=================================================================== RCS file: /cvs/mandoc/mandocdb.c,v retrieving revision 1.131 retrieving revision 1.139 diff -u -p -r1.131 -r1.139 --- mandoc/mandocdb.c 2014/04/04 18:23:30 1.131 +++ mandoc/mandocdb.c 2014/04/19 02:30:19 1.139 @@ -1,4 +1,4 @@ -/* $Id: mandocdb.c,v 1.131 2014/04/04 18:23:30 schwarze Exp $ */ +/* $Id: mandocdb.c,v 1.139 2014/04/19 02:30:19 schwarze Exp $ */ /* * Copyright (c) 2011, 2012 Kristaps Dzonsons * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze @@ -55,22 +55,22 @@ extern const char *const mansearch_keynames[]; #define SQL_EXEC(_v) \ if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \ - fprintf(stderr, "%s\n", sqlite3_errmsg(db)) + say("", "%s: %s", (_v), sqlite3_errmsg(db)) #define SQL_BIND_TEXT(_s, _i, _v) \ if (SQLITE_OK != sqlite3_bind_text \ ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \ - fprintf(stderr, "%s\n", sqlite3_errmsg(db)) + say(mlink->file, "%s", sqlite3_errmsg(db)) #define SQL_BIND_INT(_s, _i, _v) \ if (SQLITE_OK != sqlite3_bind_int \ ((_s), (_i)++, (_v))) \ - fprintf(stderr, "%s\n", sqlite3_errmsg(db)) + say(mlink->file, "%s", sqlite3_errmsg(db)) #define SQL_BIND_INT64(_s, _i, _v) \ if (SQLITE_OK != sqlite3_bind_int64 \ ((_s), (_i)++, (_v))) \ - fprintf(stderr, "%s\n", sqlite3_errmsg(db)) + say(mlink->file, "%s", sqlite3_errmsg(db)) #define SQL_STEP(_s) \ if (SQLITE_DONE != sqlite3_step((_s))) \ - fprintf(stderr, "%s\n", sqlite3_errmsg(db)) + say(mlink->file, "%s", sqlite3_errmsg(db)) enum op { OP_DEFAULT = 0, /* new dbs from dir list or default config */ @@ -100,7 +100,7 @@ struct inodev { struct mpage { struct inodev inodev; /* used for hashing routine */ - int64_t recno; /* id in mpages SQL table */ + int64_t pageid; /* pageid in mpages SQL table */ enum form form; /* format from file content */ char *sec; /* section from file content */ char *arch; /* architecture from file content */ @@ -126,6 +126,7 @@ enum stmt { STMT_DELETE_PAGE = 0, /* delete mpage */ STMT_INSERT_PAGE, /* insert mpage */ STMT_INSERT_LINK, /* insert mlink */ + STMT_INSERT_NAME, /* insert name */ STMT_INSERT_KEY, /* insert parsed key */ STMT__MAX }; @@ -188,9 +189,11 @@ static enum op op; /* operational mode */ static char basedir[PATH_MAX]; /* current base directory */ static struct ohash mpages; /* table of distinct manual pages */ static struct ohash mlinks; /* table of directory entries */ +static struct ohash names; /* table of all names */ static struct ohash strings; /* table of all strings */ static sqlite3 *db = NULL; /* current database */ static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */ +static uint64_t name_mask; static const struct mdoc_handler mdocs[MDOC_MAX] = { { NULL, 0 }, /* Ap */ @@ -224,7 +227,7 @@ static const struct mdoc_handler mdocs[MDOC_MAX] = { { NULL, TYPE_Ic }, /* Ic */ { NULL, TYPE_In }, /* In */ { NULL, TYPE_Li }, /* Li */ - { parse_mdoc_Nd, TYPE_Nd }, /* Nd */ + { parse_mdoc_Nd, 0 }, /* Nd */ { parse_mdoc_Nm, 0 }, /* Nm */ { NULL, 0 }, /* Op */ { NULL, 0 }, /* Ot */ @@ -439,15 +442,25 @@ main(int argc, char *argv[]) */ if (0 == set_basedir(path_arg)) goto out; - for (i = 0; i < argc; i++) - filescan(argv[i]); - if (0 == dbopen(1)) - goto out; - if (OP_TEST != op) - dbprune(); + if (dbopen(1)) { + for (i = 0; i < argc; i++) + filescan(argv[i]); + if (OP_TEST != op) + dbprune(); + } else { + /* + * Database missing or corrupt. + * Recreate from scratch. + */ + op = OP_DEFAULT; + if (0 == treescan()) + goto out; + if (0 == dbopen(0)) + goto out; + } if (OP_DELETE != op) mpages_merge(mc, mp); - dbclose(1); + dbclose(OP_DEFAULT == op ? 0 : 1); } else { /* * If we have arguments, use them as our manpaths. @@ -546,6 +559,7 @@ usage: static int treescan(void) { + char buf[PATH_MAX]; FTS *f; FTSENT *ff; struct mlink *mlink; @@ -557,11 +571,8 @@ treescan(void) argv[0] = "."; argv[1] = (char *)NULL; - /* - * Walk through all components under the directory, using the - * logical descent of files. - */ - f = fts_open((char * const *)argv, FTS_LOGICAL, NULL); + f = fts_open((char * const *)argv, + FTS_PHYSICAL | FTS_NOCHDIR, NULL); if (NULL == f) { exitcode = (int)MANDOCLEVEL_SYSERR; say("", "&fts_open"); @@ -573,11 +584,36 @@ treescan(void) while (NULL != (ff = fts_read(f))) { path = ff->fts_path + 2; + switch (ff->fts_info) { + /* + * Symbolic links require various sanity checks, + * then get handled just like regular files. + */ + case (FTS_SL): + if (NULL == realpath(path, buf)) { + if (warnings) + say(path, "&realpath"); + continue; + } + if (strstr(buf, basedir) != buf) { + if (warnings) say("", + "%s: outside base directory", buf); + continue; + } + /* Use logical inode to avoid mpages dupe. */ + if (-1 == stat(path, ff->fts_statp)) { + if (warnings) + say(path, "&stat"); + continue; + } + /* FALLTHROUGH */ + + /* * If we're a regular file, add an mlink by using the * stored directory data and handling the filename. */ - if (FTS_F == ff->fts_info) { + case (FTS_F): if (0 == strcmp(path, MANDOC_DB)) continue; if ( ! use_all && ff->fts_level < 2) { @@ -633,8 +669,13 @@ treescan(void) mlink->gzip = gzip; mlink_add(mlink, ff->fts_statp); continue; - } else if (FTS_D != ff->fts_info && - FTS_DP != ff->fts_info) { + + case (FTS_D): + /* FALLTHROUGH */ + case (FTS_DP): + break; + + default: if (warnings) say(path, "Not a regular file"); continue; @@ -723,6 +764,27 @@ filescan(const char *file) if (0 == strncmp(file, "./", 2)) file += 2; + /* + * We have to do lstat(2) before realpath(3) loses + * the information whether this is a symbolic link. + * We need to know that because for symbolic links, + * we want to use the orginal file name, while for + * regular files, we want to use the real path. + */ + if (-1 == lstat(file, &st)) { + exitcode = (int)MANDOCLEVEL_BADARG; + say(file, "&lstat"); + return; + } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) { + exitcode = (int)MANDOCLEVEL_BADARG; + say(file, "Not a regular file"); + return; + } + + /* + * We have to resolve the file name to the real path + * in any case for the base directory check. + */ if (NULL == realpath(file, buf)) { exitcode = (int)MANDOCLEVEL_BADARG; say(file, "&realpath"); @@ -739,14 +801,24 @@ filescan(const char *file) return; } - if (-1 == stat(buf, &st)) { - exitcode = (int)MANDOCLEVEL_BADARG; - say(file, "&stat"); - return; - } else if ( ! (S_IFREG & st.st_mode)) { - exitcode = (int)MANDOCLEVEL_BADARG; - say(file, "Not a regular file"); - return; + /* + * Now we are sure the file is inside our tree. + * If it is a symbolic link, ignore the real path + * and use the original name. + * This implies passing stuff like "cat1/../man1/foo.1" + * on the command line won't work. So don't do that. + * Note the stat(2) can still fail if the link target + * doesn't exist. + */ + if (S_IFLNK & st.st_mode) { + if (-1 == stat(buf, &st)) { + exitcode = (int)MANDOCLEVEL_BADARG; + say(file, "&stat"); + return; + } + strlcpy(buf, file, sizeof(buf)); + start = strstr(buf, basedir) == buf ? + buf + strlen(basedir) + 1 : buf; } mlink = mandoc_calloc(1, sizeof(struct mlink)); @@ -963,7 +1035,7 @@ mlink_check(struct mpage *mpage, struct mlink *mlink) /* * XXX - * parse_cat() doesn't set TYPE_Nm and TYPE_NAME yet. + * parse_cat() doesn't set NAME_TITLE yet. */ if (FORM_CAT == mpage->form) @@ -974,10 +1046,10 @@ mlink_check(struct mpage *mpage, struct mlink *mlink) * appears as a name in the NAME section. */ - slot = ohash_qlookup(&strings, mlink->name); - str = ohash_find(&strings, slot); + slot = ohash_qlookup(&names, mlink->name); + str = ohash_find(&names, slot); assert(NULL != str); - if ( ! (TYPE_NAME & str->mask)) + if ( ! (NAME_TITLE & str->mask)) say(mlink->file, "Name missing in NAME section"); } @@ -1021,6 +1093,8 @@ mpages_merge(struct mchars *mc, struct mparse *mp) continue; } + name_mask = NAME_MASK; + ohash_init(&names, 4, &str_info); ohash_init(&strings, 6, &str_info); mparse_reset(mp); mdoc = NULL; @@ -1094,7 +1168,7 @@ mpages_merge(struct mchars *mc, struct mparse *mp) * to the target. */ - if (mpage_dest->recno) + if (mpage_dest->pageid) dbadd_mlink(mlink); if (NULL == mlink->next) @@ -1146,20 +1220,20 @@ mpages_merge(struct mchars *mc, struct mparse *mp) putkey(mpage, mlink->fsec, TYPE_sec); putkey(mpage, '\0' == *mlink->arch ? any : mlink->arch, TYPE_arch); - putkey(mpage, mlink->name, TYPE_Nm); + putkey(mpage, mlink->name, NAME_FILE); } + assert(NULL == mpage->desc); if (NULL != mdoc) { if (NULL != (cp = mdoc_meta(mdoc)->name)) - putkey(mpage, cp, TYPE_Nm); - assert(NULL == mpage->desc); + putkey(mpage, cp, NAME_HEAD); parse_mdoc(mpage, mdoc_node(mdoc)); - putkey(mpage, NULL != mpage->desc ? - mpage->desc : mpage->mlinks->name, TYPE_Nd); } else if (NULL != man) parse_man(mpage, man_node(man)); else parse_cat(mpage, fd[0]); + if (NULL == mpage->desc) + mpage->desc = mandoc_strdup(mpage->mlinks->name); if (warnings && !use_all) for (mlink = mpage->mlinks; mlink; @@ -1186,6 +1260,7 @@ nextpage: } } ohash_delete(&strings); + ohash_delete(&names); mpage = ohash_next(&mpages, &pslot); } @@ -1198,24 +1273,23 @@ names_check(void) { sqlite3_stmt *stmt; const char *name, *sec, *arch, *key; - size_t i; int irc; sqlite3_prepare_v2(db, "SELECT name, sec, arch, key FROM (" - "SELECT key, pageid FROM keys " + "SELECT name AS key, pageid FROM names " "WHERE bits & ? AND NOT EXISTS (" "SELECT pageid FROM mlinks " - "WHERE mlinks.pageid == keys.pageid " - "AND mlinks.name == keys.key" + "WHERE mlinks.pageid == names.pageid " + "AND mlinks.name == names.name" ")" ") JOIN (" "SELECT * FROM mlinks GROUP BY pageid" ") USING (pageid);", -1, &stmt, NULL); - i = 1; - SQL_BIND_INT64(stmt, i, TYPE_NAME); + if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE)) + say("", "%s", sqlite3_errmsg(db)); while (SQLITE_ROW == (irc = sqlite3_step(stmt))) { name = sqlite3_column_text(stmt, 0); @@ -1296,9 +1370,6 @@ parse_cat(struct mpage *mpage, int fd) if (warnings) say(mpage->mlinks->file, "Cannot find NAME section"); - assert(NULL == mpage->desc); - mpage->desc = mandoc_strdup(mpage->mlinks->name); - putkey(mpage, mpage->mlinks->name, TYPE_Nd); fclose(stream); free(title); return; @@ -1337,9 +1408,7 @@ parse_cat(struct mpage *mpage, int fd) plen -= 2; } - assert(NULL == mpage->desc); mpage->desc = mandoc_strdup(p); - putkey(mpage, mpage->desc, TYPE_Nd); fclose(stream); free(title); } @@ -1442,7 +1511,7 @@ parse_man(struct mpage *mpage, const struct man_node * ('\\' == start[0] && '-' == start[1])) break; - putkey(mpage, start, TYPE_NAME | TYPE_Nm); + putkey(mpage, start, NAME_TITLE); if (' ' == byte) { start += sz + 1; @@ -1456,7 +1525,7 @@ parse_man(struct mpage *mpage, const struct man_node * } if (start == title) { - putkey(mpage, start, TYPE_NAME | TYPE_Nm); + putkey(mpage, start, NAME_TITLE); free(title); return; } @@ -1478,9 +1547,7 @@ parse_man(struct mpage *mpage, const struct man_node * while (' ' == *start) start++; - assert(NULL == mpage->desc); mpage->desc = mandoc_strdup(start); - putkey(mpage, mpage->desc, TYPE_Nd); free(title); return; } @@ -1634,9 +1701,9 @@ parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_n { if (SEC_NAME == n->sec) - putmdockey(mpage, n->child, TYPE_NAME | TYPE_Nm); + putmdockey(mpage, n->child, NAME_TITLE); else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) - putmdockey(mpage, n->child, TYPE_Nm); + putmdockey(mpage, n->child, NAME_SYN); return(0); } @@ -1670,28 +1737,35 @@ static void putkeys(const struct mpage *mpage, const char *cp, size_t sz, uint64_t v) { + struct ohash *htab; struct str *s; const char *end; - uint64_t mask; unsigned int slot; int i; if (0 == sz) return; - if (debug > 1) { - for (i = 0, mask = 1; - i < mansearch_keymax; - i++, mask <<= 1) - if (mask & v) - break; - say(mpage->mlinks->file, "Adding key %s=%*s", - mansearch_keynames[i], sz, cp); + if (TYPE_Nm & v) { + htab = &names; + v &= name_mask; + name_mask &= ~NAME_FIRST; + if (debug > 1) + say(mpage->mlinks->file, + "Adding name %*s", sz, cp); + } else { + htab = &strings; + if (debug > 1) + for (i = 0; i < mansearch_keymax; i++) + if (1 << i & v) + say(mpage->mlinks->file, + "Adding key %s=%*s", + mansearch_keynames[i], sz, cp); } end = cp + sz; - slot = ohash_qlookupi(&strings, cp, &end); - s = ohash_find(&strings, slot); + slot = ohash_qlookupi(htab, cp, &end); + s = ohash_find(htab, slot); if (NULL != s && mpage == s->mpage) { s->mask |= v; @@ -1699,7 +1773,7 @@ putkeys(const struct mpage *mpage, } else if (NULL == s) { s = mandoc_calloc(sizeof(struct str) + sz + 1, 1); memcpy(s->key, cp, sz); - ohash_insert(&strings, slot, s); + ohash_insert(htab, slot, s); } s->mpage = mpage; s->mask = v; @@ -1887,7 +1961,7 @@ dbadd_mlink(const struct mlink *mlink) SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec); SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch); SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name); - SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->recno); + SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid); SQL_STEP(stmts[STMT_INSERT_LINK]); sqlite3_reset(stmts[STMT_INSERT_LINK]); } @@ -1928,16 +2002,7 @@ dbadd(struct mpage *mpage, struct mchars *mc) if (NULL != mlink) fputs(", ", stdout); } - for (key = ohash_first(&strings, &slot); NULL != key; - key = ohash_next(&strings, &slot)) { - if (TYPE_Nd & key->mask) { - if (NULL == key->rendered) - render_key(mc, key); - printf(" - %s", key->rendered); - break; - } - } - putchar('\n'); + printf(" - %s\n", mpage->desc); return; } @@ -1945,16 +2010,33 @@ dbadd(struct mpage *mpage, struct mchars *mc) say(mlink->file, "Adding to database"); i = 1; + SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->desc); SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form); SQL_STEP(stmts[STMT_INSERT_PAGE]); - mpage->recno = sqlite3_last_insert_rowid(db); + mpage->pageid = sqlite3_last_insert_rowid(db); sqlite3_reset(stmts[STMT_INSERT_PAGE]); while (NULL != mlink) { dbadd_mlink(mlink); mlink = mlink->next; } + mlink = mpage->mlinks; + for (key = ohash_first(&names, &slot); NULL != key; + key = ohash_next(&names, &slot)) { + assert(key->mpage == mpage); + if (NULL == key->rendered) + render_key(mc, key); + i = 1; + SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask); + SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered); + SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid); + SQL_STEP(stmts[STMT_INSERT_NAME]); + sqlite3_reset(stmts[STMT_INSERT_NAME]); + if (key->rendered != key->key) + free(key->rendered); + free(key); + } for (key = ohash_first(&strings, &slot); NULL != key; key = ohash_next(&strings, &slot)) { assert(key->mpage == mpage); @@ -1963,7 +2045,7 @@ dbadd(struct mpage *mpage, struct mchars *mc) i = 1; SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask); SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered); - SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->recno); + SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid); SQL_STEP(stmts[STMT_INSERT_KEY]); sqlite3_reset(stmts[STMT_INSERT_KEY]); if (key->rendered != key->key) @@ -2157,24 +2239,34 @@ dbopen(int real) create_tables: sql = "CREATE TABLE \"mpages\" (\n" + " \"desc\" TEXT NOT NULL,\n" " \"form\" INTEGER NOT NULL,\n" - " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" + " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" ");\n" "\n" "CREATE TABLE \"mlinks\" (\n" " \"sec\" TEXT NOT NULL,\n" " \"arch\" TEXT NOT NULL,\n" " \"name\" TEXT NOT NULL,\n" - " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) " + " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) " "ON DELETE CASCADE\n" ");\n" + "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n" "\n" + "CREATE TABLE \"names\" (\n" + " \"bits\" INTEGER NOT NULL,\n" + " \"name\" TEXT NOT NULL,\n" + " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) " + "ON DELETE CASCADE\n" + ");\n" + "\n" "CREATE TABLE \"keys\" (\n" " \"bits\" INTEGER NOT NULL,\n" " \"key\" TEXT NOT NULL,\n" - " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) " + " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) " "ON DELETE CASCADE\n" - ");\n"; + ");\n" + "CREATE INDEX keys_pageid_idx ON keys (pageid);\n"; if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) { exitcode = (int)MANDOCLEVEL_SYSERR; @@ -2184,16 +2276,19 @@ create_tables: prepare_statements: SQL_EXEC("PRAGMA foreign_keys = ON"); - sql = "DELETE FROM mpages WHERE id IN " + sql = "DELETE FROM mpages WHERE pageid IN " "(SELECT pageid FROM mlinks WHERE " "sec=? AND arch=? AND name=?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL); sql = "INSERT INTO mpages " - "(form) VALUES (?)"; + "(desc,form) VALUES (?,?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL); sql = "INSERT INTO mlinks " "(sec,arch,name,pageid) VALUES (?,?,?,?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL); + sql = "INSERT INTO names " + "(bits,name,pageid) VALUES (?,?,?)"; + sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL); sql = "INSERT INTO keys " "(bits,key,pageid) VALUES (?,?,?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);