=================================================================== RCS file: /cvs/mandoc/mandocdb.c,v retrieving revision 1.103 retrieving revision 1.119 diff -u -p -r1.103 -r1.119 --- mandoc/mandocdb.c 2014/01/06 03:02:46 1.103 +++ mandoc/mandocdb.c 2014/03/23 11:25:26 1.119 @@ -1,4 +1,4 @@ -/* $Id: mandocdb.c,v 1.103 2014/01/06 03:02:46 schwarze Exp $ */ +/* $Id: mandocdb.c,v 1.119 2014/03/23 11:25:26 schwarze Exp $ */ /* * Copyright (c) 2011, 2012 Kristaps Dzonsons * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze @@ -20,6 +20,7 @@ #endif #include +#include #include #include @@ -45,9 +46,13 @@ #include "mdoc.h" #include "man.h" #include "mandoc.h" +#include "mandoc_aux.h" #include "manpath.h" #include "mansearch.h" +extern int mansearch_keymax; +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)) @@ -95,6 +100,7 @@ struct inodev { struct mpage { struct inodev inodev; /* used for hashing routine */ + int64_t recno; /* id in mpages SQL table */ enum form form; /* format from file content */ char *sec; /* section from file content */ char *arch; /* architecture from file content */ @@ -112,6 +118,7 @@ struct mlink { char *name; /* name from file name (not empty) */ char *fsec; /* section from file name suffix */ struct mlink *next; /* singly linked list */ + struct mpage *mpage; /* parent */ }; enum stmt { @@ -130,7 +137,8 @@ struct mdoc_handler { }; static void dbclose(int); -static void dbindex(const struct mpage *, struct mchars *); +static void dbadd(struct mpage *, struct mchars *); +static void dbadd_mlink(const struct mlink *mlink); static int dbopen(int); static void dbprune(void); static void filescan(const char *); @@ -154,8 +162,7 @@ static int parse_mdoc_Nd(struct mpage *, const struct static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *); static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *); static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *); -static void putkey(const struct mpage *, - const char *, uint64_t); +static void putkey(const struct mpage *, char *, uint64_t); static void putkeys(const struct mpage *, const char *, size_t, uint64_t); static void putmdockey(const struct mpage *, @@ -166,9 +173,10 @@ static int set_basedir(const char *); static int treescan(void); static size_t utf8(unsigned int, char [7]); +static char tempfilename[32]; static char *progname; static int nodb; /* no database changes */ -static int quick; /* abort the parse early */ +static int mparse_options; /* abort the parse early */ static int use_all; /* use all found files */ static int verb; /* print what we're doing */ static int warnings; /* warn about crap */ @@ -367,7 +375,7 @@ main(int argc, char *argv[]) nodb = 1; break; case ('Q'): - quick = 1; + mparse_options |= MPARSE_QUICK; break; case ('T'): if (strcmp(optarg, "utf8")) { @@ -407,8 +415,7 @@ main(int argc, char *argv[]) } exitcode = (int)MANDOCLEVEL_OK; - mp = mparse_alloc(MPARSE_AUTO, - MANDOCLEVEL_FATAL, NULL, NULL, quick); + mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL, NULL); mc = mchars_alloc(); ohash_init(&mpages, 6, &mpages_info); @@ -700,11 +707,19 @@ filescan(const char *file) exitcode = (int)MANDOCLEVEL_BADARG; say(file, NULL); return; - } else if (OP_TEST != op && strstr(buf, basedir) != buf) { + } + + if (strstr(buf, basedir) == buf) + start = buf + strlen(basedir) + 1; + else if (OP_TEST == op) + start = buf; + else { exitcode = (int)MANDOCLEVEL_BADARG; say("", "%s: outside base directory", buf); return; - } else if (-1 == stat(buf, &st)) { + } + + if (-1 == stat(buf, &st)) { exitcode = (int)MANDOCLEVEL_BADARG; say(file, NULL); return; @@ -713,7 +728,7 @@ filescan(const char *file) say(file, "Not a regular file"); return; } - start = buf + strlen(basedir); + mlink = mandoc_calloc(1, sizeof(struct mlink)); strlcpy(mlink->file, start, sizeof(mlink->file)); @@ -806,6 +821,7 @@ mlink_add(struct mlink *mlink, const struct stat *st) } else mlink->next = mpage->mlinks; mpage->mlinks = mlink; + mlink->mpage = mpage; } static void @@ -946,12 +962,14 @@ mlink_check(struct mpage *mpage, struct mlink *mlink) static void mpages_merge(struct mchars *mc, struct mparse *mp) { + char any[] = "any"; struct ohash_info str_info; - struct mpage *mpage; - struct mlink *mlink; + struct mpage *mpage, *mpage_dest; + struct mlink *mlink, *mlink_dest; struct mdoc *mdoc; struct man *man; - const char *cp; + char *sodest; + char *cp; int match; unsigned int pslot; enum mandoclevel lvl; @@ -961,6 +979,9 @@ mpages_merge(struct mchars *mc, struct mparse *mp) str_info.hfree = hash_free; str_info.key_offset = offsetof(struct str, key); + if (0 == nodb) + SQL_EXEC("BEGIN TRANSACTION"); + mpage = ohash_first(&mpages, &pslot); while (NULL != mpage) { mlinks_undupe(mpage); @@ -983,10 +1004,48 @@ mpages_merge(struct mchars *mc, struct mparse *mp) FORM_CAT != mpage->mlinks->fform) { lvl = mparse_readfd(mp, -1, mpage->mlinks->file); if (lvl < MANDOCLEVEL_FATAL) - mparse_result(mp, &mdoc, &man); + mparse_result(mp, &mdoc, &man, &sodest); } - if (NULL != mdoc) { + if (NULL != sodest) { + mlink_dest = ohash_find(&mlinks, + ohash_qlookup(&mlinks, sodest)); + if (NULL != mlink_dest) { + + /* The .so target exists. */ + + mpage_dest = mlink_dest->mpage; + mlink = mpage->mlinks; + while (1) { + mlink->mpage = mpage_dest; + + /* + * If the target was already + * processed, add the links + * to the database now. + * Otherwise, this will + * happen when we come + * to the target. + */ + + if (mpage_dest->recno) + dbadd_mlink(mlink); + + if (NULL == mlink->next) + break; + mlink = mlink->next; + } + + /* Move all links to the target. */ + + mlink->next = mlink_dest->next; + mlink_dest->next = mpage->mlinks; + mpage->mlinks = NULL; + } + ohash_delete(&strings); + mpage = ohash_next(&mpages, &pslot); + continue; + } else if (NULL != mdoc) { mpage->form = FORM_SRC; mpage->sec = mandoc_strdup(mdoc_meta(mdoc)->msec); @@ -1014,7 +1073,7 @@ mpages_merge(struct mchars *mc, struct mparse *mp) } putkey(mpage, mpage->sec, TYPE_sec); putkey(mpage, '\0' == *mpage->arch ? - "any" : mpage->arch, TYPE_arch); + any : mpage->arch, TYPE_arch); for (mlink = mpage->mlinks; mlink; mlink = mlink->next) { if ('\0' != *mlink->dsec) @@ -1022,7 +1081,7 @@ mpages_merge(struct mchars *mc, struct mparse *mp) if ('\0' != *mlink->fsec) putkey(mpage, mlink->fsec, TYPE_sec); putkey(mpage, '\0' == *mlink->arch ? - "any" : mlink->arch, TYPE_arch); + any : mlink->arch, TYPE_arch); putkey(mpage, mlink->name, TYPE_Nm); } @@ -1047,10 +1106,13 @@ mpages_merge(struct mchars *mc, struct mparse *mp) else parse_cat(mpage); - dbindex(mpage, mc); + dbadd(mpage, mc); ohash_delete(&strings); mpage = ohash_next(&mpages, &pslot); } + + if (0 == nodb) + SQL_EXEC("END TRANSACTION"); } static void @@ -1169,10 +1231,15 @@ parse_cat(struct mpage *mpage) * Put a type/word pair into the word database for this particular file. */ static void -putkey(const struct mpage *mpage, const char *value, uint64_t type) +putkey(const struct mpage *mpage, char *value, uint64_t type) { + char *cp; assert(NULL != value); + if (TYPE_arch == type) + for (cp = value; *cp; cp++) + if (isupper((unsigned char)*cp)) + *cp = _tolower((unsigned char)*cp); putkeys(mpage, value, strlen(value), type); } @@ -1288,6 +1355,15 @@ parse_man(struct mpage *mpage, const struct man_node * byte = start[sz]; start[sz] = '\0'; + /* + * Assume a stray trailing comma in the + * name list if a name begins with a dash. + */ + + if ('-' == start[0] || + ('\\' == start[0] && '-' == start[1])) + break; + putkey(mpage, start, TYPE_Nm); if (' ' == byte) { @@ -1417,7 +1493,7 @@ parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_n static int parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n) { - const char *cp; + char *cp; if (NULL == (n = n->child) || MDOC_TEXT != n->type) return(0); @@ -1539,12 +1615,24 @@ putkeys(const struct mpage *mpage, const char *cp, size_t sz, uint64_t v) { struct str *s; - unsigned int slot; const char *end; + uint64_t mask; + unsigned int slot; + int i; if (0 == sz) return; + if (verb > 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); + } + end = cp + sz; slot = ohash_qlookupi(&strings, cp, &end); s = ohash_find(&strings, slot); @@ -1621,7 +1709,7 @@ static void render_key(struct mchars *mc, struct str *key) { size_t sz, bsz, pos; - char utfbuf[7], res[5]; + char utfbuf[7], res[6]; char *buf; const char *seq, *cpp, *val; int len, u; @@ -1633,7 +1721,8 @@ render_key(struct mchars *mc, struct str *key) res[1] = '\t'; res[2] = ASCII_NBRSP; res[3] = ASCII_HYPH; - res[4] = '\0'; + res[4] = ASCII_BREAK; + res[5] = '\0'; val = key->key; bsz = strlen(val); @@ -1664,16 +1753,24 @@ render_key(struct mchars *mc, struct str *key) val += sz; } - if (ASCII_HYPH == *val) { + switch (*val) { + case (ASCII_HYPH): buf[pos++] = '-'; val++; continue; - } else if ('\t' == *val || ASCII_NBRSP == *val) { + case ('\t'): + /* FALLTHROUGH */ + case (ASCII_NBRSP): buf[pos++] = ' '; val++; + /* FALLTHROUGH */ + case (ASCII_BREAK): continue; - } else if ('\\' != *val) + default: break; + } + if ('\\' != *val) + break; /* Read past the slash. */ @@ -1725,6 +1822,20 @@ render_key(struct mchars *mc, struct str *key) key->rendered = buf; } +static void +dbadd_mlink(const struct mlink *mlink) +{ + size_t i; + + i = 1; + 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_STEP(stmts[STMT_INSERT_LINK]); + sqlite3_reset(stmts[STMT_INSERT_LINK]); +} + /* * Flush the current page's terms (and their bits) into the database. * Wrap the entire set of additions in a transaction to make sqlite be a @@ -1732,38 +1843,27 @@ render_key(struct mchars *mc, struct str *key) * Also, handle escape sequences at the last possible moment. */ static void -dbindex(const struct mpage *mpage, struct mchars *mc) +dbadd(struct mpage *mpage, struct mchars *mc) { struct mlink *mlink; struct str *key; - int64_t recno; size_t i; unsigned int slot; if (verb) - say(mpage->mlinks->file, "Adding to index"); + say(mpage->mlinks->file, "Adding to database"); if (nodb) return; - SQL_EXEC("BEGIN TRANSACTION"); - i = 1; SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form); SQL_STEP(stmts[STMT_INSERT_PAGE]); - recno = sqlite3_last_insert_rowid(db); + mpage->recno = sqlite3_last_insert_rowid(db); sqlite3_reset(stmts[STMT_INSERT_PAGE]); - for (mlink = mpage->mlinks; mlink; mlink = mlink->next) { - i = 1; - SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->file); - 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, recno); - SQL_STEP(stmts[STMT_INSERT_LINK]); - sqlite3_reset(stmts[STMT_INSERT_LINK]); - } + for (mlink = mpage->mlinks; mlink; mlink = mlink->next) + dbadd_mlink(mlink); for (key = ohash_first(&strings, &slot); NULL != key; key = ohash_next(&strings, &slot)) { @@ -1773,15 +1873,13 @@ dbindex(const 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, recno); + SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->recno); SQL_STEP(stmts[STMT_INSERT_KEY]); sqlite3_reset(stmts[STMT_INSERT_KEY]); if (key->rendered != key->key) free(key->rendered); free(key); } - - SQL_EXEC("END TRANSACTION"); } static void @@ -1792,20 +1890,31 @@ dbprune(void) size_t i; unsigned int slot; - if (nodb) - return; + if (0 == nodb) + SQL_EXEC("BEGIN TRANSACTION"); - mpage = ohash_first(&mpages, &slot); - while (NULL != mpage) { + for (mpage = ohash_first(&mpages, &slot); NULL != mpage; + mpage = ohash_next(&mpages, &slot)) { mlink = mpage->mlinks; - i = 1; - SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], i, mlink->file); - SQL_STEP(stmts[STMT_DELETE_PAGE]); - sqlite3_reset(stmts[STMT_DELETE_PAGE]); if (verb) - say(mlink->file, "Deleted from index"); - mpage = ohash_next(&mpages, &slot); + say(mlink->file, "Deleting from database"); + if (nodb) + continue; + for ( ; NULL != mlink; mlink = mlink->next) { + i = 1; + SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], + i, mlink->dsec); + SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], + i, mlink->arch); + SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], + i, mlink->name); + SQL_STEP(stmts[STMT_DELETE_PAGE]); + sqlite3_reset(stmts[STMT_DELETE_PAGE]); + } } + + if (0 == nodb) + SQL_EXEC("END TRANSACTION"); } /* @@ -1816,6 +1925,8 @@ static void dbclose(int real) { size_t i; + int status; + pid_t child; if (nodb) return; @@ -1831,10 +1942,60 @@ dbclose(int real) if (real) return; - if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) { + if ('\0' == *tempfilename) { + if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(MANDOC_DB, "%s", strerror(errno)); + } + return; + } + + switch (child = fork()) { + case (-1): exitcode = (int)MANDOCLEVEL_SYSERR; - say(MANDOC_DB, NULL); + say("fork cmp", "%s", strerror(errno)); + return; + case (0): + execlp("cmp", "cmp", "-s", + tempfilename, MANDOC_DB, NULL); + say("exec cmp", "%s", strerror(errno)); + exit(0); + default: + break; } + if (-1 == waitpid(child, &status, 0)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("wait cmp", "%s", strerror(errno)); + } else if (WIFSIGNALED(status)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("cmp", "Died from a signal"); + } else if (WEXITSTATUS(status)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(MANDOC_DB, + "Data changed, but cannot replace database"); + } + + *strrchr(tempfilename, '/') = '\0'; + switch (child = fork()) { + case (-1): + exitcode = (int)MANDOCLEVEL_SYSERR; + say("fork rm", "%s", strerror(errno)); + return; + case (0): + execlp("rm", "rm", "-rf", tempfilename, NULL); + say("exec rm", "%s", strerror(errno)); + exit((int)MANDOCLEVEL_SYSERR); + default: + break; + } + if (-1 == waitpid(child, &status, 0)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("wait rm", "%s", strerror(errno)); + } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(tempfilename, + "Cannot remove temporary directory"); + } } /* @@ -1848,82 +2009,99 @@ dbclose(int real) static int dbopen(int real) { - const char *file, *sql; + const char *sql; int rc, ofl; if (nodb) return(1); + *tempfilename = '\0'; ofl = SQLITE_OPEN_READWRITE; - if (0 == real) { - file = MANDOC_DB "~"; - if (-1 == remove(file) && ENOENT != errno) { + + if (real) { + rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL); + if (SQLITE_OK != rc) { exitcode = (int)MANDOCLEVEL_SYSERR; - say(file, NULL); + say(MANDOC_DB, "%s", sqlite3_errmsg(db)); return(0); } - ofl |= SQLITE_OPEN_EXCLUSIVE; - } else - file = MANDOC_DB; + goto prepare_statements; + } - rc = sqlite3_open_v2(file, &db, ofl, NULL); + ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE; + + remove(MANDOC_DB "~"); + rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL); if (SQLITE_OK == rc) - goto prepare_statements; - if (SQLITE_CANTOPEN != rc) { + goto create_tables; + if (MPARSE_QUICK & mparse_options) { exitcode = (int)MANDOCLEVEL_SYSERR; - say(file, NULL); + say(MANDOC_DB "~", "%s", sqlite3_errmsg(db)); return(0); } - sqlite3_close(db); - db = NULL; - - if (SQLITE_OK != (rc = sqlite3_open(file, &db))) { + if (strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX", + sizeof(tempfilename)) >= sizeof(tempfilename)) { exitcode = (int)MANDOCLEVEL_SYSERR; - say(file, NULL); + say("/tmp/mandocdb.XXXXXX", "Filename too long"); return(0); } + if (NULL == mkdtemp(tempfilename)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(tempfilename, "%s", strerror(errno)); + return(0); + } + if (strlcat(tempfilename, "/" MANDOC_DB, + sizeof(tempfilename)) >= sizeof(tempfilename)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(tempfilename, "Filename too long"); + return(0); + } + rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL); + if (SQLITE_OK != rc) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(tempfilename, "%s", sqlite3_errmsg(db)); + return(0); + } +create_tables: sql = "CREATE TABLE \"mpages\" (\n" " \"form\" INTEGER NOT NULL,\n" " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" ");\n" "\n" "CREATE TABLE \"mlinks\" (\n" - " \"file\" TEXT NOT NULL,\n" " \"sec\" TEXT NOT NULL,\n" " \"arch\" TEXT NOT NULL,\n" " \"name\" TEXT NOT NULL,\n" " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) " - "ON DELETE CASCADE,\n" - " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" + "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) " - "ON DELETE CASCADE,\n" - " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" - ");\n" - "\n" - "CREATE INDEX \"key_index\" ON keys (key);\n"; + "ON DELETE CASCADE\n" + ");\n"; if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) { exitcode = (int)MANDOCLEVEL_SYSERR; - say(file, "%s", sqlite3_errmsg(db)); + say(MANDOC_DB, "%s", sqlite3_errmsg(db)); return(0); } prepare_statements: SQL_EXEC("PRAGMA foreign_keys = ON"); - sql = "DELETE FROM mpages where file=?"; + sql = "DELETE FROM mpages WHERE id 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 (?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL); sql = "INSERT INTO mlinks " - "(file,sec,arch,name,pageid) VALUES (?,?,?,?,?)"; + "(sec,arch,name,pageid) VALUES (?,?,?,?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL); sql = "INSERT INTO keys " "(bits,key,pageid) VALUES (?,?,?)";