=================================================================== RCS file: /cvs/mandoc/mandocdb.c,v retrieving revision 1.103 retrieving revision 1.124 diff -u -p -r1.103 -r1.124 --- mandoc/mandocdb.c 2014/01/06 03:02:46 1.103 +++ mandoc/mandocdb.c 2014/03/26 21:39:38 1.124 @@ -1,4 +1,4 @@ -/* $Id: mandocdb.c,v 1.103 2014/01/06 03:02:46 schwarze Exp $ */ +/* $Id: mandocdb.c,v 1.124 2014/03/26 21:39:38 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 +118,8 @@ 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 */ + int gzip; /* filename has a .gz suffix */ }; enum stmt { @@ -130,7 +138,8 @@ 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 *); @@ -143,7 +152,7 @@ static void mlink_free(struct mlink *); static void mlinks_undupe(struct mpage *); static void mpages_free(void); static void mpages_merge(struct mchars *, struct mparse *); -static void parse_cat(struct mpage *); +static void parse_cat(struct mpage *, int); static void parse_man(struct mpage *, const struct man_node *); static void parse_mdoc(struct mpage *, const struct mdoc_node *); static int parse_mdoc_body(struct mpage *, const struct mdoc_node *); @@ -154,8 +163,7 @@ 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 +174,10 @@ 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 +376,7 @@ 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 +416,7 @@ 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); @@ -529,7 +537,7 @@ treescan(void) FTS *f; FTSENT *ff; struct mlink *mlink; - int dform; + int dform, gzip; char *dsec, *arch, *fsec, *cp; const char *path; const char *argv[2]; @@ -544,7 +552,7 @@ treescan(void) f = fts_open((char * const *)argv, FTS_LOGICAL, NULL); if (NULL == f) { exitcode = (int)MANDOCLEVEL_SYSERR; - say("", NULL); + say("", "&fts_open"); return(0); } @@ -564,8 +572,18 @@ treescan(void) if (warnings) say(path, "Extraneous file"); continue; - } else if (NULL == (fsec = - strrchr(ff->fts_name, '.'))) { + } + gzip = 0; + fsec = NULL; + while (NULL == fsec) { + fsec = strrchr(ff->fts_name, '.'); + if (NULL == fsec || strcmp(fsec+1, "gz")) + break; + gzip = 1; + *fsec = '\0'; + fsec = NULL; + } + if (NULL == fsec) { if ( ! use_all) { if (warnings) say(path, @@ -576,10 +594,6 @@ treescan(void) if (warnings) say(path, "Skip html"); continue; - } else if (0 == strcmp(fsec, "gz")) { - if (warnings) - say(path, "Skip gz"); - continue; } else if (0 == strcmp(fsec, "ps")) { if (warnings) say(path, "Skip ps"); @@ -604,6 +618,7 @@ treescan(void) mlink->arch = arch; mlink->name = ff->fts_name; mlink->fsec = fsec; + mlink->gzip = gzip; mlink_add(mlink, ff->fts_statp); continue; } else if (FTS_D != ff->fts_info && @@ -698,22 +713,30 @@ filescan(const char *file) if (NULL == realpath(file, buf)) { exitcode = (int)MANDOCLEVEL_BADARG; - say(file, NULL); + say(file, "&realpath"); 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); + say(file, "&stat"); return; } else if ( ! (S_IFREG & st.st_mode)) { exitcode = (int)MANDOCLEVEL_BADARG; 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 +829,7 @@ mlink_add(struct mlink *mlink, const struct stat *st) } else mlink->next = mpage->mlinks; mpage->mlinks = mlink; + mlink->mpage = mpage; } static void @@ -946,13 +970,17 @@ 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; + int fd[2]; + struct mpage *mpage, *mpage_dest; + struct mlink *mlink, *mlink_dest; struct mdoc *mdoc; struct man *man; - const char *cp; - int match; + char *sodest; + char *cp; + pid_t child_pid; + int match, status; unsigned int pslot; enum mandoclevel lvl; @@ -961,6 +989,9 @@ 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); @@ -973,7 +1004,42 @@ mpages_merge(struct mchars *mc, struct mparse *mp) mparse_reset(mp); mdoc = NULL; man = NULL; + sodest = NULL; + child_pid = 0; + fd[0] = -1; + fd[1] = -1; + if (mpage->mlinks->gzip) { + if (-1 == pipe(fd)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(mpage->mlinks->file, "&pipe gunzip"); + goto nextpage; + } + switch (child_pid = fork()) { + case (-1): + exitcode = (int)MANDOCLEVEL_SYSERR; + say(mpage->mlinks->file, "&fork gunzip"); + child_pid = 0; + close(fd[1]); + close(fd[0]); + goto nextpage; + case (0): + close(fd[0]); + if (-1 == dup2(fd[1], STDOUT_FILENO)) { + say(mpage->mlinks->file, + "&dup gunzip"); + exit(1); + } + execlp("gunzip", "gunzip", "-c", + mpage->mlinks->file, NULL); + say(mpage->mlinks->file, "&exec gunzip"); + exit(1); + default: + close(fd[1]); + break; + } + } + /* * Try interpreting the file as mdoc(7) or man(7) * source code, unless it is already known to be @@ -981,12 +1047,48 @@ mpages_merge(struct mchars *mc, struct mparse *mp) */ if (FORM_CAT != mpage->mlinks->dform || FORM_CAT != mpage->mlinks->fform) { - lvl = mparse_readfd(mp, -1, mpage->mlinks->file); + lvl = mparse_readfd(mp, fd[0], 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; + } + goto nextpage; + } else if (NULL != mdoc) { mpage->form = FORM_SRC; mpage->sec = mandoc_strdup(mdoc_meta(mdoc)->msec); @@ -1014,7 +1116,7 @@ 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 +1124,7 @@ 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); } @@ -1045,24 +1147,48 @@ mpages_merge(struct mchars *mc, struct mparse *mp) } else if (NULL != man) parse_man(mpage, man_node(man)); else - parse_cat(mpage); + parse_cat(mpage, fd[0]); - dbindex(mpage, mc); + dbadd(mpage, mc); + +nextpage: + if (child_pid) { + if (-1 == waitpid(child_pid, &status, 0)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(mpage->mlinks->file, "&wait gunzip"); + } else if (WIFSIGNALED(status)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(mpage->mlinks->file, + "gunzip died from signal %d", + WTERMSIG(status)); + } else if (WEXITSTATUS(status)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(mpage->mlinks->file, + "gunzip failed with code %d", + WEXITSTATUS(status)); + } + } ohash_delete(&strings); mpage = ohash_next(&mpages, &pslot); } + + if (0 == nodb) + SQL_EXEC("END TRANSACTION"); } static void -parse_cat(struct mpage *mpage) +parse_cat(struct mpage *mpage, int fd) { FILE *stream; char *line, *p, *title; size_t len, plen, titlesz; - if (NULL == (stream = fopen(mpage->mlinks->file, "r"))) { + stream = (-1 == fd) ? + fopen(mpage->mlinks->file, "r") : + fdopen(fd, "r"); + if (NULL == stream) { if (warnings) - say(mpage->mlinks->file, NULL); + say(mpage->mlinks->file, "&fopen"); return; } @@ -1169,10 +1295,15 @@ 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); } @@ -1196,9 +1327,9 @@ static void parse_man(struct mpage *mpage, const struct man_node *n) { const struct man_node *head, *body; - char *start, *sv, *title; + char *start, *title; char byte; - size_t sz, titlesz; + size_t sz; if (NULL == n) return; @@ -1218,12 +1349,8 @@ parse_man(struct mpage *mpage, const struct man_node * NULL != (head = (head->child)) && MAN_TEXT == head->type && 0 == strcmp(head->string, "NAME") && - NULL != (body = body->child) && - MAN_TEXT == body->type) { + NULL != body->child) { - title = NULL; - titlesz = 0; - /* * Suck the entire NAME section into memory. * Yes, we might run away. @@ -1231,47 +1358,11 @@ parse_man(struct mpage *mpage, const struct man_node * * NAME sections over many lines. */ - for ( ; NULL != body; body = body->next) { - if (MAN_TEXT != body->type) - break; - if (0 == (sz = strlen(body->string))) - continue; - title = mandoc_realloc - (title, titlesz + sz + 1); - memcpy(title + titlesz, body->string, sz); - titlesz += sz + 1; - title[titlesz - 1] = ' '; - } + title = NULL; + man_deroff(&title, body); if (NULL == title) return; - title = mandoc_realloc(title, titlesz + 1); - title[titlesz] = '\0'; - - /* Skip leading space. */ - - sv = title; - while (isspace((unsigned char)*sv)) - sv++; - - if (0 == (sz = strlen(sv))) { - free(title); - return; - } - - /* Erase trailing space. */ - - start = &sv[sz - 1]; - while (start > sv && isspace((unsigned char)*start)) - *start-- = '\0'; - - if (start == sv) { - free(title); - return; - } - - start = sv; - /* * Go through a special heuristic dance here. * Conventionally, one or more manual names are @@ -1280,6 +1371,7 @@ parse_man(struct mpage *mpage, const struct man_node * * the name parts here. */ + start = title; for ( ;; ) { sz = strcspn(start, " ,"); if ('\0' == start[sz]) @@ -1288,6 +1380,15 @@ 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) { @@ -1301,7 +1402,7 @@ parse_man(struct mpage *mpage, const struct man_node * start++; } - if (sv == start) { + if (start == title) { putkey(mpage, start, TYPE_Nm); free(title); return; @@ -1417,7 +1518,7 @@ 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); @@ -1460,10 +1561,7 @@ parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_n return(0); } - if (-1 == asprintf(&cp, "%s(%s)", n->string, n->next->string)) { - perror(NULL); - exit((int)MANDOCLEVEL_SYSERR); - } + mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string); putkey(mpage, cp, TYPE_Xr); free(cp); return(0); @@ -1472,32 +1570,10 @@ parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_n static int parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n) { - size_t sz; - if (MDOC_BODY != n->type) - return(0); - - /* - * Special-case the `Nd' because we need to put the description - * into the document table. - */ - - for (n = n->child; NULL != n; n = n->next) { - if (MDOC_TEXT == n->type) { - if (NULL != mpage->desc) { - sz = strlen(mpage->desc) + - strlen(n->string) + 2; - mpage->desc = mandoc_realloc( - mpage->desc, sz); - strlcat(mpage->desc, " ", sz); - strlcat(mpage->desc, n->string, sz); - } else - mpage->desc = mandoc_strdup(n->string); - } - if (NULL != n->child) - parse_mdoc_Nd(mpage, n); - } - return(1); + if (MDOC_BODY == n->type) + mdoc_deroff(&mpage->desc, n); + return(0); } static int @@ -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, "&rename"); + } + return; + } + + switch (child = fork()) { + case (-1): exitcode = (int)MANDOCLEVEL_SYSERR; - say(MANDOC_DB, NULL); + say("", "&fork cmp"); + return; + case (0): + execlp("cmp", "cmp", "-s", + tempfilename, MANDOC_DB, NULL); + say("", "&exec cmp"); + exit(0); + default: + break; } + if (-1 == waitpid(child, &status, 0)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "&wait cmp"); + } else if (WIFSIGNALED(status)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "cmp died from signal %d", WTERMSIG(status)); + } 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"); + return; + case (0): + execlp("rm", "rm", "-rf", tempfilename, NULL); + say("", "&exec rm"); + exit((int)MANDOCLEVEL_SYSERR); + default: + break; + } + if (-1 == waitpid(child, &status, 0)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "&wait rm"); + } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "%s: Cannot remove temporary directory", + tempfilename); + } } /* @@ -1848,82 +2009,100 @@ 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("", "&%s", tempfilename); + return(0); + } + if (strlcat(tempfilename, "/" MANDOC_DB, + sizeof(tempfilename)) >= sizeof(tempfilename)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "%s/" MANDOC_DB ": Filename too long", + tempfilename); + return(0); + } + rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL); + if (SQLITE_OK != rc) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "%s: %s", tempfilename, 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 (?,?,?)"; @@ -1979,12 +2158,12 @@ set_basedir(const char *targetdir) if (NULL == getcwd(startdir, PATH_MAX)) { exitcode = (int)MANDOCLEVEL_SYSERR; if (NULL != targetdir) - say(".", NULL); + say("", "&getcwd"); return(0); } if (-1 == (fd = open(startdir, O_RDONLY, 0))) { exitcode = (int)MANDOCLEVEL_SYSERR; - say(startdir, NULL); + say("", "&open %s", startdir); return(0); } if (NULL == targetdir) @@ -1996,7 +2175,7 @@ set_basedir(const char *targetdir) close(fd); basedir[0] = '\0'; exitcode = (int)MANDOCLEVEL_SYSERR; - say(startdir, NULL); + say("", "&chdir %s", startdir); return(0); } if (NULL == targetdir) { @@ -2007,11 +2186,11 @@ set_basedir(const char *targetdir) if (NULL == realpath(targetdir, basedir)) { basedir[0] = '\0'; exitcode = (int)MANDOCLEVEL_BADARG; - say(targetdir, NULL); + say("", "&%s: realpath", targetdir); return(0); } else if (-1 == chdir(basedir)) { exitcode = (int)MANDOCLEVEL_BADARG; - say("", NULL); + say("", "&chdir"); return(0); } return(1); @@ -2021,6 +2200,7 @@ static void say(const char *file, const char *format, ...) { va_list ap; + int use_errno; if ('\0' != *basedir) fprintf(stderr, "%s", basedir); @@ -2028,16 +2208,32 @@ say(const char *file, const char *format, ...) fputs("//", stderr); if ('\0' != *file) fprintf(stderr, "%s", file); - fputs(": ", stderr); - if (NULL == format) { - perror(NULL); - return; + use_errno = 1; + if (NULL != format) { + switch (*format) { + case ('&'): + format++; + break; + case ('\0'): + format = NULL; + break; + default: + use_errno = 0; + break; + } } - - va_start(ap, format); - vfprintf(stderr, format, ap); - va_end(ap); - - fputc('\n', stderr); + if (NULL != format) { + if ('\0' != *basedir || '\0' != *file) + fputs(": ", stderr); + va_start(ap, format); + vfprintf(stderr, format, ap); + va_end(ap); + } + if (use_errno) { + if ('\0' != *basedir || '\0' != *file || NULL != format) + fputs(": ", stderr); + perror(NULL); + } else + fputc('\n', stderr); }