=================================================================== RCS file: /cvs/mandoc/mandocdb.c,v retrieving revision 1.84 retrieving revision 1.93 diff -u -p -r1.84 -r1.93 --- mandoc/mandocdb.c 2013/12/27 15:39:03 1.84 +++ mandoc/mandocdb.c 2014/01/02 18:52:15 1.93 @@ -1,4 +1,4 @@ -/* $Id: mandocdb.c,v 1.84 2013/12/27 15:39:03 schwarze Exp $ */ +/* $Id: mandocdb.c,v 1.93 2014/01/02 18:52:15 schwarze Exp $ */ /* * Copyright (c) 2011, 2012 Kristaps Dzonsons * Copyright (c) 2011, 2012, 2013 Ingo Schwarze @@ -111,6 +111,7 @@ struct mlink { char *arch; /* architecture from directory */ char *name; /* name from file name (not empty) */ char *fsec; /* section from file name suffix */ + struct mlink *next; /* singly linked list */ }; struct title { @@ -141,9 +142,10 @@ static void filescan(const char *); static void *hash_alloc(size_t, void *); static void hash_free(void *, size_t, void *); static void *hash_halloc(size_t, void *); -static int inocheck(const struct stat *); static void mlink_add(struct mlink *, const struct stat *); +static int mlink_check(struct mpage *, struct mlink *); 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 *, int); static void parse_cat(struct mpage *); @@ -153,11 +155,9 @@ static int parse_mdoc_body(struct mpage *, const stru static int parse_mdoc_head(struct mpage *, const struct mdoc_node *); static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *); static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *); -static int parse_mdoc_In(struct mpage *, const struct mdoc_node *); static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *); 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_St(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); @@ -215,7 +215,7 @@ static const struct mdoc_handler mdocs[MDOC_MAX] = { { parse_mdoc_Fn, 0 }, /* Fn */ { NULL, TYPE_Ft }, /* Ft */ { NULL, TYPE_Ic }, /* Ic */ - { parse_mdoc_In, TYPE_In }, /* In */ + { NULL, TYPE_In }, /* In */ { NULL, TYPE_Li }, /* Li */ { parse_mdoc_Nd, TYPE_Nd }, /* Nd */ { parse_mdoc_Nm, TYPE_Nm }, /* Nm */ @@ -223,7 +223,7 @@ static const struct mdoc_handler mdocs[MDOC_MAX] = { { NULL, 0 }, /* Ot */ { NULL, TYPE_Pa }, /* Pa */ { NULL, 0 }, /* Rv */ - { parse_mdoc_St, 0 }, /* St */ + { NULL, TYPE_St }, /* St */ { NULL, TYPE_Va }, /* Va */ { parse_mdoc_body, TYPE_Va }, /* Vt */ { parse_mdoc_Xr, 0 }, /* Xr */ @@ -548,7 +548,6 @@ treescan(void) /* * If we're a regular file, add an mlink by using the * stored directory data and handling the filename. - * Disallow duplicate (hard-linked) files. */ if (FTS_F == ff->fts_info) { if (0 == strcmp(path, MANDOC_DB)) @@ -557,10 +556,6 @@ treescan(void) if (warnings) say(path, "Extraneous file"); continue; - } else if (inocheck(ff->fts_statp)) { - if (warnings) - say(path, "Duplicate file"); - continue; } else if (NULL == (fsec = strrchr(ff->fts_name, '.'))) { if ( ! use_all) { @@ -667,7 +662,7 @@ treescan(void) } /* - * Add a file to the file vector. + * Add a file to the mlinks table. * Do not verify that it's a "valid" looking manpage (we'll do that * later). * @@ -678,7 +673,6 @@ treescan(void) * or * [./]cat
[/]/.0 * - * Stuff this information directly into the mlink vector. * See treescan() for the fts(3) version of this. */ static void @@ -710,10 +704,6 @@ filescan(const char *file) exitcode = (int)MANDOCLEVEL_BADARG; say(file, "Not a regular file"); return; - } else if (inocheck(&st)) { - if (warnings) - say(file, "Duplicate file"); - return; } start = buf + strlen(basedir); mlink = mandoc_calloc(1, sizeof(struct mlink)); @@ -770,20 +760,6 @@ filescan(const char *file) mlink_add(mlink, &st); } -static int -inocheck(const struct stat *st) -{ - struct inodev inodev; - uint32_t hash; - - memset(&inodev, 0, sizeof(inodev)); - inodev.st_ino = hash = st->st_ino; - inodev.st_dev = st->st_dev; - - return(NULL != ohash_find(&mpages, ohash_lookup_memory( - &mpages, (char *)&inodev, sizeof(inodev), hash))); -} - static void mlink_add(struct mlink *mlink, const struct stat *st) { @@ -826,7 +802,7 @@ mlink_add(struct mlink *mlink, const struct stat *st) mpage->inodev.st_dev = inodev.st_dev; ohash_insert(&mpages, slot, mpage); } else - abort(); + mlink->next = mpage->mlinks; mpage->mlinks = mlink; } @@ -851,7 +827,7 @@ mpages_free(void) mpage = ohash_first(&mpages, &slot); while (NULL != mpage) { while (NULL != (mlink = mpage->mlinks)) { - mpage->mlinks = NULL; + mpage->mlinks = mlink->next; mlink_free(mlink); } free(mpage->sec); @@ -864,6 +840,101 @@ mpages_free(void) } /* + * For each mlink to the mpage, check whether the path looks like + * it is formatted, and if it does, check whether a source manual + * exists by the same name, ignoring the suffix. + * If both conditions hold, drop the mlink. + */ +static void +mlinks_undupe(struct mpage *mpage) +{ + char buf[PATH_MAX]; + struct mlink **prev; + struct mlink *mlink; + char *bufp; + + mpage->form = FORM_CAT; + prev = &mpage->mlinks; + while (NULL != (mlink = *prev)) { + if (FORM_CAT != mlink->dform) { + mpage->form = FORM_NONE; + goto nextlink; + } + if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) { + if (warnings) + say(mlink->file, "Filename too long"); + goto nextlink; + } + bufp = strstr(buf, "cat"); + assert(NULL != bufp); + memcpy(bufp, "man", 3); + if (NULL != (bufp = strrchr(buf, '.'))) + *++bufp = '\0'; + strlcat(buf, mlink->dsec, PATH_MAX); + if (NULL == ohash_find(&mlinks, + ohash_qlookup(&mlinks, buf))) + goto nextlink; + if (warnings) + say(mlink->file, "Man source exists: %s", buf); + if (use_all) + goto nextlink; + *prev = mlink->next; + mlink_free(mlink); + continue; +nextlink: + prev = &(*prev)->next; + } +} + +static int +mlink_check(struct mpage *mpage, struct mlink *mlink) +{ + int match; + + match = 1; + + /* + * Check whether the manual section given in a file + * agrees with the directory where the file is located. + * Some manuals have suffixes like (3p) on their + * section number either inside the file or in the + * directory name, some are linked into more than one + * section, like encrypt(1) = makekey(8). + */ + + if (FORM_SRC == mpage->form && + strcasecmp(mpage->sec, mlink->dsec)) { + match = 0; + say(mlink->file, "Section \"%s\" manual in %s directory", + mpage->sec, mlink->dsec); + } + + /* + * Manual page directories exist for each kernel + * architecture as returned by machine(1). + * However, many manuals only depend on the + * application architecture as returned by arch(1). + * For example, some (2/ARM) manuals are shared + * across the "armish" and "zaurus" kernel + * architectures. + * A few manuals are even shared across completely + * different architectures, for example fdformat(1) + * on amd64, i386, sparc, and sparc64. + */ + + if (strcasecmp(mpage->arch, mlink->arch)) { + match = 0; + say(mlink->file, "Architecture \"%s\" manual in " + "\"%s\" directory", mpage->arch, mlink->arch); + } + + if (strcasecmp(mpage->title, mlink->name)) + match = 0; + + return(match); +} + +/* * Run through the files in the global vector "mpages" * and add them to the database specified in "basedir". * @@ -875,14 +946,13 @@ mpages_merge(struct mchars *mc, struct mparse *mp, int { struct ohash title_table; struct ohash_info title_info, str_info; - char buf[PATH_MAX]; struct mpage *mpage; + struct mlink *mlink; struct mdoc *mdoc; struct man *man; struct title *title_entry; - char *bufp, *title_str; + char *title_str; const char *cp; - size_t sz; int match; unsigned int pslot, tslot; enum mandoclevel lvl; @@ -902,43 +972,16 @@ mpages_merge(struct mchars *mc, struct mparse *mp, int mpage = ohash_first(&mpages, &pslot); while (NULL != mpage) { - /* - * If we're a catpage (as defined by our path), then see - * if a manpage exists by the same name (ignoring the - * suffix). - * If it does, then we want to use it instead of our - * own. - */ - if ( ! use_all && FORM_CAT == mpage->mlinks->dform) { - sz = strlcpy(buf, mpage->mlinks->file, PATH_MAX); - if (sz >= PATH_MAX) { - if (warnings) - say(mpage->mlinks->file, - "Filename too long"); - mpage = ohash_next(&mpages, &pslot); - continue; - } - bufp = strstr(buf, "cat"); - assert(NULL != bufp); - memcpy(bufp, "man", 3); - if (NULL != (bufp = strrchr(buf, '.'))) - *++bufp = '\0'; - strlcat(buf, mpage->mlinks->dsec, PATH_MAX); - if (NULL != ohash_find(&mlinks, - ohash_qlookup(&mlinks, buf))) { - if (warnings) - say(mpage->mlinks->file, "Man " - "source exists: %s", buf); - mpage = ohash_next(&mpages, &pslot); - continue; - } + mlinks_undupe(mpage); + if (NULL == mpage->mlinks) { + mpage = ohash_next(&mpages, &pslot); + continue; } ohash_init(&strings, 6, &str_info); mparse_reset(mp); mdoc = NULL; man = NULL; - match = 1; /* * Try interpreting the file as mdoc(7) or man(7) @@ -979,50 +1022,18 @@ mpages_merge(struct mchars *mc, struct mparse *mp, int mandoc_strdup(mpage->mlinks->name); } - /* - * Check whether the manual section given in a file - * agrees with the directory where the file is located. - * Some manuals have suffixes like (3p) on their - * section number either inside the file or in the - * directory name, some are linked into more than one - * section, like encrypt(1) = makekey(8). Do not skip - * manuals for such reasons. - */ - if (warnings && !use_all && FORM_SRC == mpage->form && - strcasecmp(mpage->sec, mpage->mlinks->dsec)) { - match = 0; - say(mpage->mlinks->file, "Section \"%s\" " - "manual in %s directory", - mpage->sec, mpage->mlinks->dsec); - } + for (mlink = mpage->mlinks; mlink; mlink = mlink->next) + putkey(mpage, mlink->name, TYPE_Nm); - /* - * Manual page directories exist for each kernel - * architecture as returned by machine(1). - * However, many manuals only depend on the - * application architecture as returned by arch(1). - * For example, some (2/ARM) manuals are shared - * across the "armish" and "zaurus" kernel - * architectures. - * A few manuals are even shared across completely - * different architectures, for example fdformat(1) - * on amd64, i386, sparc, and sparc64. - * Thus, warn about architecture mismatches, - * but don't skip manuals for this reason. - */ - if (warnings && !use_all && - strcasecmp(mpage->arch, mpage->mlinks->arch)) { + if (warnings && !use_all) { match = 0; - say(mpage->mlinks->file, "Architecture \"%s\" " - "manual in \"%s\" directory", - mpage->arch, mpage->mlinks->arch); - } - if (warnings && !use_all && - strcasecmp(mpage->title, mpage->mlinks->name)) - match = 0; + for (mlink = mpage->mlinks; mlink; + mlink = mlink->next) + if (mlink_check(mpage, mlink)) + match = 1; + } else + match = 1; - putkey(mpage, mpage->mlinks->name, TYPE_Nm); - if (NULL != mdoc) { if (NULL != (cp = mdoc_meta(mdoc)->name)) putkey(mpage, cp, TYPE_Nm); @@ -1445,21 +1456,10 @@ parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_n if (end > start) putkeys(mpage, start, end - start + 1, TYPE_In); - return(1); + return(0); } static int -parse_mdoc_In(struct mpage *mpage, const struct mdoc_node *n) -{ - - if (NULL != n->child && MDOC_TEXT == n->child->type) - return(0); - - putkey(mpage, n->child->string, TYPE_In); - return(1); -} - -static int parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n) { const char *cp; @@ -1493,17 +1493,6 @@ parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_n } static int -parse_mdoc_St(struct mpage *mpage, const struct mdoc_node *n) -{ - - if (NULL == n->child || MDOC_TEXT != n->child->type) - return(0); - - putkey(mpage, n->child->string, TYPE_St); - return(1); -} - -static int parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n) { char *cp; @@ -1560,12 +1549,8 @@ static int parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n) { - if (SEC_NAME == n->sec) - return(1); - else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type) - return(0); - - return(1); + return(SEC_NAME == n->sec || + (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)); } static int @@ -1738,7 +1723,6 @@ utf8key(struct mchars *mc, struct str *key) /* Read past the slash. */ val++; - u = 0; /* * Parse the escape sequence and see if it's a @@ -1786,6 +1770,7 @@ utf8key(struct mchars *mc, struct str *key) static void dbindex(const struct mpage *mpage, struct mchars *mc) { + struct mlink *mlink; struct str *key; const char *desc; int64_t recno; @@ -1825,13 +1810,16 @@ dbindex(const struct mpage *mpage, struct mchars *mc) recno = sqlite3_last_insert_rowid(db); sqlite3_reset(stmts[STMT_INSERT_PAGE]); - i = 1; - SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mpage->mlinks->dsec); - SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mpage->mlinks->arch); - SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mpage->mlinks->file); - 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) { + 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 (key = ohash_first(&strings, &slot); NULL != key; key = ohash_next(&strings, &slot)) { @@ -1967,6 +1955,7 @@ dbopen(int real) ");\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" @@ -1999,7 +1988,7 @@ prepare_statements: "(file,sec,arch,desc,form) VALUES (?,?,?,?,?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL); sql = "INSERT INTO mlinks " - "(sec,arch,name,pageid) VALUES (?,?,?,?)"; + "(file,sec,arch,name,pageid) VALUES (?,?,?,?,?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL); sql = "INSERT INTO keys " "(bits,key,pageid) VALUES (?,?,?)";