=================================================================== RCS file: /cvs/mandoc/mandocdb.c,v retrieving revision 1.62 retrieving revision 1.74 diff -u -p -r1.62 -r1.74 --- mandoc/mandocdb.c 2013/06/06 02:40:37 1.62 +++ mandoc/mandocdb.c 2013/10/27 16:09:44 1.74 @@ -1,4 +1,4 @@ -/* $Id: mandocdb.c,v 1.62 2013/06/06 02:40:37 schwarze Exp $ */ +/* $Id: mandocdb.c,v 1.74 2013/10/27 16:09:44 schwarze Exp $ */ /* * Copyright (c) 2011, 2012 Kristaps Dzonsons * Copyright (c) 2011, 2012, 2013 Ingo Schwarze @@ -23,6 +23,7 @@ #include #include +#include #include #include #include @@ -83,7 +84,6 @@ enum form { struct str { char *utf8; /* key in UTF-8 form */ const struct of *of; /* if set, the owning parse */ - struct str *next; /* next in owning parse sequence */ uint64_t mask; /* bitmask in sequence */ char key[]; /* the string itself */ }; @@ -99,13 +99,18 @@ struct of { enum form dform; /* path-cued form */ enum form sform; /* suffix-cued form */ char file[PATH_MAX]; /* filename rel. to manpath */ - const char *desc; /* parsed description */ - const char *sec; /* suffix-cued section (or empty) */ - const char *dsec; /* path-cued section (or empty) */ - const char *arch; /* path-cued arch. (or empty) */ - const char *name; /* name (from filename) (not empty) */ + char *desc; /* parsed description */ + char *name; /* name (from filename) (not empty) */ + char *sec; /* suffix-cued section (or empty) */ + char *dsec; /* path-cued section (or empty) */ + char *arch; /* path-cued arch. (or empty) */ }; +struct title { + char *title; /* name(sec/arch) given inside the file */ + char *file; /* file name in case of mismatch */ +}; + enum stmt { STMT_DELETE = 0, /* delete manpage */ STMT_INSERT_DOC, /* insert manpage */ @@ -118,8 +123,6 @@ typedef int (*mdoc_fp)(struct of *, const struct mdoc_ struct mdoc_handler { mdoc_fp fp; /* optional handler */ uint64_t mask; /* set unless handler returns 0 */ - int flags; /* for use by pmdoc_node */ -#define MDOCF_CHILD 0x01 /* automatically index child nodes */ }; static void dbclose(int); @@ -129,7 +132,6 @@ static void dbprune(void); static void fileadd(struct of *); static int filecheck(const char *); static void filescan(const char *); -static struct str *hashget(const char *, size_t); static void *hash_alloc(size_t, void *); static void hash_free(void *, size_t, void *); static void *hash_halloc(size_t, void *); @@ -138,7 +140,7 @@ static int inocheck(const struct stat *); static void ofadd(int, const char *, const char *, const char *, const char *, const char *, const struct stat *); static void offree(void); -static void ofmerge(struct mchars *, struct mparse *); +static void ofmerge(struct mchars *, struct mparse *, int); static void parse_catpage(struct of *); static void parse_man(struct of *, const struct man_node *); static void parse_mdoc(struct of *, const struct mdoc_node *); @@ -156,17 +158,13 @@ static int set_basedir(const char *); static void putkey(const struct of *, const char *, uint64_t); static void putkeys(const struct of *, - const char *, int, uint64_t); + const char *, size_t, uint64_t); static void putmdockey(const struct of *, const struct mdoc_node *, uint64_t); static void say(const char *, const char *, ...); -static char *stradd(const char *); -static char *straddbuf(const char *, size_t); static int treescan(void); static size_t utf8(unsigned int, char [7]); static void utf8key(struct mchars *, struct str *); -static void wordaddbuf(const struct of *, - const char *, size_t, uint64_t); static char *progname; static int use_all; /* use all found files */ @@ -180,161 +178,154 @@ static struct ohash inos; /* table of inodes/devices static struct ohash filenames; /* table of filenames */ static struct ohash strings; /* table of all strings */ static struct of *ofs = NULL; /* vector of files to parse */ -static struct str *words = NULL; /* word list in current parse */ static sqlite3 *db = NULL; /* current database */ static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */ static const struct mdoc_handler mdocs[MDOC_MAX] = { - { NULL, 0, 0 }, /* Ap */ - { NULL, 0, 0 }, /* Dd */ - { NULL, 0, 0 }, /* Dt */ - { NULL, 0, 0 }, /* Os */ - { parse_mdoc_Sh, TYPE_Sh, MDOCF_CHILD }, /* Sh */ - { parse_mdoc_head, TYPE_Ss, MDOCF_CHILD }, /* Ss */ - { NULL, 0, 0 }, /* Pp */ - { NULL, 0, 0 }, /* D1 */ - { NULL, 0, 0 }, /* Dl */ - { NULL, 0, 0 }, /* Bd */ - { NULL, 0, 0 }, /* Ed */ - { NULL, 0, 0 }, /* Bl */ - { NULL, 0, 0 }, /* El */ - { NULL, 0, 0 }, /* It */ - { NULL, 0, 0 }, /* Ad */ - { NULL, TYPE_An, MDOCF_CHILD }, /* An */ - { NULL, TYPE_Ar, MDOCF_CHILD }, /* Ar */ - { NULL, TYPE_Cd, MDOCF_CHILD }, /* Cd */ - { NULL, TYPE_Cm, MDOCF_CHILD }, /* Cm */ - { NULL, TYPE_Dv, MDOCF_CHILD }, /* Dv */ - { NULL, TYPE_Er, MDOCF_CHILD }, /* Er */ - { NULL, TYPE_Ev, MDOCF_CHILD }, /* Ev */ - { NULL, 0, 0 }, /* Ex */ - { NULL, TYPE_Fa, MDOCF_CHILD }, /* Fa */ - { parse_mdoc_Fd, TYPE_In, 0 }, /* Fd */ - { NULL, TYPE_Fl, MDOCF_CHILD }, /* Fl */ - { parse_mdoc_Fn, 0, 0 }, /* Fn */ - { NULL, TYPE_Ft, MDOCF_CHILD }, /* Ft */ - { NULL, TYPE_Ic, MDOCF_CHILD }, /* Ic */ - { parse_mdoc_In, TYPE_In, MDOCF_CHILD }, /* In */ - { NULL, TYPE_Li, MDOCF_CHILD }, /* Li */ - { parse_mdoc_Nd, TYPE_Nd, MDOCF_CHILD }, /* Nd */ - { parse_mdoc_Nm, TYPE_Nm, MDOCF_CHILD }, /* Nm */ - { NULL, 0, 0 }, /* Op */ - { NULL, 0, 0 }, /* Ot */ - { NULL, TYPE_Pa, MDOCF_CHILD }, /* Pa */ - { NULL, 0, 0 }, /* Rv */ - { parse_mdoc_St, TYPE_St, 0 }, /* St */ - { NULL, TYPE_Va, MDOCF_CHILD }, /* Va */ - { parse_mdoc_body, TYPE_Va, MDOCF_CHILD }, /* Vt */ - { parse_mdoc_Xr, TYPE_Xr, 0 }, /* Xr */ - { NULL, 0, 0 }, /* %A */ - { NULL, 0, 0 }, /* %B */ - { NULL, 0, 0 }, /* %D */ - { NULL, 0, 0 }, /* %I */ - { NULL, 0, 0 }, /* %J */ - { NULL, 0, 0 }, /* %N */ - { NULL, 0, 0 }, /* %O */ - { NULL, 0, 0 }, /* %P */ - { NULL, 0, 0 }, /* %R */ - { NULL, 0, 0 }, /* %T */ - { NULL, 0, 0 }, /* %V */ - { NULL, 0, 0 }, /* Ac */ - { NULL, 0, 0 }, /* Ao */ - { NULL, 0, 0 }, /* Aq */ - { NULL, TYPE_At, MDOCF_CHILD }, /* At */ - { NULL, 0, 0 }, /* Bc */ - { NULL, 0, 0 }, /* Bf */ - { NULL, 0, 0 }, /* Bo */ - { NULL, 0, 0 }, /* Bq */ - { NULL, TYPE_Bsx, MDOCF_CHILD }, /* Bsx */ - { NULL, TYPE_Bx, MDOCF_CHILD }, /* Bx */ - { NULL, 0, 0 }, /* Db */ - { NULL, 0, 0 }, /* Dc */ - { NULL, 0, 0 }, /* Do */ - { NULL, 0, 0 }, /* Dq */ - { NULL, 0, 0 }, /* Ec */ - { NULL, 0, 0 }, /* Ef */ - { NULL, TYPE_Em, MDOCF_CHILD }, /* Em */ - { NULL, 0, 0 }, /* Eo */ - { NULL, TYPE_Fx, MDOCF_CHILD }, /* Fx */ - { NULL, TYPE_Ms, MDOCF_CHILD }, /* Ms */ - { NULL, 0, 0 }, /* No */ - { NULL, 0, 0 }, /* Ns */ - { NULL, TYPE_Nx, MDOCF_CHILD }, /* Nx */ - { NULL, TYPE_Ox, MDOCF_CHILD }, /* Ox */ - { NULL, 0, 0 }, /* Pc */ - { NULL, 0, 0 }, /* Pf */ - { NULL, 0, 0 }, /* Po */ - { NULL, 0, 0 }, /* Pq */ - { NULL, 0, 0 }, /* Qc */ - { NULL, 0, 0 }, /* Ql */ - { NULL, 0, 0 }, /* Qo */ - { NULL, 0, 0 }, /* Qq */ - { NULL, 0, 0 }, /* Re */ - { NULL, 0, 0 }, /* Rs */ - { NULL, 0, 0 }, /* Sc */ - { NULL, 0, 0 }, /* So */ - { NULL, 0, 0 }, /* Sq */ - { NULL, 0, 0 }, /* Sm */ - { NULL, 0, 0 }, /* Sx */ - { NULL, TYPE_Sy, MDOCF_CHILD }, /* Sy */ - { NULL, TYPE_Tn, MDOCF_CHILD }, /* Tn */ - { NULL, 0, 0 }, /* Ux */ - { NULL, 0, 0 }, /* Xc */ - { NULL, 0, 0 }, /* Xo */ - { parse_mdoc_head, TYPE_Fn, 0 }, /* Fo */ - { NULL, 0, 0 }, /* Fc */ - { NULL, 0, 0 }, /* Oo */ - { NULL, 0, 0 }, /* Oc */ - { NULL, 0, 0 }, /* Bk */ - { NULL, 0, 0 }, /* Ek */ - { NULL, 0, 0 }, /* Bt */ - { NULL, 0, 0 }, /* Hf */ - { NULL, 0, 0 }, /* Fr */ - { NULL, 0, 0 }, /* Ud */ - { NULL, TYPE_Lb, MDOCF_CHILD }, /* Lb */ - { NULL, 0, 0 }, /* Lp */ - { NULL, TYPE_Lk, MDOCF_CHILD }, /* Lk */ - { NULL, TYPE_Mt, MDOCF_CHILD }, /* Mt */ - { NULL, 0, 0 }, /* Brq */ - { NULL, 0, 0 }, /* Bro */ - { NULL, 0, 0 }, /* Brc */ - { NULL, 0, 0 }, /* %C */ - { NULL, 0, 0 }, /* Es */ - { NULL, 0, 0 }, /* En */ - { NULL, TYPE_Dx, MDOCF_CHILD }, /* Dx */ - { NULL, 0, 0 }, /* %Q */ - { NULL, 0, 0 }, /* br */ - { NULL, 0, 0 }, /* sp */ - { NULL, 0, 0 }, /* %U */ - { NULL, 0, 0 }, /* Ta */ + { NULL, 0 }, /* Ap */ + { NULL, 0 }, /* Dd */ + { NULL, 0 }, /* Dt */ + { NULL, 0 }, /* Os */ + { parse_mdoc_Sh, TYPE_Sh }, /* Sh */ + { parse_mdoc_head, TYPE_Ss }, /* Ss */ + { NULL, 0 }, /* Pp */ + { NULL, 0 }, /* D1 */ + { NULL, 0 }, /* Dl */ + { NULL, 0 }, /* Bd */ + { NULL, 0 }, /* Ed */ + { NULL, 0 }, /* Bl */ + { NULL, 0 }, /* El */ + { NULL, 0 }, /* It */ + { NULL, 0 }, /* Ad */ + { NULL, TYPE_An }, /* An */ + { NULL, TYPE_Ar }, /* Ar */ + { NULL, TYPE_Cd }, /* Cd */ + { NULL, TYPE_Cm }, /* Cm */ + { NULL, TYPE_Dv }, /* Dv */ + { NULL, TYPE_Er }, /* Er */ + { NULL, TYPE_Ev }, /* Ev */ + { NULL, 0 }, /* Ex */ + { NULL, TYPE_Fa }, /* Fa */ + { parse_mdoc_Fd, 0 }, /* Fd */ + { NULL, TYPE_Fl }, /* Fl */ + { parse_mdoc_Fn, 0 }, /* Fn */ + { NULL, TYPE_Ft }, /* Ft */ + { NULL, TYPE_Ic }, /* Ic */ + { parse_mdoc_In, TYPE_In }, /* In */ + { NULL, TYPE_Li }, /* Li */ + { parse_mdoc_Nd, TYPE_Nd }, /* Nd */ + { parse_mdoc_Nm, TYPE_Nm }, /* Nm */ + { NULL, 0 }, /* Op */ + { NULL, 0 }, /* Ot */ + { NULL, TYPE_Pa }, /* Pa */ + { NULL, 0 }, /* Rv */ + { parse_mdoc_St, 0 }, /* St */ + { NULL, TYPE_Va }, /* Va */ + { parse_mdoc_body, TYPE_Va }, /* Vt */ + { parse_mdoc_Xr, 0 }, /* Xr */ + { NULL, 0 }, /* %A */ + { NULL, 0 }, /* %B */ + { NULL, 0 }, /* %D */ + { NULL, 0 }, /* %I */ + { NULL, 0 }, /* %J */ + { NULL, 0 }, /* %N */ + { NULL, 0 }, /* %O */ + { NULL, 0 }, /* %P */ + { NULL, 0 }, /* %R */ + { NULL, 0 }, /* %T */ + { NULL, 0 }, /* %V */ + { NULL, 0 }, /* Ac */ + { NULL, 0 }, /* Ao */ + { NULL, 0 }, /* Aq */ + { NULL, TYPE_At }, /* At */ + { NULL, 0 }, /* Bc */ + { NULL, 0 }, /* Bf */ + { NULL, 0 }, /* Bo */ + { NULL, 0 }, /* Bq */ + { NULL, TYPE_Bsx }, /* Bsx */ + { NULL, TYPE_Bx }, /* Bx */ + { NULL, 0 }, /* Db */ + { NULL, 0 }, /* Dc */ + { NULL, 0 }, /* Do */ + { NULL, 0 }, /* Dq */ + { NULL, 0 }, /* Ec */ + { NULL, 0 }, /* Ef */ + { NULL, TYPE_Em }, /* Em */ + { NULL, 0 }, /* Eo */ + { NULL, TYPE_Fx }, /* Fx */ + { NULL, TYPE_Ms }, /* Ms */ + { NULL, 0 }, /* No */ + { NULL, 0 }, /* Ns */ + { NULL, TYPE_Nx }, /* Nx */ + { NULL, TYPE_Ox }, /* Ox */ + { NULL, 0 }, /* Pc */ + { NULL, 0 }, /* Pf */ + { NULL, 0 }, /* Po */ + { NULL, 0 }, /* Pq */ + { NULL, 0 }, /* Qc */ + { NULL, 0 }, /* Ql */ + { NULL, 0 }, /* Qo */ + { NULL, 0 }, /* Qq */ + { NULL, 0 }, /* Re */ + { NULL, 0 }, /* Rs */ + { NULL, 0 }, /* Sc */ + { NULL, 0 }, /* So */ + { NULL, 0 }, /* Sq */ + { NULL, 0 }, /* Sm */ + { NULL, 0 }, /* Sx */ + { NULL, TYPE_Sy }, /* Sy */ + { NULL, TYPE_Tn }, /* Tn */ + { NULL, 0 }, /* Ux */ + { NULL, 0 }, /* Xc */ + { NULL, 0 }, /* Xo */ + { parse_mdoc_head, 0 }, /* Fo */ + { NULL, 0 }, /* Fc */ + { NULL, 0 }, /* Oo */ + { NULL, 0 }, /* Oc */ + { NULL, 0 }, /* Bk */ + { NULL, 0 }, /* Ek */ + { NULL, 0 }, /* Bt */ + { NULL, 0 }, /* Hf */ + { NULL, 0 }, /* Fr */ + { NULL, 0 }, /* Ud */ + { NULL, TYPE_Lb }, /* Lb */ + { NULL, 0 }, /* Lp */ + { NULL, TYPE_Lk }, /* Lk */ + { NULL, TYPE_Mt }, /* Mt */ + { NULL, 0 }, /* Brq */ + { NULL, 0 }, /* Bro */ + { NULL, 0 }, /* Brc */ + { NULL, 0 }, /* %C */ + { NULL, 0 }, /* Es */ + { NULL, 0 }, /* En */ + { NULL, TYPE_Dx }, /* Dx */ + { NULL, 0 }, /* %Q */ + { NULL, 0 }, /* br */ + { NULL, 0 }, /* sp */ + { NULL, 0 }, /* %U */ + { NULL, 0 }, /* Ta */ }; int main(int argc, char *argv[]) { int ch, i; - unsigned int index; size_t j, sz; const char *path_arg; - struct str *s; struct mchars *mc; struct manpaths dirs; struct mparse *mp; - struct ohash_info ino_info, filename_info, str_info; + struct ohash_info ino_info, filename_info; memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *)); memset(&dirs, 0, sizeof(struct manpaths)); - ino_info.halloc = filename_info.halloc = - str_info.halloc = hash_halloc; - ino_info.hfree = filename_info.hfree = - str_info.hfree = hash_free; - ino_info.alloc = filename_info.alloc = - str_info.alloc = hash_alloc; + ino_info.alloc = filename_info.alloc = hash_alloc; + ino_info.halloc = filename_info.halloc = hash_halloc; + ino_info.hfree = filename_info.hfree = hash_free; ino_info.key_offset = offsetof(struct of, id); filename_info.key_offset = offsetof(struct of, file); - str_info.key_offset = offsetof(struct str, key); progname = strrchr(argv[0], '/'); if (progname == NULL) @@ -408,7 +399,6 @@ main(int argc, char *argv[]) MANDOCLEVEL_FATAL, NULL, NULL, NULL); mc = mchars_alloc(); - ohash_init(&strings, 6, &str_info); ohash_init(&inos, 6, &ino_info); ohash_init(&filenames, 6, &filename_info); @@ -432,7 +422,7 @@ main(int argc, char *argv[]) if (OP_TEST != op) dbprune(); if (OP_DELETE != op) - ofmerge(mc, mp); + ofmerge(mc, mp, 0); dbclose(1); } else { /* @@ -450,9 +440,8 @@ main(int argc, char *argv[]) manpath_parse(&dirs, path_arg, NULL, NULL); /* - * First scan the tree rooted at a base directory. - * Then whak its database (if one exists), parse, and - * build up the database. + * First scan the tree rooted at a base directory, then + * build a new database and finally move it into place. * Ignore zero-length directories and strip trailing * slashes. */ @@ -462,6 +451,12 @@ main(int argc, char *argv[]) dirs.paths[j][--sz] = '\0'; if (0 == sz) continue; + + if (j) { + ohash_init(&inos, 6, &ino_info); + ohash_init(&filenames, 6, &filename_info); + } + if (0 == set_basedir(dirs.paths[j])) goto out; if (0 == treescan()) @@ -471,22 +466,14 @@ main(int argc, char *argv[]) if (0 == dbopen(0)) goto out; - /* - * Since we're opening up a new database, we can - * turn off synchronous mode for much better - * performance. - */ -#ifndef __APPLE__ - SQL_EXEC("PRAGMA synchronous = OFF"); -#endif - - ofmerge(mc, mp); + ofmerge(mc, mp, warnings && !use_all); dbclose(0); - offree(); - ohash_delete(&inos); - ohash_init(&inos, 6, &ino_info); - ohash_delete(&filenames); - ohash_init(&filenames, 6, &filename_info); + + if (j + 1 < dirs.sz) { + ohash_delete(&inos); + ohash_delete(&filenames); + offree(); + } } } out: @@ -494,13 +481,6 @@ out: manpath_free(&dirs); mchars_free(mc); mparse_free(mp); - for (s = ohash_first(&strings, &index); - NULL != s; s = ohash_next(&strings, &index)) { - if (s->utf8 != s->key) - free(s->utf8); - free(s); - } - ohash_delete(&strings); ohash_delete(&inos); ohash_delete(&filenames); offree(); @@ -538,7 +518,7 @@ treescan(void) FTSENT *ff; int dform; char *sec; - const char *dsec, *arch, *cp, *name, *path; + const char *dsec, *arch, *cp, *path; const char *argv[2]; argv[0] = "."; @@ -606,13 +586,10 @@ treescan(void) if (warnings) say(path, "Wrong filename suffix"); continue; - } else { + } else sec[-1] = '\0'; - sec = stradd(sec); - } - name = stradd(ff->fts_name); - ofadd(dform, path, - name, dsec, sec, arch, ff->fts_statp); + ofadd(dform, path, ff->fts_name, dsec, sec, + arch, ff->fts_statp); continue; } else if (FTS_D != ff->fts_info && FTS_DP != ff->fts_info) { @@ -639,10 +616,10 @@ treescan(void) if (0 == strncmp(cp, "man", 3)) { dform = FORM_SRC; - dsec = stradd(cp + 3); + dsec = cp + 3; } else if (0 == strncmp(cp, "cat", 3)) { dform = FORM_CAT; - dsec = stradd(cp + 3); + dsec = cp + 3; } if (NULL != dsec || use_all) @@ -659,7 +636,7 @@ treescan(void) */ arch = NULL; if (FTS_DP != ff->fts_info && NULL != dsec) - arch = stradd(ff->fts_name); + arch = ff->fts_name; break; default: if (FTS_DP == ff->fts_info || use_all) @@ -708,7 +685,7 @@ filescan(const char *file) exitcode = (int)MANDOCLEVEL_BADARG; say(file, NULL); return; - } else if (strstr(buf, basedir) != buf) { + } else if (OP_TEST != op && strstr(buf, basedir) != buf) { exitcode = (int)MANDOCLEVEL_BADARG; say("", "%s: outside base directory", buf); return; @@ -785,10 +762,9 @@ filescan(const char *file) static int filecheck(const char *name) { - unsigned int index; - index = ohash_qlookup(&filenames, name); - return(NULL != ohash_find(&filenames, index)); + return(NULL != ohash_find(&filenames, + ohash_qlookup(&filenames, name))); } /* @@ -798,11 +774,11 @@ filecheck(const char *name) static void fileadd(struct of *of) { - unsigned int index; + unsigned int slot; - index = ohash_qlookup(&filenames, of->file); - assert(NULL == ohash_find(&filenames, index)); - ohash_insert(&filenames, index, of); + slot = ohash_qlookup(&filenames, of->file); + assert(NULL == ohash_find(&filenames, slot)); + ohash_insert(&filenames, slot, of); } /* @@ -813,15 +789,13 @@ inocheck(const struct stat *st) { struct id id; uint32_t hash; - unsigned int index; memset(&id, 0, sizeof(id)); id.ino = hash = st->st_ino; id.dev = st->st_dev; - index = ohash_lookup_memory - (&inos, (char *)&id, sizeof(id), hash); - return(NULL != ohash_find(&inos, index)); + return(NULL != ohash_find(&inos, ohash_lookup_memory( + &inos, (char *)&id, sizeof(id), hash))); } /* @@ -833,15 +807,15 @@ static void inoadd(const struct stat *st, struct of *of) { uint32_t hash; - unsigned int index; + unsigned int slot; of->id.ino = hash = st->st_ino; of->id.dev = st->st_dev; - index = ohash_lookup_memory + slot = ohash_lookup_memory (&inos, (char *)&of->id, sizeof(of->id), hash); - assert(NULL == ohash_find(&inos, index)); - ohash_insert(&inos, index, of); + assert(NULL == ohash_find(&inos, slot)); + ohash_insert(&inos, slot, of); } static void @@ -862,20 +836,20 @@ ofadd(int dform, const char *file, const char *name, c if (NULL == arch) arch = ""; - sform = FORM_NONE; - if (NULL != sec && *sec <= '9' && *sec >= '1') - sform = FORM_SRC; - else if (NULL != sec && *sec == '0') { + if ('0' == *sec) { sec = dsec; sform = FORM_CAT; - } + } else if ('1' <= *sec && '9' >= *sec) + sform = FORM_SRC; + else + sform = FORM_NONE; of = mandoc_calloc(1, sizeof(struct of)); strlcpy(of->file, file, PATH_MAX); - of->name = name; - of->sec = sec; - of->dsec = dsec; - of->arch = arch; + of->name = mandoc_strdup(name); + of->sec = mandoc_strdup(sec); + of->dsec = mandoc_strdup(dsec); + of->arch = mandoc_strdup(arch); of->sform = sform; of->dform = dform; of->next = ofs; @@ -897,6 +871,10 @@ offree(void) while (NULL != (of = ofs)) { ofs = of->next; + free(of->name); + free(of->sec); + free(of->dsec); + free(of->arch); free(of); } } @@ -909,18 +887,36 @@ offree(void) * and filename to determine whether the file is parsable or not. */ static void -ofmerge(struct mchars *mc, struct mparse *mp) +ofmerge(struct mchars *mc, struct mparse *mp, int check_reachable) { - int form; - size_t sz; - struct mdoc *mdoc; - struct man *man; - char buf[PATH_MAX]; - char *bufp; - const char *msec, *march, *mtitle, *cp; - struct of *of; - enum mandoclevel lvl; + struct ohash title_table; + struct ohash_info title_info, str_info; + char buf[PATH_MAX]; + struct of *of; + struct mdoc *mdoc; + struct man *man; + struct title *title_entry; + char *bufp, *title_str; + const char *msec, *march, *mtitle, *cp; + size_t sz; + int form; + int match; + unsigned int slot; + enum mandoclevel lvl; + str_info.alloc = hash_alloc; + str_info.halloc = hash_halloc; + str_info.hfree = hash_free; + str_info.key_offset = offsetof(struct str, key); + + if (check_reachable) { + title_info.alloc = hash_alloc; + title_info.halloc = hash_halloc; + title_info.hfree = hash_free; + title_info.key_offset = offsetof(struct title, title); + ohash_init(&title_table, 6, &title_info); + } + for (of = ofs; NULL != of; of = of->next) { /* * If we're a catpage (as defined by our path), then see @@ -950,7 +946,7 @@ ofmerge(struct mchars *mc, struct mparse *mp) } } - words = NULL; + ohash_init(&strings, 6, &str_info); mparse_reset(mp); mdoc = NULL; man = NULL; @@ -958,6 +954,7 @@ ofmerge(struct mchars *mc, struct mparse *mp) msec = of->dsec; march = of->arch; mtitle = of->name; + match = 1; /* * Try interpreting the file as mdoc(7) or man(7) @@ -999,10 +996,12 @@ ofmerge(struct mchars *mc, struct mparse *mp) * manuals for such reasons. */ if (warnings && !use_all && form && - strcasecmp(msec, of->dsec)) + strcasecmp(msec, of->dsec)) { + match = 0; say(of->file, "Section \"%s\" " "manual in %s directory", msec, of->dsec); + } /* * Manual page directories exist for each kernel @@ -1018,24 +1017,76 @@ ofmerge(struct mchars *mc, struct mparse *mp) * Thus, warn about architecture mismatches, * but don't skip manuals for this reason. */ - if (warnings && !use_all && strcasecmp(march, of->arch)) + if (warnings && !use_all && strcasecmp(march, of->arch)) { + match = 0; say(of->file, "Architecture \"%s\" " "manual in \"%s\" directory", march, of->arch); + } + if (warnings && !use_all && strcasecmp(mtitle, of->name)) + match = 0; putkey(of, of->name, TYPE_Nm); if (NULL != mdoc) { if (NULL != (cp = mdoc_meta(mdoc)->name)) putkey(of, cp, TYPE_Nm); + assert(NULL == of->desc); parse_mdoc(of, mdoc_node(mdoc)); + putkey(of, NULL != of->desc ? + of->desc : of->name, TYPE_Nd); } else if (NULL != man) parse_man(of, man_node(man)); else parse_catpage(of); + /* + * Build a title string for the file. If it matches + * the location of the file, remember the title as + * found; else, remember it as missing. + */ + + if (check_reachable) { + if (-1 == asprintf(&title_str, "%s(%s%s%s)", mtitle, + msec, '\0' == *march ? "" : "/", march)) { + perror(NULL); + exit((int)MANDOCLEVEL_SYSERR); + } + slot = ohash_qlookup(&title_table, title_str); + title_entry = ohash_find(&title_table, slot); + if (NULL == title_entry) { + title_entry = mandoc_malloc( + sizeof(struct title)); + title_entry->title = title_str; + title_entry->file = mandoc_strdup( + match ? "" : of->file); + ohash_insert(&title_table, slot, + title_entry); + } else { + if (match) + *title_entry->file = '\0'; + free(title_str); + } + } + dbindex(mc, form, of); + ohash_delete(&strings); } + + if (check_reachable) { + title_entry = ohash_first(&title_table, &slot); + while (NULL != title_entry) { + if ('\0' != *title_entry->file) + say(title_entry->file, + "Probably unreachable, title is %s", + title_entry->title); + free(title_entry->title); + free(title_entry->file); + free(title_entry); + title_entry = ohash_next(&title_table, &slot); + } + ohash_delete(&title_table); + } } static void @@ -1101,6 +1152,8 @@ parse_catpage(struct of *of) if (NULL == title || '\0' == *title) { if (warnings) say(of->file, "Cannot find NAME section"); + assert(NULL == of->desc); + of->desc = mandoc_strdup(of->name); putkey(of, of->name, TYPE_Nd); fclose(stream); free(title); @@ -1139,8 +1192,9 @@ parse_catpage(struct of *of) plen -= 2; } - of->desc = stradd(p); - putkey(of, p, TYPE_Nd); + assert(NULL == of->desc); + of->desc = mandoc_strdup(p); + putkey(of, of->desc, TYPE_Nd); fclose(stream); free(title); } @@ -1153,20 +1207,10 @@ putkey(const struct of *of, const char *value, uint64_ { assert(NULL != value); - wordaddbuf(of, value, strlen(value), type); + putkeys(of, value, strlen(value), type); } /* - * Like putkey() but for unterminated strings. - */ -static void -putkeys(const struct of *of, const char *value, int sz, uint64_t type) -{ - - wordaddbuf(of, value, sz, type); -} - -/* * Grok all nodes at or below a certain mdoc node into putkey(). */ static void @@ -1314,8 +1358,8 @@ parse_man(struct of *of, const struct man_node *n) start++; assert(NULL == of->desc); - of->desc = stradd(start); - putkey(of, start, TYPE_Nd); + of->desc = mandoc_strdup(start); + putkey(of, of->desc, TYPE_Nd); free(title); return; } @@ -1344,8 +1388,7 @@ parse_mdoc(struct of *of, const struct mdoc_node *n) if (NULL != mdocs[n->tok].fp) if (0 == (*mdocs[n->tok].fp)(of, n)) break; - - if (MDOCF_CHILD & mdocs[n->tok].flags) + if (mdocs[n->tok].mask) putmdockey(of, n->child, mdocs[n->tok].mask); break; default: @@ -1458,19 +1501,29 @@ parse_mdoc_St(struct of *of, const struct mdoc_node *n static int parse_mdoc_Xr(struct of *of, const struct mdoc_node *n) { + char *cp; if (NULL == (n = n->child)) return(0); - putkey(of, n->string, TYPE_Xr); - return(1); + if (NULL == n->next) { + putkey(of, n->string, TYPE_Xr); + return(0); + } + + if (-1 == asprintf(&cp, "%s(%s)", n->string, n->next->string)) { + perror(NULL); + exit((int)MANDOCLEVEL_SYSERR); + } + putkey(of, cp, TYPE_Xr); + free(cp); + return(0); } static int parse_mdoc_Nd(struct of *of, const struct mdoc_node *n) { size_t sz; - char *sv, *desc; if (MDOC_BODY != n->type) return(0); @@ -1480,25 +1533,20 @@ parse_mdoc_Nd(struct of *of, const struct mdoc_node *n * into the document table. */ - desc = NULL; for (n = n->child; NULL != n; n = n->next) { if (MDOC_TEXT == n->type) { - sz = strlen(n->string) + 1; - if (NULL != (sv = desc)) - sz += strlen(desc) + 1; - desc = mandoc_realloc(desc, sz); - if (NULL != sv) - strlcat(desc, " ", sz); - else - *desc = '\0'; - strlcat(desc, n->string, sz); + if (NULL != of->desc) { + sz = strlen(of->desc) + + strlen(n->string) + 2; + of->desc = mandoc_realloc(of->desc, sz); + strlcat(of->desc, " ", sz); + strlcat(of->desc, n->string, sz); + } else + of->desc = mandoc_strdup(n->string); } if (NULL != n->child) parse_mdoc_Nd(of, n); } - - of->desc = NULL != desc ? stradd(desc) : NULL; - free(desc); return(1); } @@ -1536,76 +1584,23 @@ parse_mdoc_body(struct of *of, const struct mdoc_node } /* - * See straddbuf(). + * Add a string to the hash table for the current manual. + * Each string has a bitmask telling which macros it belongs to. + * When we finish the manual, we'll dump the table. */ -static char * -stradd(const char *cp) -{ - - return(straddbuf(cp, strlen(cp))); -} - -/* - * This looks up or adds a string to the string table. - * The string table is a table of all strings encountered during parse - * or file scan. - * In using it, we avoid having thousands of (e.g.) "cat1" string - * allocations for the "of" table. - * We also have a layer atop the string table for keeping track of words - * in a parse sequence (see wordaddbuf()). - */ -static char * -straddbuf(const char *cp, size_t sz) -{ - struct str *s; - unsigned int index; - const char *end; - - if (NULL != (s = hashget(cp, sz))) - return(s->key); - - s = mandoc_calloc(sizeof(struct str) + sz + 1, 1); - memcpy(s->key, cp, sz); - - end = cp + sz; - index = ohash_qlookupi(&strings, cp, &end); - assert(NULL == ohash_find(&strings, index)); - ohash_insert(&strings, index, s); - return(s->key); -} - -static struct str * -hashget(const char *cp, size_t sz) -{ - unsigned int index; - const char *end; - - end = cp + sz; - index = ohash_qlookupi(&strings, cp, &end); - return(ohash_find(&strings, index)); -} - -/* - * Add a word to the current parse sequence. - * Within the hashtable of strings, we maintain a list of strings that - * are currently indexed. - * Each of these ("words") has a bitmask modified within the parse. - * When we finish a parse, we'll dump the list, then remove the head - * entry -- since the next parse will have a new "of", it can keep track - * of its entries without conflict. - */ static void -wordaddbuf(const struct of *of, - const char *cp, size_t sz, uint64_t v) +putkeys(const struct of *of, const char *cp, size_t sz, uint64_t v) { struct str *s; - unsigned int index; + unsigned int slot; const char *end; if (0 == sz) return; - s = hashget(cp, sz); + end = cp + sz; + slot = ohash_qlookupi(&strings, cp, &end); + s = ohash_find(&strings, slot); if (NULL != s && of == s->of) { s->mask |= v; @@ -1613,16 +1608,10 @@ wordaddbuf(const struct of *of, } else if (NULL == s) { s = mandoc_calloc(sizeof(struct str) + sz + 1, 1); memcpy(s->key, cp, sz); - end = cp + sz; - index = ohash_qlookupi(&strings, cp, &end); - assert(NULL == ohash_find(&strings, index)); - ohash_insert(&strings, index, s); + ohash_insert(&strings, slot, s); } - - s->next = words; s->of = of; s->mask = v; - words = s; } /* @@ -1794,6 +1783,7 @@ dbindex(struct mchars *mc, int form, const struct of * const char *desc; int64_t recno; size_t i; + unsigned int slot; if (verb) say(of->file, "Adding to index"); @@ -1802,8 +1792,9 @@ dbindex(struct mchars *mc, int form, const struct of * return; desc = ""; - if (NULL != of->desc) { - key = hashget(of->desc, strlen(of->desc)); + if (NULL != of->desc && '\0' != *of->desc) { + key = ohash_find(&strings, + ohash_qlookup(&strings, of->desc)); assert(NULL != key); if (NULL == key->utf8) utf8key(mc, key); @@ -1822,7 +1813,8 @@ dbindex(struct mchars *mc, int form, const struct of * recno = sqlite3_last_insert_rowid(db); sqlite3_reset(stmts[STMT_INSERT_DOC]); - for (key = words; NULL != key; key = key->next) { + for (key = ohash_first(&strings, &slot); NULL != key; + key = ohash_next(&strings, &slot)) { assert(key->of == of); if (NULL == key->utf8) utf8key(mc, key); @@ -1832,6 +1824,9 @@ dbindex(struct mchars *mc, int form, const struct of * SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno); SQL_STEP(stmts[STMT_INSERT_KEY]); sqlite3_reset(stmts[STMT_INSERT_KEY]); + if (key->utf8 != key->key) + free(key->utf8); + free(key); } SQL_EXEC("END TRANSACTION"); @@ -1864,7 +1859,6 @@ static void dbclose(int real) { size_t i; - char file[PATH_MAX]; if (nodb) return; @@ -1880,9 +1874,7 @@ dbclose(int real) if (real) return; - strlcpy(file, MANDOC_DB, PATH_MAX); - strlcat(file, "~", PATH_MAX); - if (-1 == rename(file, MANDOC_DB)) { + if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) { exitcode = (int)MANDOCLEVEL_SYSERR; say(MANDOC_DB, NULL); } @@ -1899,29 +1891,24 @@ dbclose(int real) static int dbopen(int real) { - char file[PATH_MAX]; - const char *sql; + const char *file, *sql; int rc, ofl; - size_t sz; if (nodb) return(1); - sz = strlcpy(file, MANDOC_DB, PATH_MAX); - if ( ! real) - sz = strlcat(file, "~", PATH_MAX); + ofl = SQLITE_OPEN_READWRITE; + if (0 == real) { + file = MANDOC_DB "~"; + if (-1 == remove(file) && ENOENT != errno) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(file, NULL); + return(0); + } + ofl |= SQLITE_OPEN_EXCLUSIVE; + } else + file = MANDOC_DB; - if (sz >= PATH_MAX) { - fprintf(stderr, "%s: Path too long\n", file); - return(0); - } - - if ( ! real) - remove(file); - - ofl = SQLITE_OPEN_READWRITE | - (0 == real ? SQLITE_OPEN_EXCLUSIVE : 0); - rc = sqlite3_open_v2(file, &db, ofl, NULL); if (SQLITE_OK == rc) goto prepare_statements; @@ -1975,6 +1962,17 @@ prepare_statements: sql = "INSERT INTO keys " "(bits,key,docid) VALUES (?,?,?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL); + +#ifndef __APPLE__ + /* + * When opening a new database, we can turn off + * synchronous mode for much better performance. + */ + + if (real) + SQL_EXEC("PRAGMA synchronous = OFF"); +#endif + return(1); }