_mon; |
438 | |
439 | if (unit[0] == 'm') |
440 | ym -= nunits; |
441 | else |
442 | ym -= 12 * nunits; |
443 | |
444 | tm.tm_year = ym / 12; |
445 | tm.tm_mon = ym % 12; |
446 | |
447 | t = mktime(&tm); |
448 | } |
449 | |
450 | return t; |
451 | } |
452 | |
70322ae3 |
453 | int main(int argc, char **argv) |
454 | { |
455 | int fd, count; |
456 | struct ctx actx, *ctx = &actx; |
457 | struct stat st; |
458 | off_t totalsize, realsize; |
459 | void *mappedfile; |
460 | triewalk *tw; |
461 | indexbuild *ib; |
14601b5d |
462 | const struct trie_file *tf, *prevtf; |
bf53e756 |
463 | char *filename = PNAME ".dat"; |
70322ae3 |
464 | int doing_opts = 1; |
355c3af7 |
465 | enum { TEXT, HTML, SCAN, DUMP, SCANDUMP, LOAD, HTTPD, REMOVE }; |
444c684c |
466 | struct action { |
467 | int mode; |
468 | char *arg; |
469 | } *actions = NULL; |
470 | int nactions = 0, actionsize = 0, action; |
f2e52893 |
471 | time_t now = time(NULL); |
472 | time_t textcutoff = now, htmlnewest = now, htmloldest = now; |
473 | int htmlautoagerange = 1; |
1e8d78b9 |
474 | const char *httpserveraddr = NULL; |
475 | int httpserverport = 0; |
476 | const char *httpauthdata = NULL; |
812e4bf2 |
477 | int auth = HTTPD_AUTH_MAGIC | HTTPD_AUTH_BASIC; |
8b1f55d6 |
478 | int progress = 1; |
9d0b9596 |
479 | struct inclusion_exclusion *inex = NULL; |
480 | int ninex = 0, inexsize = 0; |
481 | int crossfs = 0; |
16e591d6 |
482 | int tqdepth = 1; |
05b0f827 |
483 | int fakediratimes = 1; |
f59a5d34 |
484 | int mtime = 0; |
70322ae3 |
485 | |
56fa1896 |
486 | #ifdef DEBUG_MAD_OPTION_PARSING_MACROS |
487 | { |
488 | static const char *const optnames[NOPTIONS] = { |
489 | OPTIONS(STRING,STRING,IGNORE,IGNORE) |
490 | }; |
491 | int i; |
492 | for (i = 0; i < NSHORTOPTS; i++) |
493 | printf("-%c == %s [%s]\n", shortopts[i], optnames[shortvals[i]], |
494 | opthasval[shortvals[i]] ? "value" : "no value"); |
495 | for (i = 0; i < NLONGOPTS; i++) |
496 | printf("--%s == %s [%s]\n", longopts[i], optnames[longvals[i]], |
497 | opthasval[longvals[i]] ? "value" : "no value"); |
498 | } |
499 | #endif |
500 | |
70322ae3 |
501 | while (--argc > 0) { |
502 | char *p = *++argv; |
70322ae3 |
503 | |
504 | if (doing_opts && *p == '-') { |
56fa1896 |
505 | int wordstart = 1; |
506 | |
70322ae3 |
507 | if (!strcmp(p, "--")) { |
508 | doing_opts = 0; |
56fa1896 |
509 | continue; |
510 | } |
511 | |
512 | p++; |
513 | while (*p) { |
514 | int optid = -1; |
515 | int i; |
516 | char *optval; |
517 | |
518 | if (wordstart && *p == '-') { |
70322ae3 |
519 | /* |
56fa1896 |
520 | * GNU-style long option. |
70322ae3 |
521 | */ |
56fa1896 |
522 | p++; |
523 | optval = strchr(p, '='); |
524 | if (optval) |
525 | *optval++ = '\0'; |
526 | |
527 | for (i = 0; i < NLONGOPTS; i++) { |
528 | const char *opt = longopts[i], *s = p; |
529 | int match = 1; |
530 | /* |
531 | * The underscores in the option names |
532 | * defined above may be given by the user |
533 | * as underscores or dashes, or omitted |
534 | * entirely. |
535 | */ |
536 | while (*opt) { |
537 | if (*opt == '_') { |
538 | if (*s == '-' || *s == '_') |
539 | s++; |
540 | } else { |
541 | if (*opt != *s) { |
542 | match = 0; |
543 | break; |
544 | } |
545 | s++; |
546 | } |
547 | opt++; |
548 | } |
549 | if (match && !*s) { |
550 | optid = longvals[i]; |
551 | break; |
70322ae3 |
552 | } |
553 | } |
56fa1896 |
554 | |
555 | if (optid < 0) { |
556 | fprintf(stderr, "%s: unrecognised option '--%s'\n", |
557 | PNAME, p); |
558 | return 1; |
559 | } |
560 | |
561 | if (!opthasval[optid]) { |
562 | if (optval) { |
563 | fprintf(stderr, "%s: unexpected argument to option" |
564 | " '--%s'\n", PNAME, p); |
812e4bf2 |
565 | return 1; |
566 | } |
56fa1896 |
567 | } else { |
568 | if (!optval) { |
569 | if (--argc > 0) { |
570 | optval = *++argv; |
571 | } else { |
572 | fprintf(stderr, "%s: option '--%s' expects" |
573 | " an argument\n", PNAME, p); |
574 | return 1; |
575 | } |
9d0b9596 |
576 | } |
70322ae3 |
577 | } |
56fa1896 |
578 | |
579 | p += strlen(p); /* finished with this argument word */ |
70322ae3 |
580 | } else { |
56fa1896 |
581 | /* |
582 | * Short option. |
583 | */ |
70322ae3 |
584 | char c = *p++; |
585 | |
56fa1896 |
586 | for (i = 0; i < NSHORTOPTS; i++) |
587 | if (c == shortopts[i]) { |
588 | optid = shortvals[i]; |
589 | break; |
590 | } |
591 | |
592 | if (optid < 0) { |
593 | fprintf(stderr, "%s: unrecognised option '-%c'\n", |
594 | PNAME, c); |
595 | return 1; |
596 | } |
597 | |
598 | if (opthasval[optid]) { |
70322ae3 |
599 | if (*p) { |
600 | optval = p; |
601 | p += strlen(p); |
602 | } else if (--argc > 0) { |
603 | optval = *++argv; |
604 | } else { |
56fa1896 |
605 | fprintf(stderr, "%s: option '-%c' expects" |
70322ae3 |
606 | " an argument\n", PNAME, c); |
607 | return 1; |
608 | } |
56fa1896 |
609 | } else { |
610 | optval = NULL; |
611 | } |
612 | } |
613 | |
614 | wordstart = 0; |
615 | |
616 | /* |
617 | * Now actually process the option. |
618 | */ |
619 | switch (optid) { |
620 | case OPT_HELP: |
e9e7a1bf |
621 | usage(stdout); |
56fa1896 |
622 | return 0; |
623 | case OPT_VERSION: |
e6fde1f7 |
624 | #ifdef PACKAGE_VERSION |
625 | printf("%s, revision %s\n", PNAME, PACKAGE_VERSION); |
626 | #else |
627 | printf("%s: version number not available when not built" |
628 | " via automake\n", PNAME); |
629 | #endif |
56fa1896 |
630 | return 0; |
631 | case OPT_LICENCE: |
5a29503d |
632 | { |
633 | extern const char *const licence[]; |
634 | int i; |
635 | |
636 | for (i = 0; licence[i]; i++) |
637 | fputs(licence[i], stdout); |
638 | |
639 | return 0; |
640 | } |
56fa1896 |
641 | return 0; |
642 | case OPT_SCAN: |
444c684c |
643 | if (nactions >= actionsize) { |
644 | actionsize = nactions * 3 / 2 + 16; |
645 | actions = sresize(actions, actionsize, struct action); |
646 | } |
647 | actions[nactions].mode = SCAN; |
648 | actions[nactions].arg = optval; |
649 | nactions++; |
56fa1896 |
650 | break; |
84849cbd |
651 | case OPT_SCANDUMP: |
444c684c |
652 | if (nactions >= actionsize) { |
653 | actionsize = nactions * 3 / 2 + 16; |
654 | actions = sresize(actions, actionsize, struct action); |
655 | } |
656 | actions[nactions].mode = SCANDUMP; |
657 | actions[nactions].arg = optval; |
658 | nactions++; |
84849cbd |
659 | break; |
56fa1896 |
660 | case OPT_DUMP: |
444c684c |
661 | if (nactions >= actionsize) { |
662 | actionsize = nactions * 3 / 2 + 16; |
663 | actions = sresize(actions, actionsize, struct action); |
664 | } |
665 | actions[nactions].mode = DUMP; |
666 | actions[nactions].arg = NULL; |
667 | nactions++; |
56fa1896 |
668 | break; |
84849cbd |
669 | case OPT_LOAD: |
444c684c |
670 | if (nactions >= actionsize) { |
671 | actionsize = nactions * 3 / 2 + 16; |
672 | actions = sresize(actions, actionsize, struct action); |
673 | } |
674 | actions[nactions].mode = LOAD; |
675 | actions[nactions].arg = NULL; |
676 | nactions++; |
84849cbd |
677 | break; |
56fa1896 |
678 | case OPT_TEXT: |
444c684c |
679 | if (nactions >= actionsize) { |
680 | actionsize = nactions * 3 / 2 + 16; |
681 | actions = sresize(actions, actionsize, struct action); |
682 | } |
683 | actions[nactions].mode = TEXT; |
684 | actions[nactions].arg = optval; |
685 | nactions++; |
56fa1896 |
686 | break; |
687 | case OPT_HTML: |
444c684c |
688 | if (nactions >= actionsize) { |
689 | actionsize = nactions * 3 / 2 + 16; |
690 | actions = sresize(actions, actionsize, struct action); |
691 | } |
692 | actions[nactions].mode = HTML; |
693 | actions[nactions].arg = optval; |
694 | nactions++; |
56fa1896 |
695 | break; |
696 | case OPT_HTTPD: |
444c684c |
697 | if (nactions >= actionsize) { |
698 | actionsize = nactions * 3 / 2 + 16; |
699 | actions = sresize(actions, actionsize, struct action); |
700 | } |
701 | actions[nactions].mode = HTTPD; |
702 | actions[nactions].arg = NULL; |
703 | nactions++; |
56fa1896 |
704 | break; |
355c3af7 |
705 | case OPT_REMOVE: |
706 | if (nactions >= actionsize) { |
707 | actionsize = nactions * 3 / 2 + 16; |
708 | actions = sresize(actions, actionsize, struct action); |
709 | } |
710 | actions[nactions].mode = REMOVE; |
711 | actions[nactions].arg = NULL; |
712 | nactions++; |
713 | break; |
56fa1896 |
714 | case OPT_PROGRESS: |
715 | progress = 2; |
716 | break; |
717 | case OPT_NOPROGRESS: |
718 | progress = 0; |
719 | break; |
720 | case OPT_TTYPROGRESS: |
721 | progress = 1; |
722 | break; |
723 | case OPT_CROSSFS: |
724 | crossfs = 1; |
725 | break; |
726 | case OPT_NOCROSSFS: |
727 | crossfs = 0; |
728 | break; |
05b0f827 |
729 | case OPT_DIRATIME: |
730 | fakediratimes = 0; |
731 | break; |
732 | case OPT_NODIRATIME: |
733 | fakediratimes = 1; |
734 | break; |
f59a5d34 |
735 | case OPT_MTIME: |
736 | mtime = 1; |
737 | break; |
56fa1896 |
738 | case OPT_DATAFILE: |
739 | filename = optval; |
740 | break; |
16e591d6 |
741 | case OPT_TQDEPTH: |
742 | tqdepth = atoi(optval); |
743 | break; |
56fa1896 |
744 | case OPT_MINAGE: |
f2e52893 |
745 | textcutoff = parse_age(now, optval); |
746 | break; |
747 | case OPT_AGERANGE: |
748 | if (!strcmp(optval, "auto")) { |
749 | htmlautoagerange = 1; |
750 | } else { |
751 | char *q = optval + strcspn(optval, "-:"); |
752 | if (*q) |
753 | *q++ = '\0'; |
754 | htmloldest = parse_age(now, optval); |
755 | htmlnewest = *q ? parse_age(now, q) : now; |
756 | htmlautoagerange = 0; |
757 | } |
56fa1896 |
758 | break; |
1e8d78b9 |
759 | case OPT_SERVERADDR: |
760 | { |
761 | char *port; |
762 | if (optval[0] == '[' && |
763 | (port = strchr(optval, ']')) != NULL) |
764 | port++; |
765 | else |
766 | port = optval; |
767 | port += strcspn(port, ":"); |
768 | if (port) |
769 | *port++ = '\0'; |
770 | httpserveraddr = optval; |
771 | httpserverport = atoi(port); |
772 | } |
773 | break; |
56fa1896 |
774 | case OPT_AUTH: |
775 | if (!strcmp(optval, "magic")) |
776 | auth = HTTPD_AUTH_MAGIC; |
777 | else if (!strcmp(optval, "basic")) |
778 | auth = HTTPD_AUTH_BASIC; |
779 | else if (!strcmp(optval, "none")) |
780 | auth = HTTPD_AUTH_NONE; |
781 | else if (!strcmp(optval, "default")) |
782 | auth = HTTPD_AUTH_MAGIC | HTTPD_AUTH_BASIC; |
f2e52893 |
783 | else if (!strcmp(optval, "help") || |
784 | !strcmp(optval, "list")) { |
bf53e756 |
785 | printf(PNAME ": supported HTTP authentication types" |
f2e52893 |
786 | " are:\n" |
787 | " magic use Linux /proc/net/tcp to" |
788 | " determine owner of peer socket\n" |
789 | " basic HTTP Basic username and" |
790 | " password authentication\n" |
791 | " default use 'magic' if possible, " |
792 | " otherwise fall back to 'basic'\n" |
793 | " none unauthenticated HTTP (if" |
794 | " the data file is non-confidential)\n"); |
795 | return 0; |
796 | } else { |
56fa1896 |
797 | fprintf(stderr, "%s: unrecognised authentication" |
798 | " type '%s'\n%*s options are 'magic'," |
799 | " 'basic', 'none', 'default'\n", |
800 | PNAME, optval, (int)strlen(PNAME), ""); |
801 | return 1; |
802 | } |
803 | break; |
1e8d78b9 |
804 | case OPT_AUTHFILE: |
805 | case OPT_AUTHFD: |
806 | { |
807 | int fd; |
808 | char namebuf[40]; |
809 | const char *name; |
810 | char *authbuf; |
811 | int authlen, authsize; |
812 | int ret; |
813 | |
814 | if (optid == OPT_AUTHFILE) { |
815 | fd = open(optval, O_RDONLY); |
816 | if (fd < 0) { |
817 | fprintf(stderr, "%s: %s: open: %s\n", PNAME, |
818 | optval, strerror(errno)); |
819 | return 1; |
820 | } |
821 | name = optval; |
822 | } else { |
823 | fd = atoi(optval); |
824 | name = namebuf; |
825 | sprintf(namebuf, "fd %d", fd); |
826 | } |
827 | |
828 | authlen = 0; |
829 | authsize = 256; |
830 | authbuf = snewn(authsize, char); |
831 | while ((ret = read(fd, authbuf+authlen, |
832 | authsize-authlen)) > 0) { |
833 | authlen += ret; |
834 | if ((authsize - authlen) < (authsize / 16)) { |
835 | authsize = authlen * 3 / 2 + 4096; |
836 | authbuf = sresize(authbuf, authsize, char); |
837 | } |
838 | } |
839 | if (ret < 0) { |
840 | fprintf(stderr, "%s: %s: read: %s\n", PNAME, |
841 | name, strerror(errno)); |
842 | return 1; |
843 | } |
844 | if (optid == OPT_AUTHFILE) |
845 | close(fd); |
846 | httpauthdata = authbuf; |
847 | } |
848 | break; |
56fa1896 |
849 | case OPT_INCLUDE: |
850 | case OPT_INCLUDEPATH: |
851 | case OPT_EXCLUDE: |
852 | case OPT_EXCLUDEPATH: |
0ba55302 |
853 | case OPT_PRUNE: |
854 | case OPT_PRUNEPATH: |
56fa1896 |
855 | if (ninex >= inexsize) { |
856 | inexsize = ninex * 3 / 2 + 16; |
857 | inex = sresize(inex, inexsize, |
858 | struct inclusion_exclusion); |
859 | } |
860 | inex[ninex].path = (optid == OPT_INCLUDEPATH || |
0ba55302 |
861 | optid == OPT_EXCLUDEPATH || |
862 | optid == OPT_PRUNEPATH); |
863 | inex[ninex].type = (optid == OPT_INCLUDE ? 1 : |
864 | optid == OPT_INCLUDEPATH ? 1 : |
865 | optid == OPT_EXCLUDE ? 0 : |
866 | optid == OPT_EXCLUDEPATH ? 0 : |
867 | optid == OPT_PRUNE ? -1 : |
868 | /* optid == OPT_PRUNEPATH ? */ -1); |
56fa1896 |
869 | inex[ninex].wildcard = optval; |
870 | ninex++; |
871 | break; |
872 | } |
873 | } |
70322ae3 |
874 | } else { |
e9e7a1bf |
875 | fprintf(stderr, "%s: unexpected argument '%s'\n", PNAME, p); |
876 | return 1; |
70322ae3 |
877 | } |
878 | } |
879 | |
444c684c |
880 | if (nactions == 0) { |
e9e7a1bf |
881 | usage(stderr); |
882 | return 1; |
444c684c |
883 | } |
884 | |
885 | for (action = 0; action < nactions; action++) { |
886 | int mode = actions[action].mode; |
887 | |
888 | if (mode == SCAN || mode == SCANDUMP || mode == LOAD) { |
889 | const char *scandir = actions[action].arg; |
14601b5d |
890 | |
444c684c |
891 | if (mode == LOAD) { |
892 | char *buf = fgetline(stdin); |
893 | unsigned newpathsep; |
894 | buf[strcspn(buf, "\r\n")] = '\0'; |
bf53e756 |
895 | if (1 != sscanf(buf, DUMPHDR "%x", |
444c684c |
896 | &newpathsep)) { |
897 | fprintf(stderr, "%s: header in dump file not recognised\n", |
898 | PNAME); |
899 | return 1; |
900 | } |
901 | pathsep = (char)newpathsep; |
902 | sfree(buf); |
84849cbd |
903 | } |
70322ae3 |
904 | |
444c684c |
905 | if (mode == SCAN || mode == LOAD) { |
906 | /* |
907 | * Prepare to write out the index file. |
908 | */ |
cc7db507 |
909 | fd = open(filename, O_RDWR | O_TRUNC | O_CREAT, |
910 | S_IRUSR | S_IWUSR); |
444c684c |
911 | if (fd < 0) { |
912 | fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, |
913 | strerror(errno)); |
914 | return 1; |
915 | } |
916 | if (fstat(fd, &st) < 0) { |
bf53e756 |
917 | perror(PNAME ": fstat"); |
444c684c |
918 | return 1; |
919 | } |
920 | ctx->datafile_dev = st.st_dev; |
921 | ctx->datafile_ino = st.st_ino; |
922 | ctx->straight_to_dump = 0; |
923 | } else { |
924 | ctx->datafile_dev = -1; |
925 | ctx->datafile_ino = -1; |
926 | ctx->straight_to_dump = 1; |
84849cbd |
927 | } |
444c684c |
928 | |
929 | if (mode == SCAN || mode == SCANDUMP) { |
930 | if (stat(scandir, &st) < 0) { |
931 | fprintf(stderr, "%s: %s: stat: %s\n", PNAME, scandir, |
932 | strerror(errno)); |
933 | return 1; |
934 | } |
935 | ctx->filesystem_dev = crossfs ? 0 : st.st_dev; |
84849cbd |
936 | } |
70322ae3 |
937 | |
444c684c |
938 | ctx->inex = inex; |
939 | ctx->ninex = ninex; |
940 | ctx->crossfs = crossfs; |
05b0f827 |
941 | ctx->fakeatimes = fakediratimes; |
f59a5d34 |
942 | ctx->usemtime = mtime; |
444c684c |
943 | |
944 | ctx->last_output_update = time(NULL); |
945 | |
946 | /* progress==1 means report progress only if stderr is a tty */ |
947 | if (progress == 1) |
948 | progress = isatty(2) ? 2 : 0; |
949 | ctx->progress = progress; |
950 | { |
951 | struct winsize ws; |
952 | if (progress && ioctl(2, TIOCGWINSZ, &ws) == 0) |
953 | ctx->progwidth = ws.ws_col - 1; |
954 | else |
955 | ctx->progwidth = 79; |
84849cbd |
956 | } |
84849cbd |
957 | |
444c684c |
958 | if (mode == SCANDUMP) |
bf53e756 |
959 | printf(DUMPHDR "%02x\n", (unsigned char)pathsep); |
8b1f55d6 |
960 | |
444c684c |
961 | /* |
962 | * Scan the directory tree, and write out the trie component |
963 | * of the data file. |
964 | */ |
965 | if (mode != SCANDUMP) { |
966 | ctx->tb = triebuild_new(fd); |
967 | } |
968 | if (mode == LOAD) { |
969 | char *buf; |
970 | int line = 2; |
971 | while ((buf = fgetline(stdin)) != NULL) { |
972 | struct trie_file tf; |
973 | char *p, *q; |
974 | |
975 | buf[strcspn(buf, "\r\n")] = '\0'; |
976 | |
977 | p = buf; |
978 | q = p; |
979 | while (*p && *p != ' ') p++; |
980 | if (!*p) { |
981 | fprintf(stderr, "%s: dump file line %d: expected at least" |
982 | " three fields\n", PNAME, line); |
983 | return 1; |
984 | } |
985 | *p++ = '\0'; |
986 | tf.size = strtoull(q, NULL, 10); |
987 | q = p; |
988 | while (*p && *p != ' ') p++; |
989 | if (!*p) { |
990 | fprintf(stderr, "%s: dump file line %d: expected at least" |
991 | " three fields\n", PNAME, line); |
992 | return 1; |
993 | } |
994 | *p++ = '\0'; |
995 | tf.atime = strtoull(q, NULL, 10); |
996 | q = buf; |
997 | while (*p) { |
998 | int c = *p; |
999 | if (*p == '%') { |
1000 | int i; |
1001 | p++; |
1002 | c = 0; |
1003 | for (i = 0; i < 2; i++) { |
de693987 |
1004 | c *= 16; |
444c684c |
1005 | if (*p >= '0' && *p <= '9') |
1006 | c += *p - '0'; |
1007 | else if (*p >= 'A' && *p <= 'F') |
1008 | c += *p - ('A' - 10); |
1009 | else if (*p >= 'a' && *p <= 'f') |
1010 | c += *p - ('a' - 10); |
1011 | else { |
1012 | fprintf(stderr, "%s: dump file line %d: unable" |
1013 | " to parse hex escape\n", PNAME, line); |
1014 | } |
1015 | p++; |
1016 | } |
1017 | } |
1018 | *q++ = c; |
1019 | p++; |
1020 | } |
1021 | *q = '\0'; |
1022 | triebuild_add(ctx->tb, buf, &tf); |
1023 | sfree(buf); |
de693987 |
1024 | line++; |
444c684c |
1025 | } |
1026 | } else { |
09fd7619 |
1027 | du(scandir, gotdata, scan_error, ctx); |
444c684c |
1028 | } |
1029 | if (mode != SCANDUMP) { |
14601b5d |
1030 | size_t maxpathlen; |
522edd92 |
1031 | size_t delta; |
14601b5d |
1032 | char *buf, *prevbuf; |
1033 | |
444c684c |
1034 | count = triebuild_finish(ctx->tb); |
1035 | triebuild_free(ctx->tb); |
84849cbd |
1036 | |
444c684c |
1037 | if (ctx->progress) { |
1038 | fprintf(stderr, "%-*s\r", ctx->progwidth, ""); |
1039 | fflush(stderr); |
1040 | } |
84849cbd |
1041 | |
444c684c |
1042 | /* |
1043 | * Work out how much space the cumulative index trees |
1044 | * will take; enlarge the file, and memory-map it. |
1045 | */ |
1046 | if (fstat(fd, &st) < 0) { |
bf53e756 |
1047 | perror(PNAME ": fstat"); |
444c684c |
1048 | return 1; |
1049 | } |
84849cbd |
1050 | |
522edd92 |
1051 | printf("Built pathname index, %d entries," |
1052 | " %llu bytes of index\n", count, |
50e82fdc |
1053 | (unsigned long long)st.st_size); |
444c684c |
1054 | |
522edd92 |
1055 | totalsize = index_initial_size(st.st_size, count); |
1056 | totalsize += totalsize / 10; |
444c684c |
1057 | |
1058 | if (lseek(fd, totalsize-1, SEEK_SET) < 0) { |
bf53e756 |
1059 | perror(PNAME ": lseek"); |
84849cbd |
1060 | return 1; |
1061 | } |
444c684c |
1062 | if (write(fd, "\0", 1) < 1) { |
bf53e756 |
1063 | perror(PNAME ": write"); |
84849cbd |
1064 | return 1; |
1065 | } |
444c684c |
1066 | |
444c684c |
1067 | mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0); |
1068 | if (!mappedfile) { |
bf53e756 |
1069 | perror(PNAME ": mmap"); |
444c684c |
1070 | return 1; |
84849cbd |
1071 | } |
444c684c |
1072 | |
05b0f827 |
1073 | if (fakediratimes) { |
1074 | printf("Faking directory atimes\n"); |
1075 | trie_fake_dir_atimes(mappedfile); |
1076 | } |
1077 | |
1078 | printf("Building index\n"); |
522edd92 |
1079 | ib = indexbuild_new(mappedfile, st.st_size, count, &delta); |
14601b5d |
1080 | maxpathlen = trie_maxpathlen(mappedfile); |
1081 | buf = snewn(maxpathlen, char); |
1082 | prevbuf = snewn(maxpathlen, char); |
444c684c |
1083 | tw = triewalk_new(mappedfile); |
14601b5d |
1084 | prevbuf[0] = '\0'; |
1085 | tf = triewalk_next(tw, buf); |
1086 | assert(tf); |
1087 | while (1) { |
1088 | int i; |
1089 | |
522edd92 |
1090 | if (totalsize - indexbuild_realsize(ib) < delta) { |
1091 | /* |
1092 | * Unmap the file, grow it, and remap it. |
1093 | */ |
1094 | munmap(mappedfile, totalsize); |
1095 | |
1096 | totalsize += delta; |
1097 | totalsize += totalsize / 10; |
1098 | |
1099 | if (lseek(fd, totalsize-1, SEEK_SET) < 0) { |
1100 | perror(PNAME ": lseek"); |
1101 | return 1; |
1102 | } |
1103 | if (write(fd, "\0", 1) < 1) { |
1104 | perror(PNAME ": write"); |
1105 | return 1; |
1106 | } |
1107 | |
1108 | mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0); |
1109 | if (!mappedfile) { |
1110 | perror(PNAME ": mmap"); |
1111 | return 1; |
1112 | } |
1113 | |
1114 | indexbuild_rebase(ib, mappedfile); |
1115 | triewalk_rebase(tw, mappedfile); |
1116 | } |
1117 | |
14601b5d |
1118 | /* |
1119 | * Get the next file from the index. So we are |
1120 | * currently holding, and have not yet |
1121 | * indexed, prevtf (with pathname prevbuf) and |
1122 | * tf (with pathname buf). |
1123 | */ |
1124 | prevtf = tf; |
1125 | memcpy(prevbuf, buf, maxpathlen); |
1126 | tf = triewalk_next(tw, buf); |
1127 | |
1128 | if (!tf) |
1129 | buf[0] = '\0'; |
1130 | |
1131 | /* |
1132 | * Find the first differing character position |
1133 | * between our two pathnames. |
1134 | */ |
1135 | for (i = 0; prevbuf[i] && prevbuf[i] == buf[i]; i++); |
1136 | |
1137 | /* |
1138 | * If prevbuf was a directory name and buf is |
1139 | * something inside that directory, then |
1140 | * trie_before() will be called on prevbuf |
1141 | * itself. Hence we must drop a tag before it, |
1142 | * so that the resulting index is usable. |
1143 | */ |
1144 | if ((!prevbuf[i] && (buf[i] == pathsep || |
1145 | (i > 0 && buf[i-1] == pathsep)))) |
1146 | indexbuild_tag(ib); |
1147 | |
1148 | /* |
1149 | * Add prevtf to the index. |
1150 | */ |
1151 | indexbuild_add(ib, prevtf); |
1152 | |
1153 | if (!tf) { |
1154 | /* |
1155 | * Drop an unconditional final tag, and |
1156 | * get out of this loop. |
1157 | */ |
1158 | indexbuild_tag(ib); |
1159 | break; |
1160 | } |
14601b5d |
1161 | |
1162 | /* |
1163 | * If prevbuf was a filename inside some |
1164 | * directory which buf is outside, then |
1165 | * trie_before() will be called on some |
1166 | * pathname either equal to buf or epsilon |
1167 | * less than it. Either way, we're going to |
1168 | * need to drop a tag after prevtf. |
1169 | */ |
1170 | if (strchr(prevbuf+i, pathsep) || !tf) |
1171 | indexbuild_tag(ib); |
1172 | } |
1173 | |
444c684c |
1174 | triewalk_free(tw); |
1175 | realsize = indexbuild_realsize(ib); |
1176 | indexbuild_free(ib); |
1177 | |
1178 | munmap(mappedfile, totalsize); |
1179 | ftruncate(fd, realsize); |
1180 | close(fd); |
522edd92 |
1181 | printf("Final index file size = %llu bytes\n", |
50e82fdc |
1182 | (unsigned long long)realsize); |
84849cbd |
1183 | } |
444c684c |
1184 | } else if (mode == TEXT) { |
1185 | char *querydir = actions[action].arg; |
1186 | size_t pathlen; |
70322ae3 |
1187 | |
444c684c |
1188 | fd = open(filename, O_RDONLY); |
1189 | if (fd < 0) { |
1190 | fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, |
1191 | strerror(errno)); |
1192 | return 1; |
1193 | } |
1194 | if (fstat(fd, &st) < 0) { |
bf53e756 |
1195 | perror(PNAME ": fstat"); |
444c684c |
1196 | return 1; |
1197 | } |
1198 | totalsize = st.st_size; |
1199 | mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); |
1200 | if (!mappedfile) { |
bf53e756 |
1201 | perror(PNAME ": mmap"); |
444c684c |
1202 | return 1; |
84849cbd |
1203 | } |
444c684c |
1204 | pathsep = trie_pathsep(mappedfile); |
70322ae3 |
1205 | |
84849cbd |
1206 | /* |
444c684c |
1207 | * Trim trailing slash, just in case. |
84849cbd |
1208 | */ |
444c684c |
1209 | pathlen = strlen(querydir); |
1210 | if (pathlen > 0 && querydir[pathlen-1] == pathsep) |
1211 | querydir[--pathlen] = '\0'; |
1212 | |
16e591d6 |
1213 | text_query(mappedfile, querydir, textcutoff, tqdepth); |
56cae6e1 |
1214 | |
1215 | munmap(mappedfile, totalsize); |
444c684c |
1216 | } else if (mode == HTML) { |
1217 | char *querydir = actions[action].arg; |
1218 | size_t pathlen; |
1219 | struct html_config cfg; |
1220 | unsigned long xi; |
1221 | char *html; |
1222 | |
1223 | fd = open(filename, O_RDONLY); |
1224 | if (fd < 0) { |
1225 | fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, |
1226 | strerror(errno)); |
1227 | return 1; |
1228 | } |
84849cbd |
1229 | if (fstat(fd, &st) < 0) { |
bf53e756 |
1230 | perror(PNAME ": fstat"); |
84849cbd |
1231 | return 1; |
1232 | } |
444c684c |
1233 | totalsize = st.st_size; |
1234 | mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); |
1235 | if (!mappedfile) { |
bf53e756 |
1236 | perror(PNAME ": mmap"); |
444c684c |
1237 | return 1; |
1238 | } |
1239 | pathsep = trie_pathsep(mappedfile); |
70322ae3 |
1240 | |
444c684c |
1241 | /* |
1242 | * Trim trailing slash, just in case. |
1243 | */ |
1244 | pathlen = strlen(querydir); |
1245 | if (pathlen > 0 && querydir[pathlen-1] == pathsep) |
1246 | querydir[--pathlen] = '\0'; |
1247 | |
1248 | xi = trie_before(mappedfile, querydir); |
1249 | cfg.format = NULL; |
1250 | cfg.autoage = htmlautoagerange; |
1251 | cfg.oldest = htmloldest; |
1252 | cfg.newest = htmlnewest; |
1253 | html = html_query(mappedfile, xi, &cfg); |
1254 | fputs(html, stdout); |
56cae6e1 |
1255 | |
1256 | munmap(mappedfile, totalsize); |
444c684c |
1257 | } else if (mode == DUMP) { |
1258 | size_t maxpathlen; |
1259 | char *buf; |
70322ae3 |
1260 | |
444c684c |
1261 | fd = open(filename, O_RDONLY); |
1262 | if (fd < 0) { |
1263 | fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, |
1264 | strerror(errno)); |
84849cbd |
1265 | return 1; |
1266 | } |
444c684c |
1267 | if (fstat(fd, &st) < 0) { |
bf53e756 |
1268 | perror(PNAME ": fstat"); |
84849cbd |
1269 | return 1; |
1270 | } |
444c684c |
1271 | totalsize = st.st_size; |
1272 | mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); |
84849cbd |
1273 | if (!mappedfile) { |
bf53e756 |
1274 | perror(PNAME ": mmap"); |
84849cbd |
1275 | return 1; |
1276 | } |
444c684c |
1277 | pathsep = trie_pathsep(mappedfile); |
1278 | |
1279 | maxpathlen = trie_maxpathlen(mappedfile); |
1280 | buf = snewn(maxpathlen, char); |
84849cbd |
1281 | |
bf53e756 |
1282 | printf(DUMPHDR "%02x\n", (unsigned char)pathsep); |
84849cbd |
1283 | tw = triewalk_new(mappedfile); |
444c684c |
1284 | while ((tf = triewalk_next(tw, buf)) != NULL) |
1285 | dump_line(buf, tf); |
84849cbd |
1286 | triewalk_free(tw); |
56cae6e1 |
1287 | |
1288 | munmap(mappedfile, totalsize); |
444c684c |
1289 | } else if (mode == HTTPD) { |
1290 | struct html_config pcfg; |
1291 | struct httpd_config dcfg; |
70322ae3 |
1292 | |
444c684c |
1293 | fd = open(filename, O_RDONLY); |
1294 | if (fd < 0) { |
1295 | fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, |
1296 | strerror(errno)); |
1297 | return 1; |
1298 | } |
1299 | if (fstat(fd, &st) < 0) { |
bf53e756 |
1300 | perror(PNAME ": fstat"); |
444c684c |
1301 | return 1; |
1302 | } |
1303 | totalsize = st.st_size; |
1304 | mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); |
1305 | if (!mappedfile) { |
bf53e756 |
1306 | perror(PNAME ": mmap"); |
444c684c |
1307 | return 1; |
1308 | } |
1309 | pathsep = trie_pathsep(mappedfile); |
1310 | |
1311 | dcfg.address = httpserveraddr; |
1312 | dcfg.port = httpserverport; |
1313 | dcfg.basicauthdata = httpauthdata; |
1314 | pcfg.format = NULL; |
1315 | pcfg.autoage = htmlautoagerange; |
1316 | pcfg.oldest = htmloldest; |
1317 | pcfg.newest = htmlnewest; |
1318 | run_httpd(mappedfile, auth, &dcfg, &pcfg); |
56cae6e1 |
1319 | munmap(mappedfile, totalsize); |
355c3af7 |
1320 | } else if (mode == REMOVE) { |
1321 | if (remove(filename) < 0) { |
1322 | fprintf(stderr, "%s: %s: remove: %s\n", PNAME, filename, |
1323 | strerror(errno)); |
1324 | return 1; |
1325 | } |
70322ae3 |
1326 | } |
70322ae3 |
1327 | } |
1328 | |
1329 | return 0; |
1330 | } |