=================================================================== RCS file: /cvs/mandoc/mansearch.c,v retrieving revision 1.24 retrieving revision 1.41 diff -u -p -r1.24 -r1.41 --- mandoc/mansearch.c 2014/03/23 12:11:18 1.24 +++ mandoc/mansearch.c 2014/08/05 14:43:10 1.41 @@ -1,4 +1,4 @@ -/* $Id: mansearch.c,v 1.24 2014/03/23 12:11:18 schwarze Exp $ */ +/* $Id: mansearch.c,v 1.41 2014/08/05 14:43:10 schwarze Exp $ */ /* * Copyright (c) 2012 Kristaps Dzonsons * Copyright (c) 2013, 2014 Ingo Schwarze @@ -19,6 +19,7 @@ #include "config.h" #endif +#include #include #include #include @@ -37,6 +38,9 @@ #include "compat_ohash.h" #endif #include +#ifndef SQLITE_DETERMINISTIC +#define SQLITE_DETERMINISTIC 0 +#endif #include "mandoc.h" #include "mandoc_aux.h" @@ -63,17 +67,19 @@ extern const char *const mansearch_keynames[]; } while (0) struct expr { - uint64_t bits; /* type-mask */ - const char *substr; /* to search for, if applicable */ regex_t regexp; /* compiled regexp, if applicable */ + const char *substr; /* to search for, if applicable */ + struct expr *next; /* next in sequence */ + uint64_t bits; /* type-mask */ + int equal; /* equality, not subsring match */ int open; /* opening parentheses before */ int and; /* logical AND before */ int close; /* closing parentheses after */ - struct expr *next; /* next in sequence */ }; struct match { - uint64_t id; /* identifier in database */ + uint64_t pageid; /* identifier in database */ + char *desc; /* manual page description */ int form; /* 0 == catpage */ }; @@ -83,14 +89,15 @@ static void buildnames(struct manpage *, sqlite3 *, static char *buildoutput(sqlite3 *, sqlite3_stmt *, uint64_t, uint64_t); static void *hash_alloc(size_t, void *); -static void hash_free(void *, size_t, void *); -static void *hash_halloc(size_t, void *); -static struct expr *exprcomp(const struct mansearch *, +static void hash_free(void *, void *); +static void *hash_calloc(size_t, size_t, void *); +static struct expr *exprcomp(const struct mansearch *, int, char *[]); static void exprfree(struct expr *); static struct expr *exprspec(struct expr *, uint64_t, const char *, const char *); static struct expr *exprterm(const struct mansearch *, char *, int); +static int manpage_compare(const void *, const void *); static void sql_append(char **sql, size_t *sz, const char *newstr, int count); static void sql_match(sqlite3_context *context, @@ -99,7 +106,55 @@ static void sql_regexp(sqlite3_context *context, int argc, sqlite3_value **argv); static char *sql_statement(const struct expr *); + int +mansearch_setup(int start) +{ + static void *pagecache; + int c; + +#define PC_PAGESIZE 1280 +#define PC_NUMPAGES 256 + + if (start) { + if (NULL != pagecache) { + fprintf(stderr, "pagecache already enabled\n"); + return((int)MANDOCLEVEL_BADARG); + } + + pagecache = mmap(NULL, PC_PAGESIZE * PC_NUMPAGES, + PROT_READ | PROT_WRITE, MAP_ANON, -1, 0); + + if (MAP_FAILED == pagecache) { + perror("mmap"); + pagecache = NULL; + return((int)MANDOCLEVEL_SYSERR); + } + + c = sqlite3_config(SQLITE_CONFIG_PAGECACHE, + pagecache, PC_PAGESIZE, PC_NUMPAGES); + + if (SQLITE_OK == c) + return((int)MANDOCLEVEL_OK); + + fprintf(stderr, "pagecache: %s\n", sqlite3_errstr(c)); + + } else if (NULL == pagecache) { + fprintf(stderr, "pagecache missing\n"); + return((int)MANDOCLEVEL_BADARG); + } + + if (-1 == munmap(pagecache, PC_PAGESIZE * PC_NUMPAGES)) { + perror("munmap"); + pagecache = NULL; + return((int)MANDOCLEVEL_SYSERR); + } + + pagecache = NULL; + return((int)MANDOCLEVEL_OK); +} + +int mansearch(const struct mansearch *search, const struct manpaths *paths, int argc, char *argv[], @@ -107,7 +162,7 @@ mansearch(const struct mansearch *search, struct manpage **res, size_t *sz) { int fd, rc, c, indexbit; - int64_t id; + int64_t pageid; uint64_t outbit, iterbit; char buf[PATH_MAX]; char *sql; @@ -121,12 +176,10 @@ mansearch(const struct mansearch *search, unsigned int idx; size_t i, j, cur, maxres; - memset(&info, 0, sizeof(struct ohash_info)); - - info.halloc = hash_halloc; + info.calloc = hash_calloc; info.alloc = hash_alloc; - info.hfree = hash_free; - info.key_offset = offsetof(struct match, id); + info.free = hash_free; + info.key_offset = offsetof(struct match, pageid); *sz = cur = maxres = 0; sql = NULL; @@ -161,7 +214,7 @@ mansearch(const struct mansearch *search, */ if (NULL == getcwd(buf, PATH_MAX)) { - perror(NULL); + perror("getcwd"); goto out; } else if (-1 == (fd = open(buf, O_RDONLY, 0))) { perror(buf); @@ -186,11 +239,10 @@ mansearch(const struct mansearch *search, } else if (-1 == chdir(paths->paths[i])) { perror(paths->paths[i]); continue; - } + } - c = sqlite3_open_v2 - (MANDOC_DB, &db, - SQLITE_OPEN_READONLY, NULL); + c = sqlite3_open_v2(MANDOC_DB, &db, + SQLITE_OPEN_READONLY, NULL); if (SQLITE_OK != c) { perror(MANDOC_DB); @@ -204,10 +256,12 @@ mansearch(const struct mansearch *search, */ c = sqlite3_create_function(db, "match", 2, - SQLITE_ANY, NULL, sql_match, NULL, NULL); + SQLITE_UTF8 | SQLITE_DETERMINISTIC, + NULL, sql_match, NULL, NULL); assert(SQLITE_OK == c); c = sqlite3_create_function(db, "regexp", 2, - SQLITE_ANY, NULL, sql_regexp, NULL, NULL); + SQLITE_UTF8 | SQLITE_DETERMINISTIC, + NULL, sql_regexp, NULL, NULL); assert(SQLITE_OK == c); j = 1; @@ -220,7 +274,8 @@ mansearch(const struct mansearch *search, SQL_BIND_BLOB(db, s, j, ep->regexp); } else SQL_BIND_TEXT(db, s, j, ep->substr); - SQL_BIND_INT64(db, s, j, ep->bits); + if (0 == ((TYPE_Nd | TYPE_Nm) & ep->bits)) + SQL_BIND_INT64(db, s, j, ep->bits); } memset(&htab, 0, sizeof(struct ohash)); @@ -235,17 +290,20 @@ mansearch(const struct mansearch *search, * distribution of buckets in the table. */ while (SQLITE_ROW == (c = sqlite3_step(s))) { - id = sqlite3_column_int64(s, 1); - idx = ohash_lookup_memory - (&htab, (char *)&id, - sizeof(uint64_t), (uint32_t)id); + pageid = sqlite3_column_int64(s, 2); + idx = ohash_lookup_memory(&htab, + (char *)&pageid, sizeof(uint64_t), + (uint32_t)pageid); if (NULL != ohash_find(&htab, idx)) continue; mp = mandoc_calloc(1, sizeof(struct match)); - mp->id = id; - mp->form = sqlite3_column_int(s, 0); + mp->pageid = pageid; + mp->form = sqlite3_column_int(s, 1); + if (TYPE_Nd == outbit) + mp->desc = mandoc_strdup((const char *) + sqlite3_column_text(s, 0)); ohash_insert(&htab, idx, mp); } @@ -254,15 +312,16 @@ mansearch(const struct mansearch *search, sqlite3_finalize(s); - c = sqlite3_prepare_v2(db, - "SELECT * FROM mlinks WHERE pageid=?" - " ORDER BY sec, arch, name", + c = sqlite3_prepare_v2(db, + "SELECT sec, arch, name, pageid FROM mlinks " + "WHERE pageid=? ORDER BY sec, arch, name", -1, &s, NULL); if (SQLITE_OK != c) fprintf(stderr, "%s\n", sqlite3_errmsg(db)); c = sqlite3_prepare_v2(db, - "SELECT * FROM keys WHERE pageid=? AND bits & ?", + "SELECT bits, key, pageid FROM keys " + "WHERE pageid=? AND bits & ?", -1, &s2, NULL); if (SQLITE_OK != c) fprintf(stderr, "%s\n", sqlite3_errmsg(db)); @@ -272,15 +331,17 @@ mansearch(const struct mansearch *search, mp = ohash_next(&htab, &idx)) { if (cur + 1 > maxres) { maxres += 1024; - *res = mandoc_realloc - (*res, maxres * sizeof(struct manpage)); + *res = mandoc_reallocarray(*res, + maxres, sizeof(struct manpage)); } mpage = *res + cur; + mpage->sec = 10; mpage->form = mp->form; - buildnames(mpage, db, s, mp->id, + buildnames(mpage, db, s, mp->pageid, paths->paths[i], mp->form); - mpage->output = outbit ? - buildoutput(db, s2, mp->id, outbit) : NULL; + mpage->output = TYPE_Nd & outbit ? + mp->desc : outbit ? + buildoutput(db, s2, mp->pageid, outbit) : NULL; free(mp); cur++; @@ -291,29 +352,46 @@ mansearch(const struct mansearch *search, sqlite3_close(db); ohash_delete(&htab); } + qsort(*res, cur, sizeof(struct manpage), manpage_compare); rc = 1; out: - exprfree(e); - if (-1 != fd) + if (-1 != fd) { + if (-1 == fchdir(fd)) + perror(buf); close(fd); + } + exprfree(e); free(sql); *sz = cur; return(rc); } +static int +manpage_compare(const void *vp1, const void *vp2) +{ + const struct manpage *mp1, *mp2; + int diff; + + mp1 = vp1; + mp2 = vp2; + diff = mp1->sec - mp2->sec; + return(diff ? diff : strcasecmp(mp1->names, mp2->names)); +} + static void buildnames(struct manpage *mpage, sqlite3 *db, sqlite3_stmt *s, - uint64_t id, const char *path, int form) + uint64_t pageid, const char *path, int form) { char *newnames, *prevsec, *prevarch; const char *oldnames, *sep1, *name, *sec, *sep2, *arch, *fsec; size_t i; int c; + mpage->file = NULL; mpage->names = NULL; prevsec = prevarch = NULL; i = 1; - SQL_BIND_INT64(db, s, i, id); + SQL_BIND_INT64(db, s, i, pageid); while (SQLITE_ROW == (c = sqlite3_step(s))) { /* Decide whether we already have some names. */ @@ -328,10 +406,15 @@ buildnames(struct manpage *mpage, sqlite3 *db, sqlite3 /* Fetch the next name. */ - sec = sqlite3_column_text(s, 0); - arch = sqlite3_column_text(s, 1); - name = sqlite3_column_text(s, 2); + sec = (const char *)sqlite3_column_text(s, 0); + arch = (const char *)sqlite3_column_text(s, 1); + name = (const char *)sqlite3_column_text(s, 2); + /* Remember the first section found. */ + + if (9 < mpage->sec && '1' <= *sec && '9' >= *sec) + mpage->sec = (*sec - '1') + 1; + /* If the section changed, append the old one. */ if (NULL != prevsec && @@ -395,7 +478,7 @@ buildnames(struct manpage *mpage, sqlite3 *db, sqlite3 } static char * -buildoutput(sqlite3 *db, sqlite3_stmt *s, uint64_t id, uint64_t outbit) +buildoutput(sqlite3 *db, sqlite3_stmt *s, uint64_t pageid, uint64_t outbit) { char *output, *newoutput; const char *oldoutput, *sep1, *data; @@ -404,7 +487,7 @@ buildoutput(sqlite3 *db, sqlite3_stmt *s, uint64_t id, output = NULL; i = 1; - SQL_BIND_INT64(db, s, i, id); + SQL_BIND_INT64(db, s, i, pageid); SQL_BIND_INT64(db, s, i, outbit); while (SQLITE_ROW == (c = sqlite3_step(s))) { if (NULL == output) { @@ -414,7 +497,7 @@ buildoutput(sqlite3 *db, sqlite3_stmt *s, uint64_t id, oldoutput = output; sep1 = " # "; } - data = sqlite3_column_text(s, 1); + data = (const char *)sqlite3_column_text(s, 1); mandoc_asprintf(&newoutput, "%s%s%s", oldoutput, sep1, data); free(output); @@ -482,7 +565,8 @@ sql_statement(const struct expr *e) size_t sz; int needop; - sql = mandoc_strdup("SELECT * FROM mpages WHERE "); + sql = mandoc_strdup( + "SELECT desc, form, pageid FROM mpages WHERE "); sz = strlen(sql); for (needop = 0; NULL != e; e = e->next) { @@ -492,11 +576,25 @@ sql_statement(const struct expr *e) sql_append(&sql, &sz, " OR ", 1); if (e->open) sql_append(&sql, &sz, "(", e->open); - sql_append(&sql, &sz, NULL == e->substr ? - "id IN (SELECT pageid FROM keys " - "WHERE key REGEXP ? AND bits & ?)" : - "id IN (SELECT pageid FROM keys " - "WHERE key MATCH ? AND bits & ?)", 1); + sql_append(&sql, &sz, + TYPE_Nd & e->bits + ? (NULL == e->substr + ? "desc REGEXP ?" + : "desc MATCH ?") + : TYPE_Nm == e->bits + ? (NULL == e->substr + ? "pageid IN (SELECT pageid FROM names " + "WHERE name REGEXP ?)" + : e->equal + ? "pageid IN (SELECT pageid FROM names " + "WHERE name = ?)" + : "pageid IN (SELECT pageid FROM names " + "WHERE name MATCH ?)") + : (NULL == e->substr + ? "pageid IN (SELECT pageid FROM keys " + "WHERE key REGEXP ? AND bits & ?)" + : "pageid IN (SELECT pageid FROM keys " + "WHERE key MATCH ? AND bits & ?)"), 1); if (e->close) sql_append(&sql, &sz, ")", e->close); needop = 1; @@ -513,12 +611,13 @@ sql_statement(const struct expr *e) static struct expr * exprcomp(const struct mansearch *search, int argc, char *argv[]) { + uint64_t mask; int i, toopen, logic, igncase, toclose; - struct expr *first, *next, *cur; + struct expr *first, *prev, *cur, *next; first = cur = NULL; logic = igncase = toclose = 0; - toopen = 1; + toopen = NULL != search->sec || NULL != search->arch; for (i = 0; i < argc; i++) { if (0 == strcmp("(", argv[i])) { @@ -553,21 +652,46 @@ exprcomp(const struct mansearch *search, int argc, cha next = exprterm(search, argv[i], !igncase); if (NULL == next) goto fail; - next->open = toopen; - next->and = (1 == logic); - if (NULL != first) { + if (NULL == first) + first = next; + else cur->next = next; - cur = next; - } else - cur = first = next; + prev = cur = next; + + /* + * Searching for descriptions must be split out + * because they are stored in the mpages table, + * not in the keys table. + */ + + for (mask = TYPE_Nm; mask <= TYPE_Nd; mask <<= 1) { + if (mask & cur->bits && ~mask & cur->bits) { + next = mandoc_calloc(1, + sizeof(struct expr)); + memcpy(next, cur, sizeof(struct expr)); + prev->open = 1; + cur->bits = mask; + cur->next = next; + cur = next; + cur->bits &= ~mask; + } + } + prev->and = (1 == logic); + prev->open += toopen; + if (cur != prev) + cur->close = 1; + toopen = logic = igncase = 0; } if (toopen || logic || igncase || toclose) goto fail; - cur->close++; - cur = exprspec(cur, TYPE_arch, search->arch, "^(%s|any)$"); - exprspec(cur, TYPE_sec, search->sec, "^%s$"); + if (NULL != search->sec || NULL != search->arch) + cur->close++; + if (NULL != search->arch) + cur = exprspec(cur, TYPE_arch, search->arch, "^(%s|any)$"); + if (NULL != search->sec) + exprspec(cur, TYPE_sec, search->sec, "^%s$"); return(first); @@ -585,9 +709,6 @@ exprspec(struct expr *cur, uint64_t key, const char *v char *cp; int irc; - if (NULL == value) - return(cur); - mandoc_asprintf(&cp, format, value); cur->next = mandoc_calloc(1, sizeof(struct expr)); cur = cur->next; @@ -608,7 +729,7 @@ exprterm(const struct mansearch *search, char *buf, in { char errbuf[BUFSIZ]; struct expr *e; - char *key, *v; + char *key, *val; uint64_t iterbit; int i, irc; @@ -617,41 +738,65 @@ exprterm(const struct mansearch *search, char *buf, in e = mandoc_calloc(1, sizeof(struct expr)); - /*"whatis" mode uses an opaque string and default fields. */ - - if (MANSEARCH_WHATIS & search->flags) { - e->substr = buf; + if (MANSEARCH_MAN & search->flags) { e->bits = search->deftype; + e->substr = buf; + e->equal = 1; return(e); } /* - * If no =~ is specified, search with equality over names and - * descriptions. - * If =~ begins the phrase, use name and description fields. + * Look for an '=' or '~' operator, + * unless forced to some fixed macro keys. */ - if (NULL == (v = strpbrk(buf, "=~"))) { - e->substr = buf; + if (MANSEARCH_WHATIS & search->flags) + val = NULL; + else + val = strpbrk(buf, "=~"); + + if (NULL == val) { e->bits = search->deftype; - return(e); - } else if (v == buf) - e->bits = search->deftype; + e->substr = buf; - if ('~' == *v++) { + /* + * Found an operator. + * Regexp search is requested by !e->substr. + */ + + } else { + if (val == buf) + e->bits = search->deftype; + if ('=' == *val) + e->substr = val + 1; + *val++ = '\0'; if (NULL != strstr(buf, "arch")) cs = 0; - if (0 != (irc = regcomp(&e->regexp, v, - REG_EXTENDED | REG_NOSUB | (cs ? 0 : REG_ICASE)))) { + } + + /* Compile regular expressions. */ + + if (MANSEARCH_WHATIS & search->flags) { + e->substr = NULL; + mandoc_asprintf(&val, "[[:<:]]%s[[:>:]]", buf); + } + + if (NULL == e->substr) { + irc = regcomp(&e->regexp, val, + REG_EXTENDED | REG_NOSUB | (cs ? 0 : REG_ICASE)); + if (MANSEARCH_WHATIS & search->flags) + free(val); + if (irc) { regerror(irc, &e->regexp, errbuf, sizeof(errbuf)); fprintf(stderr, "regcomp: %s\n", errbuf); free(e); return(NULL); } - } else - e->substr = v; - v[-1] = '\0'; + } + if (e->bits) + return(e); + /* * Parse out all possible fields. * If the field doesn't resolve, bail. @@ -694,10 +839,10 @@ exprfree(struct expr *p) } static void * -hash_halloc(size_t sz, void *arg) +hash_calloc(size_t nmemb, size_t sz, void *arg) { - return(mandoc_calloc(sz, 1)); + return(mandoc_calloc(nmemb, sz)); } static void * @@ -708,7 +853,7 @@ hash_alloc(size_t sz, void *arg) } static void -hash_free(void *p, size_t sz, void *arg) +hash_free(void *p, void *arg) { free(p);