2 * Main program for agedu.
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.
25 void fatal(const char *fmt
, ...)
28 fprintf(stderr
, "%s: ", PNAME
);
30 vfprintf(stderr
, fmt
, ap
);
32 fprintf(stderr
, "\n");
36 struct inclusion_exclusion
{
44 dev_t datafile_dev
, filesystem_dev
;
46 time_t last_output_update
;
47 int progress
, progwidth
;
49 struct inclusion_exclusion
*inex
;
56 static void dump_line(const char *pathname
, const struct trie_file
*tf
)
59 if (printf("%llu %llu ", tf
->size
, tf
->atime
) < 0) goto error
;
60 for (p
= pathname
; *p
; p
++) {
61 if (*p
>= ' ' && *p
< 127 && *p
!= '%') {
62 if (putchar(*p
) == EOF
) goto error
;
64 if (printf("%%%02x", (unsigned char)*p
) < 0) goto error
;
67 if (putchar('\n') == EOF
) goto error
;
70 fatal("standard output: %s", strerror(errno
));
73 static int gotdata(void *vctx
, const char *pathname
, const STRUCT_STAT
*st
)
75 struct ctx
*ctx
= (struct ctx
*)vctx
;
76 struct trie_file file
;
82 * Filter out our own data file.
84 if (st
->st_dev
== ctx
->datafile_dev
&& st
->st_ino
== ctx
->datafile_ino
)
88 * Don't cross the streams^W^Wany file system boundary.
90 if (!ctx
->crossfs
&& st
->st_dev
!= ctx
->filesystem_dev
)
93 file
.size
= (unsigned long long)512 * st
->st_blocks
;
94 if (ctx
->usemtime
|| (ctx
->fakeatimes
&& S_ISDIR(st
->st_mode
)))
95 file
.atime
= st
->st_mtime
;
97 file
.atime
= max(st
->st_mtime
, st
->st_atime
);
100 * Filter based on wildcards.
103 filename
= strrchr(pathname
, pathsep
);
108 for (i
= 0; i
< ctx
->ninex
; i
++) {
109 if (fnmatch(ctx
->inex
[i
].wildcard
,
110 ctx
->inex
[i
].path ? pathname
: filename
, 0) == 0)
111 include
= ctx
->inex
[i
].type
;
114 return 0; /* ignore this entry and any subdirs */
117 * Here we are supposed to be filtering an entry out, but
118 * still recursing into it if it's a directory. However,
119 * we can't actually leave out any directory whose
120 * subdirectories we then look at. So we cheat, in that
121 * case, by setting the size to zero.
123 if (!S_ISDIR(st
->st_mode
))
124 return 0; /* just ignore */
129 if (ctx
->straight_to_dump
)
130 dump_line(pathname
, &file
);
132 triebuild_add(ctx
->tb
, pathname
, &file
);
136 if (t
!= ctx
->last_output_update
) {
137 fprintf(stderr
, "%-*.*s\r", ctx
->progwidth
, ctx
->progwidth
,
140 ctx
->last_output_update
= t
;
147 static void scan_error(void *vctx
, const char *fmt
, ...)
149 struct ctx
*ctx
= (struct ctx
*)vctx
;
153 fprintf(stderr
, "%-*s\r", ctx
->progwidth
, "");
157 fprintf(stderr
, "%s: ", PNAME
);
159 vfprintf(stderr
, fmt
, ap
);
162 ctx
->last_output_update
--; /* force a progress report next time */
165 static void text_query(const void *mappedfile
, const char *querydir
,
166 time_t t
, int showfiles
, int depth
, FILE *fp
)
170 unsigned long xi1
, xi2
;
171 unsigned long long size
;
173 maxpathlen
= trie_maxpathlen(mappedfile
);
174 pathbuf
= snewn(maxpathlen
+ 1, char);
177 * We want to query everything between the supplied filename
178 * (inclusive) and that filename with a ^A on the end
179 * (exclusive). So find the x indices for each.
181 strcpy(pathbuf
, querydir
);
182 make_successor(pathbuf
);
183 xi1
= trie_before(mappedfile
, querydir
);
184 xi2
= trie_before(mappedfile
, pathbuf
);
186 if (!showfiles
&& xi2
- xi1
== 1)
187 return; /* file, or empty dir => no display */
190 * Now do the lookups in the age index.
192 if (xi2
- xi1
== 1) {
194 * We are querying an individual file, so we should not
195 * depend on the index entries either side of the node,
196 * since they almost certainly don't both exist. Instead,
197 * just look up the file's size and atime in the main trie.
199 const struct trie_file
*f
= trie_getfile(mappedfile
, xi1
);
205 unsigned long long s1
, s2
;
206 s1
= index_query(mappedfile
, xi1
, t
);
207 s2
= index_query(mappedfile
, xi2
, t
);
212 return; /* no space taken up => no display */
216 * Now scan for first-level subdirectories and report
219 int newdepth
= (depth
> 0 ? depth
- 1 : depth
);
222 trie_getpath(mappedfile
, xi1
, pathbuf
);
223 text_query(mappedfile
, pathbuf
, t
, showfiles
, newdepth
, fp
);
224 make_successor(pathbuf
);
225 xi1
= trie_before(mappedfile
, pathbuf
);
229 /* Display in units of 1Kb */
230 fprintf(fp
, "%-11llu %s\n", (size
) / 1024, querydir
);
234 * Largely frivolous way to define all my command-line options. I
235 * present here a parametric macro which declares a series of
236 * _logical_ option identifiers, and for each one declares zero or
237 * more short option characters and zero or more long option
238 * words. Then I repeatedly invoke that macro with its arguments
239 * defined to be various other macros, which allows me to
242 * - define an enum allocating a distinct integer value to each
244 * - define a string consisting of precisely all the short option
246 * - define a string array consisting of all the long option
248 * - define (with help from auxiliary enums) integer arrays
249 * parallel to both of the above giving the logical option id
250 * for each physical short and long option
251 * - define an array indexed by logical option id indicating
252 * whether the option in question takes a value
253 * - define a function which prints out brief online help for all
256 * It's not at all clear to me that this trickery is actually
257 * particularly _efficient_ - it still, after all, requires going
258 * linearly through the option list at run time and doing a
259 * strcmp, whereas in an ideal world I'd have liked the lists of
260 * long and short options to be pre-sorted so that a binary search
261 * or some other more efficient lookup was possible. (Not that
262 * asymptotic algorithmic complexity is remotely vital in option
263 * parsing, but if I were doing this in, say, Lisp or something
264 * with an equivalently powerful preprocessor then once I'd had
265 * the idea of preparing the option-parsing data structures at
266 * compile time I would probably have made the effort to prepare
267 * them _properly_. I could have Perl generate me a source file
268 * from some sort of description, I suppose, but that would seem
269 * like overkill. And in any case, it's more of a challenge to
270 * achieve as much as possible by cunning use of cpp and enum than
271 * to just write some sensible and logical code in a Turing-
272 * complete language. I said it was largely frivolous :-)
274 * This approach does have the virtue that it brings together the
275 * option ids, option spellings and help text into a single
276 * combined list and defines them all in exactly one place. If I
277 * want to add a new option, or a new spelling for an option, I
278 * only have to modify the main OPTHELP macro below and then add
279 * code to process the new logical id.
281 * (Though, really, even that isn't ideal, since it still involves
282 * modifying the source file in more than one place. In a
283 * _properly_ ideal world, I'd be able to interleave the option
284 * definitions with the code fragments that process them. And then
285 * not bother defining logical identifiers for them at all - those
286 * would be automatically generated, since I wouldn't have any
287 * need to specify them manually in another part of the code.)
289 * One other helpful consequence of the enum-based structure here
290 * is that it causes a compiler error if I accidentally try to
291 * define the same option (short or long) twice.
294 #define OPTHELP(NOVAL, VAL, SHORT, LONG, HELPPFX, HELPARG, HELPLINE, HELPOPT) \
295 HELPPFX("usage") HELPLINE(PNAME " [options] action [action...]") \
297 VAL(SCAN) SHORT(s) LONG(scan) \
298 HELPARG("directory") HELPOPT("scan and index a directory") \
299 NOVAL(HTTPD) SHORT(w) LONG(web) LONG(server) LONG(httpd) \
300 HELPOPT("serve HTML reports from a temporary web server") \
301 VAL(TEXT) SHORT(t) LONG(text) \
302 HELPARG("subdir") HELPOPT("print a plain text report on a subdirectory") \
303 NOVAL(REMOVE) SHORT(R) LONG(remove) LONG(delete) LONG(unlink) \
304 HELPOPT("remove the index file") \
305 NOVAL(DUMP) SHORT(D) LONG(dump) HELPOPT("dump the index file on stdout") \
306 NOVAL(LOAD) SHORT(L) LONG(load) \
307 HELPOPT("load and index a dump file") \
308 VAL(SCANDUMP) SHORT(S) LONG(scan_dump) \
309 HELPARG("directory") HELPOPT("scan only, generating a dump") \
310 VAL(HTML) SHORT(H) LONG(html) \
311 HELPARG("subdir") HELPOPT("print an HTML report on a subdirectory") \
313 VAL(DATAFILE) SHORT(f) LONG(file) \
314 HELPARG("filename") HELPOPT("[most modes] specify index file") \
315 NOVAL(CROSSFS) LONG(cross_fs) \
316 HELPOPT("[--scan] cross filesystem boundaries") \
317 NOVAL(NOCROSSFS) LONG(no_cross_fs) \
318 HELPOPT("[--scan] stick to one filesystem") \
319 VAL(PRUNE) LONG(prune) \
320 HELPARG("wildcard") HELPOPT("[--scan] prune files matching pattern") \
321 VAL(PRUNEPATH) LONG(prune_path) \
322 HELPARG("wildcard") HELPOPT("[--scan] prune pathnames matching pattern") \
323 VAL(EXCLUDE) LONG(exclude) \
324 HELPARG("wildcard") HELPOPT("[--scan] exclude files matching pattern") \
325 VAL(EXCLUDEPATH) LONG(exclude_path) \
326 HELPARG("wildcard") HELPOPT("[--scan] exclude pathnames matching pattern") \
327 VAL(INCLUDE) LONG(include) \
328 HELPARG("wildcard") HELPOPT("[--scan] include files matching pattern") \
329 VAL(INCLUDEPATH) LONG(include_path) \
330 HELPARG("wildcard") HELPOPT("[--scan] include pathnames matching pattern") \
331 NOVAL(PROGRESS) LONG(progress) LONG(scan_progress) \
332 HELPOPT("[--scan] report progress on stderr") \
333 NOVAL(NOPROGRESS) LONG(no_progress) LONG(no_scan_progress) \
334 HELPOPT("[--scan] do not report progress") \
335 NOVAL(TTYPROGRESS) LONG(tty_progress) LONG(tty_scan_progress) \
336 LONG(progress_tty) LONG(scan_progress_tty) \
337 HELPOPT("[--scan] report progress if stderr is a tty") \
338 NOVAL(DIRATIME) LONG(dir_atime) LONG(dir_atimes) \
339 HELPOPT("[--scan,--load] keep real atimes on directories") \
340 NOVAL(NODIRATIME) LONG(no_dir_atime) LONG(no_dir_atimes) \
341 HELPOPT("[--scan,--load] fake atimes on directories") \
342 NOVAL(MTIME) LONG(mtime) \
343 HELPOPT("[--scan] use mtime instead of atime") \
344 NOVAL(SHOWFILES) LONG(files) \
345 HELPOPT("[--web,--html,--text] list individual files") \
346 VAL(AGERANGE) SHORT(r) LONG(age_range) LONG(range) LONG(ages) \
347 HELPARG("age[-age]") HELPOPT("[--web,--html] set limits of colour coding") \
348 VAL(OUTFILE) SHORT(o) LONG(output) \
349 HELPARG("filename") HELPOPT("[--html] specify output file or directory name") \
350 VAL(SERVERADDR) LONG(address) LONG(addr) LONG(server_address) \
352 HELPARG("addr[:port]") HELPOPT("[--web] specify HTTP server address") \
353 VAL(AUTH) LONG(auth) LONG(http_auth) LONG(httpd_auth) \
354 LONG(server_auth) LONG(web_auth) \
355 HELPARG("type") HELPOPT("[--web] specify HTTP authentication method") \
356 VAL(AUTHFILE) LONG(auth_file) \
357 HELPARG("filename") HELPOPT("[--web] read HTTP Basic user/pass from file") \
358 VAL(AUTHFD) LONG(auth_fd) \
359 HELPARG("fd") HELPOPT("[--web] read HTTP Basic user/pass from fd") \
360 VAL(DEPTH) SHORT(d) LONG(depth) LONG(max_depth) LONG(maximum_depth) \
361 HELPARG("levels") HELPOPT("[--text,--html] recurse to this many levels") \
362 VAL(MINAGE) SHORT(a) LONG(age) LONG(min_age) LONG(minimum_age) \
363 HELPARG("age") HELPOPT("[--text] include only files older than this") \
365 NOVAL(HELP) SHORT(h) LONG(help) HELPOPT("display this help text") \
366 NOVAL(VERSION) SHORT(V) LONG(version) HELPOPT("report version number") \
367 NOVAL(LICENCE) LONG(licence) LONG(license) \
368 HELPOPT("display (MIT) licence text") \
371 #define DEFENUM(x) OPT_ ## x,
374 #define STRING(x) #x ,
375 #define STRINGNOCOMMA(x) #x
376 #define SHORTNEWOPT(x) SHORTtmp_ ## x = OPT_ ## x,
377 #define SHORTTHISOPT(x) SHORTtmp2_ ## x, SHORTVAL_ ## x = SHORTtmp2_ ## x - 1,
378 #define SHORTOPTVAL(x) SHORTVAL_ ## x,
379 #define SHORTTMP(x) SHORTtmp3_ ## x,
380 #define LONGNEWOPT(x) LONGtmp_ ## x = OPT_ ## x,
381 #define LONGTHISOPT(x) LONGtmp2_ ## x, LONGVAL_ ## x = LONGtmp2_ ## x - 1,
382 #define LONGOPTVAL(x) LONGVAL_ ## x,
383 #define LONGTMP(x) SHORTtmp3_ ## x,
385 #define OPTIONS(NOVAL, VAL, SHORT, LONG) \
386 OPTHELP(NOVAL, VAL, SHORT, LONG, IGNORE, IGNORE, IGNORE, IGNORE)
388 enum { OPTIONS(DEFENUM
,DEFENUM
,IGNORE
,IGNORE
) NOPTIONS
};
389 enum { OPTIONS(IGNORE
,IGNORE
,SHORTTMP
,IGNORE
) NSHORTOPTS
};
390 enum { OPTIONS(IGNORE
,IGNORE
,IGNORE
,LONGTMP
) NLONGOPTS
};
391 static const int opthasval
[NOPTIONS
] = {OPTIONS(ZERO
,ONE
,IGNORE
,IGNORE
)};
392 static const char shortopts
[] = {OPTIONS(IGNORE
,IGNORE
,STRINGNOCOMMA
,IGNORE
)};
393 static const char *const longopts
[] = {OPTIONS(IGNORE
,IGNORE
,IGNORE
,STRING
)};
394 enum { OPTIONS(SHORTNEWOPT
,SHORTNEWOPT
,SHORTTHISOPT
,IGNORE
) };
395 enum { OPTIONS(LONGNEWOPT
,LONGNEWOPT
,IGNORE
,LONGTHISOPT
) };
396 static const int shortvals
[] = {OPTIONS(IGNORE
,IGNORE
,SHORTOPTVAL
,IGNORE
)};
397 static const int longvals
[] = {OPTIONS(IGNORE
,IGNORE
,IGNORE
,LONGOPTVAL
)};
399 static void usage(FILE *fp
)
402 const char *prefix
, *shortopt
, *longopt
, *optarg
;
405 #define HELPRESET prefix = shortopt = longopt = optarg = NULL, optex = -1
406 #define HELPNOVAL(s) optex = 0;
407 #define HELPVAL(s) optex = 1;
408 #define HELPSHORT(s) if (!shortopt) shortopt = "-" #s;
409 #define HELPLONG(s) if (!longopt) { \
410 strcpy(longbuf, "--" #s); longopt = longbuf; \
411 for (i = 0; longbuf[i]; i++) if (longbuf[i] == '_') longbuf[i] = '-'; }
412 #define HELPPFX(s) prefix = s;
413 #define HELPARG(s) optarg = s;
414 #define HELPLINE(s) assert(optex == -1); \
415 fprintf(fp, "%7s%c %s\n", prefix?prefix:"", prefix?':':' ', s); \
417 #define HELPOPT(s) assert((optex == 1 && optarg) || (optex == 0 && !optarg)); \
418 assert(shortopt || longopt); \
419 i = fprintf(fp, "%7s%c %s%s%s%s%s", prefix?prefix:"", prefix?':':' ', \
420 shortopt?shortopt:"", shortopt&&longopt?", ":"", longopt?longopt:"", \
421 optarg?" ":"", optarg?optarg:""); \
422 fprintf(fp, "%*s %s\n", i<32?32-i:0,"",s); HELPRESET;
425 OPTHELP(HELPNOVAL
, HELPVAL
, HELPSHORT
, HELPLONG
,
426 HELPPFX
, HELPARG
, HELPLINE
, HELPOPT
);
439 static time_t parse_age(time_t now
, const char *agestr
)
448 if (2 != sscanf(agestr
, "%d%1[DdWwMmYy]", &nunits
, unit
)) {
449 fprintf(stderr
, "%s: age specification should be a number followed by"
450 " one of d,w,m,y\n", PNAME
);
454 if (unit
[0] == 'd') {
456 } else if (unit
[0] == 'w') {
457 t
-= 86400 * 7 * nunits
;
462 ym
= tm
.tm_year
* 12 + tm
.tm_mon
;
469 tm
.tm_year
= ym
/ 12;
478 int main(int argc
, char **argv
)
481 struct ctx actx
, *ctx
= &actx
;
483 off_t totalsize
, realsize
;
487 const struct trie_file
*tf
, *prevtf
;
488 char *filename
= PNAME
".dat";
490 enum { TEXT
, HTML
, SCAN
, DUMP
, SCANDUMP
, LOAD
, HTTPD
, REMOVE
};
495 int nactions
= 0, actionsize
= 0, action
;
496 time_t now
= time(NULL
);
497 time_t textcutoff
= now
, htmlnewest
= now
, htmloldest
= now
;
498 int htmlautoagerange
= 1;
499 const char *httpserveraddr
= NULL
;
500 int httpserverport
= 0;
501 const char *httpauthdata
= NULL
;
502 const char *outfile
= NULL
;
503 int auth
= HTTPD_AUTH_MAGIC
| HTTPD_AUTH_BASIC
;
505 struct inclusion_exclusion
*inex
= NULL
;
506 int ninex
= 0, inexsize
= 0;
508 int depth
= -1, gotdepth
= 0;
509 int fakediratimes
= 1;
513 #ifdef DEBUG_MAD_OPTION_PARSING_MACROS
515 static const char *const optnames
[NOPTIONS
] = {
516 OPTIONS(STRING
,STRING
,IGNORE
,IGNORE
)
519 for (i
= 0; i
< NSHORTOPTS
; i
++)
520 printf("-%c == %s [%s]\n", shortopts
[i
], optnames
[shortvals
[i
]],
521 opthasval
[shortvals
[i
]] ?
"value" : "no value");
522 for (i
= 0; i
< NLONGOPTS
; i
++)
523 printf("--%s == %s [%s]\n", longopts
[i
], optnames
[longvals
[i
]],
524 opthasval
[longvals
[i
]] ?
"value" : "no value");
531 if (doing_opts
&& *p
== '-') {
534 if (!strcmp(p
, "--")) {
545 if (wordstart
&& *p
== '-') {
547 * GNU-style long option.
550 optval
= strchr(p
, '=');
554 for (i
= 0; i
< NLONGOPTS
; i
++) {
555 const char *opt
= longopts
[i
], *s
= p
;
558 * The underscores in the option names
559 * defined above may be given by the user
560 * as underscores or dashes, or omitted
565 if (*s
== '-' || *s
== '_')
583 fprintf(stderr
, "%s: unrecognised option '--%s'\n",
588 if (!opthasval
[optid
]) {
590 fprintf(stderr
, "%s: unexpected argument to option"
591 " '--%s'\n", PNAME
, p
);
599 fprintf(stderr
, "%s: option '--%s' expects"
600 " an argument\n", PNAME
, p
);
606 p
+= strlen(p
); /* finished with this argument word */
613 for (i
= 0; i
< NSHORTOPTS
; i
++)
614 if (c
== shortopts
[i
]) {
615 optid
= shortvals
[i
];
620 fprintf(stderr
, "%s: unrecognised option '-%c'\n",
625 if (opthasval
[optid
]) {
629 } else if (--argc
> 0) {
632 fprintf(stderr
, "%s: option '-%c' expects"
633 " an argument\n", PNAME
, c
);
644 * Now actually process the option.
651 #ifdef PACKAGE_VERSION
652 printf("%s, revision %s\n", PNAME
, PACKAGE_VERSION
);
654 printf("%s: version number not available when not built"
655 " via automake\n", PNAME
);
660 extern const char *const licence
[];
663 for (i
= 0; licence
[i
]; i
++)
664 fputs(licence
[i
], stdout
);
670 if (nactions
>= actionsize
) {
671 actionsize
= nactions
* 3 / 2 + 16;
672 actions
= sresize(actions
, actionsize
, struct action
);
674 actions
[nactions
].mode
= SCAN
;
675 actions
[nactions
].arg
= optval
;
679 if (nactions
>= actionsize
) {
680 actionsize
= nactions
* 3 / 2 + 16;
681 actions
= sresize(actions
, actionsize
, struct action
);
683 actions
[nactions
].mode
= SCANDUMP
;
684 actions
[nactions
].arg
= optval
;
688 if (nactions
>= actionsize
) {
689 actionsize
= nactions
* 3 / 2 + 16;
690 actions
= sresize(actions
, actionsize
, struct action
);
692 actions
[nactions
].mode
= DUMP
;
693 actions
[nactions
].arg
= NULL
;
697 if (nactions
>= actionsize
) {
698 actionsize
= nactions
* 3 / 2 + 16;
699 actions
= sresize(actions
, actionsize
, struct action
);
701 actions
[nactions
].mode
= LOAD
;
702 actions
[nactions
].arg
= NULL
;
706 if (nactions
>= actionsize
) {
707 actionsize
= nactions
* 3 / 2 + 16;
708 actions
= sresize(actions
, actionsize
, struct action
);
710 actions
[nactions
].mode
= TEXT
;
711 actions
[nactions
].arg
= optval
;
715 if (nactions
>= actionsize
) {
716 actionsize
= nactions
* 3 / 2 + 16;
717 actions
= sresize(actions
, actionsize
, struct action
);
719 actions
[nactions
].mode
= HTML
;
720 actions
[nactions
].arg
= optval
;
724 if (nactions
>= actionsize
) {
725 actionsize
= nactions
* 3 / 2 + 16;
726 actions
= sresize(actions
, actionsize
, struct action
);
728 actions
[nactions
].mode
= HTTPD
;
729 actions
[nactions
].arg
= NULL
;
733 if (nactions
>= actionsize
) {
734 actionsize
= nactions
* 3 / 2 + 16;
735 actions
= sresize(actions
, actionsize
, struct action
);
737 actions
[nactions
].mode
= REMOVE
;
738 actions
[nactions
].arg
= NULL
;
747 case OPT_TTYPROGRESS
:
772 if (!strcasecmp(optval
, "unlimited") ||
773 !strcasecmp(optval
, "infinity") ||
774 !strcasecmp(optval
, "infinite") ||
775 !strcasecmp(optval
, "inf") ||
776 !strcasecmp(optval
, "maximum") ||
777 !strcasecmp(optval
, "max"))
780 depth
= atoi(optval
);
787 textcutoff
= parse_age(now
, optval
);
790 if (!strcmp(optval
, "auto")) {
791 htmlautoagerange
= 1;
793 char *q
= optval
+ strcspn(optval
, "-:");
796 htmloldest
= parse_age(now
, optval
);
797 htmlnewest
= *q ?
parse_age(now
, q
) : now
;
798 htmlautoagerange
= 0;
804 if (optval
[0] == '[' &&
805 (port
= strchr(optval
, ']')) != NULL
)
809 port
+= strcspn(port
, ":");
812 httpserveraddr
= optval
;
813 httpserverport
= atoi(port
);
817 if (!strcmp(optval
, "magic"))
818 auth
= HTTPD_AUTH_MAGIC
;
819 else if (!strcmp(optval
, "basic"))
820 auth
= HTTPD_AUTH_BASIC
;
821 else if (!strcmp(optval
, "none"))
822 auth
= HTTPD_AUTH_NONE
;
823 else if (!strcmp(optval
, "default"))
824 auth
= HTTPD_AUTH_MAGIC
| HTTPD_AUTH_BASIC
;
825 else if (!strcmp(optval
, "help") ||
826 !strcmp(optval
, "list")) {
827 printf(PNAME
": supported HTTP authentication types"
829 " magic use Linux /proc/net/tcp to"
830 " determine owner of peer socket\n"
831 " basic HTTP Basic username and"
832 " password authentication\n"
833 " default use 'magic' if possible, "
834 " otherwise fall back to 'basic'\n"
835 " none unauthenticated HTTP (if"
836 " the data file is non-confidential)\n");
839 fprintf(stderr
, "%s: unrecognised authentication"
840 " type '%s'\n%*s options are 'magic',"
841 " 'basic', 'none', 'default'\n",
842 PNAME
, optval
, (int)strlen(PNAME
), "");
853 int authlen
, authsize
;
856 if (optid
== OPT_AUTHFILE
) {
857 fd
= open(optval
, O_RDONLY
);
859 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
,
860 optval
, strerror(errno
));
867 sprintf(namebuf
, "fd %d", fd
);
872 authbuf
= snewn(authsize
, char);
873 while ((ret
= read(fd
, authbuf
+authlen
,
874 authsize
-authlen
)) > 0) {
876 if ((authsize
- authlen
) < (authsize
/ 16)) {
877 authsize
= authlen
* 3 / 2 + 4096;
878 authbuf
= sresize(authbuf
, authsize
, char);
882 fprintf(stderr
, "%s: %s: read: %s\n", PNAME
,
883 name
, strerror(errno
));
886 if (optid
== OPT_AUTHFILE
)
888 httpauthdata
= authbuf
;
892 case OPT_INCLUDEPATH
:
894 case OPT_EXCLUDEPATH
:
897 if (ninex
>= inexsize
) {
898 inexsize
= ninex
* 3 / 2 + 16;
899 inex
= sresize(inex
, inexsize
,
900 struct inclusion_exclusion
);
902 inex
[ninex
].path
= (optid
== OPT_INCLUDEPATH
||
903 optid
== OPT_EXCLUDEPATH
||
904 optid
== OPT_PRUNEPATH
);
905 inex
[ninex
].type
= (optid
== OPT_INCLUDE ?
1 :
906 optid
== OPT_INCLUDEPATH ?
1 :
907 optid
== OPT_EXCLUDE ?
0 :
908 optid
== OPT_EXCLUDEPATH ?
0 :
909 optid
== OPT_PRUNE ?
-1 :
910 /* optid == OPT_PRUNEPATH ? */ -1);
911 inex
[ninex
].wildcard
= optval
;
917 fprintf(stderr
, "%s: unexpected argument '%s'\n", PNAME
, p
);
927 for (action
= 0; action
< nactions
; action
++) {
928 int mode
= actions
[action
].mode
;
930 if (mode
== SCAN
|| mode
== SCANDUMP
|| mode
== LOAD
) {
931 const char *scandir
= actions
[action
].arg
;
934 char *buf
= fgetline(stdin
);
936 buf
[strcspn(buf
, "\r\n")] = '\0';
937 if (1 != sscanf(buf
, DUMPHDR
"%x",
939 fprintf(stderr
, "%s: header in dump file not recognised\n",
943 pathsep
= (char)newpathsep
;
947 if (mode
== SCAN
|| mode
== LOAD
) {
949 * Prepare to write out the index file.
951 fd
= open(filename
, O_RDWR
| O_TRUNC
| O_CREAT
,
954 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
, filename
,
958 if (fstat(fd
, &st
) < 0) {
959 perror(PNAME
": fstat");
962 ctx
->datafile_dev
= st
.st_dev
;
963 ctx
->datafile_ino
= st
.st_ino
;
964 ctx
->straight_to_dump
= 0;
966 ctx
->datafile_dev
= -1;
967 ctx
->datafile_ino
= -1;
968 ctx
->straight_to_dump
= 1;
971 if (mode
== SCAN
|| mode
== SCANDUMP
) {
972 if (stat(scandir
, &st
) < 0) {
973 fprintf(stderr
, "%s: %s: stat: %s\n", PNAME
, scandir
,
977 ctx
->filesystem_dev
= crossfs ?
0 : st
.st_dev
;
982 ctx
->crossfs
= crossfs
;
983 ctx
->fakeatimes
= fakediratimes
;
984 ctx
->usemtime
= mtime
;
986 ctx
->last_output_update
= time(NULL
);
988 /* progress==1 means report progress only if stderr is a tty */
990 progress
= isatty(2) ?
2 : 0;
991 ctx
->progress
= progress
;
995 ioctl(2, TIOCGWINSZ
, &ws
) == 0 &&
997 ctx
->progwidth
= ws
.ws_col
- 1;
1002 if (mode
== SCANDUMP
)
1003 printf(DUMPHDR
"%02x\n", (unsigned char)pathsep
);
1006 * Scan the directory tree, and write out the trie component
1009 if (mode
!= SCANDUMP
) {
1010 ctx
->tb
= triebuild_new(fd
);
1015 while ((buf
= fgetline(stdin
)) != NULL
) {
1016 struct trie_file tf
;
1019 buf
[strcspn(buf
, "\r\n")] = '\0';
1023 while (*p
&& *p
!= ' ') p
++;
1025 fprintf(stderr
, "%s: dump file line %d: expected at least"
1026 " three fields\n", PNAME
, line
);
1030 tf
.size
= strtoull(q
, NULL
, 10);
1032 while (*p
&& *p
!= ' ') p
++;
1034 fprintf(stderr
, "%s: dump file line %d: expected at least"
1035 " three fields\n", PNAME
, line
);
1039 tf
.atime
= strtoull(q
, NULL
, 10);
1047 for (i
= 0; i
< 2; i
++) {
1049 if (*p
>= '0' && *p
<= '9')
1051 else if (*p
>= 'A' && *p
<= 'F')
1052 c
+= *p
- ('A' - 10);
1053 else if (*p
>= 'a' && *p
<= 'f')
1054 c
+= *p
- ('a' - 10);
1056 fprintf(stderr
, "%s: dump file line %d: unable"
1057 " to parse hex escape\n", PNAME
, line
);
1067 triebuild_add(ctx
->tb
, buf
, &tf
);
1072 du(scandir
, gotdata
, scan_error
, ctx
);
1074 if (mode
!= SCANDUMP
) {
1077 char *buf
, *prevbuf
;
1079 count
= triebuild_finish(ctx
->tb
);
1080 triebuild_free(ctx
->tb
);
1082 if (ctx
->progress
) {
1083 fprintf(stderr
, "%-*s\r", ctx
->progwidth
, "");
1088 * Work out how much space the cumulative index trees
1089 * will take; enlarge the file, and memory-map it.
1091 if (fstat(fd
, &st
) < 0) {
1092 perror(PNAME
": fstat");
1096 printf("Built pathname index, %d entries,"
1097 " %llu bytes of index\n", count
,
1098 (unsigned long long)st
.st_size
);
1100 totalsize
= index_initial_size(st
.st_size
, count
);
1101 totalsize
+= totalsize
/ 10;
1103 if (lseek(fd
, totalsize
-1, SEEK_SET
) < 0) {
1104 perror(PNAME
": lseek");
1107 if (write(fd
, "\0", 1) < 1) {
1108 perror(PNAME
": write");
1112 mappedfile
= mmap(NULL
, totalsize
, PROT_READ
|PROT_WRITE
,MAP_SHARED
, fd
, 0);
1114 perror(PNAME
": mmap");
1118 if (fakediratimes
) {
1119 printf("Faking directory atimes\n");
1120 trie_fake_dir_atimes(mappedfile
);
1123 printf("Building index\n");
1124 ib
= indexbuild_new(mappedfile
, st
.st_size
, count
, &delta
);
1125 maxpathlen
= trie_maxpathlen(mappedfile
);
1126 buf
= snewn(maxpathlen
, char);
1127 prevbuf
= snewn(maxpathlen
, char);
1128 tw
= triewalk_new(mappedfile
);
1130 tf
= triewalk_next(tw
, buf
);
1135 if (totalsize
- indexbuild_realsize(ib
) < delta
) {
1136 const void *oldfile
= mappedfile
;
1140 * Unmap the file, grow it, and remap it.
1142 munmap(mappedfile
, totalsize
);
1145 totalsize
+= totalsize
/ 10;
1147 if (lseek(fd
, totalsize
-1, SEEK_SET
) < 0) {
1148 perror(PNAME
": lseek");
1151 if (write(fd
, "\0", 1) < 1) {
1152 perror(PNAME
": write");
1156 mappedfile
= mmap(NULL
, totalsize
, PROT_READ
|PROT_WRITE
,MAP_SHARED
, fd
, 0);
1158 perror(PNAME
": mmap");
1162 indexbuild_rebase(ib
, mappedfile
);
1163 triewalk_rebase(tw
, mappedfile
);
1164 diff
= (const unsigned char *)mappedfile
-
1165 (const unsigned char *)oldfile
;
1167 prevtf
= (const struct trie_file
*)
1168 (((const unsigned char *)prevtf
) + diff
);
1170 tf
= (const struct trie_file
*)
1171 (((const unsigned char *)tf
) + diff
);
1175 * Get the next file from the index. So we are
1176 * currently holding, and have not yet
1177 * indexed, prevtf (with pathname prevbuf) and
1178 * tf (with pathname buf).
1181 memcpy(prevbuf
, buf
, maxpathlen
);
1182 tf
= triewalk_next(tw
, buf
);
1188 * Find the first differing character position
1189 * between our two pathnames.
1191 for (i
= 0; prevbuf
[i
] && prevbuf
[i
] == buf
[i
]; i
++);
1194 * If prevbuf was a directory name and buf is
1195 * something inside that directory, then
1196 * trie_before() will be called on prevbuf
1197 * itself. Hence we must drop a tag before it,
1198 * so that the resulting index is usable.
1200 if ((!prevbuf
[i
] && (buf
[i
] == pathsep
||
1201 (i
> 0 && buf
[i
-1] == pathsep
))))
1205 * Add prevtf to the index.
1207 indexbuild_add(ib
, prevtf
);
1211 * Drop an unconditional final tag, and
1212 * get out of this loop.
1219 * If prevbuf was a filename inside some
1220 * directory which buf is outside, then
1221 * trie_before() will be called on some
1222 * pathname either equal to buf or epsilon
1223 * less than it. Either way, we're going to
1224 * need to drop a tag after prevtf.
1226 if (strchr(prevbuf
+i
, pathsep
) || !tf
)
1231 realsize
= indexbuild_realsize(ib
);
1232 indexbuild_free(ib
);
1234 munmap(mappedfile
, totalsize
);
1235 ftruncate(fd
, realsize
);
1237 printf("Final index file size = %llu bytes\n",
1238 (unsigned long long)realsize
);
1240 } else if (mode
== TEXT
) {
1241 char *querydir
= actions
[action
].arg
;
1244 fd
= open(filename
, O_RDONLY
);
1246 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
, filename
,
1250 if (fstat(fd
, &st
) < 0) {
1251 perror(PNAME
": fstat");
1254 totalsize
= st
.st_size
;
1255 mappedfile
= mmap(NULL
, totalsize
, PROT_READ
, MAP_SHARED
, fd
, 0);
1257 perror(PNAME
": mmap");
1260 pathsep
= trie_pathsep(mappedfile
);
1263 * Trim trailing slash, just in case.
1265 pathlen
= strlen(querydir
);
1266 if (pathlen
> 0 && querydir
[pathlen
-1] == pathsep
)
1267 querydir
[--pathlen
] = '\0';
1270 depth
= 1; /* default for text mode */
1271 if (outfile
!= NULL
) {
1272 FILE *fp
= fopen(outfile
, "w");
1274 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
,
1275 outfile
, strerror(errno
));
1278 text_query(mappedfile
, querydir
, textcutoff
, showfiles
,
1282 text_query(mappedfile
, querydir
, textcutoff
, showfiles
,
1286 munmap(mappedfile
, totalsize
);
1287 } else if (mode
== HTML
) {
1288 char *querydir
= actions
[action
].arg
;
1289 size_t pathlen
, maxpathlen
;
1291 struct html_config cfg
;
1295 fd
= open(filename
, O_RDONLY
);
1297 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
, filename
,
1301 if (fstat(fd
, &st
) < 0) {
1302 perror(PNAME
": fstat");
1305 totalsize
= st
.st_size
;
1306 mappedfile
= mmap(NULL
, totalsize
, PROT_READ
, MAP_SHARED
, fd
, 0);
1308 perror(PNAME
": mmap");
1311 pathsep
= trie_pathsep(mappedfile
);
1313 maxpathlen
= trie_maxpathlen(mappedfile
);
1314 pathbuf
= snewn(maxpathlen
, char);
1317 * Trim trailing slash, just in case.
1319 pathlen
= strlen(querydir
);
1320 if (pathlen
> 0 && querydir
[pathlen
-1] == pathsep
)
1321 querydir
[--pathlen
] = '\0';
1323 xi
= trie_before(mappedfile
, querydir
);
1324 if (xi
>= trie_count(mappedfile
) ||
1325 (trie_getpath(mappedfile
, xi
, pathbuf
),
1326 strcmp(pathbuf
, querydir
))) {
1327 fprintf(stderr
, "%s: pathname '%s' does not exist in index\n"
1328 "%*s(check it is spelled exactly as it is in the "
1329 "index, including\n%*sany leading './')\n",
1331 (int)(1+sizeof(PNAME
)), "",
1332 (int)(1+sizeof(PNAME
)), "");
1333 } else if (!index_has_root(mappedfile
, xi
)) {
1334 fprintf(stderr
, "%s: pathname '%s' is"
1335 " a file, not a directory\n", PNAME
, querydir
);
1336 } else if (!gotdepth
) {
1338 * Single output file.
1341 cfg
.rootpage
= NULL
;
1342 cfg
.autoage
= htmlautoagerange
;
1343 cfg
.oldest
= htmloldest
;
1344 cfg
.newest
= htmlnewest
;
1345 cfg
.showfiles
= showfiles
;
1346 html
= html_query(mappedfile
, xi
, &cfg
, 0);
1347 if (outfile
!= NULL
) {
1348 FILE *fp
= fopen(outfile
, "w");
1350 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
,
1351 outfile
, strerror(errno
));
1353 } else if (fputs(html
, fp
) < 0) {
1354 fprintf(stderr
, "%s: %s: write: %s\n", PNAME
,
1355 outfile
, strerror(errno
));
1358 } else if (fclose(fp
) < 0) {
1359 fprintf(stderr
, "%s: %s: fclose: %s\n", PNAME
,
1360 outfile
, strerror(errno
));
1364 fputs(html
, stdout
);
1368 * Multiple output files.
1370 int dirlen
= outfile ?
2+strlen(outfile
) : 3;
1371 char prefix
[dirlen
];
1373 if (mkdir(outfile
, 0777) < 0 && errno
!= EEXIST
) {
1374 fprintf(stderr
, "%s: %s: mkdir: %s\n", PNAME
,
1375 outfile
, strerror(errno
));
1378 snprintf(prefix
, dirlen
, "%s/", outfile
);
1380 snprintf(prefix
, dirlen
, "./");
1383 make_successor(pathbuf
);
1384 xi2
= trie_before(mappedfile
, pathbuf
);
1386 cfg
.format
= "%lu.html";
1387 cfg
.rootpage
= "index.html";
1388 cfg
.autoage
= htmlautoagerange
;
1389 cfg
.oldest
= htmloldest
;
1390 cfg
.newest
= htmlnewest
;
1391 cfg
.showfiles
= showfiles
;
1392 if (html_dump(mappedfile
, xi
, xi2
, depth
, &cfg
, prefix
))
1396 munmap(mappedfile
, totalsize
);
1398 } else if (mode
== DUMP
) {
1402 fd
= open(filename
, O_RDONLY
);
1404 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
, filename
,
1408 if (fstat(fd
, &st
) < 0) {
1409 perror(PNAME
": fstat");
1412 totalsize
= st
.st_size
;
1413 mappedfile
= mmap(NULL
, totalsize
, PROT_READ
, MAP_SHARED
, fd
, 0);
1415 perror(PNAME
": mmap");
1418 pathsep
= trie_pathsep(mappedfile
);
1420 maxpathlen
= trie_maxpathlen(mappedfile
);
1421 buf
= snewn(maxpathlen
, char);
1423 printf(DUMPHDR
"%02x\n", (unsigned char)pathsep
);
1424 tw
= triewalk_new(mappedfile
);
1425 while ((tf
= triewalk_next(tw
, buf
)) != NULL
)
1429 munmap(mappedfile
, totalsize
);
1430 } else if (mode
== HTTPD
) {
1431 struct html_config pcfg
;
1432 struct httpd_config dcfg
;
1434 fd
= open(filename
, O_RDONLY
);
1436 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
, filename
,
1440 if (fstat(fd
, &st
) < 0) {
1441 perror(PNAME
": fstat");
1444 totalsize
= st
.st_size
;
1445 mappedfile
= mmap(NULL
, totalsize
, PROT_READ
, MAP_SHARED
, fd
, 0);
1447 perror(PNAME
": mmap");
1450 pathsep
= trie_pathsep(mappedfile
);
1452 dcfg
.address
= httpserveraddr
;
1453 dcfg
.port
= httpserverport
;
1454 dcfg
.basicauthdata
= httpauthdata
;
1456 pcfg
.rootpage
= NULL
;
1457 pcfg
.autoage
= htmlautoagerange
;
1458 pcfg
.oldest
= htmloldest
;
1459 pcfg
.newest
= htmlnewest
;
1460 pcfg
.showfiles
= showfiles
;
1461 run_httpd(mappedfile
, auth
, &dcfg
, &pcfg
);
1462 munmap(mappedfile
, totalsize
);
1463 } else if (mode
== REMOVE
) {
1464 if (remove(filename
) < 0) {
1465 fprintf(stderr
, "%s: %s: remove: %s\n", PNAME
, filename
,