X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/agedu/blobdiff_plain/e9e7a1bfea4a4620be655143f8cce84a3cace46e..HEAD:/agedu.c diff --git a/agedu.c b/agedu.c index 6e3a2e5..1847e2c 100644 --- a/agedu.c +++ b/agedu.c @@ -2,34 +2,25 @@ * Main program for agedu. */ -#define _GNU_SOURCE -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include -#include -#include -#include +#include "agedu.h" #include "du.h" #include "trie.h" #include "index.h" -#include "malloc.h" +#include "alloc.h" #include "html.h" #include "httpd.h" +#include "fgetline.h" -#define PNAME "agedu" - -#define lenof(x) (sizeof((x))/sizeof(*(x))) +/* + * Path separator. This global variable affects the behaviour of + * various parts of the code when they need to deal with path + * separators. The path separator appropriate to a particular data + * set is encoded in the index file storing that data set; data + * sets generated on Unix will of course have the default '/', but + * foreign data sets are conceivable and must be handled correctly. + */ +char pathsep = '/'; void fatal(const char *fmt, ...) { @@ -43,7 +34,7 @@ void fatal(const char *fmt, ...) } struct inclusion_exclusion { - int include; + int type; const char *wildcard; int path; }; @@ -54,12 +45,32 @@ struct ctx { ino_t datafile_ino; time_t last_output_update; int progress, progwidth; + int straight_to_dump; struct inclusion_exclusion *inex; int ninex; int crossfs; + int usemtime; + int fakeatimes; }; -static int gotdata(void *vctx, const char *pathname, const struct stat64 *st) +static void dump_line(const char *pathname, const struct trie_file *tf) +{ + const char *p; + if (printf("%llu %llu ", tf->size, tf->atime) < 0) goto error; + for (p = pathname; *p; p++) { + if (*p >= ' ' && *p < 127 && *p != '%') { + if (putchar(*p) == EOF) goto error; + } else { + if (printf("%%%02x", (unsigned char)*p) < 0) goto error; + } + } + if (putchar('\n') == EOF) goto error; + return; + error: + fatal("standard output: %s", strerror(errno)); +} + +static int gotdata(void *vctx, const char *pathname, const STRUCT_STAT *st) { struct ctx *ctx = (struct ctx *)vctx; struct trie_file file; @@ -79,48 +90,85 @@ static int gotdata(void *vctx, const char *pathname, const struct stat64 *st) if (!ctx->crossfs && st->st_dev != ctx->filesystem_dev) return 0; + file.size = (unsigned long long)512 * st->st_blocks; + if (ctx->usemtime || (ctx->fakeatimes && S_ISDIR(st->st_mode))) + file.atime = st->st_mtime; + else + file.atime = max(st->st_mtime, st->st_atime); + /* * Filter based on wildcards. */ include = 1; - filename = strrchr(pathname, '/'); + filename = strrchr(pathname, pathsep); if (!filename) filename = pathname; else filename++; for (i = 0; i < ctx->ninex; i++) { if (fnmatch(ctx->inex[i].wildcard, - ctx->inex[i].path ? pathname : filename, - FNM_PATHNAME) == 0) - include = ctx->inex[i].include; + ctx->inex[i].path ? pathname : filename, 0) == 0) + include = ctx->inex[i].type; + } + if (include == -1) + return 0; /* ignore this entry and any subdirs */ + if (include == 0) { + /* + * Here we are supposed to be filtering an entry out, but + * still recursing into it if it's a directory. However, + * we can't actually leave out any directory whose + * subdirectories we then look at. So we cheat, in that + * case, by setting the size to zero. + */ + if (!S_ISDIR(st->st_mode)) + return 0; /* just ignore */ + else + file.size = 0; } - if (!include) - return 1; /* filter, but don't prune */ - file.blocks = st->st_blocks; - file.atime = st->st_atime; - triebuild_add(ctx->tb, pathname, &file); + if (ctx->straight_to_dump) + dump_line(pathname, &file); + else + triebuild_add(ctx->tb, pathname, &file); - t = time(NULL); - if (t != ctx->last_output_update) { - if (ctx->progress) { + if (ctx->progress) { + t = time(NULL); + if (t != ctx->last_output_update) { fprintf(stderr, "%-*.*s\r", ctx->progwidth, ctx->progwidth, pathname); fflush(stderr); + ctx->last_output_update = t; } - ctx->last_output_update = t; } return 1; } +static void scan_error(void *vctx, const char *fmt, ...) +{ + struct ctx *ctx = (struct ctx *)vctx; + va_list ap; + + if (ctx->progress) { + fprintf(stderr, "%-*s\r", ctx->progwidth, ""); + fflush(stderr); + } + + fprintf(stderr, "%s: ", PNAME); + va_start(ap, fmt); + vfprintf(stderr, fmt, ap); + va_end(ap); + + ctx->last_output_update--; /* force a progress report next time */ +} + static void text_query(const void *mappedfile, const char *querydir, - time_t t, int depth) + time_t t, int showfiles, int depth, FILE *fp) { size_t maxpathlen; char *pathbuf; unsigned long xi1, xi2; - unsigned long long s1, s2; + unsigned long long size; maxpathlen = trie_maxpathlen(mappedfile); pathbuf = snewn(maxpathlen + 1, char); @@ -130,32 +178,56 @@ static void text_query(const void *mappedfile, const char *querydir, * (inclusive) and that filename with a ^A on the end * (exclusive). So find the x indices for each. */ - sprintf(pathbuf, "%s\001", querydir); + strcpy(pathbuf, querydir); + make_successor(pathbuf); xi1 = trie_before(mappedfile, querydir); xi2 = trie_before(mappedfile, pathbuf); + if (!showfiles && xi2 - xi1 == 1) + return; /* file, or empty dir => no display */ + /* * Now do the lookups in the age index. */ - s1 = index_query(mappedfile, xi1, t); - s2 = index_query(mappedfile, xi2, t); + if (xi2 - xi1 == 1) { + /* + * We are querying an individual file, so we should not + * depend on the index entries either side of the node, + * since they almost certainly don't both exist. Instead, + * just look up the file's size and atime in the main trie. + */ + const struct trie_file *f = trie_getfile(mappedfile, xi1); + if (f->atime < t) + size = f->size; + else + size = 0; + } else { + unsigned long long s1, s2; + s1 = index_query(mappedfile, xi1, t); + s2 = index_query(mappedfile, xi2, t); + size = s2 - s1; + } - /* Display in units of 2 512-byte blocks = 1Kb */ - printf("%-11llu %s\n", (s2 - s1) / 2, querydir); + if (size == 0) + return; /* no space taken up => no display */ - if (depth > 0) { + if (depth != 0) { /* * Now scan for first-level subdirectories and report * those too. */ + int newdepth = (depth > 0 ? depth - 1 : depth); xi1++; while (xi1 < xi2) { trie_getpath(mappedfile, xi1, pathbuf); - text_query(mappedfile, pathbuf, t, depth-1); - strcat(pathbuf, "\001"); + text_query(mappedfile, pathbuf, t, showfiles, newdepth, fp); + make_successor(pathbuf); xi1 = trie_before(mappedfile, pathbuf); } } + + /* Display in units of 1Kb */ + fprintf(fp, "%-11llu %s\n", (size) / 1024, querydir); } /* @@ -213,47 +285,88 @@ static void text_query(const void *mappedfile, const char *querydir, * not bother defining logical identifiers for them at all - those * would be automatically generated, since I wouldn't have any * need to specify them manually in another part of the code.) + * + * One other helpful consequence of the enum-based structure here + * is that it causes a compiler error if I accidentally try to + * define the same option (short or long) twice. */ #define OPTHELP(NOVAL, VAL, SHORT, LONG, HELPPFX, HELPARG, HELPLINE, HELPOPT) \ - HELPPFX("usage") HELPLINE("agedu [options] action") \ + HELPPFX("usage") HELPLINE(PNAME " [options] action [action...]") \ HELPPFX("actions") \ VAL(SCAN) SHORT(s) LONG(scan) \ HELPARG("directory") HELPOPT("scan and index a directory") \ - NOVAL(DUMP) SHORT(d) LONG(dump) HELPOPT("dump the index file") \ + NOVAL(HTTPD) SHORT(w) LONG(web) LONG(server) LONG(httpd) \ + HELPOPT("serve HTML reports from a temporary web server") \ VAL(TEXT) SHORT(t) LONG(text) \ HELPARG("subdir") HELPOPT("print a plain text report on a subdirectory") \ + NOVAL(REMOVE) SHORT(R) LONG(remove) LONG(delete) LONG(unlink) \ + HELPOPT("remove the index file") \ + NOVAL(DUMP) SHORT(D) LONG(dump) HELPOPT("dump the index file on stdout") \ + NOVAL(LOAD) SHORT(L) LONG(load) \ + HELPOPT("load and index a dump file") \ + VAL(SCANDUMP) SHORT(S) LONG(scan_dump) \ + HELPARG("directory") HELPOPT("scan only, generating a dump") \ VAL(HTML) SHORT(H) LONG(html) \ HELPARG("subdir") HELPOPT("print an HTML report on a subdirectory") \ - NOVAL(HTTPD) SHORT(w) LONG(web) LONG(server) LONG(httpd) \ - HELPOPT("serve reports from a temporary web server") \ + NOVAL(CGI) LONG(cgi) \ + HELPOPT("do the right thing when run from a CGI script") \ HELPPFX("options") \ VAL(DATAFILE) SHORT(f) LONG(file) \ - HELPARG("filename") HELPOPT("[all modes] specify index file") \ - NOVAL(PROGRESS) LONG(progress) LONG(scan_progress) \ - HELPOPT("[--scan] report progress on stderr") \ - NOVAL(NOPROGRESS) LONG(no_progress) LONG(no_scan_progress) \ - HELPOPT("[--scan] do not report progress") \ - NOVAL(TTYPROGRESS) LONG(tty_progress) LONG(tty_scan_progress) \ - LONG(progress_tty) LONG(scan_progress_tty) \ - HELPOPT("[--scan] report progress if stderr is a tty") \ + HELPARG("filename") HELPOPT("[most modes] specify index file") \ NOVAL(CROSSFS) LONG(cross_fs) \ HELPOPT("[--scan] cross filesystem boundaries") \ NOVAL(NOCROSSFS) LONG(no_cross_fs) \ HELPOPT("[--scan] stick to one filesystem") \ - VAL(INCLUDE) LONG(include) \ - HELPARG("wildcard") HELPOPT("[--scan] include files matching pattern") \ - VAL(INCLUDEPATH) LONG(include_path) \ - HELPARG("wildcard") HELPOPT("[--scan] include pathnames matching pattern") \ + VAL(PRUNE) LONG(prune) \ + HELPARG("wildcard") HELPOPT("[--scan] prune files matching pattern") \ + VAL(PRUNEPATH) LONG(prune_path) \ + HELPARG("wildcard") HELPOPT("[--scan] prune pathnames matching pattern") \ VAL(EXCLUDE) LONG(exclude) \ HELPARG("wildcard") HELPOPT("[--scan] exclude files matching pattern") \ VAL(EXCLUDEPATH) LONG(exclude_path) \ HELPARG("wildcard") HELPOPT("[--scan] exclude pathnames matching pattern") \ - VAL(MINAGE) SHORT(a) LONG(age) LONG(min_age) LONG(minimum_age) \ - HELPARG("age") HELPOPT("[--text] include only files older than this") \ + VAL(INCLUDE) LONG(include) \ + HELPARG("wildcard") HELPOPT("[--scan] include files matching pattern") \ + VAL(INCLUDEPATH) LONG(include_path) \ + HELPARG("wildcard") HELPOPT("[--scan] include pathnames matching pattern") \ + NOVAL(PROGRESS) LONG(progress) LONG(scan_progress) \ + HELPOPT("[--scan] report progress on stderr") \ + NOVAL(NOPROGRESS) LONG(no_progress) LONG(no_scan_progress) \ + HELPOPT("[--scan] do not report progress") \ + NOVAL(TTYPROGRESS) LONG(tty_progress) LONG(tty_scan_progress) \ + LONG(progress_tty) LONG(scan_progress_tty) \ + HELPOPT("[--scan] report progress if stderr is a tty") \ + NOVAL(DIRATIME) LONG(dir_atime) LONG(dir_atimes) \ + HELPOPT("[--scan,--load] keep real atimes on directories") \ + NOVAL(NODIRATIME) LONG(no_dir_atime) LONG(no_dir_atimes) \ + HELPOPT("[--scan,--load] fake atimes on directories") \ + NOVAL(NOEOF) LONG(no_eof) LONG(noeof) \ + HELPOPT("[--web] do not close web server on EOF") \ + NOVAL(MTIME) LONG(mtime) \ + HELPOPT("[--scan] use mtime instead of atime") \ + NOVAL(SHOWFILES) LONG(files) \ + HELPOPT("[--web,--html,--text] list individual files") \ + VAL(AGERANGE) SHORT(r) LONG(age_range) LONG(range) LONG(ages) \ + HELPARG("age[-age]") HELPOPT("[--web,--html] set limits of colour coding") \ + VAL(OUTFILE) SHORT(o) LONG(output) \ + HELPARG("filename") HELPOPT("[--html] specify output file or directory name") \ + VAL(SERVERADDR) LONG(address) LONG(addr) LONG(server_address) \ + LONG(server_addr) \ + HELPARG("addr[:port]") HELPOPT("[--web] specify HTTP server address") \ VAL(AUTH) LONG(auth) LONG(http_auth) LONG(httpd_auth) \ LONG(server_auth) LONG(web_auth) \ HELPARG("type") HELPOPT("[--web] specify HTTP authentication method") \ + VAL(AUTHFILE) LONG(auth_file) \ + HELPARG("filename") HELPOPT("[--web] read HTTP Basic user/pass from file") \ + VAL(AUTHFD) LONG(auth_fd) \ + HELPARG("fd") HELPOPT("[--web] read HTTP Basic user/pass from fd") \ + VAL(HTMLTITLE) LONG(title) \ + HELPARG("title") HELPOPT("[--web,--html] title prefix for web pages") \ + VAL(DEPTH) SHORT(d) LONG(depth) LONG(max_depth) LONG(maximum_depth) \ + HELPARG("levels") HELPOPT("[--text,--html] recurse to this many levels") \ + VAL(MINAGE) SHORT(a) LONG(age) LONG(min_age) LONG(minimum_age) \ + HELPARG("age") HELPOPT("[--text] include only files older than this") \ HELPPFX("also") \ NOVAL(HELP) SHORT(h) LONG(help) HELPOPT("display this help text") \ NOVAL(VERSION) SHORT(V) LONG(version) HELPOPT("report version number") \ @@ -284,8 +397,8 @@ enum { OPTIONS(IGNORE,IGNORE,IGNORE,LONGTMP) NLONGOPTS }; static const int opthasval[NOPTIONS] = {OPTIONS(ZERO,ONE,IGNORE,IGNORE)}; static const char shortopts[] = {OPTIONS(IGNORE,IGNORE,STRINGNOCOMMA,IGNORE)}; static const char *const longopts[] = {OPTIONS(IGNORE,IGNORE,IGNORE,STRING)}; -enum { OPTIONS(SHORTNEWOPT,SHORTNEWOPT,SHORTTHISOPT,IGNORE) }; -enum { OPTIONS(LONGNEWOPT,LONGNEWOPT,IGNORE,LONGTHISOPT) }; +enum { OPTIONS(SHORTNEWOPT,SHORTNEWOPT,SHORTTHISOPT,IGNORE) UNUSEDENUMVAL1 }; +enum { OPTIONS(LONGNEWOPT,LONGNEWOPT,IGNORE,LONGTHISOPT) UNUSEDENUMVAL2 }; static const int shortvals[] = {OPTIONS(IGNORE,IGNORE,SHORTOPTVAL,IGNORE)}; static const int longvals[] = {OPTIONS(IGNORE,IGNORE,IGNORE,LONGOPTVAL)}; @@ -329,6 +442,45 @@ static void usage(FILE *fp) #undef HELPOPT } +static time_t parse_age(time_t now, const char *agestr) +{ + time_t t; + struct tm tm; + int nunits; + char unit[2]; + + t = now; + + if (2 != sscanf(agestr, "%d%1[DdWwMmYy]", &nunits, unit)) { + fprintf(stderr, "%s: age specification should be a number followed by" + " one of d,w,m,y\n", PNAME); + exit(1); + } + + if (unit[0] == 'd') { + t -= 86400 * nunits; + } else if (unit[0] == 'w') { + t -= 86400 * 7 * nunits; + } else { + int ym; + + tm = *localtime(&t); + ym = tm.tm_year * 12 + tm.tm_mon; + + if (unit[0] == 'm') + ym -= nunits; + else + ym -= 12 * nunits; + + tm.tm_year = ym / 12; + tm.tm_mon = ym % 12; + + t = mktime(&tm); + } + + return t; +} + int main(int argc, char **argv) { int fd, count; @@ -338,18 +490,33 @@ int main(int argc, char **argv) void *mappedfile; triewalk *tw; indexbuild *ib; - const struct trie_file *tf; - char *filename = "agedu.dat"; - char *scandir = NULL; - char *querydir = NULL; + const struct trie_file *tf, *prevtf; + char *filename = PNAME ".dat"; int doing_opts = 1; - enum { USAGE, TEXT, HTML, SCAN, DUMP, HTTPD } mode = USAGE; - char *minage = "0d"; + enum { TEXT, HTML, SCAN, DUMP, SCANDUMP, LOAD, HTTPD, REMOVE }; + struct action { + int mode; + char *arg; + } *actions = NULL; + int nactions = 0, actionsize = 0, action; + time_t now = time(NULL); + time_t textcutoff = now, htmlnewest = now, htmloldest = now; + int htmlautoagerange = 1; + const char *httpserveraddr = "localhost"; + const char *httpserverport = NULL; + const char *httpauthdata = NULL; + const char *outfile = NULL; + const char *html_title = PNAME; int auth = HTTPD_AUTH_MAGIC | HTTPD_AUTH_BASIC; int progress = 1; struct inclusion_exclusion *inex = NULL; int ninex = 0, inexsize = 0; int crossfs = 0; + int depth = -1, gotdepth = 0; + int fakediratimes = 1; + int mtime = 0; + int closeoneof = 1; + int showfiles = 0; #ifdef DEBUG_MAD_OPTION_PARSING_MACROS { @@ -489,28 +656,95 @@ int main(int argc, char **argv) usage(stdout); return 0; case OPT_VERSION: - printf("FIXME: version();\n"); +#ifdef PACKAGE_VERSION + printf("%s, revision %s\n", PNAME, PACKAGE_VERSION); +#else + printf("%s: version number not available when not built" + " via automake\n", PNAME); +#endif return 0; case OPT_LICENCE: - printf("FIXME: licence();\n"); + { + extern const char *const licence[]; + int i; + + for (i = 0; licence[i]; i++) + fputs(licence[i], stdout); + } return 0; case OPT_SCAN: - mode = SCAN; - scandir = optval; + if (nactions >= actionsize) { + actionsize = nactions * 3 / 2 + 16; + actions = sresize(actions, actionsize, struct action); + } + actions[nactions].mode = SCAN; + actions[nactions].arg = optval; + nactions++; + break; + case OPT_SCANDUMP: + if (nactions >= actionsize) { + actionsize = nactions * 3 / 2 + 16; + actions = sresize(actions, actionsize, struct action); + } + actions[nactions].mode = SCANDUMP; + actions[nactions].arg = optval; + nactions++; break; case OPT_DUMP: - mode = DUMP; + if (nactions >= actionsize) { + actionsize = nactions * 3 / 2 + 16; + actions = sresize(actions, actionsize, struct action); + } + actions[nactions].mode = DUMP; + actions[nactions].arg = NULL; + nactions++; + break; + case OPT_LOAD: + if (nactions >= actionsize) { + actionsize = nactions * 3 / 2 + 16; + actions = sresize(actions, actionsize, struct action); + } + actions[nactions].mode = LOAD; + actions[nactions].arg = NULL; + nactions++; break; case OPT_TEXT: - querydir = optval; - mode = TEXT; + if (nactions >= actionsize) { + actionsize = nactions * 3 / 2 + 16; + actions = sresize(actions, actionsize, struct action); + } + actions[nactions].mode = TEXT; + actions[nactions].arg = optval; + nactions++; break; case OPT_HTML: - mode = HTML; - querydir = optval; + case OPT_CGI: + if (nactions >= actionsize) { + actionsize = nactions * 3 / 2 + 16; + actions = sresize(actions, actionsize, struct action); + } + actions[nactions].mode = HTML; + actions[nactions].arg = (optid == OPT_HTML ? optval : + NULL); + nactions++; break; case OPT_HTTPD: - mode = HTTPD; + if (nactions >= actionsize) { + actionsize = nactions * 3 / 2 + 16; + actions = sresize(actions, actionsize, struct action); + } + actions[nactions].mode = HTTPD; + actions[nactions].arg = NULL; + nactions++; + break; + case OPT_REMOVE: + if (nactions >= actionsize) { + actionsize = nactions * 3 / 2 + 16; + actions = sresize(actions, actionsize, struct action); + } + actions[nactions].mode = REMOVE; + actions[nactions].arg = NULL; + nactions++; break; case OPT_PROGRESS: progress = 2; @@ -527,11 +761,74 @@ int main(int argc, char **argv) case OPT_NOCROSSFS: crossfs = 0; break; + case OPT_DIRATIME: + fakediratimes = 0; + break; + case OPT_NODIRATIME: + fakediratimes = 1; + break; + case OPT_SHOWFILES: + showfiles = 1; + break; + case OPT_MTIME: + mtime = 1; + break; + case OPT_NOEOF: + closeoneof = 0; + break; case OPT_DATAFILE: filename = optval; break; + case OPT_DEPTH: + if (!strcasecmp(optval, "unlimited") || + !strcasecmp(optval, "infinity") || + !strcasecmp(optval, "infinite") || + !strcasecmp(optval, "inf") || + !strcasecmp(optval, "maximum") || + !strcasecmp(optval, "max")) + depth = -1; + else + depth = atoi(optval); + gotdepth = 1; + break; + case OPT_OUTFILE: + outfile = optval; + break; + case OPT_HTMLTITLE: + html_title = optval; + break; case OPT_MINAGE: - minage = optval; + textcutoff = parse_age(now, optval); + break; + case OPT_AGERANGE: + if (!strcmp(optval, "auto")) { + htmlautoagerange = 1; + } else { + char *q = optval + strcspn(optval, "-:"); + if (*q) + *q++ = '\0'; + htmloldest = parse_age(now, optval); + htmlnewest = *q ? parse_age(now, q) : now; + htmlautoagerange = 0; + } + break; + case OPT_SERVERADDR: + { + char *port; + if (optval[0] == '[' && + (port = strchr(optval, ']')) != NULL) + port++; + else + port = optval; + port += strcspn(port, ":"); + if (port && *port) + *port++ = '\0'; + if (!strcmp(optval, "ANY")) + httpserveraddr = NULL; + else + httpserveraddr = optval; + httpserverport = port; + } break; case OPT_AUTH: if (!strcmp(optval, "magic")) @@ -542,7 +839,20 @@ int main(int argc, char **argv) auth = HTTPD_AUTH_NONE; else if (!strcmp(optval, "default")) auth = HTTPD_AUTH_MAGIC | HTTPD_AUTH_BASIC; - else { + else if (!strcmp(optval, "help") || + !strcmp(optval, "list")) { + printf(PNAME ": supported HTTP authentication types" + " are:\n" + " magic use Linux /proc/net/tcp to" + " determine owner of peer socket\n" + " basic HTTP Basic username and" + " password authentication\n" + " default use 'magic' if possible, " + " otherwise fall back to 'basic'\n" + " none unauthenticated HTTP (if" + " the data file is non-confidential)\n"); + return 0; + } else { fprintf(stderr, "%s: unrecognised authentication" " type '%s'\n%*s options are 'magic'," " 'basic', 'none', 'default'\n", @@ -550,19 +860,71 @@ int main(int argc, char **argv) return 1; } break; + case OPT_AUTHFILE: + case OPT_AUTHFD: + { + int fd; + char namebuf[40]; + const char *name; + char *authbuf; + int authlen, authsize; + int ret; + + if (optid == OPT_AUTHFILE) { + fd = open(optval, O_RDONLY); + if (fd < 0) { + fprintf(stderr, "%s: %s: open: %s\n", PNAME, + optval, strerror(errno)); + return 1; + } + name = optval; + } else { + fd = atoi(optval); + name = namebuf; + sprintf(namebuf, "fd %d", fd); + } + + authlen = 0; + authsize = 256; + authbuf = snewn(authsize, char); + while ((ret = read(fd, authbuf+authlen, + authsize-authlen)) > 0) { + authlen += ret; + if ((authsize - authlen) < (authsize / 16)) { + authsize = authlen * 3 / 2 + 4096; + authbuf = sresize(authbuf, authsize, char); + } + } + if (ret < 0) { + fprintf(stderr, "%s: %s: read: %s\n", PNAME, + name, strerror(errno)); + return 1; + } + if (optid == OPT_AUTHFILE) + close(fd); + httpauthdata = authbuf; + } + break; case OPT_INCLUDE: case OPT_INCLUDEPATH: case OPT_EXCLUDE: case OPT_EXCLUDEPATH: + case OPT_PRUNE: + case OPT_PRUNEPATH: if (ninex >= inexsize) { inexsize = ninex * 3 / 2 + 16; inex = sresize(inex, inexsize, struct inclusion_exclusion); } inex[ninex].path = (optid == OPT_INCLUDEPATH || - optid == OPT_EXCLUDEPATH); - inex[ninex].include = (optid == OPT_INCLUDE || - optid == OPT_INCLUDEPATH); + optid == OPT_EXCLUDEPATH || + optid == OPT_PRUNEPATH); + inex[ninex].type = (optid == OPT_INCLUDE ? 1 : + optid == OPT_INCLUDEPATH ? 1 : + optid == OPT_EXCLUDE ? 0 : + optid == OPT_EXCLUDEPATH ? 0 : + optid == OPT_PRUNE ? -1 : + /* optid == OPT_PRUNEPATH ? */ -1); inex[ninex].wildcard = optval; ninex++; break; @@ -574,248 +936,710 @@ int main(int argc, char **argv) } } - if (mode == USAGE) { + if (nactions == 0) { usage(stderr); return 1; - } else if (mode == SCAN) { + } - fd = open(filename, O_RDWR | O_TRUNC | O_CREAT, S_IRWXU); - if (fd < 0) { - fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, - strerror(errno)); - return 1; - } + for (action = 0; action < nactions; action++) { + int mode = actions[action].mode; + + if (mode == SCAN || mode == SCANDUMP || mode == LOAD) { + const char *scandir = actions[action].arg; + + if (mode == LOAD) { + char *buf = fgetline(stdin); + unsigned newpathsep; + buf[strcspn(buf, "\r\n")] = '\0'; + if (1 != sscanf(buf, DUMPHDR "%x", + &newpathsep)) { + fprintf(stderr, "%s: header in dump file not recognised\n", + PNAME); + return 1; + } + pathsep = (char)newpathsep; + sfree(buf); + } - if (stat(scandir, &st) < 0) { - fprintf(stderr, "%s: %s: stat: %s\n", PNAME, scandir, - strerror(errno)); - return 1; - } - ctx->filesystem_dev = crossfs ? 0 : st.st_dev; + if (mode == SCAN || mode == LOAD) { + /* + * Prepare to write out the index file. + */ + fd = open(filename, O_RDWR | O_TRUNC | O_CREAT, + S_IRUSR | S_IWUSR); + if (fd < 0) { + fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, + strerror(errno)); + return 1; + } + if (fstat(fd, &st) < 0) { + perror(PNAME ": fstat"); + return 1; + } + ctx->datafile_dev = st.st_dev; + ctx->datafile_ino = st.st_ino; + ctx->straight_to_dump = 0; + } else { + ctx->datafile_dev = -1; + ctx->datafile_ino = -1; + ctx->straight_to_dump = 1; + } - if (fstat(fd, &st) < 0) { - perror("agedu: fstat"); - return 1; - } - ctx->datafile_dev = st.st_dev; - ctx->datafile_ino = st.st_ino; - ctx->inex = inex; - ctx->ninex = ninex; - ctx->crossfs = crossfs; - - ctx->last_output_update = time(NULL); - - /* progress==1 means report progress only if stderr is a tty */ - if (progress == 1) - progress = isatty(2) ? 2 : 0; - ctx->progress = progress; - { - struct winsize ws; - if (progress && ioctl(2, TIOCGWINSZ, &ws) == 0) - ctx->progwidth = ws.ws_col - 1; - else - ctx->progwidth = 79; - } + if (mode == SCAN || mode == SCANDUMP) { + if (stat(scandir, &st) < 0) { + fprintf(stderr, "%s: %s: stat: %s\n", PNAME, scandir, + strerror(errno)); + return 1; + } + ctx->filesystem_dev = crossfs ? 0 : st.st_dev; + } - /* - * Scan the directory tree, and write out the trie component - * of the data file. - */ - ctx->tb = triebuild_new(fd); - du(scandir, gotdata, ctx); - count = triebuild_finish(ctx->tb); - triebuild_free(ctx->tb); + ctx->inex = inex; + ctx->ninex = ninex; + ctx->crossfs = crossfs; + ctx->fakeatimes = fakediratimes; + ctx->usemtime = mtime; + + ctx->last_output_update = time(NULL); + + /* progress==1 means report progress only if stderr is a tty */ + if (progress == 1) + progress = isatty(2) ? 2 : 0; + ctx->progress = progress; + { + struct winsize ws; + if (progress && + ioctl(2, TIOCGWINSZ, &ws) == 0 && + ws.ws_col > 0) + ctx->progwidth = ws.ws_col - 1; + else + ctx->progwidth = 79; + } - if (ctx->progress) { - fprintf(stderr, "%-*s\r", ctx->progwidth, ""); - fflush(stderr); - } + if (mode == SCANDUMP) + printf(DUMPHDR "%02x\n", (unsigned char)pathsep); - /* - * Work out how much space the cumulative index trees will - * take; enlarge the file, and memory-map it. - */ - if (fstat(fd, &st) < 0) { - perror("agedu: fstat"); - return 1; - } + /* + * Scan the directory tree, and write out the trie component + * of the data file. + */ + if (mode != SCANDUMP) { + ctx->tb = triebuild_new(fd); + } + if (mode == LOAD) { + char *buf; + int line = 2; + while ((buf = fgetline(stdin)) != NULL) { + struct trie_file tf; + char *p, *q; + + buf[strcspn(buf, "\r\n")] = '\0'; + + p = buf; + q = p; + while (*p && *p != ' ') p++; + if (!*p) { + fprintf(stderr, "%s: dump file line %d: expected at least" + " three fields\n", PNAME, line); + return 1; + } + *p++ = '\0'; + tf.size = strtoull(q, NULL, 10); + q = p; + while (*p && *p != ' ') p++; + if (!*p) { + fprintf(stderr, "%s: dump file line %d: expected at least" + " three fields\n", PNAME, line); + return 1; + } + *p++ = '\0'; + tf.atime = strtoull(q, NULL, 10); + q = buf; + while (*p) { + int c = *p; + if (*p == '%') { + int i; + p++; + c = 0; + for (i = 0; i < 2; i++) { + c *= 16; + if (*p >= '0' && *p <= '9') + c += *p - '0'; + else if (*p >= 'A' && *p <= 'F') + c += *p - ('A' - 10); + else if (*p >= 'a' && *p <= 'f') + c += *p - ('a' - 10); + else { + fprintf(stderr, "%s: dump file line %d: unable" + " to parse hex escape\n", PNAME, line); + } + p++; + } + } else { + p++; + } + *q++ = c; + } + *q = '\0'; + triebuild_add(ctx->tb, buf, &tf); + sfree(buf); + line++; + } + } else { + du(scandir, gotdata, scan_error, ctx); + } + if (mode != SCANDUMP) { + size_t maxpathlen; + size_t delta; + char *buf, *prevbuf; - printf("Built pathname index, %d entries, %ju bytes\n", count, - (intmax_t)st.st_size); + count = triebuild_finish(ctx->tb); + triebuild_free(ctx->tb); - totalsize = index_compute_size(st.st_size, count); + if (ctx->progress) { + fprintf(stderr, "%-*s\r", ctx->progwidth, ""); + fflush(stderr); + } - if (lseek(fd, totalsize-1, SEEK_SET) < 0) { - perror("agedu: lseek"); - return 1; - } - if (write(fd, "\0", 1) < 1) { - perror("agedu: write"); - return 1; - } + /* + * Work out how much space the cumulative index trees + * will take; enlarge the file, and memory-map it. + */ + if (fstat(fd, &st) < 0) { + perror(PNAME ": fstat"); + return 1; + } - printf("Upper bound on index file size = %ju bytes\n", - (intmax_t)totalsize); + printf("Built pathname index, %d entries," + " %llu bytes of index\n", count, + (unsigned long long)st.st_size); - mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0); - if (!mappedfile) { - perror("agedu: mmap"); - return 1; - } + totalsize = index_initial_size(st.st_size, count); + totalsize += totalsize / 10; - ib = indexbuild_new(mappedfile, st.st_size, count); - tw = triewalk_new(mappedfile); - while ((tf = triewalk_next(tw, NULL)) != NULL) - indexbuild_add(ib, tf); - triewalk_free(tw); - realsize = indexbuild_realsize(ib); - indexbuild_free(ib); - - munmap(mappedfile, totalsize); - ftruncate(fd, realsize); - close(fd); - printf("Actual index file size = %ju bytes\n", (intmax_t)realsize); - } else if (mode == TEXT) { - time_t t; - struct tm tm; - int nunits; - char unit[2]; - size_t pathlen; + if (lseek(fd, totalsize-1, SEEK_SET) < 0) { + perror(PNAME ": lseek"); + return 1; + } + if (write(fd, "\0", 1) < 1) { + perror(PNAME ": write"); + return 1; + } - t = time(NULL); + mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0); + if (!mappedfile) { + perror(PNAME ": mmap"); + return 1; + } - if (2 != sscanf(minage, "%d%1[DdWwMmYy]", &nunits, unit)) { - fprintf(stderr, "%s: minimum age should be a number followed by" - " one of d,w,m,y\n", PNAME); - return 1; - } + if (fakediratimes) { + printf("Faking directory atimes\n"); + trie_fake_dir_atimes(mappedfile); + } - if (unit[0] == 'd') { - t -= 86400 * nunits; - } else if (unit[0] == 'w') { - t -= 86400 * 7 * nunits; - } else { - int ym; + printf("Building index\n"); + ib = indexbuild_new(mappedfile, st.st_size, count, &delta); + maxpathlen = trie_maxpathlen(mappedfile); + buf = snewn(maxpathlen, char); + prevbuf = snewn(maxpathlen, char); + tw = triewalk_new(mappedfile); + prevbuf[0] = '\0'; + tf = triewalk_next(tw, buf); + assert(tf); + while (1) { + int i; + + if (totalsize - indexbuild_realsize(ib) < delta) { + const void *oldfile = mappedfile; + ptrdiff_t diff; - tm = *localtime(&t); - ym = tm.tm_year * 12 + tm.tm_mon; + /* + * Unmap the file, grow it, and remap it. + */ + munmap(mappedfile, totalsize); - if (unit[0] == 'm') - ym -= nunits; - else - ym -= 12 * nunits; + totalsize += delta; + totalsize += totalsize / 10; - tm.tm_year = ym / 12; - tm.tm_mon = ym % 12; + if (lseek(fd, totalsize-1, SEEK_SET) < 0) { + perror(PNAME ": lseek"); + return 1; + } + if (write(fd, "\0", 1) < 1) { + perror(PNAME ": write"); + return 1; + } - t = mktime(&tm); - } + mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0); + if (!mappedfile) { + perror(PNAME ": mmap"); + return 1; + } - fd = open(filename, O_RDONLY); - if (fd < 0) { - fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, - strerror(errno)); - return 1; - } - if (fstat(fd, &st) < 0) { - perror("agedu: fstat"); - return 1; - } - totalsize = st.st_size; - mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); - if (!mappedfile) { - perror("agedu: mmap"); - return 1; - } + indexbuild_rebase(ib, mappedfile); + triewalk_rebase(tw, mappedfile); + diff = (const unsigned char *)mappedfile - + (const unsigned char *)oldfile; + if (tf) + tf = (const struct trie_file *) + (((const unsigned char *)tf) + diff); + } - /* - * Trim trailing slash, just in case. - */ - pathlen = strlen(querydir); - if (pathlen > 0 && querydir[pathlen-1] == '/') - querydir[--pathlen] = '\0'; - - text_query(mappedfile, querydir, t, 1); - } else if (mode == HTML) { - size_t pathlen; - unsigned long xi; - char *html; - - fd = open(filename, O_RDONLY); - if (fd < 0) { - fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, - strerror(errno)); - return 1; - } - if (fstat(fd, &st) < 0) { - perror("agedu: fstat"); - return 1; - } - totalsize = st.st_size; - mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); - if (!mappedfile) { - perror("agedu: mmap"); - return 1; - } + /* + * Get the next file from the index. So we are + * currently holding, and have not yet + * indexed, prevtf (with pathname prevbuf) and + * tf (with pathname buf). + */ + prevtf = tf; + memcpy(prevbuf, buf, maxpathlen); + tf = triewalk_next(tw, buf); - /* - * Trim trailing slash, just in case. - */ - pathlen = strlen(querydir); - if (pathlen > 0 && querydir[pathlen-1] == '/') - querydir[--pathlen] = '\0'; - - xi = trie_before(mappedfile, querydir); - html = html_query(mappedfile, xi, NULL); - fputs(html, stdout); - } else if (mode == DUMP) { - size_t maxpathlen; - char *buf; - - fd = open(filename, O_RDONLY); - if (fd < 0) { - fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, - strerror(errno)); - return 1; - } - if (fstat(fd, &st) < 0) { - perror("agedu: fstat"); - return 1; - } - totalsize = st.st_size; - mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); - if (!mappedfile) { - perror("agedu: mmap"); - return 1; - } + if (!tf) + buf[0] = '\0'; - maxpathlen = trie_maxpathlen(mappedfile); - buf = snewn(maxpathlen, char); + /* + * Find the first differing character position + * between our two pathnames. + */ + for (i = 0; prevbuf[i] && prevbuf[i] == buf[i]; i++); - tw = triewalk_new(mappedfile); - while ((tf = triewalk_next(tw, buf)) != NULL) { - printf("%s: %llu %llu\n", buf, tf->blocks, tf->atime); - } - triewalk_free(tw); - } else if (mode == HTTPD) { - fd = open(filename, O_RDONLY); - if (fd < 0) { - fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, - strerror(errno)); - return 1; - } - if (fstat(fd, &st) < 0) { - perror("agedu: fstat"); - return 1; - } - totalsize = st.st_size; - mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); - if (!mappedfile) { - perror("agedu: mmap"); - return 1; - } + /* + * If prevbuf was a directory name and buf is + * something inside that directory, then + * trie_before() will be called on prevbuf + * itself. Hence we must drop a tag before it, + * so that the resulting index is usable. + */ + if ((!prevbuf[i] && (buf[i] == pathsep || + (i > 0 && buf[i-1] == pathsep)))) + indexbuild_tag(ib); + + /* + * Add prevtf to the index. + */ + indexbuild_add(ib, prevtf); + + if (!tf) { + /* + * Drop an unconditional final tag, and + * get out of this loop. + */ + indexbuild_tag(ib); + break; + } + + /* + * If prevbuf was a filename inside some + * directory which buf is outside, then + * trie_before() will be called on some + * pathname either equal to buf or epsilon + * less than it. Either way, we're going to + * need to drop a tag after prevtf. + */ + if (strchr(prevbuf+i, pathsep) || !tf) + indexbuild_tag(ib); + } + + triewalk_free(tw); + realsize = indexbuild_realsize(ib); + indexbuild_free(ib); + + munmap(mappedfile, totalsize); + if (ftruncate(fd, realsize) < 0) + fatal("%s: truncate: %s\n", filename, strerror(errno)); + close(fd); + printf("Final index file size = %llu bytes\n", + (unsigned long long)realsize); + } + } else if (mode == TEXT) { + char *querydir = actions[action].arg; + size_t pathlen; + + fd = open(filename, O_RDONLY); + if (fd < 0) { + fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, + strerror(errno)); + return 1; + } + if (fstat(fd, &st) < 0) { + perror(PNAME ": fstat"); + return 1; + } + totalsize = st.st_size; + mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); + if (!mappedfile) { + perror(PNAME ": mmap"); + return 1; + } + if (!trie_check_magic(mappedfile)) { + fprintf(stderr, "%s: %s: magic numbers did not match\n" + "%s: check that the index was built by this version of agedu on this platform\n", PNAME, filename, PNAME); + return 1; + } + pathsep = trie_pathsep(mappedfile); + + /* + * Trim trailing slash, just in case. + */ + pathlen = strlen(querydir); + if (pathlen > 0 && querydir[pathlen-1] == pathsep) + querydir[--pathlen] = '\0'; + + if (!gotdepth) + depth = 1; /* default for text mode */ + if (outfile != NULL) { + FILE *fp = fopen(outfile, "w"); + if (!fp) { + fprintf(stderr, "%s: %s: open: %s\n", PNAME, + outfile, strerror(errno)); + return 1; + } + text_query(mappedfile, querydir, textcutoff, showfiles, + depth, fp); + fclose(fp); + } else { + text_query(mappedfile, querydir, textcutoff, showfiles, + depth, stdout); + } + + munmap(mappedfile, totalsize); + } else if (mode == HTML) { + char *querydir = actions[action].arg; + size_t pathlen, maxpathlen; + char *pathbuf; + struct html_config cfg; + unsigned long xi; + char *html; + + fd = open(filename, O_RDONLY); + if (fd < 0) { + fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, + strerror(errno)); + if (!querydir) { + printf("Status: 500\nContent-type: text/html\n\n" + "" + "500 Internal Server Error" + "" + "

500 Internal Server Error

" + "

agedu suffered an internal error." + "\n"); + return 0; + } + return 1; + } + if (fstat(fd, &st) < 0) { + fprintf(stderr, "%s: %s: fstat: %s\n", PNAME, filename, + strerror(errno)); + if (!querydir) { + printf("Status: 500\nContent-type: text/html\n\n" + "" + "500 Internal Server Error" + "" + "

500 Internal Server Error

" + "

agedu suffered an internal error." + "\n"); + return 0; + } + return 1; + } + totalsize = st.st_size; + mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); + if (!mappedfile) { + fprintf(stderr, "%s: %s: mmap: %s\n", PNAME, filename, + strerror(errno)); + if (!querydir) { + printf("Status: 500\nContent-type: text/html\n\n" + "" + "500 Internal Server Error" + "" + "

500 Internal Server Error

" + "

agedu suffered an internal error." + "\n"); + return 0; + } + return 1; + } + if (!trie_check_magic(mappedfile)) { + fprintf(stderr, "%s: %s: magic numbers did not match\n" + "%s: check that the index was built by this version of agedu on this platform\n", PNAME, filename, PNAME); + if (!querydir) { + printf("Status: 500\nContent-type: text/html\n\n" + "" + "500 Internal Server Error" + "" + "

500 Internal Server Error

" + "

agedu suffered an internal error." + "\n"); + return 0; + } + return 1; + } + pathsep = trie_pathsep(mappedfile); + + maxpathlen = trie_maxpathlen(mappedfile); + pathbuf = snewn(maxpathlen, char); + + if (!querydir || !gotdepth) { + /* + * Single output file. + */ + if (!querydir) { + cfg.uriformat = "/%|/%p/%|%|/%p"; + } else { + cfg.uriformat = NULL; + } + cfg.autoage = htmlautoagerange; + cfg.oldest = htmloldest; + cfg.newest = htmlnewest; + cfg.showfiles = showfiles; + } else { + cfg.uriformat = "/index.html%|/%/p.html"; + cfg.fileformat = "/index.html%|/%/p.html"; + cfg.autoage = htmlautoagerange; + cfg.oldest = htmloldest; + cfg.newest = htmlnewest; + cfg.showfiles = showfiles; + } + cfg.html_title = html_title; + + if (!querydir) { + /* + * If we're run in --cgi mode, read PATH_INFO to get + * a numeric pathname index. + */ + char *path_info = getenv("PATH_INFO"); + + if (!path_info) + path_info = ""; + + /* + * Parse the path. + */ + if (!html_parse_path(mappedfile, path_info, &cfg, &xi)) { + printf("Status: 404\nContent-type: text/html\n\n" + "" + "404 Not Found" + "" + "

400 Not Found

" + "

Invalid agedu pathname." + "\n"); + return 0; + } + + /* + * If the path was parseable but not canonically + * expressed, return a redirect to the canonical + * version. + */ + char *canonpath = html_format_path(mappedfile, &cfg, xi); + if (strcmp(canonpath, path_info)) { + char *servername = getenv("SERVER_NAME"); + char *scriptname = getenv("SCRIPT_NAME"); + if (!servername || !scriptname) { + if (servername) + fprintf(stderr, "%s: SCRIPT_NAME unset\n", PNAME); + else if (scriptname) + fprintf(stderr, "%s: SCRIPT_NAME unset\n", PNAME); + else + fprintf(stderr, "%s: SERVER_NAME and " + "SCRIPT_NAME both unset\n", PNAME); + printf("Status: 500\nContent-type: text/html\n\n" + "" + "500 Internal Server Error" + "" + "

500 Internal Server Error

" + "

agedu suffered an internal " + "error." + "\n"); + return 0; + } + printf("Status: 301\n" + "Location: http://%s/%s%s\n" + "Content-type: text/html\n\n" + "" + "301 Moved" + "" + "

301 Moved

" + "

Moved." + "\n", + servername, scriptname, canonpath); + return 0; + } - run_httpd(mappedfile, auth); + } else { + /* + * In ordinary --html mode, process a query + * directory passed in on the command line. + */ + + /* + * Trim trailing slash, just in case. + */ + pathlen = strlen(querydir); + if (pathlen > 0 && querydir[pathlen-1] == pathsep) + querydir[--pathlen] = '\0'; + + xi = trie_before(mappedfile, querydir); + if (xi >= trie_count(mappedfile) || + (trie_getpath(mappedfile, xi, pathbuf), + strcmp(pathbuf, querydir))) { + fprintf(stderr, "%s: pathname '%s' does not exist in index\n" + "%*s(check it is spelled exactly as it is in the " + "index, including\n%*sany leading './')\n", + PNAME, querydir, + (int)(1+sizeof(PNAME)), "", + (int)(1+sizeof(PNAME)), ""); + return 1; + } else if (!index_has_root(mappedfile, xi)) { + fprintf(stderr, "%s: pathname '%s' is" + " a file, not a directory\n", PNAME, querydir); + return 1; + } + } + + if (!querydir || !gotdepth) { + /* + * Single output file. + */ + html = html_query(mappedfile, xi, &cfg, 1); + if (querydir && outfile != NULL) { + FILE *fp = fopen(outfile, "w"); + if (!fp) { + fprintf(stderr, "%s: %s: open: %s\n", PNAME, + outfile, strerror(errno)); + return 1; + } else if (fputs(html, fp) < 0) { + fprintf(stderr, "%s: %s: write: %s\n", PNAME, + outfile, strerror(errno)); + fclose(fp); + return 1; + } else if (fclose(fp) < 0) { + fprintf(stderr, "%s: %s: fclose: %s\n", PNAME, + outfile, strerror(errno)); + return 1; + } + } else { + if (!querydir) { + printf("Content-type: text/html\n\n"); + } + fputs(html, stdout); + } + } else { + /* + * Multiple output files. + */ + int dirlen = outfile ? 2+strlen(outfile) : 3; + char prefix[dirlen]; + if (outfile) { + if (mkdir(outfile, 0777) < 0 && errno != EEXIST) { + fprintf(stderr, "%s: %s: mkdir: %s\n", PNAME, + outfile, strerror(errno)); + return 1; + } + snprintf(prefix, dirlen, "%s/", outfile); + } else + snprintf(prefix, dirlen, "./"); + + unsigned long xi2; + /* + * pathbuf is only set up in the plain-HTML case and + * not in the CGI case; but that's OK, because the + * CGI case can't come to this branch of the if + * anyway. + */ + make_successor(pathbuf); + xi2 = trie_before(mappedfile, pathbuf); + + if (html_dump(mappedfile, xi, xi2, depth, &cfg, prefix)) + return 1; + } + + munmap(mappedfile, totalsize); + sfree(pathbuf); + } else if (mode == DUMP) { + size_t maxpathlen; + char *buf; + + fd = open(filename, O_RDONLY); + if (fd < 0) { + fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, + strerror(errno)); + return 1; + } + if (fstat(fd, &st) < 0) { + perror(PNAME ": fstat"); + return 1; + } + totalsize = st.st_size; + mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); + if (!mappedfile) { + perror(PNAME ": mmap"); + return 1; + } + if (!trie_check_magic(mappedfile)) { + fprintf(stderr, "%s: %s: magic numbers did not match\n" + "%s: check that the index was built by this version of agedu on this platform\n", PNAME, filename, PNAME); + return 1; + } + pathsep = trie_pathsep(mappedfile); + + maxpathlen = trie_maxpathlen(mappedfile); + buf = snewn(maxpathlen, char); + + printf(DUMPHDR "%02x\n", (unsigned char)pathsep); + tw = triewalk_new(mappedfile); + while ((tf = triewalk_next(tw, buf)) != NULL) + dump_line(buf, tf); + triewalk_free(tw); + + munmap(mappedfile, totalsize); + } else if (mode == HTTPD) { + struct html_config pcfg; + struct httpd_config dcfg; + + fd = open(filename, O_RDONLY); + if (fd < 0) { + fprintf(stderr, "%s: %s: open: %s\n", PNAME, filename, + strerror(errno)); + return 1; + } + if (fstat(fd, &st) < 0) { + perror(PNAME ": fstat"); + return 1; + } + totalsize = st.st_size; + mappedfile = mmap(NULL, totalsize, PROT_READ, MAP_SHARED, fd, 0); + if (!mappedfile) { + perror(PNAME ": mmap"); + return 1; + } + if (!trie_check_magic(mappedfile)) { + fprintf(stderr, "%s: %s: magic numbers did not match\n" + "%s: check that the index was built by this version of agedu on this platform\n", PNAME, filename, PNAME); + return 1; + } + pathsep = trie_pathsep(mappedfile); + + dcfg.address = httpserveraddr; + dcfg.port = httpserverport; + dcfg.closeoneof = closeoneof; + dcfg.basicauthdata = httpauthdata; + pcfg.uriformat = "/%|/%p/%|%|/%p"; + pcfg.autoage = htmlautoagerange; + pcfg.oldest = htmloldest; + pcfg.newest = htmlnewest; + pcfg.showfiles = showfiles; + pcfg.html_title = html_title; + run_httpd(mappedfile, auth, &dcfg, &pcfg); + munmap(mappedfile, totalsize); + } else if (mode == REMOVE) { + if (remove(filename) < 0) { + fprintf(stderr, "%s: %s: remove: %s\n", PNAME, filename, + strerror(errno)); + return 1; + } + } } return 0;