=================================================================== RCS file: /cvs/mandoc/mandocdb.c,v retrieving revision 1.113 retrieving revision 1.134 diff -u -p -r1.113 -r1.134 --- mandoc/mandocdb.c 2014/01/19 22:48:16 1.113 +++ mandoc/mandocdb.c 2014/04/13 20:55:45 1.134 @@ -1,4 +1,4 @@ -/* $Id: mandocdb.c,v 1.113 2014/01/19 22:48:16 schwarze Exp $ */ +/* $Id: mandocdb.c,v 1.134 2014/04/13 20:55:45 schwarze Exp $ */ /* * Copyright (c) 2011, 2012 Kristaps Dzonsons * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze @@ -20,6 +20,7 @@ #endif #include +#include #include #include @@ -45,6 +46,7 @@ #include "mdoc.h" #include "man.h" #include "mandoc.h" +#include "mandoc_aux.h" #include "manpath.h" #include "mansearch.h" @@ -53,22 +55,22 @@ extern const char *const mansearch_keynames[]; #define SQL_EXEC(_v) \ if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \ - fprintf(stderr, "%s\n", sqlite3_errmsg(db)) + say("", "%s: %s", (_v), sqlite3_errmsg(db)) #define SQL_BIND_TEXT(_s, _i, _v) \ if (SQLITE_OK != sqlite3_bind_text \ ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \ - fprintf(stderr, "%s\n", sqlite3_errmsg(db)) + say(mlink->file, "%s", sqlite3_errmsg(db)) #define SQL_BIND_INT(_s, _i, _v) \ if (SQLITE_OK != sqlite3_bind_int \ ((_s), (_i)++, (_v))) \ - fprintf(stderr, "%s\n", sqlite3_errmsg(db)) + say(mlink->file, "%s", sqlite3_errmsg(db)) #define SQL_BIND_INT64(_s, _i, _v) \ if (SQLITE_OK != sqlite3_bind_int64 \ ((_s), (_i)++, (_v))) \ - fprintf(stderr, "%s\n", sqlite3_errmsg(db)) + say(mlink->file, "%s", sqlite3_errmsg(db)) #define SQL_STEP(_s) \ if (SQLITE_DONE != sqlite3_step((_s))) \ - fprintf(stderr, "%s\n", sqlite3_errmsg(db)) + say(mlink->file, "%s", sqlite3_errmsg(db)) enum op { OP_DEFAULT = 0, /* new dbs from dir list or default config */ @@ -98,6 +100,7 @@ struct inodev { struct mpage { struct inodev inodev; /* used for hashing routine */ + int64_t recno; /* id in mpages SQL table */ enum form form; /* format from file content */ char *sec; /* section from file content */ char *arch; /* architecture from file content */ @@ -115,12 +118,15 @@ struct mlink { char *name; /* name from file name (not empty) */ char *fsec; /* section from file name suffix */ struct mlink *next; /* singly linked list */ + struct mpage *mpage; /* parent */ + int gzip; /* filename has a .gz suffix */ }; enum stmt { STMT_DELETE_PAGE = 0, /* delete mpage */ STMT_INSERT_PAGE, /* insert mpage */ STMT_INSERT_LINK, /* insert mlink */ + STMT_INSERT_NAME, /* insert name */ STMT_INSERT_KEY, /* insert parsed key */ STMT__MAX }; @@ -133,7 +139,8 @@ struct mdoc_handler { }; static void dbclose(int); -static void dbadd(const struct mpage *, struct mchars *); +static void dbadd(struct mpage *, struct mchars *); +static void dbadd_mlink(const struct mlink *mlink); static int dbopen(int); static void dbprune(void); static void filescan(const char *); @@ -141,12 +148,13 @@ static void *hash_alloc(size_t, void *); static void hash_free(void *, size_t, void *); static void *hash_halloc(size_t, void *); static void mlink_add(struct mlink *, const struct stat *); -static int mlink_check(struct mpage *, struct mlink *); +static void 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 *); -static void parse_cat(struct mpage *); +static void names_check(void); +static void parse_cat(struct mpage *, int); static void parse_man(struct mpage *, const struct man_node *); static void parse_mdoc(struct mpage *, const struct mdoc_node *); static int parse_mdoc_body(struct mpage *, const struct mdoc_node *); @@ -168,11 +176,12 @@ static int set_basedir(const char *); static int treescan(void); static size_t utf8(unsigned int, char [7]); +static char tempfilename[32]; static char *progname; static int nodb; /* no database changes */ -static int quick; /* abort the parse early */ +static int mparse_options; /* abort the parse early */ static int use_all; /* use all found files */ -static int verb; /* print what we're doing */ +static int debug; /* print what we're doing */ static int warnings; /* warn about crap */ static int write_utf8; /* write UTF-8 output; else ASCII */ static int exitcode; /* to be returned by main */ @@ -180,9 +189,11 @@ static enum op op; /* operational mode */ static char basedir[PATH_MAX]; /* current base directory */ static struct ohash mpages; /* table of distinct manual pages */ static struct ohash mlinks; /* table of directory entries */ +static struct ohash names; /* table of all names */ static struct ohash strings; /* table of all strings */ static sqlite3 *db = NULL; /* current database */ static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */ +static uint64_t name_mask; static const struct mdoc_handler mdocs[MDOC_MAX] = { { NULL, 0 }, /* Ap */ @@ -216,8 +227,8 @@ static const struct mdoc_handler mdocs[MDOC_MAX] = { { NULL, TYPE_Ic }, /* Ic */ { NULL, TYPE_In }, /* In */ { NULL, TYPE_Li }, /* Li */ - { parse_mdoc_Nd, TYPE_Nd }, /* Nd */ - { parse_mdoc_Nm, TYPE_Nm }, /* Nm */ + { parse_mdoc_Nd, 0 }, /* Nd */ + { parse_mdoc_Nm, 0 }, /* Nm */ { NULL, 0 }, /* Op */ { NULL, 0 }, /* Ot */ { NULL, TYPE_Pa }, /* Pa */ @@ -350,7 +361,7 @@ main(int argc, char *argv[]) path_arg = NULL; op = OP_DEFAULT; - while (-1 != (ch = getopt(argc, argv, "aC:d:nQT:tu:vW"))) + while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v"))) switch (ch) { case ('a'): use_all = 1; @@ -360,6 +371,9 @@ main(int argc, char *argv[]) path_arg = optarg; op = OP_CONFFILE; break; + case ('D'): + debug++; + break; case ('d'): CHECKOP(op, ch); path_arg = optarg; @@ -368,8 +382,11 @@ main(int argc, char *argv[]) case ('n'): nodb = 1; break; + case ('p'): + warnings = 1; + break; case ('Q'): - quick = 1; + mparse_options |= MPARSE_QUICK; break; case ('T'): if (strcmp(optarg, "utf8")) { @@ -391,11 +408,8 @@ main(int argc, char *argv[]) op = OP_DELETE; break; case ('v'): - verb++; + /* Compatibility with espie@'s makewhatis. */ break; - case ('W'): - warnings = 1; - break; default: goto usage; } @@ -409,8 +423,7 @@ main(int argc, char *argv[]) } exitcode = (int)MANDOCLEVEL_OK; - mp = mparse_alloc(MPARSE_AUTO, - MANDOCLEVEL_FATAL, NULL, NULL, quick); + mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL, NULL); mc = mchars_alloc(); ohash_init(&mpages, 6, &mpages_info); @@ -453,6 +466,11 @@ main(int argc, char *argv[]) } else manpath_parse(&dirs, path_arg, NULL, NULL); + if (0 == dirs.sz) { + exitcode = (int)MANDOCLEVEL_BADARG; + say("", "Empty manpath"); + } + /* * First scan the tree rooted at a base directory, then * build a new database and finally move it into place. @@ -481,6 +499,9 @@ main(int argc, char *argv[]) goto out; mpages_merge(mc, mp); + if (warnings && + ! (MPARSE_QUICK & mparse_options)) + names_check(); dbclose(0); if (j + 1 < dirs.sz) { @@ -500,10 +521,10 @@ out: ohash_delete(&mlinks); return(exitcode); usage: - fprintf(stderr, "usage: %s [-anQvW] [-C file] [-Tutf8]\n" - " %s [-anQvW] [-Tutf8] dir ...\n" - " %s [-nQvW] [-Tutf8] -d dir [file ...]\n" - " %s [-nvW] -u dir [file ...]\n" + fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n" + " %s [-aDnpQ] [-Tutf8] dir ...\n" + " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n" + " %s [-Dnp] -u dir [file ...]\n" " %s [-Q] -t file ...\n", progname, progname, progname, progname, progname); @@ -531,7 +552,7 @@ treescan(void) FTS *f; FTSENT *ff; struct mlink *mlink; - int dform; + int dform, gzip; char *dsec, *arch, *fsec, *cp; const char *path; const char *argv[2]; @@ -546,7 +567,7 @@ treescan(void) f = fts_open((char * const *)argv, FTS_LOGICAL, NULL); if (NULL == f) { exitcode = (int)MANDOCLEVEL_SYSERR; - say("", NULL); + say("", "&fts_open"); return(0); } @@ -566,8 +587,18 @@ treescan(void) if (warnings) say(path, "Extraneous file"); continue; - } else if (NULL == (fsec = - strrchr(ff->fts_name, '.'))) { + } + gzip = 0; + fsec = NULL; + while (NULL == fsec) { + fsec = strrchr(ff->fts_name, '.'); + if (NULL == fsec || strcmp(fsec+1, "gz")) + break; + gzip = 1; + *fsec = '\0'; + fsec = NULL; + } + if (NULL == fsec) { if ( ! use_all) { if (warnings) say(path, @@ -578,10 +609,6 @@ treescan(void) if (warnings) say(path, "Skip html"); continue; - } else if (0 == strcmp(fsec, "gz")) { - if (warnings) - say(path, "Skip gz"); - continue; } else if (0 == strcmp(fsec, "ps")) { if (warnings) say(path, "Skip ps"); @@ -606,6 +633,7 @@ treescan(void) mlink->arch = arch; mlink->name = ff->fts_name; mlink->fsec = fsec; + mlink->gzip = gzip; mlink_add(mlink, ff->fts_statp); continue; } else if (FTS_D != ff->fts_info && @@ -700,7 +728,7 @@ filescan(const char *file) if (NULL == realpath(file, buf)) { exitcode = (int)MANDOCLEVEL_BADARG; - say(file, NULL); + say(file, "&realpath"); return; } @@ -716,7 +744,7 @@ filescan(const char *file) if (-1 == stat(buf, &st)) { exitcode = (int)MANDOCLEVEL_BADARG; - say(file, NULL); + say(file, "&stat"); return; } else if ( ! (S_IFREG & st.st_mode)) { exitcode = (int)MANDOCLEVEL_BADARG; @@ -816,6 +844,7 @@ mlink_add(struct mlink *mlink, const struct stat *st) } else mlink->next = mpage->mlinks; mpage->mlinks = mlink; + mlink->mpage = mpage; } static void @@ -898,13 +927,12 @@ nextlink: } } -static int +static void mlink_check(struct mpage *mpage, struct mlink *mlink) { - int match; + struct str *str; + unsigned int slot; - match = 1; - /* * Check whether the manual section given in a file * agrees with the directory where the file is located. @@ -915,11 +943,9 @@ mlink_check(struct mpage *mpage, struct mlink *mlink) */ if (FORM_SRC == mpage->form && - strcasecmp(mpage->sec, mlink->dsec)) { - match = 0; + strcasecmp(mpage->sec, mlink->dsec)) say(mlink->file, "Section \"%s\" manual in %s directory", mpage->sec, mlink->dsec); - } /* * Manual page directories exist for each kernel @@ -934,16 +960,28 @@ mlink_check(struct mpage *mpage, struct mlink *mlink) * on amd64, i386, sparc, and sparc64. */ - if (strcasecmp(mpage->arch, mlink->arch)) { - match = 0; + if (strcasecmp(mpage->arch, mlink->arch)) say(mlink->file, "Architecture \"%s\" manual in " "\"%s\" directory", mpage->arch, mlink->arch); - } - if (strcasecmp(mpage->title, mlink->name)) - match = 0; + /* + * XXX + * parse_cat() doesn't set NAME_TITLE yet. + */ - return(match); + if (FORM_CAT == mpage->form) + return; + + /* + * Check whether this mlink + * appears as a name in the NAME section. + */ + + slot = ohash_qlookup(&names, mlink->name); + str = ohash_find(&names, slot); + assert(NULL != str); + if ( ! (NAME_TITLE & str->mask)) + say(mlink->file, "Name missing in NAME section"); } /* @@ -958,12 +996,15 @@ mpages_merge(struct mchars *mc, struct mparse *mp) { char any[] = "any"; struct ohash_info str_info; - struct mpage *mpage; - struct mlink *mlink; + int fd[2]; + struct mpage *mpage, *mpage_dest; + struct mlink *mlink, *mlink_dest; struct mdoc *mdoc; struct man *man; + char *sodest; char *cp; - int match; + pid_t child_pid; + int status; unsigned int pslot; enum mandoclevel lvl; @@ -983,11 +1024,48 @@ mpages_merge(struct mchars *mc, struct mparse *mp) continue; } + name_mask = NAME_MASK; + ohash_init(&names, 4, &str_info); ohash_init(&strings, 6, &str_info); mparse_reset(mp); mdoc = NULL; man = NULL; + sodest = NULL; + child_pid = 0; + fd[0] = -1; + fd[1] = -1; + if (mpage->mlinks->gzip) { + if (-1 == pipe(fd)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(mpage->mlinks->file, "&pipe gunzip"); + goto nextpage; + } + switch (child_pid = fork()) { + case (-1): + exitcode = (int)MANDOCLEVEL_SYSERR; + say(mpage->mlinks->file, "&fork gunzip"); + child_pid = 0; + close(fd[1]); + close(fd[0]); + goto nextpage; + case (0): + close(fd[0]); + if (-1 == dup2(fd[1], STDOUT_FILENO)) { + say(mpage->mlinks->file, + "&dup gunzip"); + exit(1); + } + execlp("gunzip", "gunzip", "-c", + mpage->mlinks->file, NULL); + say(mpage->mlinks->file, "&exec gunzip"); + exit(1); + default: + close(fd[1]); + break; + } + } + /* * Try interpreting the file as mdoc(7) or man(7) * source code, unless it is already known to be @@ -995,12 +1073,48 @@ mpages_merge(struct mchars *mc, struct mparse *mp) */ if (FORM_CAT != mpage->mlinks->dform || FORM_CAT != mpage->mlinks->fform) { - lvl = mparse_readfd(mp, -1, mpage->mlinks->file); + lvl = mparse_readfd(mp, fd[0], mpage->mlinks->file); if (lvl < MANDOCLEVEL_FATAL) - mparse_result(mp, &mdoc, &man); + mparse_result(mp, &mdoc, &man, &sodest); } - if (NULL != mdoc) { + if (NULL != sodest) { + mlink_dest = ohash_find(&mlinks, + ohash_qlookup(&mlinks, sodest)); + if (NULL != mlink_dest) { + + /* The .so target exists. */ + + mpage_dest = mlink_dest->mpage; + mlink = mpage->mlinks; + while (1) { + mlink->mpage = mpage_dest; + + /* + * If the target was already + * processed, add the links + * to the database now. + * Otherwise, this will + * happen when we come + * to the target. + */ + + if (mpage_dest->recno) + dbadd_mlink(mlink); + + if (NULL == mlink->next) + break; + mlink = mlink->next; + } + + /* Move all links to the target. */ + + mlink->next = mlink_dest->next; + mlink_dest->next = mpage->mlinks; + mpage->mlinks = NULL; + } + goto nextpage; + } else if (NULL != mdoc) { mpage->form = FORM_SRC; mpage->sec = mandoc_strdup(mdoc_meta(mdoc)->msec); @@ -1037,32 +1151,48 @@ mpages_merge(struct mchars *mc, struct mparse *mp) putkey(mpage, mlink->fsec, TYPE_sec); putkey(mpage, '\0' == *mlink->arch ? any : mlink->arch, TYPE_arch); - putkey(mpage, mlink->name, TYPE_Nm); + putkey(mpage, mlink->name, NAME_FILE); } - if (warnings && !use_all) { - match = 0; - for (mlink = mpage->mlinks; mlink; - mlink = mlink->next) - if (mlink_check(mpage, mlink)) - match = 1; - } else - match = 1; - if (NULL != mdoc) { if (NULL != (cp = mdoc_meta(mdoc)->name)) - putkey(mpage, cp, TYPE_Nm); + putkey(mpage, cp, NAME_HEAD); assert(NULL == mpage->desc); parse_mdoc(mpage, mdoc_node(mdoc)); - putkey(mpage, NULL != mpage->desc ? - mpage->desc : mpage->mlinks->name, TYPE_Nd); + if (NULL == mpage->desc) + mpage->desc = mandoc_strdup( + mpage->mlinks->name); } else if (NULL != man) parse_man(mpage, man_node(man)); else - parse_cat(mpage); + parse_cat(mpage, fd[0]); + if (warnings && !use_all) + for (mlink = mpage->mlinks; mlink; + mlink = mlink->next) + mlink_check(mpage, mlink); + dbadd(mpage, mc); + +nextpage: + if (child_pid) { + if (-1 == waitpid(child_pid, &status, 0)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(mpage->mlinks->file, "&wait gunzip"); + } else if (WIFSIGNALED(status)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(mpage->mlinks->file, + "gunzip died from signal %d", + WTERMSIG(status)); + } else if (WEXITSTATUS(status)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(mpage->mlinks->file, + "gunzip failed with code %d", + WEXITSTATUS(status)); + } + } ohash_delete(&strings); + ohash_delete(&names); mpage = ohash_next(&mpages, &pslot); } @@ -1071,15 +1201,53 @@ mpages_merge(struct mchars *mc, struct mparse *mp) } static void -parse_cat(struct mpage *mpage) +names_check(void) { + sqlite3_stmt *stmt; + const char *name, *sec, *arch, *key; + int irc; + + sqlite3_prepare_v2(db, + "SELECT name, sec, arch, key FROM (" + "SELECT name AS key, pageid FROM names " + "WHERE bits & ? AND NOT EXISTS (" + "SELECT pageid FROM mlinks " + "WHERE mlinks.pageid == names.pageid " + "AND mlinks.name == names.name" + ")" + ") JOIN (" + "SELECT * FROM mlinks GROUP BY pageid" + ") USING (pageid);", + -1, &stmt, NULL); + + if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE)) + say("", "%s", sqlite3_errmsg(db)); + + while (SQLITE_ROW == (irc = sqlite3_step(stmt))) { + name = sqlite3_column_text(stmt, 0); + sec = sqlite3_column_text(stmt, 1); + arch = sqlite3_column_text(stmt, 2); + key = sqlite3_column_text(stmt, 3); + say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec, + '\0' == *arch ? "" : "/", + '\0' == *arch ? "" : arch, key); + } + sqlite3_finalize(stmt); +} + +static void +parse_cat(struct mpage *mpage, int fd) +{ FILE *stream; char *line, *p, *title; size_t len, plen, titlesz; - if (NULL == (stream = fopen(mpage->mlinks->file, "r"))) { + stream = (-1 == fd) ? + fopen(mpage->mlinks->file, "r") : + fdopen(fd, "r"); + if (NULL == stream) { if (warnings) - say(mpage->mlinks->file, NULL); + say(mpage->mlinks->file, "&fopen"); return; } @@ -1136,7 +1304,6 @@ parse_cat(struct mpage *mpage) "Cannot find NAME section"); assert(NULL == mpage->desc); mpage->desc = mandoc_strdup(mpage->mlinks->name); - putkey(mpage, mpage->mlinks->name, TYPE_Nd); fclose(stream); free(title); return; @@ -1177,7 +1344,6 @@ parse_cat(struct mpage *mpage) assert(NULL == mpage->desc); mpage->desc = mandoc_strdup(p); - putkey(mpage, mpage->desc, TYPE_Nd); fclose(stream); free(title); } @@ -1218,9 +1384,9 @@ static void parse_man(struct mpage *mpage, const struct man_node *n) { const struct man_node *head, *body; - char *start, *sv, *title; + char *start, *title; char byte; - size_t sz, titlesz; + size_t sz; if (NULL == n) return; @@ -1240,12 +1406,8 @@ parse_man(struct mpage *mpage, const struct man_node * NULL != (head = (head->child)) && MAN_TEXT == head->type && 0 == strcmp(head->string, "NAME") && - NULL != (body = body->child) && - MAN_TEXT == body->type) { + NULL != body->child) { - title = NULL; - titlesz = 0; - /* * Suck the entire NAME section into memory. * Yes, we might run away. @@ -1253,47 +1415,11 @@ parse_man(struct mpage *mpage, const struct man_node * * NAME sections over many lines. */ - for ( ; NULL != body; body = body->next) { - if (MAN_TEXT != body->type) - break; - if (0 == (sz = strlen(body->string))) - continue; - title = mandoc_realloc - (title, titlesz + sz + 1); - memcpy(title + titlesz, body->string, sz); - titlesz += sz + 1; - title[titlesz - 1] = ' '; - } + title = NULL; + man_deroff(&title, body); if (NULL == title) return; - title = mandoc_realloc(title, titlesz + 1); - title[titlesz] = '\0'; - - /* Skip leading space. */ - - sv = title; - while (isspace((unsigned char)*sv)) - sv++; - - if (0 == (sz = strlen(sv))) { - free(title); - return; - } - - /* Erase trailing space. */ - - start = &sv[sz - 1]; - while (start > sv && isspace((unsigned char)*start)) - *start-- = '\0'; - - if (start == sv) { - free(title); - return; - } - - start = sv; - /* * Go through a special heuristic dance here. * Conventionally, one or more manual names are @@ -1302,6 +1428,7 @@ parse_man(struct mpage *mpage, const struct man_node * * the name parts here. */ + start = title; for ( ;; ) { sz = strcspn(start, " ,"); if ('\0' == start[sz]) @@ -1319,7 +1446,7 @@ parse_man(struct mpage *mpage, const struct man_node * ('\\' == start[0] && '-' == start[1])) break; - putkey(mpage, start, TYPE_Nm); + putkey(mpage, start, NAME_TITLE); if (' ' == byte) { start += sz + 1; @@ -1332,8 +1459,8 @@ parse_man(struct mpage *mpage, const struct man_node * start++; } - if (sv == start) { - putkey(mpage, start, TYPE_Nm); + if (start == title) { + putkey(mpage, start, NAME_TITLE); free(title); return; } @@ -1357,7 +1484,6 @@ parse_man(struct mpage *mpage, const struct man_node * assert(NULL == mpage->desc); mpage->desc = mandoc_strdup(start); - putkey(mpage, mpage->desc, TYPE_Nd); free(title); return; } @@ -1491,10 +1617,7 @@ parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_n return(0); } - if (-1 == asprintf(&cp, "%s(%s)", n->string, n->next->string)) { - perror(NULL); - exit((int)MANDOCLEVEL_SYSERR); - } + mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string); putkey(mpage, cp, TYPE_Xr); free(cp); return(0); @@ -1503,40 +1626,21 @@ parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_n static int parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n) { - size_t sz; - if (MDOC_BODY != n->type) - return(0); - - /* - * Special-case the `Nd' because we need to put the description - * into the document table. - */ - - for (n = n->child; NULL != n; n = n->next) { - if (MDOC_TEXT == n->type) { - if (NULL != mpage->desc) { - sz = strlen(mpage->desc) + - strlen(n->string) + 2; - mpage->desc = mandoc_realloc( - mpage->desc, sz); - strlcat(mpage->desc, " ", sz); - strlcat(mpage->desc, n->string, sz); - } else - mpage->desc = mandoc_strdup(n->string); - } - if (NULL != n->child) - parse_mdoc_Nd(mpage, n); - } - return(1); + if (MDOC_BODY == n->type) + mdoc_deroff(&mpage->desc, n); + return(0); } static int parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n) { - return(SEC_NAME == n->sec || - (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)); + if (SEC_NAME == n->sec) + putmdockey(mpage, n->child, NAME_TITLE); + else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) + putmdockey(mpage, n->child, NAME_SYN); + return(0); } static int @@ -1569,28 +1673,35 @@ static void putkeys(const struct mpage *mpage, const char *cp, size_t sz, uint64_t v) { + struct ohash *htab; struct str *s; const char *end; - uint64_t mask; unsigned int slot; int i; if (0 == sz) return; - if (verb > 1) { - for (i = 0, mask = 1; - i < mansearch_keymax; - i++, mask <<= 1) - if (mask & v) - break; - say(mpage->mlinks->file, "Adding key %s=%*s", - mansearch_keynames[i], sz, cp); + if (TYPE_Nm & v) { + htab = &names; + v &= name_mask; + name_mask &= ~NAME_FIRST; + if (debug > 1) + say(mpage->mlinks->file, + "Adding name %*s", sz, cp); + } else { + htab = &strings; + if (debug > 1) + for (i = 0; i < mansearch_keymax; i++) + if (1 << i & v) + say(mpage->mlinks->file, + "Adding key %s=%*s", + mansearch_keynames[i], sz, cp); } end = cp + sz; - slot = ohash_qlookupi(&strings, cp, &end); - s = ohash_find(&strings, slot); + slot = ohash_qlookupi(htab, cp, &end); + s = ohash_find(htab, slot); if (NULL != s && mpage == s->mpage) { s->mask |= v; @@ -1598,7 +1709,7 @@ putkeys(const struct mpage *mpage, } else if (NULL == s) { s = mandoc_calloc(sizeof(struct str) + sz + 1, 1); memcpy(s->key, cp, sz); - ohash_insert(&strings, slot, s); + ohash_insert(htab, slot, s); } s->mpage = mpage; s->mask = v; @@ -1664,7 +1775,7 @@ static void render_key(struct mchars *mc, struct str *key) { size_t sz, bsz, pos; - char utfbuf[7], res[5]; + char utfbuf[7], res[6]; char *buf; const char *seq, *cpp, *val; int len, u; @@ -1676,7 +1787,8 @@ render_key(struct mchars *mc, struct str *key) res[1] = '\t'; res[2] = ASCII_NBRSP; res[3] = ASCII_HYPH; - res[4] = '\0'; + res[4] = ASCII_BREAK; + res[5] = '\0'; val = key->key; bsz = strlen(val); @@ -1707,16 +1819,24 @@ render_key(struct mchars *mc, struct str *key) val += sz; } - if (ASCII_HYPH == *val) { + switch (*val) { + case (ASCII_HYPH): buf[pos++] = '-'; val++; continue; - } else if ('\t' == *val || ASCII_NBRSP == *val) { + case ('\t'): + /* FALLTHROUGH */ + case (ASCII_NBRSP): buf[pos++] = ' '; val++; + /* FALLTHROUGH */ + case (ASCII_BREAK): continue; - } else if ('\\' != *val) + default: break; + } + if ('\\' != *val) + break; /* Read past the slash. */ @@ -1768,6 +1888,20 @@ render_key(struct mchars *mc, struct str *key) key->rendered = buf; } +static void +dbadd_mlink(const struct mlink *mlink) +{ + size_t i; + + i = 1; + SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec); + SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch); + SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name); + SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->recno); + SQL_STEP(stmts[STMT_INSERT_LINK]); + sqlite3_reset(stmts[STMT_INSERT_LINK]); +} + /* * Flush the current page's terms (and their bits) into the database. * Wrap the entire set of additions in a transaction to make sqlite be a @@ -1775,36 +1909,70 @@ render_key(struct mchars *mc, struct str *key) * Also, handle escape sequences at the last possible moment. */ static void -dbadd(const struct mpage *mpage, struct mchars *mc) +dbadd(struct mpage *mpage, struct mchars *mc) { struct mlink *mlink; struct str *key; - int64_t recno; size_t i; unsigned int slot; - if (verb) - say(mpage->mlinks->file, "Adding to database"); + mlink = mpage->mlinks; - if (nodb) + if (nodb) { + while (NULL != mlink) { + fputs(mlink->name, stdout); + if (NULL == mlink->next || + strcmp(mlink->dsec, mlink->next->dsec) || + strcmp(mlink->fsec, mlink->next->fsec) || + strcmp(mlink->arch, mlink->next->arch)) { + putchar('('); + if ('\0' == *mlink->dsec) + fputs(mlink->fsec, stdout); + else + fputs(mlink->dsec, stdout); + if ('\0' != *mlink->arch) + printf("/%s", mlink->arch); + putchar(')'); + } + mlink = mlink->next; + if (NULL != mlink) + fputs(", ", stdout); + } + printf(" - %s\n", mpage->desc); return; + } + if (debug) + say(mlink->file, "Adding to database"); + i = 1; + SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->desc); SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form); SQL_STEP(stmts[STMT_INSERT_PAGE]); - recno = sqlite3_last_insert_rowid(db); + mpage->recno = sqlite3_last_insert_rowid(db); sqlite3_reset(stmts[STMT_INSERT_PAGE]); - for (mlink = mpage->mlinks; mlink; mlink = mlink->next) { - i = 1; - SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->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]); + while (NULL != mlink) { + dbadd_mlink(mlink); + mlink = mlink->next; } + mlink = mpage->mlinks; + for (key = ohash_first(&names, &slot); NULL != key; + key = ohash_next(&names, &slot)) { + assert(key->mpage == mpage); + if (NULL == key->rendered) + render_key(mc, key); + i = 1; + SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask); + SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered); + SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->recno); + SQL_STEP(stmts[STMT_INSERT_NAME]); + sqlite3_reset(stmts[STMT_INSERT_NAME]); + if (key->rendered != key->key) + free(key->rendered); + free(key); + } for (key = ohash_first(&strings, &slot); NULL != key; key = ohash_next(&strings, &slot)) { assert(key->mpage == mpage); @@ -1813,7 +1981,7 @@ dbadd(const struct mpage *mpage, struct mchars *mc) i = 1; SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask); SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered); - SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno); + SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->recno); SQL_STEP(stmts[STMT_INSERT_KEY]); sqlite3_reset(stmts[STMT_INSERT_KEY]); if (key->rendered != key->key) @@ -1836,7 +2004,7 @@ dbprune(void) for (mpage = ohash_first(&mpages, &slot); NULL != mpage; mpage = ohash_next(&mpages, &slot)) { mlink = mpage->mlinks; - if (verb) + if (debug) say(mlink->file, "Deleting from database"); if (nodb) continue; @@ -1865,6 +2033,8 @@ static void dbclose(int real) { size_t i; + int status; + pid_t child; if (nodb) return; @@ -1880,10 +2050,60 @@ dbclose(int real) if (real) return; - if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) { + if ('\0' == *tempfilename) { + if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(MANDOC_DB, "&rename"); + } + return; + } + + switch (child = fork()) { + case (-1): exitcode = (int)MANDOCLEVEL_SYSERR; - say(MANDOC_DB, NULL); + say("", "&fork cmp"); + return; + case (0): + execlp("cmp", "cmp", "-s", + tempfilename, MANDOC_DB, NULL); + say("", "&exec cmp"); + exit(0); + default: + break; } + if (-1 == waitpid(child, &status, 0)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "&wait cmp"); + } else if (WIFSIGNALED(status)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "cmp died from signal %d", WTERMSIG(status)); + } else if (WEXITSTATUS(status)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say(MANDOC_DB, + "Data changed, but cannot replace database"); + } + + *strrchr(tempfilename, '/') = '\0'; + switch (child = fork()) { + case (-1): + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "&fork rm"); + return; + case (0): + execlp("rm", "rm", "-rf", tempfilename, NULL); + say("", "&exec rm"); + exit((int)MANDOCLEVEL_SYSERR); + default: + break; + } + if (-1 == waitpid(child, &status, 0)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "&wait rm"); + } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "%s: Cannot remove temporary directory", + tempfilename); + } } /* @@ -1897,43 +2117,65 @@ dbclose(int real) static int dbopen(int real) { - const char *file, *sql; + const char *sql; int rc, ofl; if (nodb) return(1); + *tempfilename = '\0'; ofl = SQLITE_OPEN_READWRITE; - if (0 == real) { - file = MANDOC_DB "~"; - if (-1 == remove(file) && ENOENT != errno) { + + if (real) { + rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL); + if (SQLITE_OK != rc) { exitcode = (int)MANDOCLEVEL_SYSERR; - say(file, NULL); + say(MANDOC_DB, "%s", sqlite3_errmsg(db)); return(0); } - ofl |= SQLITE_OPEN_EXCLUSIVE; - } else - file = MANDOC_DB; + goto prepare_statements; + } - rc = sqlite3_open_v2(file, &db, ofl, NULL); + ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE; + + remove(MANDOC_DB "~"); + rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL); if (SQLITE_OK == rc) - goto prepare_statements; - if (SQLITE_CANTOPEN != rc) { + goto create_tables; + if (MPARSE_QUICK & mparse_options) { exitcode = (int)MANDOCLEVEL_SYSERR; - say(file, NULL); + say(MANDOC_DB "~", "%s", sqlite3_errmsg(db)); return(0); } - sqlite3_close(db); - db = NULL; - - if (SQLITE_OK != (rc = sqlite3_open(file, &db))) { + if (strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX", + sizeof(tempfilename)) >= sizeof(tempfilename)) { exitcode = (int)MANDOCLEVEL_SYSERR; - say(file, NULL); + say("", "/tmp/mandocdb.XXXXXX: Filename too long"); return(0); } + if (NULL == mkdtemp(tempfilename)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "&%s", tempfilename); + return(0); + } + if (strlcat(tempfilename, "/" MANDOC_DB, + sizeof(tempfilename)) >= sizeof(tempfilename)) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "%s/" MANDOC_DB ": Filename too long", + tempfilename); + return(0); + } + rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL); + if (SQLITE_OK != rc) { + exitcode = (int)MANDOCLEVEL_SYSERR; + say("", "%s: %s", tempfilename, sqlite3_errmsg(db)); + return(0); + } +create_tables: sql = "CREATE TABLE \"mpages\" (\n" + " \"desc\" TEXT NOT NULL,\n" " \"form\" INTEGER NOT NULL,\n" " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" ");\n" @@ -1946,6 +2188,13 @@ dbopen(int real) "ON DELETE CASCADE\n" ");\n" "\n" + "CREATE TABLE \"names\" (\n" + " \"bits\" INTEGER NOT NULL,\n" + " \"name\" TEXT NOT NULL,\n" + " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) " + "ON DELETE CASCADE\n" + ");\n" + "\n" "CREATE TABLE \"keys\" (\n" " \"bits\" INTEGER NOT NULL,\n" " \"key\" TEXT NOT NULL,\n" @@ -1955,7 +2204,7 @@ dbopen(int real) if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) { exitcode = (int)MANDOCLEVEL_SYSERR; - say(file, "%s", sqlite3_errmsg(db)); + say(MANDOC_DB, "%s", sqlite3_errmsg(db)); return(0); } @@ -1966,11 +2215,14 @@ prepare_statements: "sec=? AND arch=? AND name=?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL); sql = "INSERT INTO mpages " - "(form) VALUES (?)"; + "(desc,form) VALUES (?,?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL); sql = "INSERT INTO mlinks " "(sec,arch,name,pageid) VALUES (?,?,?,?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL); + sql = "INSERT INTO names " + "(bits,name,pageid) VALUES (?,?,?)"; + sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL); sql = "INSERT INTO keys " "(bits,key,pageid) VALUES (?,?,?)"; sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL); @@ -2025,12 +2277,12 @@ set_basedir(const char *targetdir) if (NULL == getcwd(startdir, PATH_MAX)) { exitcode = (int)MANDOCLEVEL_SYSERR; if (NULL != targetdir) - say(".", NULL); + say("", "&getcwd"); return(0); } if (-1 == (fd = open(startdir, O_RDONLY, 0))) { exitcode = (int)MANDOCLEVEL_SYSERR; - say(startdir, NULL); + say("", "&open %s", startdir); return(0); } if (NULL == targetdir) @@ -2042,7 +2294,7 @@ set_basedir(const char *targetdir) close(fd); basedir[0] = '\0'; exitcode = (int)MANDOCLEVEL_SYSERR; - say(startdir, NULL); + say("", "&chdir %s", startdir); return(0); } if (NULL == targetdir) { @@ -2053,11 +2305,11 @@ set_basedir(const char *targetdir) if (NULL == realpath(targetdir, basedir)) { basedir[0] = '\0'; exitcode = (int)MANDOCLEVEL_BADARG; - say(targetdir, NULL); + say("", "&%s: realpath", targetdir); return(0); } else if (-1 == chdir(basedir)) { exitcode = (int)MANDOCLEVEL_BADARG; - say("", NULL); + say("", "&chdir"); return(0); } return(1); @@ -2067,6 +2319,7 @@ static void say(const char *file, const char *format, ...) { va_list ap; + int use_errno; if ('\0' != *basedir) fprintf(stderr, "%s", basedir); @@ -2074,16 +2327,32 @@ say(const char *file, const char *format, ...) fputs("//", stderr); if ('\0' != *file) fprintf(stderr, "%s", file); - fputs(": ", stderr); - if (NULL == format) { - perror(NULL); - return; + use_errno = 1; + if (NULL != format) { + switch (*format) { + case ('&'): + format++; + break; + case ('\0'): + format = NULL; + break; + default: + use_errno = 0; + break; + } } - - va_start(ap, format); - vfprintf(stderr, format, ap); - va_end(ap); - - fputc('\n', stderr); + if (NULL != format) { + if ('\0' != *basedir || '\0' != *file) + fputs(": ", stderr); + va_start(ap, format); + vfprintf(stderr, format, ap); + va_end(ap); + } + if (use_errno) { + if ('\0' != *basedir || '\0' != *file || NULL != format) + fputs(": ", stderr); + perror(NULL); + } else + fputc('\n', stderr); }