X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/agedu/blobdiff_plain/f2e528936aac5fb2f68693ce8cda348c2a7c1a91..9c6e61f23dd65086378606be16d3740d909477d4:/agedu.c diff --git a/agedu.c b/agedu.c index 607d425..c3ac6a7 100644 --- a/agedu.c +++ b/agedu.c @@ -20,16 +20,24 @@ #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, ...) { @@ -54,12 +62,27 @@ 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 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; + printf("%llu %llu ", tf->size, tf->atime); + for (p = pathname; *p; p++) { + if (*p >= ' ' && *p < 127 && *p != '%') + putchar(*p); + else + printf("%%%02x", (unsigned char)*p); + } + putchar('\n'); +} + +static int gotdata(void *vctx, const char *pathname, const STRUCT_STAT *st) { struct ctx *ctx = (struct ctx *)vctx; struct trie_file file; @@ -79,14 +102,17 @@ static int gotdata(void *vctx, const char *pathname, const struct stat64 *st) if (!ctx->crossfs && st->st_dev != ctx->filesystem_dev) return 0; - file.blocks = st->st_blocks; - file.atime = st->st_atime; + file.size = (unsigned long long)512 * st->st_blocks; + if (ctx->fakeatimes && S_ISDIR(st->st_mode)) + file.atime = st->st_mtime; + else + file.atime = st->st_atime; /* * Filter based on wildcards. */ include = 1; - filename = strrchr(pathname, '/'); + filename = strrchr(pathname, pathsep); if (!filename) filename = pathname; else @@ -109,19 +135,22 @@ static int gotdata(void *vctx, const char *pathname, const struct stat64 *st) if (!S_ISDIR(st->st_mode)) return 0; /* just ignore */ else - file.blocks = 0; + file.size = 0; } - 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; @@ -143,10 +172,14 @@ 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 (xi2 - xi1 == 1) + return; /* file, or empty dir => no display */ + /* * Now do the lookups in the age index. */ @@ -156,9 +189,6 @@ static void text_query(const void *mappedfile, const char *querydir, if (s1 == s2) return; /* no space taken up => no display */ - /* Display in units of 2 512-byte blocks = 1Kb */ - printf("%-11llu %s\n", (s2 - s1) / 2, querydir); - if (depth > 0) { /* * Now scan for first-level subdirectories and report @@ -168,10 +198,13 @@ static void text_query(const void *mappedfile, const char *querydir, while (xi1 < xi2) { trie_getpath(mappedfile, xi1, pathbuf); text_query(mappedfile, pathbuf, t, depth-1); - strcat(pathbuf, "\001"); + make_successor(pathbuf); xi1 = trie_before(mappedfile, pathbuf); } } + + /* Display in units of 1Kb */ + printf("%-11llu %s\n", (s2 - s1) / 1024, querydir); } /* @@ -229,53 +262,76 @@ 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") \ 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(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(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(MINAGE) SHORT(a) LONG(age) LONG(min_age) LONG(minimum_age) \ - HELPARG("age") HELPOPT("[--text] include only files older than this") \ + 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(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") \ VAL(AGERANGE) SHORT(r) LONG(age_range) LONG(range) LONG(ages) \ - HELPARG("age[-age]") HELPOPT("[--html,--web] set limits of colour coding") \ + HELPARG("age[-age]") HELPOPT("[--web,--html] set limits of colour coding") \ + 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(TQDEPTH) SHORT(d) LONG(depth) LONG(max_depth) LONG(maximum_depth) \ + HELPARG("levels") HELPOPT("[--text] 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") \ @@ -400,19 +456,27 @@ int main(int argc, char **argv) triewalk *tw; indexbuild *ib; const struct trie_file *tf; - char *filename = "agedu.dat"; - char *scandir = NULL; - char *querydir = NULL; + char *filename = PNAME ".dat"; int doing_opts = 1; - enum { USAGE, TEXT, HTML, SCAN, DUMP, HTTPD } mode = USAGE; + 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 = NULL; + int httpserverport = 0; + const char *httpauthdata = NULL; 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 tqdepth = 1; + int fakediratimes = 1; #ifdef DEBUG_MAD_OPTION_PARSING_MACROS { @@ -552,28 +616,95 @@ 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; + } 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; + if (nactions >= actionsize) { + actionsize = nactions * 3 / 2 + 16; + actions = sresize(actions, actionsize, struct action); + } + actions[nactions].mode = HTML; + actions[nactions].arg = optval; + 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; @@ -590,9 +721,18 @@ 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_DATAFILE: filename = optval; break; + case OPT_TQDEPTH: + tqdepth = atoi(optval); + break; case OPT_MINAGE: textcutoff = parse_age(now, optval); break; @@ -608,6 +748,21 @@ int main(int argc, char **argv) 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++ = '\0'; + httpserveraddr = optval; + httpserverport = atoi(port); + } + break; case OPT_AUTH: if (!strcmp(optval, "magic")) auth = HTTPD_AUTH_MAGIC; @@ -619,7 +774,7 @@ int main(int argc, char **argv) auth = HTTPD_AUTH_MAGIC | HTTPD_AUTH_BASIC; else if (!strcmp(optval, "help") || !strcmp(optval, "list")) { - printf("agedu: supported HTTP authentication types" + printf(PNAME ": supported HTTP authentication types" " are:\n" " magic use Linux /proc/net/tcp to" " determine owner of peer socket\n" @@ -638,6 +793,51 @@ 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: @@ -669,226 +869,348 @@ 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; - } + } - 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; + 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 (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 == LOAD) { + /* + * Prepare to write out the index file. + */ + 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; + } + 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; + } - /* - * 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); + 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; + } - if (ctx->progress) { - fprintf(stderr, "%-*s\r", ctx->progwidth, ""); - fflush(stderr); - } + ctx->inex = inex; + ctx->ninex = ninex; + ctx->crossfs = crossfs; + ctx->fakeatimes = fakediratimes; + + 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; + } - /* - * 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; - } + if (mode == SCANDUMP) + printf(DUMPHDR "%02x\n", (unsigned char)pathsep); - printf("Built pathname index, %d entries, %ju bytes\n", count, - (intmax_t)st.st_size); + /* + * 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++; + } + } + *q++ = c; + p++; + } + *q = '\0'; + triebuild_add(ctx->tb, buf, &tf); + sfree(buf); + line++; + } + } else { + du(scandir, gotdata, ctx); + } + if (mode != SCANDUMP) { + 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, %ju bytes\n", count, + (intmax_t)st.st_size); - mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0); - if (!mappedfile) { - perror("agedu: mmap"); - return 1; - } + totalsize = index_compute_size(st.st_size, count); - 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) { - 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("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 (lseek(fd, totalsize-1, SEEK_SET) < 0) { + perror(PNAME ": lseek"); + return 1; + } + if (write(fd, "\0", 1) < 1) { + perror(PNAME ": write"); + return 1; + } - /* - * Trim trailing slash, just in case. - */ - pathlen = strlen(querydir); - if (pathlen > 0 && querydir[pathlen-1] == '/') - querydir[--pathlen] = '\0'; - - text_query(mappedfile, querydir, textcutoff, 1); - } else if (mode == HTML) { - size_t pathlen; - 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)); - 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; - } + printf("Upper bound on index file size = %ju bytes\n", + (intmax_t)totalsize); - /* - * Trim trailing slash, just in case. - */ - pathlen = strlen(querydir); - if (pathlen > 0 && querydir[pathlen-1] == '/') - querydir[--pathlen] = '\0'; - - xi = trie_before(mappedfile, querydir); - cfg.format = NULL; - cfg.autoage = htmlautoagerange; - cfg.oldest = htmloldest; - cfg.newest = htmlnewest; - html = html_query(mappedfile, xi, &cfg); - 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; - } + mappedfile = mmap(NULL, totalsize, PROT_READ|PROT_WRITE,MAP_SHARED, fd, 0); + if (!mappedfile) { + perror(PNAME ": mmap"); + return 1; + } - maxpathlen = trie_maxpathlen(mappedfile); - buf = snewn(maxpathlen, char); + if (fakediratimes) { + printf("Faking directory atimes\n"); + trie_fake_dir_atimes(mappedfile); + } - 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) { - struct html_config cfg; - - 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; + printf("Building index\n"); + 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) { + 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; + } + pathsep = trie_pathsep(mappedfile); + + /* + * Trim trailing slash, just in case. + */ + pathlen = strlen(querydir); + if (pathlen > 0 && querydir[pathlen-1] == pathsep) + querydir[--pathlen] = '\0'; + + text_query(mappedfile, querydir, textcutoff, tqdepth); + } else if (mode == HTML) { + char *querydir = actions[action].arg; + size_t pathlen; + 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)); + 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; + } + pathsep = trie_pathsep(mappedfile); + + /* + * Trim trailing slash, just in case. + */ + pathlen = strlen(querydir); + if (pathlen > 0 && querydir[pathlen-1] == pathsep) + querydir[--pathlen] = '\0'; + + xi = trie_before(mappedfile, querydir); + cfg.format = NULL; + cfg.autoage = htmlautoagerange; + cfg.oldest = htmloldest; + cfg.newest = htmlnewest; + html = html_query(mappedfile, xi, &cfg); + 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(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; + } + 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); + } 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; + } + pathsep = trie_pathsep(mappedfile); + + dcfg.address = httpserveraddr; + dcfg.port = httpserverport; + dcfg.basicauthdata = httpauthdata; + pcfg.format = NULL; + pcfg.autoage = htmlautoagerange; + pcfg.oldest = htmloldest; + pcfg.newest = htmlnewest; + run_httpd(mappedfile, auth, &dcfg, &pcfg); + } else if (mode == REMOVE) { + if (remove(filename) < 0) { + fprintf(stderr, "%s: %s: remove: %s\n", PNAME, filename, + strerror(errno)); + return 1; + } } - - cfg.format = NULL; - cfg.autoage = htmlautoagerange; - cfg.oldest = htmloldest; - cfg.newest = htmlnewest; - run_httpd(mappedfile, auth, &cfg); } return 0;