=================================================================== RCS file: /cvs/mandoc/mandocdb.c,v retrieving revision 1.56 retrieving revision 1.78 diff -u -p -r1.56 -r1.78 --- mandoc/mandocdb.c 2013/06/03 19:01:31 1.56 +++ mandoc/mandocdb.c 2013/12/26 22:12:46 1.78 @@ -1,4 +1,4 @@ -/* $Id: mandocdb.c,v 1.56 2013/06/03 19:01:31 schwarze Exp $ */ +/* $Id: mandocdb.c,v 1.78 2013/12/26 22:12:46 schwarze Exp $ */ /* * Copyright (c) 2011, 2012 Kristaps Dzonsons * Copyright (c) 2011, 2012, 2013 Ingo Schwarze @@ -19,7 +19,6 @@ #include "config.h" #endif -#include #include #include @@ -28,7 +27,9 @@ #include #include #include +#include #include +#include #include #include #include @@ -82,8 +83,7 @@ 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 */ + const struct mpage *mpage; /* if set, the owning parse */ uint64_t mask; /* bitmask in sequence */ char key[]; /* the string itself */ }; @@ -93,19 +93,24 @@ struct id { dev_t dev; }; -struct of { +struct mpage { struct id id; /* used for hashing routine */ - struct of *next; /* next in ofs */ + struct mpage *next; /* next in mpages_list */ enum form dform; /* path-cued form */ enum form sform; /* suffix-cued form */ - char file[MAXPATHLEN]; /* 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 file[PATH_MAX]; /* filename rel. to manpath */ + 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 */ @@ -113,230 +118,214 @@ enum stmt { STMT__MAX }; -typedef int (*mdoc_fp)(struct of *, const struct mdoc_node *); +typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *); 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(const char *, int); -static void dbindex(struct mchars *, int, - const struct of *, const char *); -static int dbopen(const char *, int); -static void dbprune(const char *); -static void fileadd(struct of *); +static void dbclose(int); +static void dbindex(struct mchars *, int, const struct mpage *); +static int dbopen(int); +static void dbprune(void); +static void fileadd(struct mpage *); static int filecheck(const char *); -static void filescan(const char *, const char *); -static struct str *hashget(const char *, size_t); +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 void inoadd(const struct stat *, struct of *); +static void inoadd(const struct stat *, struct mpage *); static int inocheck(const struct stat *); -static void ofadd(const char *, int, const char *, - const char *, const char *, const char *, - const char *, const struct stat *); -static void offree(void); -static int ofmerge(struct mchars *, struct mparse *, const char *); -static void parse_catpage(struct of *, const char *); -static int parse_man(struct of *, - const struct man_node *); -static void parse_mdoc(struct of *, const struct mdoc_node *); -static int parse_mdoc_body(struct of *, const struct mdoc_node *); -static int parse_mdoc_head(struct of *, const struct mdoc_node *); -static int parse_mdoc_Fd(struct of *, const struct mdoc_node *); -static int parse_mdoc_Fn(struct of *, const struct mdoc_node *); -static int parse_mdoc_In(struct of *, const struct mdoc_node *); -static int parse_mdoc_Nd(struct of *, const struct mdoc_node *); -static int parse_mdoc_Nm(struct of *, const struct mdoc_node *); -static int parse_mdoc_Sh(struct of *, const struct mdoc_node *); -static int parse_mdoc_St(struct of *, const struct mdoc_node *); -static int parse_mdoc_Xr(struct of *, const struct mdoc_node *); -static int path_reset(const char *, int, const char *); -static void putkey(const struct of *, +static void ofadd(int, const char *, const char *, const char *, + const char *, const char *, const struct stat *); +static void mpages_free(void); +static void mpages_merge(struct mchars *, struct mparse *, int); +static void parse_cat(struct mpage *); +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 *); +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 int set_basedir(const char *); +static void putkey(const struct mpage *, const char *, uint64_t); -static void putkeys(const struct of *, - const char *, int, uint64_t); -static void putmdockey(const struct of *, +static void putkeys(const struct mpage *, + const char *, size_t, uint64_t); +static void putmdockey(const struct mpage *, const struct mdoc_node *, uint64_t); -static void say(const char *, const char *, const char *, ...); -static char *stradd(const char *); -static char *straddbuf(const char *, size_t); -static int treescan(const char *); +static void say(const char *, const char *, ...); +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 */ static int nodb; /* no database changes */ static int verb; /* print what we're doing */ static int warnings; /* warn about crap */ +static int exitcode; /* to be returned by main */ static enum op op; /* operational mode */ -static struct ohash inos; /* table of inodes/devices */ +static char basedir[PATH_MAX]; /* current base directory */ +static struct ohash mpages; /* table of distinct manual pages */ 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 struct mpage *mpages_list = NULL; /* vector of files to 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[]) { - char cwd[MAXPATHLEN]; - int ch, rc, fd, i; - unsigned int index; + int ch, i; size_t j, sz; - const char *dir; - struct str *s; + const char *path_arg; struct mchars *mc; struct manpaths dirs; struct mparse *mp; - struct ohash_info ino_info, filename_info, str_info; + struct ohash_info mpages_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; + mpages_info.alloc = filename_info.alloc = hash_alloc; + mpages_info.halloc = filename_info.halloc = hash_halloc; + mpages_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); + mpages_info.key_offset = offsetof(struct mpage, id); + filename_info.key_offset = offsetof(struct mpage, file); progname = strrchr(argv[0], '/'); if (progname == NULL) @@ -345,20 +334,6 @@ main(int argc, char *argv[]) ++progname; /* - * Remember where we started by keeping a fd open to the origin - * path component: throughout this utility, we chdir() a lot to - * handle relative paths, and by doing this, we can return to - * the starting point. - */ - if (NULL == getcwd(cwd, MAXPATHLEN)) { - perror(NULL); - return(EXIT_FAILURE); - } else if (-1 == (fd = open(cwd, O_RDONLY, 0))) { - perror(cwd); - return(EXIT_FAILURE); - } - - /* * We accept a few different invocations. * The CHECKOP macro makes sure that invocation styles don't * clobber each other. @@ -369,7 +344,7 @@ main(int argc, char *argv[]) goto usage; \ } while (/*CONSTCOND*/0) - dir = NULL; + path_arg = NULL; op = OP_DEFAULT; while (-1 != (ch = getopt(argc, argv, "aC:d:ntu:vW"))) @@ -379,12 +354,12 @@ main(int argc, char *argv[]) break; case ('C'): CHECKOP(op, ch); - dir = optarg; + path_arg = optarg; op = OP_CONFFILE; break; case ('d'): CHECKOP(op, ch); - dir = optarg; + path_arg = optarg; op = OP_UPDATE; break; case ('n'): @@ -398,7 +373,7 @@ main(int argc, char *argv[]) break; case ('u'): CHECKOP(op, ch); - dir = optarg; + path_arg = optarg; op = OP_DELETE; break; case ('v'): @@ -419,13 +394,12 @@ main(int argc, char *argv[]) goto usage; } - rc = 1; + exitcode = (int)MANDOCLEVEL_OK; mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL, NULL); mc = mchars_alloc(); - ohash_init(&strings, 6, &str_info); - ohash_init(&inos, 6, &ino_info); + ohash_init(&mpages, 6, &mpages_info); ohash_init(&filenames, 6, &filename_info); if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) { @@ -433,24 +407,23 @@ main(int argc, char *argv[]) * Force processing all files. */ use_all = 1; - if (NULL == dir) - dir = cwd; + /* * All of these deal with a specific directory. * Jump into that directory then collect files specified * on the command-line. */ - if (0 == path_reset(cwd, fd, dir)) + if (0 == set_basedir(path_arg)) goto out; for (i = 0; i < argc; i++) - filescan(argv[i], dir); - if (0 == dbopen(dir, 1)) + filescan(argv[i]); + if (0 == dbopen(1)) goto out; if (OP_TEST != op) - dbprune(dir); + dbprune(); if (OP_DELETE != op) - rc = ofmerge(mc, mp, dir); - dbclose(dir, 1); + mpages_merge(mc, mp, 0); + dbclose(1); } else { /* * If we have arguments, use them as our manpaths. @@ -464,12 +437,11 @@ main(int argc, char *argv[]) for (i = 0; i < argc; i++) dirs.paths[i] = mandoc_strdup(argv[i]); } else - manpath_parse(&dirs, dir, NULL, NULL); + 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. */ @@ -479,50 +451,40 @@ main(int argc, char *argv[]) dirs.paths[j][--sz] = '\0'; if (0 == sz) continue; - if (0 == path_reset(cwd, fd, dirs.paths[j])) + + if (j) { + ohash_init(&mpages, 6, &mpages_info); + ohash_init(&filenames, 6, &filename_info); + } + + if (0 == set_basedir(dirs.paths[j])) goto out; - if (0 == treescan(dirs.paths[j])) + if (0 == treescan()) goto out; - if (0 == path_reset(cwd, fd, dirs.paths[j])) + if (0 == set_basedir(dirs.paths[j])) goto out; - if (0 == dbopen(dirs.paths[j], 0)) + 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 + mpages_merge(mc, mp, warnings && !use_all); + dbclose(0); - if (0 == ofmerge(mc, mp, dirs.paths[j])) - goto out; - dbclose(dirs.paths[j], 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(&mpages); + ohash_delete(&filenames); + mpages_free(); + } } } out: - close(fd); + set_basedir(NULL); 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(&mpages); ohash_delete(&filenames); - offree(); - return(rc ? EXIT_SUCCESS : EXIT_FAILURE); + mpages_free(); + return(exitcode); usage: fprintf(stderr, "usage: %s [-anvW] [-C file]\n" " %s [-anvW] dir ...\n" @@ -532,11 +494,11 @@ usage: progname, progname, progname, progname, progname); - return(EXIT_FAILURE); + return((int)MANDOCLEVEL_BADARG); } /* - * Scan a directory tree rooted at "base" for manpages. + * Scan a directory tree rooted at "basedir" for manpages. * We use fts(), scanning directory parts along the way for clues to our * section and architecture. * @@ -550,13 +512,13 @@ usage: * TODO: accomodate for multi-language directories. */ static int -treescan(const char *base) +treescan(void) { FTS *f; 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] = "."; @@ -568,7 +530,8 @@ treescan(const char *base) */ f = fts_open((char * const *)argv, FTS_LOGICAL, NULL); if (NULL == f) { - perror(base); + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", NULL); return(0); } @@ -578,58 +541,62 @@ treescan(const char *base) while (NULL != (ff = fts_read(f))) { path = ff->fts_path + 2; /* - * If we're a regular file, add an "of" by using the + * If we're a regular file, add an mpage 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)) + continue; if ( ! use_all && ff->fts_level < 2) { if (warnings) - say(base, path, "Extraneous file"); + say(path, "Extraneous file"); continue; } else if (inocheck(ff->fts_statp)) { if (warnings) - say(base, path, "Duplicate file"); + say(path, "Duplicate file"); continue; - } - - cp = ff->fts_name; - - if (0 == strcmp(cp, "mandocdb.db")) { - if (warnings) - say(base, path, "Skip database"); - continue; - } else if (NULL != (cp = strrchr(cp, '.'))) { - if (0 == strcmp(cp + 1, "html")) { + } else if (NULL == (sec = + strrchr(ff->fts_name, '.'))) { + if ( ! use_all) { if (warnings) - say(base, path, "Skip html"); + say(path, + "No filename suffix"); continue; - } else if (0 == strcmp(cp + 1, "gz")) { - if (warnings) - say(base, path, "Skip gz"); - continue; - } else if (0 == strcmp(cp + 1, "ps")) { - if (warnings) - say(base, path, "Skip ps"); - continue; - } else if (0 == strcmp(cp + 1, "pdf")) { - if (warnings) - say(base, path, "Skip pdf"); - continue; } - } - - if (NULL != (sec = strrchr(ff->fts_name, '.'))) { - *sec = '\0'; - sec = stradd(sec + 1); - } - name = stradd(ff->fts_name); - ofadd(base, dform, path, - name, dsec, sec, arch, ff->fts_statp); + } else if (0 == strcmp(++sec, "html")) { + if (warnings) + say(path, "Skip html"); + continue; + } else if (0 == strcmp(sec, "gz")) { + if (warnings) + say(path, "Skip gz"); + continue; + } else if (0 == strcmp(sec, "ps")) { + if (warnings) + say(path, "Skip ps"); + continue; + } else if (0 == strcmp(sec, "pdf")) { + if (warnings) + say(path, "Skip pdf"); + continue; + } else if ( ! use_all && + ((FORM_SRC == dform && strcmp(sec, dsec)) || + (FORM_CAT == dform && strcmp(sec, "0")))) { + if (warnings) + say(path, "Wrong filename suffix"); + continue; + } else + sec[-1] = '\0'; + 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) + FTS_DP != ff->fts_info) { + if (warnings) + say(path, "Not a regular file"); continue; + } switch (ff->fts_level) { case (0): @@ -649,17 +616,17 @@ treescan(const char *base) 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) break; if (warnings) - say(base, path, "Unknown directory part"); + say(path, "Unknown directory part"); fts_set(f, ff, FTS_SKIP); break; case (2): @@ -669,13 +636,13 @@ treescan(const char *base) */ 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) break; if (warnings) - say(base, path, "Extraneous directory part"); + say(path, "Extraneous directory part"); fts_set(f, ff, FTS_SKIP); break; } @@ -697,14 +664,15 @@ treescan(const char *base) * or * [./]cat
[/]/.0 * - * Stuff this information directly into the "of" vector. + * Stuff this information directly into the mpage vector. * See treescan() for the fts(3) version of this. */ static void -filescan(const char *file, const char *base) +filescan(const char *file) { + char buf[PATH_MAX]; const char *sec, *arch, *name, *dsec; - char *p, *start, *buf; + char *p, *start; int dform; struct stat st; @@ -713,22 +681,28 @@ filescan(const char *file, const char *base) if (0 == strncmp(file, "./", 2)) file += 2; - if (-1 == stat(file, &st)) { - if (warnings) - say(base, file, "%s", strerror(errno)); + if (NULL == realpath(file, buf)) { + exitcode = (int)MANDOCLEVEL_BADARG; + say(file, NULL); return; + } else if (OP_TEST != op && strstr(buf, basedir) != buf) { + exitcode = (int)MANDOCLEVEL_BADARG; + say("", "%s: outside base directory", buf); + return; + } else if (-1 == stat(buf, &st)) { + exitcode = (int)MANDOCLEVEL_BADARG; + say(file, NULL); + return; } else if ( ! (S_IFREG & st.st_mode)) { - if (warnings) - say(base, file, "Not a regular file"); + exitcode = (int)MANDOCLEVEL_BADARG; + say(file, "Not a regular file"); return; } else if (inocheck(&st)) { if (warnings) - say(base, file, "Duplicate file"); + say(file, "Duplicate file"); return; } - - buf = mandoc_strdup(file); - start = buf; + start = buf + strlen(basedir); sec = arch = name = dsec = NULL; dform = FORM_NONE; @@ -779,8 +753,7 @@ filescan(const char *file, const char *base) *p = '\0'; } - ofadd(base, dform, file, name, dsec, sec, arch, &st); - free(buf); + ofadd(dform, file, name, dsec, sec, arch, &st); } /* @@ -789,10 +762,9 @@ filescan(const char *file, const char *base) 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))); } /* @@ -800,13 +772,13 @@ filecheck(const char *name) * already exists. */ static void -fileadd(struct of *of) +fileadd(struct mpage *mpage) { - 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, mpage->file); + assert(NULL == ohash_find(&filenames, slot)); + ohash_insert(&filenames, slot, mpage); } /* @@ -817,15 +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(&mpages, ohash_lookup_memory( + &mpages, (char *)&id, sizeof(id), hash))); } /* @@ -834,26 +804,25 @@ inocheck(const struct stat *st) * Then when we do the lookup, use both the inode and device identifier. */ static void -inoadd(const struct stat *st, struct of *of) +inoadd(const struct stat *st, struct mpage *mpage) { 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 - (&inos, (char *)&of->id, sizeof(of->id), hash); + mpage->id.ino = hash = st->st_ino; + mpage->id.dev = st->st_dev; + slot = ohash_lookup_memory + (&mpages, (char *)&mpage->id, sizeof(mpage->id), hash); - assert(NULL == ohash_find(&inos, index)); - ohash_insert(&inos, index, of); + assert(NULL == ohash_find(&mpages, slot)); + ohash_insert(&mpages, slot, mpage); } static void -ofadd(const char *base, int dform, const char *file, - const char *name, const char *dsec, const char *sec, - const char *arch, const struct stat *st) +ofadd(int dform, const char *file, const char *name, const char *dsec, + const char *sec, const char *arch, const struct stat *st) { - struct of *of; + struct mpage *mpage; int sform; assert(NULL != file); @@ -867,66 +836,88 @@ ofadd(const char *base, int dform, const char *file, 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, MAXPATHLEN); - of->name = name; - of->sec = sec; - of->dsec = dsec; - of->arch = arch; - of->sform = sform; - of->dform = dform; - of->next = ofs; - ofs = of; + mpage = mandoc_calloc(1, sizeof(struct mpage)); + strlcpy(mpage->file, file, PATH_MAX); + mpage->name = mandoc_strdup(name); + mpage->sec = mandoc_strdup(sec); + mpage->dsec = mandoc_strdup(dsec); + mpage->arch = mandoc_strdup(arch); + mpage->sform = sform; + mpage->dform = dform; + mpage->next = mpages_list; + mpages_list = mpage; /* * Add to unique identifier hash. * Then if it's a source manual and we're going to use source in * favour of catpages, add it to that hash. */ - inoadd(st, of); - fileadd(of); + inoadd(st, mpage); + fileadd(mpage); } static void -offree(void) +mpages_free(void) { - struct of *of; + struct mpage *mpage; - while (NULL != (of = ofs)) { - ofs = of->next; - free(of); + while (NULL != (mpage = mpages_list)) { + mpages_list = mpage->next; + free(mpage->name); + free(mpage->sec); + free(mpage->dsec); + free(mpage->arch); + free(mpage); } } /* - * Run through the files in the global vector "ofs" and add them to the - * database specified in "base". + * Run through the files in the global vector "mpages_list" and add them to the + * database specified in "basedir". * * This handles the parsing scheme itself, using the cues of directory * and filename to determine whether the file is parsable or not. */ -static int -ofmerge(struct mchars *mc, struct mparse *mp, const char *base) +static void +mpages_merge(struct mchars *mc, struct mparse *mp, int check_reachable) { - int form; - size_t sz; - struct mdoc *mdoc; - struct man *man; - char buf[MAXPATHLEN]; - 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 mpage *mpage; + 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; - for (of = ofs; NULL != of; of = of->next) { + 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 (mpage = mpages_list; NULL != mpage; mpage = mpage->next) { /* * If we're a catpage (as defined by our path), then see * if a manpage exists by the same name (ignoring the @@ -934,11 +925,11 @@ ofmerge(struct mchars *mc, struct mparse *mp, const ch * If it does, then we want to use it instead of our * own. */ - if ( ! use_all && FORM_CAT == of->dform) { - sz = strlcpy(buf, of->file, MAXPATHLEN); - if (sz >= MAXPATHLEN) { + if ( ! use_all && FORM_CAT == mpage->dform) { + sz = strlcpy(buf, mpage->file, PATH_MAX); + if (sz >= PATH_MAX) { if (warnings) - say(base, of->file, + say(mpage->file, "Filename too long"); continue; } @@ -947,34 +938,31 @@ ofmerge(struct mchars *mc, struct mparse *mp, const ch memcpy(bufp, "man", 3); if (NULL != (bufp = strrchr(buf, '.'))) *++bufp = '\0'; - strlcat(buf, of->dsec, MAXPATHLEN); + strlcat(buf, mpage->dsec, PATH_MAX); if (filecheck(buf)) { if (warnings) - say(base, of->file, "Man " + say(mpage->file, "Man " "source exists: %s", buf); continue; } } - words = NULL; + ohash_init(&strings, 6, &str_info); mparse_reset(mp); mdoc = NULL; man = NULL; - form = 0; - msec = of->dsec; - march = of->arch; - mtitle = of->name; + match = 1; /* * Try interpreting the file as mdoc(7) or man(7) * source code, unless it is already known to be * formatted. Fall back to formatted mode. */ - if (FORM_SRC == of->dform || FORM_SRC == of->sform) { - lvl = mparse_readfd(mp, -1, of->file); + if (FORM_CAT != mpage->dform || FORM_CAT != mpage->sform) { + lvl = mparse_readfd(mp, -1, mpage->file); if (lvl < MANDOCLEVEL_FATAL) mparse_result(mp, &mdoc, &man); - } + } if (NULL != mdoc) { form = 1; @@ -984,15 +972,20 @@ ofmerge(struct mchars *mc, struct mparse *mp, const ch } else if (NULL != man) { form = 1; msec = man_meta(man)->msec; - march = ""; + march = mpage->arch; mtitle = man_meta(man)->title; - } + } else { + form = 0; + msec = mpage->dsec; + march = mpage->arch; + mtitle = mpage->name; + } - if (NULL == msec) + if (NULL == msec) msec = ""; - if (NULL == march) + if (NULL == march) march = ""; - if (NULL == mtitle) + if (NULL == mtitle) mtitle = ""; /* @@ -1005,10 +998,12 @@ ofmerge(struct mchars *mc, struct mparse *mp, const ch * manuals for such reasons. */ if (warnings && !use_all && form && - strcasecmp(msec, of->dsec)) - say(base, of->file, "Section \"%s\" " - "manual in %s directory", - msec, of->dsec); + strcasecmp(msec, mpage->dsec)) { + match = 0; + say(mpage->file, "Section \"%s\" " + "manual in %s directory", + msec, mpage->dsec); + } /* * Manual page directories exist for each kernel @@ -1024,38 +1019,88 @@ ofmerge(struct mchars *mc, struct mparse *mp, const ch * Thus, warn about architecture mismatches, * but don't skip manuals for this reason. */ - if (warnings && !use_all && strcasecmp(march, of->arch)) - say(base, of->file, "Architecture \"%s\" " + if (warnings && !use_all && strcasecmp(march, mpage->arch)) { + match = 0; + say(mpage->file, "Architecture \"%s\" " "manual in \"%s\" directory", - march, of->arch); + march, mpage->arch); + } + if (warnings && !use_all && strcasecmp(mtitle, mpage->name)) + match = 0; - putkey(of, of->name, TYPE_Nm); + putkey(mpage, mpage->name, TYPE_Nm); if (NULL != mdoc) { if (NULL != (cp = mdoc_meta(mdoc)->name)) - putkey(of, cp, TYPE_Nm); - parse_mdoc(of, mdoc_node(mdoc)); + putkey(mpage, cp, TYPE_Nm); + assert(NULL == mpage->desc); + parse_mdoc(mpage, mdoc_node(mdoc)); + putkey(mpage, NULL != mpage->desc ? + mpage->desc : mpage->name, TYPE_Nd); } else if (NULL != man) - parse_man(of, man_node(man)); + parse_man(mpage, man_node(man)); else - parse_catpage(of, base); + parse_cat(mpage); - dbindex(mc, form, of, base); + /* + * 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 ? "" : mpage->file); + ohash_insert(&title_table, slot, + title_entry); + } else { + if (match) + *title_entry->file = '\0'; + free(title_str); + } + } + + dbindex(mc, form, mpage); + ohash_delete(&strings); } - return(1); + 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 -parse_catpage(struct of *of, const char *base) +parse_cat(struct mpage *mpage) { FILE *stream; char *line, *p, *title; size_t len, plen, titlesz; - if (NULL == (stream = fopen(of->file, "r"))) { + if (NULL == (stream = fopen(mpage->file, "r"))) { if (warnings) - say(base, of->file, "%s", strerror(errno)); + say(mpage->file, NULL); return; } @@ -1108,7 +1153,10 @@ parse_catpage(struct of *of, const char *base) if (NULL == title || '\0' == *title) { if (warnings) - say(base, of->file, "Cannot find NAME section"); + say(mpage->file, "Cannot find NAME section"); + assert(NULL == mpage->desc); + mpage->desc = mandoc_strdup(mpage->name); + putkey(mpage, mpage->name, TYPE_Nd); fclose(stream); free(title); return; @@ -1128,7 +1176,7 @@ parse_catpage(struct of *of, const char *base) /* Skip to next word. */ ; } else { if (warnings) - say(base, of->file, "No dash in title line"); + say(mpage->file, "No dash in title line"); p = title; } @@ -1146,8 +1194,9 @@ parse_catpage(struct of *of, const char *base) plen -= 2; } - of->desc = stradd(p); - putkey(of, p, TYPE_Nd); + assert(NULL == mpage->desc); + mpage->desc = mandoc_strdup(p); + putkey(mpage, mpage->desc, TYPE_Nd); fclose(stream); free(title); } @@ -1156,40 +1205,31 @@ parse_catpage(struct of *of, const char *base) * Put a type/word pair into the word database for this particular file. */ static void -putkey(const struct of *of, const char *value, uint64_t type) +putkey(const struct mpage *mpage, const char *value, uint64_t type) { assert(NULL != value); - wordaddbuf(of, value, strlen(value), type); + putkeys(mpage, 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 -putmdockey(const struct of *of, const struct mdoc_node *n, uint64_t m) +putmdockey(const struct mpage *mpage, + const struct mdoc_node *n, uint64_t m) { for ( ; NULL != n; n = n->next) { if (NULL != n->child) - putmdockey(of, n->child, m); + putmdockey(mpage, n->child, m); if (MDOC_TEXT == n->type) - putkey(of, n->string, m); + putkey(mpage, n->string, m); } } -static int -parse_man(struct of *of, const struct man_node *n) +static void +parse_man(struct mpage *mpage, const struct man_node *n) { const struct man_node *head, *body; char *start, *sv, *title; @@ -1197,7 +1237,7 @@ parse_man(struct of *of, const struct man_node *n) size_t sz, titlesz; if (NULL == n) - return(0); + return; /* * We're only searching for one thing: the first text child in @@ -1239,7 +1279,7 @@ parse_man(struct of *of, const struct man_node *n) title[titlesz - 1] = ' '; } if (NULL == title) - return(1); + return; title = mandoc_realloc(title, titlesz + 1); title[titlesz] = '\0'; @@ -1252,7 +1292,7 @@ parse_man(struct of *of, const struct man_node *n) if (0 == (sz = strlen(sv))) { free(title); - return(1); + return; } /* Erase trailing space. */ @@ -1263,7 +1303,7 @@ parse_man(struct of *of, const struct man_node *n) if (start == sv) { free(title); - return(1); + return; } start = sv; @@ -1284,7 +1324,7 @@ parse_man(struct of *of, const struct man_node *n) byte = start[sz]; start[sz] = '\0'; - putkey(of, start, TYPE_Nm); + putkey(mpage, start, TYPE_Nm); if (' ' == byte) { start += sz + 1; @@ -1298,9 +1338,9 @@ parse_man(struct of *of, const struct man_node *n) } if (sv == start) { - putkey(of, start, TYPE_Nm); + putkey(mpage, start, TYPE_Nm); free(title); - return(1); + return; } while (isspace((unsigned char)*start)) @@ -1320,23 +1360,23 @@ parse_man(struct of *of, const struct man_node *n) while (' ' == *start) start++; - assert(NULL == of->desc); - of->desc = stradd(start); - putkey(of, start, TYPE_Nd); + assert(NULL == mpage->desc); + mpage->desc = mandoc_strdup(start); + putkey(mpage, mpage->desc, TYPE_Nd); free(title); - return(1); + return; } } - for (n = n->child; n; n = n->next) - if (parse_man(of, n)) - return(1); - - return(0); + for (n = n->child; n; n = n->next) { + if (NULL != mpage->desc) + break; + parse_man(mpage, n); + } } static void -parse_mdoc(struct of *of, const struct mdoc_node *n) +parse_mdoc(struct mpage *mpage, const struct mdoc_node *n) { assert(NULL != n); @@ -1352,23 +1392,23 @@ parse_mdoc(struct of *of, const struct mdoc_node *n) /* FALLTHROUGH */ case (MDOC_TAIL): if (NULL != mdocs[n->tok].fp) - if (0 == (*mdocs[n->tok].fp)(of, n)) + if (0 == (*mdocs[n->tok].fp)(mpage, n)) break; - - if (MDOCF_CHILD & mdocs[n->tok].flags) - putmdockey(of, n->child, mdocs[n->tok].mask); + if (mdocs[n->tok].mask) + putmdockey(mpage, n->child, + mdocs[n->tok].mask); break; default: assert(MDOC_ROOT != n->type); continue; } if (NULL != n->child) - parse_mdoc(of, n); + parse_mdoc(mpage, n); } } static int -parse_mdoc_Fd(struct of *of, const struct mdoc_node *n) +parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n) { const char *start, *end; size_t sz; @@ -1406,23 +1446,23 @@ parse_mdoc_Fd(struct of *of, const struct mdoc_node *n end--; if (end > start) - putkeys(of, start, end - start + 1, TYPE_In); + putkeys(mpage, start, end - start + 1, TYPE_In); return(1); } static int -parse_mdoc_In(struct of *of, const struct mdoc_node *n) +parse_mdoc_In(struct mpage *mpage, const struct mdoc_node *n) { if (NULL != n->child && MDOC_TEXT == n->child->type) return(0); - putkey(of, n->child->string, TYPE_In); + putkey(mpage, n->child->string, TYPE_In); return(1); } static int -parse_mdoc_Fn(struct of *of, const struct mdoc_node *n) +parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n) { const char *cp; @@ -1442,45 +1482,55 @@ parse_mdoc_Fn(struct of *of, const struct mdoc_node *n while ('*' == *cp) cp++; - putkey(of, cp, TYPE_Fn); + putkey(mpage, cp, TYPE_Fn); if (n->string < cp) - putkeys(of, n->string, cp - n->string, TYPE_Ft); + putkeys(mpage, n->string, cp - n->string, TYPE_Ft); for (n = n->next; NULL != n; n = n->next) if (MDOC_TEXT == n->type) - putkey(of, n->string, TYPE_Fa); + putkey(mpage, n->string, TYPE_Fa); return(0); } static int -parse_mdoc_St(struct of *of, const struct mdoc_node *n) +parse_mdoc_St(struct mpage *mpage, const struct mdoc_node *n) { if (NULL == n->child || MDOC_TEXT != n->child->type) return(0); - putkey(of, n->child->string, TYPE_St); + putkey(mpage, n->child->string, TYPE_St); return(1); } static int -parse_mdoc_Xr(struct of *of, const struct mdoc_node *n) +parse_mdoc_Xr(struct mpage *mpage, 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(mpage, n->string, TYPE_Xr); + return(0); + } + + if (-1 == asprintf(&cp, "%s(%s)", n->string, n->next->string)) { + perror(NULL); + exit((int)MANDOCLEVEL_SYSERR); + } + putkey(mpage, cp, TYPE_Xr); + free(cp); + return(0); } static int -parse_mdoc_Nd(struct of *of, const struct mdoc_node *n) +parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n) { size_t sz; - char *sv, *desc; if (MDOC_BODY != n->type) return(0); @@ -1490,30 +1540,26 @@ 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 != 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(of, n); + parse_mdoc_Nd(mpage, n); } - - of->desc = NULL != desc ? stradd(desc) : NULL; - free(desc); return(1); } static int -parse_mdoc_Nm(struct of *of, const struct mdoc_node *n) +parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n) { if (SEC_NAME == n->sec) @@ -1525,114 +1571,56 @@ parse_mdoc_Nm(struct of *of, const struct mdoc_node *n } static int -parse_mdoc_Sh(struct of *of, const struct mdoc_node *n) +parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n) { return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type); } static int -parse_mdoc_head(struct of *of, const struct mdoc_node *n) +parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n) { return(MDOC_HEAD == n->type); } static int -parse_mdoc_body(struct of *of, const struct mdoc_node *n) +parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n) { return(MDOC_BODY == n->type); } /* - * 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 mpage *mpage, + 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) { + if (NULL != s && mpage == s->mpage) { s->mask |= v; return; } 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->mpage = mpage; s->mask = v; - words = s; } /* @@ -1798,23 +1786,24 @@ utf8key(struct mchars *mc, struct str *key) * Also, UTF-8-encode the description at the last possible moment. */ static void -dbindex(struct mchars *mc, int form, - const struct of *of, const char *base) +dbindex(struct mchars *mc, int form, const struct mpage *mpage) { struct str *key; const char *desc; int64_t recno; size_t i; + unsigned int slot; if (verb) - say(base, of->file, "Adding to index"); + say(mpage->file, "Adding to index"); if (nodb) return; desc = ""; - if (NULL != of->desc) { - key = hashget(of->desc, strlen(of->desc)); + if (NULL != mpage->desc && '\0' != *mpage->desc) { + key = ohash_find(&strings, + ohash_qlookup(&strings, mpage->desc)); assert(NULL != key); if (NULL == key->utf8) utf8key(mc, key); @@ -1824,17 +1813,18 @@ dbindex(struct mchars *mc, int form, SQL_EXEC("BEGIN TRANSACTION"); i = 1; - SQL_BIND_TEXT(stmts[STMT_INSERT_DOC], i, of->file); - SQL_BIND_TEXT(stmts[STMT_INSERT_DOC], i, of->sec); - SQL_BIND_TEXT(stmts[STMT_INSERT_DOC], i, of->arch); + SQL_BIND_TEXT(stmts[STMT_INSERT_DOC], i, mpage->file); + SQL_BIND_TEXT(stmts[STMT_INSERT_DOC], i, mpage->sec); + SQL_BIND_TEXT(stmts[STMT_INSERT_DOC], i, mpage->arch); SQL_BIND_TEXT(stmts[STMT_INSERT_DOC], i, desc); SQL_BIND_INT(stmts[STMT_INSERT_DOC], i, form); SQL_STEP(stmts[STMT_INSERT_DOC]); recno = sqlite3_last_insert_rowid(db); sqlite3_reset(stmts[STMT_INSERT_DOC]); - for (key = words; NULL != key; key = key->next) { - assert(key->of == of); + for (key = ohash_first(&strings, &slot); NULL != key; + key = ohash_next(&strings, &slot)) { + assert(key->mpage == mpage); if (NULL == key->utf8) utf8key(mc, key); i = 1; @@ -1843,27 +1833,30 @@ dbindex(struct mchars *mc, int form, 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"); } static void -dbprune(const char *base) +dbprune(void) { - struct of *of; + struct mpage *mpage; size_t i; if (nodb) return; - for (of = ofs; NULL != of; of = of->next) { + for (mpage = mpages_list; NULL != mpage; mpage = mpage->next) { i = 1; - SQL_BIND_TEXT(stmts[STMT_DELETE], i, of->file); + SQL_BIND_TEXT(stmts[STMT_DELETE], i, mpage->file); SQL_STEP(stmts[STMT_DELETE]); sqlite3_reset(stmts[STMT_DELETE]); if (verb) - say(base, of->file, "Deleted from index"); + say(mpage->file, "Deleted from index"); } } @@ -1872,10 +1865,9 @@ dbprune(const char *base) * If "real" is not set, rename the temporary file into the real one. */ static void -dbclose(const char *base, int real) +dbclose(int real) { size_t i; - char file[MAXPATHLEN]; if (nodb) return; @@ -1891,10 +1883,10 @@ dbclose(const char *base, int real) if (real) return; - strlcpy(file, MANDOC_DB, MAXPATHLEN); - strlcat(file, "~", MAXPATHLEN); - if (-1 == rename(file, MANDOC_DB)) - perror(MANDOC_DB); + if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(MANDOC_DB, NULL); + } } /* @@ -1906,36 +1898,32 @@ dbclose(const char *base, int real) * Must be matched by dbclose(). */ static int -dbopen(const char *base, int real) +dbopen(int real) { - char file[MAXPATHLEN]; - const char *sql; + const char *file, *sql; int rc, ofl; - size_t sz; if (nodb) return(1); - sz = strlcpy(file, MANDOC_DB, MAXPATHLEN); - if ( ! real) - sz = strlcat(file, "~", MAXPATHLEN); + 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 >= MAXPATHLEN) { - 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) - return(1); + goto prepare_statements; if (SQLITE_CANTOPEN != rc) { - perror(file); + exitcode = (int)MANDOCLEVEL_SYSERR; + say(file, NULL); return(0); } @@ -1943,7 +1931,8 @@ dbopen(const char *base, int real) db = NULL; if (SQLITE_OK != (rc = sqlite3_open(file, &db))) { - perror(file); + exitcode = (int)MANDOCLEVEL_SYSERR; + say(file, NULL); return(0); } @@ -1967,10 +1956,13 @@ dbopen(const char *base, int real) "CREATE INDEX \"key_index\" ON keys (key);\n"; if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) { - perror(sqlite3_errmsg(db)); + exitcode = (int)MANDOCLEVEL_SYSERR; + say(file, "%s", sqlite3_errmsg(db)); return(0); } +prepare_statements: + SQL_EXEC("PRAGMA foreign_keys = ON"); sql = "DELETE FROM docs where file=?"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE], NULL); sql = "INSERT INTO docs " @@ -1979,6 +1971,17 @@ dbopen(const char *base, int real) 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); } @@ -2004,28 +2007,76 @@ hash_free(void *p, size_t sz, void *arg) } static int -path_reset(const char *cwd, int fd, const char *base) +set_basedir(const char *targetdir) { + static char startdir[PATH_MAX]; + static int fd; - if (-1 == fchdir(fd)) { - perror(cwd); + /* + * Remember where we started by keeping a fd open to the origin + * path component: throughout this utility, we chdir() a lot to + * handle relative paths, and by doing this, we can return to + * the starting point. + */ + if ('\0' == *startdir) { + if (NULL == getcwd(startdir, PATH_MAX)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + if (NULL != targetdir) + say(".", NULL); + return(0); + } + if (-1 == (fd = open(startdir, O_RDONLY, 0))) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(startdir, NULL); + return(0); + } + if (NULL == targetdir) + targetdir = startdir; + } else { + if (-1 == fd) + return(0); + if (-1 == fchdir(fd)) { + close(fd); + basedir[0] = '\0'; + exitcode = (int)MANDOCLEVEL_SYSERR; + say(startdir, NULL); + return(0); + } + if (NULL == targetdir) { + close(fd); + return(1); + } + } + if (NULL == realpath(targetdir, basedir)) { + basedir[0] = '\0'; + exitcode = (int)MANDOCLEVEL_BADARG; + say(targetdir, NULL); return(0); - } else if (-1 == chdir(base)) { - perror(base); + } else if (-1 == chdir(basedir)) { + exitcode = (int)MANDOCLEVEL_BADARG; + say("", NULL); return(0); } return(1); } static void -say(const char *dir, const char *file, const char *format, ...) +say(const char *file, const char *format, ...) { va_list ap; - fprintf(stderr, "%s", dir); + if ('\0' != *basedir) + fprintf(stderr, "%s", basedir); + if ('\0' != *basedir && '\0' != *file) + fputs("//", stderr); if ('\0' != *file) - fprintf(stderr, "//%s", file); + fprintf(stderr, "%s", file); fputs(": ", stderr); + + if (NULL == format) { + perror(NULL); + return; + } va_start(ap, format); vfprintf(stderr, format, ap);