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 printf("%llu %llu ", tf
->size
, tf
->atime
);
60 for (p
= pathname
; *p
; p
++) {
61 if (*p
>= ' ' && *p
< 127 && *p
!= '%')
64 printf("%%%02x", (unsigned char)*p
);
69 static int gotdata(void *vctx
, const char *pathname
, const STRUCT_STAT
*st
)
71 struct ctx
*ctx
= (struct ctx
*)vctx
;
72 struct trie_file file
;
78 * Filter out our own data file.
80 if (st
->st_dev
== ctx
->datafile_dev
&& st
->st_ino
== ctx
->datafile_ino
)
84 * Don't cross the streams^W^Wany file system boundary.
86 if (!ctx
->crossfs
&& st
->st_dev
!= ctx
->filesystem_dev
)
89 file
.size
= (unsigned long long)512 * st
->st_blocks
;
90 if (ctx
->usemtime
|| (ctx
->fakeatimes
&& S_ISDIR(st
->st_mode
)))
91 file
.atime
= st
->st_mtime
;
93 file
.atime
= st
->st_atime
;
96 * Filter based on wildcards.
99 filename
= strrchr(pathname
, pathsep
);
104 for (i
= 0; i
< ctx
->ninex
; i
++) {
105 if (fnmatch(ctx
->inex
[i
].wildcard
,
106 ctx
->inex
[i
].path ? pathname
: filename
, 0) == 0)
107 include
= ctx
->inex
[i
].type
;
110 return 0; /* ignore this entry and any subdirs */
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.
119 if (!S_ISDIR(st
->st_mode
))
120 return 0; /* just ignore */
125 if (ctx
->straight_to_dump
)
126 dump_line(pathname
, &file
);
128 triebuild_add(ctx
->tb
, pathname
, &file
);
132 if (t
!= ctx
->last_output_update
) {
133 fprintf(stderr
, "%-*.*s\r", ctx
->progwidth
, ctx
->progwidth
,
136 ctx
->last_output_update
= t
;
143 static void text_query(const void *mappedfile
, const char *querydir
,
148 unsigned long xi1
, xi2
;
149 unsigned long long s1
, s2
;
151 maxpathlen
= trie_maxpathlen(mappedfile
);
152 pathbuf
= snewn(maxpathlen
+ 1, char);
155 * We want to query everything between the supplied filename
156 * (inclusive) and that filename with a ^A on the end
157 * (exclusive). So find the x indices for each.
159 strcpy(pathbuf
, querydir
);
160 make_successor(pathbuf
);
161 xi1
= trie_before(mappedfile
, querydir
);
162 xi2
= trie_before(mappedfile
, pathbuf
);
165 return; /* file, or empty dir => no display */
168 * Now do the lookups in the age index.
170 s1
= index_query(mappedfile
, xi1
, t
);
171 s2
= index_query(mappedfile
, xi2
, t
);
174 return; /* no space taken up => no display */
178 * Now scan for first-level subdirectories and report
183 trie_getpath(mappedfile
, xi1
, pathbuf
);
184 text_query(mappedfile
, pathbuf
, t
, depth
-1);
185 make_successor(pathbuf
);
186 xi1
= trie_before(mappedfile
, pathbuf
);
190 /* Display in units of 1Kb */
191 printf("%-11llu %s\n", (s2
- s1
) / 1024, querydir
);
195 * Largely frivolous way to define all my command-line options. I
196 * present here a parametric macro which declares a series of
197 * _logical_ option identifiers, and for each one declares zero or
198 * more short option characters and zero or more long option
199 * words. Then I repeatedly invoke that macro with its arguments
200 * defined to be various other macros, which allows me to
203 * - define an enum allocating a distinct integer value to each
205 * - define a string consisting of precisely all the short option
207 * - define a string array consisting of all the long option
209 * - define (with help from auxiliary enums) integer arrays
210 * parallel to both of the above giving the logical option id
211 * for each physical short and long option
212 * - define an array indexed by logical option id indicating
213 * whether the option in question takes a value
214 * - define a function which prints out brief online help for all
217 * It's not at all clear to me that this trickery is actually
218 * particularly _efficient_ - it still, after all, requires going
219 * linearly through the option list at run time and doing a
220 * strcmp, whereas in an ideal world I'd have liked the lists of
221 * long and short options to be pre-sorted so that a binary search
222 * or some other more efficient lookup was possible. (Not that
223 * asymptotic algorithmic complexity is remotely vital in option
224 * parsing, but if I were doing this in, say, Lisp or something
225 * with an equivalently powerful preprocessor then once I'd had
226 * the idea of preparing the option-parsing data structures at
227 * compile time I would probably have made the effort to prepare
228 * them _properly_. I could have Perl generate me a source file
229 * from some sort of description, I suppose, but that would seem
230 * like overkill. And in any case, it's more of a challenge to
231 * achieve as much as possible by cunning use of cpp and enum than
232 * to just write some sensible and logical code in a Turing-
233 * complete language. I said it was largely frivolous :-)
235 * This approach does have the virtue that it brings together the
236 * option ids, option spellings and help text into a single
237 * combined list and defines them all in exactly one place. If I
238 * want to add a new option, or a new spelling for an option, I
239 * only have to modify the main OPTHELP macro below and then add
240 * code to process the new logical id.
242 * (Though, really, even that isn't ideal, since it still involves
243 * modifying the source file in more than one place. In a
244 * _properly_ ideal world, I'd be able to interleave the option
245 * definitions with the code fragments that process them. And then
246 * not bother defining logical identifiers for them at all - those
247 * would be automatically generated, since I wouldn't have any
248 * need to specify them manually in another part of the code.)
250 * One other helpful consequence of the enum-based structure here
251 * is that it causes a compiler error if I accidentally try to
252 * define the same option (short or long) twice.
255 #define OPTHELP(NOVAL, VAL, SHORT, LONG, HELPPFX, HELPARG, HELPLINE, HELPOPT) \
256 HELPPFX("usage") HELPLINE(PNAME " [options] action [action...]") \
258 VAL(SCAN) SHORT(s) LONG(scan) \
259 HELPARG("directory") HELPOPT("scan and index a directory") \
260 NOVAL(HTTPD) SHORT(w) LONG(web) LONG(server) LONG(httpd) \
261 HELPOPT("serve HTML reports from a temporary web server") \
262 VAL(TEXT) SHORT(t) LONG(text) \
263 HELPARG("subdir") HELPOPT("print a plain text report on a subdirectory") \
264 NOVAL(REMOVE) SHORT(R) LONG(remove) LONG(delete) LONG(unlink) \
265 HELPOPT("remove the index file") \
266 NOVAL(DUMP) SHORT(D) LONG(dump) HELPOPT("dump the index file on stdout") \
267 NOVAL(LOAD) SHORT(L) LONG(load) \
268 HELPOPT("load and index a dump file") \
269 VAL(SCANDUMP) SHORT(S) LONG(scan_dump) \
270 HELPARG("directory") HELPOPT("scan only, generating a dump") \
271 VAL(HTML) SHORT(H) LONG(html) \
272 HELPARG("subdir") HELPOPT("print an HTML report on a subdirectory") \
274 VAL(DATAFILE) SHORT(f) LONG(file) \
275 HELPARG("filename") HELPOPT("[most modes] specify index file") \
276 NOVAL(CROSSFS) LONG(cross_fs) \
277 HELPOPT("[--scan] cross filesystem boundaries") \
278 NOVAL(NOCROSSFS) LONG(no_cross_fs) \
279 HELPOPT("[--scan] stick to one filesystem") \
280 VAL(PRUNE) LONG(prune) \
281 HELPARG("wildcard") HELPOPT("[--scan] prune files matching pattern") \
282 VAL(PRUNEPATH) LONG(prune_path) \
283 HELPARG("wildcard") HELPOPT("[--scan] prune pathnames matching pattern") \
284 VAL(EXCLUDE) LONG(exclude) \
285 HELPARG("wildcard") HELPOPT("[--scan] exclude files matching pattern") \
286 VAL(EXCLUDEPATH) LONG(exclude_path) \
287 HELPARG("wildcard") HELPOPT("[--scan] exclude pathnames matching pattern") \
288 VAL(INCLUDE) LONG(include) \
289 HELPARG("wildcard") HELPOPT("[--scan] include files matching pattern") \
290 VAL(INCLUDEPATH) LONG(include_path) \
291 HELPARG("wildcard") HELPOPT("[--scan] include pathnames matching pattern") \
292 NOVAL(PROGRESS) LONG(progress) LONG(scan_progress) \
293 HELPOPT("[--scan] report progress on stderr") \
294 NOVAL(NOPROGRESS) LONG(no_progress) LONG(no_scan_progress) \
295 HELPOPT("[--scan] do not report progress") \
296 NOVAL(TTYPROGRESS) LONG(tty_progress) LONG(tty_scan_progress) \
297 LONG(progress_tty) LONG(scan_progress_tty) \
298 HELPOPT("[--scan] report progress if stderr is a tty") \
299 NOVAL(DIRATIME) LONG(dir_atime) LONG(dir_atimes) \
300 HELPOPT("[--scan,--load] keep real atimes on directories") \
301 NOVAL(NODIRATIME) LONG(no_dir_atime) LONG(no_dir_atimes) \
302 HELPOPT("[--scan,--load] fake atimes on directories") \
303 NOVAL(MTIME) LONG(mtime) \
304 HELPOPT("[--scan] use mtime instead of atime") \
305 NOVAL(FULL) LONG(full_index) \
306 HELPOPT("[--scan] index every file individually") \
307 VAL(AGERANGE) SHORT(r) LONG(age_range) LONG(range) LONG(ages) \
308 HELPARG("age[-age]") HELPOPT("[--web,--html] set limits of colour coding") \
309 VAL(SERVERADDR) LONG(address) LONG(addr) LONG(server_address) \
311 HELPARG("addr[:port]") HELPOPT("[--web] specify HTTP server address") \
312 VAL(AUTH) LONG(auth) LONG(http_auth) LONG(httpd_auth) \
313 LONG(server_auth) LONG(web_auth) \
314 HELPARG("type") HELPOPT("[--web] specify HTTP authentication method") \
315 VAL(AUTHFILE) LONG(auth_file) \
316 HELPARG("filename") HELPOPT("[--web] read HTTP Basic user/pass from file") \
317 VAL(AUTHFD) LONG(auth_fd) \
318 HELPARG("fd") HELPOPT("[--web] read HTTP Basic user/pass from fd") \
319 VAL(TQDEPTH) SHORT(d) LONG(depth) LONG(max_depth) LONG(maximum_depth) \
320 HELPARG("levels") HELPOPT("[--text] recurse to this many levels") \
321 VAL(MINAGE) SHORT(a) LONG(age) LONG(min_age) LONG(minimum_age) \
322 HELPARG("age") HELPOPT("[--text] include only files older than this") \
324 NOVAL(HELP) SHORT(h) LONG(help) HELPOPT("display this help text") \
325 NOVAL(VERSION) SHORT(V) LONG(version) HELPOPT("report version number") \
326 NOVAL(LICENCE) LONG(licence) LONG(license) \
327 HELPOPT("display (MIT) licence text") \
330 #define DEFENUM(x) OPT_ ## x,
333 #define STRING(x) #x ,
334 #define STRINGNOCOMMA(x) #x
335 #define SHORTNEWOPT(x) SHORTtmp_ ## x = OPT_ ## x,
336 #define SHORTTHISOPT(x) SHORTtmp2_ ## x, SHORTVAL_ ## x = SHORTtmp2_ ## x - 1,
337 #define SHORTOPTVAL(x) SHORTVAL_ ## x,
338 #define SHORTTMP(x) SHORTtmp3_ ## x,
339 #define LONGNEWOPT(x) LONGtmp_ ## x = OPT_ ## x,
340 #define LONGTHISOPT(x) LONGtmp2_ ## x, LONGVAL_ ## x = LONGtmp2_ ## x - 1,
341 #define LONGOPTVAL(x) LONGVAL_ ## x,
342 #define LONGTMP(x) SHORTtmp3_ ## x,
344 #define OPTIONS(NOVAL, VAL, SHORT, LONG) \
345 OPTHELP(NOVAL, VAL, SHORT, LONG, IGNORE, IGNORE, IGNORE, IGNORE)
347 enum { OPTIONS(DEFENUM
,DEFENUM
,IGNORE
,IGNORE
) NOPTIONS
};
348 enum { OPTIONS(IGNORE
,IGNORE
,SHORTTMP
,IGNORE
) NSHORTOPTS
};
349 enum { OPTIONS(IGNORE
,IGNORE
,IGNORE
,LONGTMP
) NLONGOPTS
};
350 static const int opthasval
[NOPTIONS
] = {OPTIONS(ZERO
,ONE
,IGNORE
,IGNORE
)};
351 static const char shortopts
[] = {OPTIONS(IGNORE
,IGNORE
,STRINGNOCOMMA
,IGNORE
)};
352 static const char *const longopts
[] = {OPTIONS(IGNORE
,IGNORE
,IGNORE
,STRING
)};
353 enum { OPTIONS(SHORTNEWOPT
,SHORTNEWOPT
,SHORTTHISOPT
,IGNORE
) };
354 enum { OPTIONS(LONGNEWOPT
,LONGNEWOPT
,IGNORE
,LONGTHISOPT
) };
355 static const int shortvals
[] = {OPTIONS(IGNORE
,IGNORE
,SHORTOPTVAL
,IGNORE
)};
356 static const int longvals
[] = {OPTIONS(IGNORE
,IGNORE
,IGNORE
,LONGOPTVAL
)};
358 static void usage(FILE *fp
)
361 const char *prefix
, *shortopt
, *longopt
, *optarg
;
364 #define HELPRESET prefix = shortopt = longopt = optarg = NULL, optex = -1
365 #define HELPNOVAL(s) optex = 0;
366 #define HELPVAL(s) optex = 1;
367 #define HELPSHORT(s) if (!shortopt) shortopt = "-" #s;
368 #define HELPLONG(s) if (!longopt) { \
369 strcpy(longbuf, "--" #s); longopt = longbuf; \
370 for (i = 0; longbuf[i]; i++) if (longbuf[i] == '_') longbuf[i] = '-'; }
371 #define HELPPFX(s) prefix = s;
372 #define HELPARG(s) optarg = s;
373 #define HELPLINE(s) assert(optex == -1); \
374 fprintf(fp, "%7s%c %s\n", prefix?prefix:"", prefix?':':' ', s); \
376 #define HELPOPT(s) assert((optex == 1 && optarg) || (optex == 0 && !optarg)); \
377 assert(shortopt || longopt); \
378 i = fprintf(fp, "%7s%c %s%s%s%s%s", prefix?prefix:"", prefix?':':' ', \
379 shortopt?shortopt:"", shortopt&&longopt?", ":"", longopt?longopt:"", \
380 optarg?" ":"", optarg?optarg:""); \
381 fprintf(fp, "%*s %s\n", i<32?32-i:0,"",s); HELPRESET;
384 OPTHELP(HELPNOVAL
, HELPVAL
, HELPSHORT
, HELPLONG
,
385 HELPPFX
, HELPARG
, HELPLINE
, HELPOPT
);
398 static time_t parse_age(time_t now
, const char *agestr
)
407 if (2 != sscanf(agestr
, "%d%1[DdWwMmYy]", &nunits
, unit
)) {
408 fprintf(stderr
, "%s: age specification should be a number followed by"
409 " one of d,w,m,y\n", PNAME
);
413 if (unit
[0] == 'd') {
415 } else if (unit
[0] == 'w') {
416 t
-= 86400 * 7 * nunits
;
421 ym
= tm
.tm_year
* 12 + tm
.tm_mon
;
428 tm
.tm_year
= ym
/ 12;
437 int main(int argc
, char **argv
)
440 struct ctx actx
, *ctx
= &actx
;
442 off_t totalsize
, realsize
;
446 const struct trie_file
*tf
, *prevtf
;
447 char *filename
= PNAME
".dat";
449 enum { TEXT
, HTML
, SCAN
, DUMP
, SCANDUMP
, LOAD
, HTTPD
, REMOVE
};
454 int nactions
= 0, actionsize
= 0, action
;
455 time_t now
= time(NULL
);
456 time_t textcutoff
= now
, htmlnewest
= now
, htmloldest
= now
;
457 int htmlautoagerange
= 1;
458 const char *httpserveraddr
= NULL
;
459 int httpserverport
= 0;
460 const char *httpauthdata
= NULL
;
461 int auth
= HTTPD_AUTH_MAGIC
| HTTPD_AUTH_BASIC
;
463 struct inclusion_exclusion
*inex
= NULL
;
464 int ninex
= 0, inexsize
= 0;
467 int fakediratimes
= 1;
471 #ifdef DEBUG_MAD_OPTION_PARSING_MACROS
473 static const char *const optnames
[NOPTIONS
] = {
474 OPTIONS(STRING
,STRING
,IGNORE
,IGNORE
)
477 for (i
= 0; i
< NSHORTOPTS
; i
++)
478 printf("-%c == %s [%s]\n", shortopts
[i
], optnames
[shortvals
[i
]],
479 opthasval
[shortvals
[i
]] ?
"value" : "no value");
480 for (i
= 0; i
< NLONGOPTS
; i
++)
481 printf("--%s == %s [%s]\n", longopts
[i
], optnames
[longvals
[i
]],
482 opthasval
[longvals
[i
]] ?
"value" : "no value");
489 if (doing_opts
&& *p
== '-') {
492 if (!strcmp(p
, "--")) {
503 if (wordstart
&& *p
== '-') {
505 * GNU-style long option.
508 optval
= strchr(p
, '=');
512 for (i
= 0; i
< NLONGOPTS
; i
++) {
513 const char *opt
= longopts
[i
], *s
= p
;
516 * The underscores in the option names
517 * defined above may be given by the user
518 * as underscores or dashes, or omitted
523 if (*s
== '-' || *s
== '_')
541 fprintf(stderr
, "%s: unrecognised option '--%s'\n",
546 if (!opthasval
[optid
]) {
548 fprintf(stderr
, "%s: unexpected argument to option"
549 " '--%s'\n", PNAME
, p
);
557 fprintf(stderr
, "%s: option '--%s' expects"
558 " an argument\n", PNAME
, p
);
564 p
+= strlen(p
); /* finished with this argument word */
571 for (i
= 0; i
< NSHORTOPTS
; i
++)
572 if (c
== shortopts
[i
]) {
573 optid
= shortvals
[i
];
578 fprintf(stderr
, "%s: unrecognised option '-%c'\n",
583 if (opthasval
[optid
]) {
587 } else if (--argc
> 0) {
590 fprintf(stderr
, "%s: option '-%c' expects"
591 " an argument\n", PNAME
, c
);
602 * Now actually process the option.
609 #ifdef PACKAGE_VERSION
610 printf("%s, revision %s\n", PNAME
, PACKAGE_VERSION
);
612 printf("%s: version number not available when not built"
613 " via automake\n", PNAME
);
618 extern const char *const licence
[];
621 for (i
= 0; licence
[i
]; i
++)
622 fputs(licence
[i
], stdout
);
628 if (nactions
>= actionsize
) {
629 actionsize
= nactions
* 3 / 2 + 16;
630 actions
= sresize(actions
, actionsize
, struct action
);
632 actions
[nactions
].mode
= SCAN
;
633 actions
[nactions
].arg
= optval
;
637 if (nactions
>= actionsize
) {
638 actionsize
= nactions
* 3 / 2 + 16;
639 actions
= sresize(actions
, actionsize
, struct action
);
641 actions
[nactions
].mode
= SCANDUMP
;
642 actions
[nactions
].arg
= optval
;
646 if (nactions
>= actionsize
) {
647 actionsize
= nactions
* 3 / 2 + 16;
648 actions
= sresize(actions
, actionsize
, struct action
);
650 actions
[nactions
].mode
= DUMP
;
651 actions
[nactions
].arg
= NULL
;
655 if (nactions
>= actionsize
) {
656 actionsize
= nactions
* 3 / 2 + 16;
657 actions
= sresize(actions
, actionsize
, struct action
);
659 actions
[nactions
].mode
= LOAD
;
660 actions
[nactions
].arg
= NULL
;
664 if (nactions
>= actionsize
) {
665 actionsize
= nactions
* 3 / 2 + 16;
666 actions
= sresize(actions
, actionsize
, struct action
);
668 actions
[nactions
].mode
= TEXT
;
669 actions
[nactions
].arg
= optval
;
673 if (nactions
>= actionsize
) {
674 actionsize
= nactions
* 3 / 2 + 16;
675 actions
= sresize(actions
, actionsize
, struct action
);
677 actions
[nactions
].mode
= HTML
;
678 actions
[nactions
].arg
= optval
;
682 if (nactions
>= actionsize
) {
683 actionsize
= nactions
* 3 / 2 + 16;
684 actions
= sresize(actions
, actionsize
, struct action
);
686 actions
[nactions
].mode
= HTTPD
;
687 actions
[nactions
].arg
= NULL
;
691 if (nactions
>= actionsize
) {
692 actionsize
= nactions
* 3 / 2 + 16;
693 actions
= sresize(actions
, actionsize
, struct action
);
695 actions
[nactions
].mode
= REMOVE
;
696 actions
[nactions
].arg
= NULL
;
705 case OPT_TTYPROGRESS
:
730 tqdepth
= atoi(optval
);
733 textcutoff
= parse_age(now
, optval
);
736 if (!strcmp(optval
, "auto")) {
737 htmlautoagerange
= 1;
739 char *q
= optval
+ strcspn(optval
, "-:");
742 htmloldest
= parse_age(now
, optval
);
743 htmlnewest
= *q ?
parse_age(now
, q
) : now
;
744 htmlautoagerange
= 0;
750 if (optval
[0] == '[' &&
751 (port
= strchr(optval
, ']')) != NULL
)
755 port
+= strcspn(port
, ":");
758 httpserveraddr
= optval
;
759 httpserverport
= atoi(port
);
763 if (!strcmp(optval
, "magic"))
764 auth
= HTTPD_AUTH_MAGIC
;
765 else if (!strcmp(optval
, "basic"))
766 auth
= HTTPD_AUTH_BASIC
;
767 else if (!strcmp(optval
, "none"))
768 auth
= HTTPD_AUTH_NONE
;
769 else if (!strcmp(optval
, "default"))
770 auth
= HTTPD_AUTH_MAGIC
| HTTPD_AUTH_BASIC
;
771 else if (!strcmp(optval
, "help") ||
772 !strcmp(optval
, "list")) {
773 printf(PNAME
": supported HTTP authentication types"
775 " magic use Linux /proc/net/tcp to"
776 " determine owner of peer socket\n"
777 " basic HTTP Basic username and"
778 " password authentication\n"
779 " default use 'magic' if possible, "
780 " otherwise fall back to 'basic'\n"
781 " none unauthenticated HTTP (if"
782 " the data file is non-confidential)\n");
785 fprintf(stderr
, "%s: unrecognised authentication"
786 " type '%s'\n%*s options are 'magic',"
787 " 'basic', 'none', 'default'\n",
788 PNAME
, optval
, (int)strlen(PNAME
), "");
799 int authlen
, authsize
;
802 if (optid
== OPT_AUTHFILE
) {
803 fd
= open(optval
, O_RDONLY
);
805 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
,
806 optval
, strerror(errno
));
813 sprintf(namebuf
, "fd %d", fd
);
818 authbuf
= snewn(authsize
, char);
819 while ((ret
= read(fd
, authbuf
+authlen
,
820 authsize
-authlen
)) > 0) {
822 if ((authsize
- authlen
) < (authsize
/ 16)) {
823 authsize
= authlen
* 3 / 2 + 4096;
824 authbuf
= sresize(authbuf
, authsize
, char);
828 fprintf(stderr
, "%s: %s: read: %s\n", PNAME
,
829 name
, strerror(errno
));
832 if (optid
== OPT_AUTHFILE
)
834 httpauthdata
= authbuf
;
838 case OPT_INCLUDEPATH
:
840 case OPT_EXCLUDEPATH
:
843 if (ninex
>= inexsize
) {
844 inexsize
= ninex
* 3 / 2 + 16;
845 inex
= sresize(inex
, inexsize
,
846 struct inclusion_exclusion
);
848 inex
[ninex
].path
= (optid
== OPT_INCLUDEPATH
||
849 optid
== OPT_EXCLUDEPATH
||
850 optid
== OPT_PRUNEPATH
);
851 inex
[ninex
].type
= (optid
== OPT_INCLUDE ?
1 :
852 optid
== OPT_INCLUDEPATH ?
1 :
853 optid
== OPT_EXCLUDE ?
0 :
854 optid
== OPT_EXCLUDEPATH ?
0 :
855 optid
== OPT_PRUNE ?
-1 :
856 /* optid == OPT_PRUNEPATH ? */ -1);
857 inex
[ninex
].wildcard
= optval
;
863 fprintf(stderr
, "%s: unexpected argument '%s'\n", PNAME
, p
);
873 for (action
= 0; action
< nactions
; action
++) {
874 int mode
= actions
[action
].mode
;
876 if (mode
== SCAN
|| mode
== SCANDUMP
|| mode
== LOAD
) {
877 const char *scandir
= actions
[action
].arg
;
880 char *buf
= fgetline(stdin
);
882 buf
[strcspn(buf
, "\r\n")] = '\0';
883 if (1 != sscanf(buf
, DUMPHDR
"%x",
885 fprintf(stderr
, "%s: header in dump file not recognised\n",
889 pathsep
= (char)newpathsep
;
893 if (mode
== SCAN
|| mode
== LOAD
) {
895 * Prepare to write out the index file.
897 fd
= open(filename
, O_RDWR
| O_TRUNC
| O_CREAT
,
900 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
, filename
,
904 if (fstat(fd
, &st
) < 0) {
905 perror(PNAME
": fstat");
908 ctx
->datafile_dev
= st
.st_dev
;
909 ctx
->datafile_ino
= st
.st_ino
;
910 ctx
->straight_to_dump
= 0;
912 ctx
->datafile_dev
= -1;
913 ctx
->datafile_ino
= -1;
914 ctx
->straight_to_dump
= 1;
917 if (mode
== SCAN
|| mode
== SCANDUMP
) {
918 if (stat(scandir
, &st
) < 0) {
919 fprintf(stderr
, "%s: %s: stat: %s\n", PNAME
, scandir
,
923 ctx
->filesystem_dev
= crossfs ?
0 : st
.st_dev
;
928 ctx
->crossfs
= crossfs
;
929 ctx
->fakeatimes
= fakediratimes
;
930 ctx
->usemtime
= mtime
;
932 ctx
->last_output_update
= time(NULL
);
934 /* progress==1 means report progress only if stderr is a tty */
936 progress
= isatty(2) ?
2 : 0;
937 ctx
->progress
= progress
;
940 if (progress
&& ioctl(2, TIOCGWINSZ
, &ws
) == 0)
941 ctx
->progwidth
= ws
.ws_col
- 1;
946 if (mode
== SCANDUMP
)
947 printf(DUMPHDR
"%02x\n", (unsigned char)pathsep
);
950 * Scan the directory tree, and write out the trie component
953 if (mode
!= SCANDUMP
) {
954 ctx
->tb
= triebuild_new(fd
);
959 while ((buf
= fgetline(stdin
)) != NULL
) {
963 buf
[strcspn(buf
, "\r\n")] = '\0';
967 while (*p
&& *p
!= ' ') p
++;
969 fprintf(stderr
, "%s: dump file line %d: expected at least"
970 " three fields\n", PNAME
, line
);
974 tf
.size
= strtoull(q
, NULL
, 10);
976 while (*p
&& *p
!= ' ') p
++;
978 fprintf(stderr
, "%s: dump file line %d: expected at least"
979 " three fields\n", PNAME
, line
);
983 tf
.atime
= strtoull(q
, NULL
, 10);
991 for (i
= 0; i
< 2; i
++) {
993 if (*p
>= '0' && *p
<= '9')
995 else if (*p
>= 'A' && *p
<= 'F')
996 c
+= *p
- ('A' - 10);
997 else if (*p
>= 'a' && *p
<= 'f')
998 c
+= *p
- ('a' - 10);
1000 fprintf(stderr
, "%s: dump file line %d: unable"
1001 " to parse hex escape\n", PNAME
, line
);
1010 triebuild_add(ctx
->tb
, buf
, &tf
);
1015 du(scandir
, gotdata
, ctx
);
1017 if (mode
!= SCANDUMP
) {
1019 char *buf
, *prevbuf
;
1021 count
= triebuild_finish(ctx
->tb
);
1022 triebuild_free(ctx
->tb
);
1024 if (ctx
->progress
) {
1025 fprintf(stderr
, "%-*s\r", ctx
->progwidth
, "");
1030 * Work out how much space the cumulative index trees
1031 * will take; enlarge the file, and memory-map it.
1033 if (fstat(fd
, &st
) < 0) {
1034 perror(PNAME
": fstat");
1038 printf("Built pathname index, %d entries, %llu bytes\n", count
,
1039 (unsigned long long)st
.st_size
);
1041 totalsize
= index_compute_size(st
.st_size
, count
);
1043 if (lseek(fd
, totalsize
-1, SEEK_SET
) < 0) {
1044 perror(PNAME
": lseek");
1047 if (write(fd
, "\0", 1) < 1) {
1048 perror(PNAME
": write");
1052 printf("Upper bound on index file size = %llu bytes\n",
1053 (unsigned long long)totalsize
);
1055 mappedfile
= mmap(NULL
, totalsize
, PROT_READ
|PROT_WRITE
,MAP_SHARED
, fd
, 0);
1057 perror(PNAME
": mmap");
1061 if (fakediratimes
) {
1062 printf("Faking directory atimes\n");
1063 trie_fake_dir_atimes(mappedfile
);
1066 printf("Building index\n");
1067 ib
= indexbuild_new(mappedfile
, st
.st_size
, count
);
1068 maxpathlen
= trie_maxpathlen(mappedfile
);
1069 buf
= snewn(maxpathlen
, char);
1070 prevbuf
= snewn(maxpathlen
, char);
1071 tw
= triewalk_new(mappedfile
);
1073 tf
= triewalk_next(tw
, buf
);
1079 * Get the next file from the index. So we are
1080 * currently holding, and have not yet
1081 * indexed, prevtf (with pathname prevbuf) and
1082 * tf (with pathname buf).
1085 memcpy(prevbuf
, buf
, maxpathlen
);
1086 tf
= triewalk_next(tw
, buf
);
1092 * Find the first differing character position
1093 * between our two pathnames.
1095 for (i
= 0; prevbuf
[i
] && prevbuf
[i
] == buf
[i
]; i
++);
1098 * If prevbuf was a directory name and buf is
1099 * something inside that directory, then
1100 * trie_before() will be called on prevbuf
1101 * itself. Hence we must drop a tag before it,
1102 * so that the resulting index is usable.
1104 if ((!prevbuf
[i
] && (buf
[i
] == pathsep
||
1105 (i
> 0 && buf
[i
-1] == pathsep
))))
1109 * Add prevtf to the index.
1111 indexbuild_add(ib
, prevtf
);
1115 * Drop an unconditional final tag, and
1116 * get out of this loop.
1123 * In full-index mode, index everything.
1129 * If prevbuf was a filename inside some
1130 * directory which buf is outside, then
1131 * trie_before() will be called on some
1132 * pathname either equal to buf or epsilon
1133 * less than it. Either way, we're going to
1134 * need to drop a tag after prevtf.
1136 if (strchr(prevbuf
+i
, pathsep
) || !tf
)
1141 realsize
= indexbuild_realsize(ib
);
1142 indexbuild_free(ib
);
1144 munmap(mappedfile
, totalsize
);
1145 ftruncate(fd
, realsize
);
1147 printf("Actual index file size = %llu bytes\n",
1148 (unsigned long long)realsize
);
1150 } else if (mode
== TEXT
) {
1151 char *querydir
= actions
[action
].arg
;
1154 fd
= open(filename
, O_RDONLY
);
1156 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
, filename
,
1160 if (fstat(fd
, &st
) < 0) {
1161 perror(PNAME
": fstat");
1164 totalsize
= st
.st_size
;
1165 mappedfile
= mmap(NULL
, totalsize
, PROT_READ
, MAP_SHARED
, fd
, 0);
1167 perror(PNAME
": mmap");
1170 pathsep
= trie_pathsep(mappedfile
);
1173 * Trim trailing slash, just in case.
1175 pathlen
= strlen(querydir
);
1176 if (pathlen
> 0 && querydir
[pathlen
-1] == pathsep
)
1177 querydir
[--pathlen
] = '\0';
1179 text_query(mappedfile
, querydir
, textcutoff
, tqdepth
);
1181 munmap(mappedfile
, totalsize
);
1182 } else if (mode
== HTML
) {
1183 char *querydir
= actions
[action
].arg
;
1185 struct html_config cfg
;
1189 fd
= open(filename
, O_RDONLY
);
1191 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
, filename
,
1195 if (fstat(fd
, &st
) < 0) {
1196 perror(PNAME
": fstat");
1199 totalsize
= st
.st_size
;
1200 mappedfile
= mmap(NULL
, totalsize
, PROT_READ
, MAP_SHARED
, fd
, 0);
1202 perror(PNAME
": mmap");
1205 pathsep
= trie_pathsep(mappedfile
);
1208 * Trim trailing slash, just in case.
1210 pathlen
= strlen(querydir
);
1211 if (pathlen
> 0 && querydir
[pathlen
-1] == pathsep
)
1212 querydir
[--pathlen
] = '\0';
1214 xi
= trie_before(mappedfile
, querydir
);
1216 cfg
.autoage
= htmlautoagerange
;
1217 cfg
.oldest
= htmloldest
;
1218 cfg
.newest
= htmlnewest
;
1219 html
= html_query(mappedfile
, xi
, &cfg
);
1220 fputs(html
, stdout
);
1222 munmap(mappedfile
, totalsize
);
1223 } else if (mode
== DUMP
) {
1227 fd
= open(filename
, O_RDONLY
);
1229 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
, filename
,
1233 if (fstat(fd
, &st
) < 0) {
1234 perror(PNAME
": fstat");
1237 totalsize
= st
.st_size
;
1238 mappedfile
= mmap(NULL
, totalsize
, PROT_READ
, MAP_SHARED
, fd
, 0);
1240 perror(PNAME
": mmap");
1243 pathsep
= trie_pathsep(mappedfile
);
1245 maxpathlen
= trie_maxpathlen(mappedfile
);
1246 buf
= snewn(maxpathlen
, char);
1248 printf(DUMPHDR
"%02x\n", (unsigned char)pathsep
);
1249 tw
= triewalk_new(mappedfile
);
1250 while ((tf
= triewalk_next(tw
, buf
)) != NULL
)
1254 munmap(mappedfile
, totalsize
);
1255 } else if (mode
== HTTPD
) {
1256 struct html_config pcfg
;
1257 struct httpd_config dcfg
;
1259 fd
= open(filename
, O_RDONLY
);
1261 fprintf(stderr
, "%s: %s: open: %s\n", PNAME
, filename
,
1265 if (fstat(fd
, &st
) < 0) {
1266 perror(PNAME
": fstat");
1269 totalsize
= st
.st_size
;
1270 mappedfile
= mmap(NULL
, totalsize
, PROT_READ
, MAP_SHARED
, fd
, 0);
1272 perror(PNAME
": mmap");
1275 pathsep
= trie_pathsep(mappedfile
);
1277 dcfg
.address
= httpserveraddr
;
1278 dcfg
.port
= httpserverport
;
1279 dcfg
.basicauthdata
= httpauthdata
;
1281 pcfg
.autoage
= htmlautoagerange
;
1282 pcfg
.oldest
= htmloldest
;
1283 pcfg
.newest
= htmlnewest
;
1284 run_httpd(mappedfile
, auth
, &dcfg
, &pcfg
);
1285 munmap(mappedfile
, totalsize
);
1286 } else if (mode
== REMOVE
) {
1287 if (remove(filename
) < 0) {
1288 fprintf(stderr
, "%s: %s: remove: %s\n", PNAME
, filename
,