1 /* $OpenBSD: mansearch.c,v 1.50 2016/07/09 15:23:36 schwarze Exp $ */
3 * Copyright (c) 2012 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2013, 2014, 2015, 2016 Ingo Schwarze <schwarze@openbsd.org>
6 * Permission to use, copy, modify, and distribute this software for any
7 * purpose with or without fee is hereby granted, provided that the above
8 * copyright notice and this permission notice appear in all copies.
10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
21 #include <sys/types.h>
40 #include "mandoc_aux.h"
41 #include "mandoc_ohash.h"
43 #include "mansearch.h"
48 struct dbm_match match; /* Match type and expression. */
49 uint64_t bits; /* Type mask. */
50 /* Used for OR and AND groups: */
51 struct expr *next; /* Next child in the parent group. */
52 struct expr *child; /* First child in this group. */
53 enum { EXPR_TERM, EXPR_OR, EXPR_AND } type;
56 const char *const mansearch_keynames[KEY_MAX] = {
57 "arch", "sec", "Xr", "Ar", "Fa", "Fl", "Dv", "Fn",
58 "Ic", "Pa", "Cm", "Li", "Em", "Cd", "Va", "Ft",
59 "Tn", "Er", "Ev", "Sy", "Sh", "In", "Ss", "Ox",
60 "An", "Mt", "St", "Bx", "At", "Nx", "Fx", "Lk",
61 "Ms", "Bsx", "Dx", "Rs", "Vt", "Lb", "Nm", "Nd"
65 static struct ohash *manmerge(struct expr *, struct ohash *);
66 static struct ohash *manmerge_term(struct expr *, struct ohash *);
67 static struct ohash *manmerge_or(struct expr *, struct ohash *);
68 static struct ohash *manmerge_and(struct expr *, struct ohash *);
69 static char *buildnames(const struct dbm_page *);
70 static char *buildoutput(size_t, int32_t);
71 static size_t lstlen(const char *);
72 static void lstcat(char *, size_t *, const char *);
73 static int lstmatch(const char *, const char *);
74 static struct expr *exprcomp(const struct mansearch *,
75 int, char *[], int *);
76 static struct expr *expr_and(const struct mansearch *,
77 int, char *[], int *);
78 static struct expr *exprterm(const struct mansearch *,
79 int, char *[], int *);
80 static void exprfree(struct expr *);
81 static int manpage_compare(const void *, const void *);
85 mansearch(const struct mansearch *search,
86 const struct manpaths *paths,
87 int argc, char *argv[],
88 struct manpage **res, size_t *sz)
93 struct dbm_page *page;
94 struct manpage *mpage;
96 size_t cur, i, maxres, outkey;
98 int argi, chdir_status, getcwd_status, im;
101 if ((e = exprcomp(search, argc, argv, &argi)) == NULL) {
110 if (search->outkey != NULL)
111 for (im = 0; im < KEY_MAX; im++)
112 if (0 == strcasecmp(search->outkey,
113 mansearch_keynames[im])) {
119 * Remember the original working directory, if possible.
120 * This will be needed if the second or a later directory
121 * is given as a relative path.
122 * Do not error out if the current directory is not
123 * searchable: Maybe it won't be needed after all.
126 if (getcwd(buf, PATH_MAX) == NULL) {
128 (void)strlcpy(buf, strerror(errno), sizeof(buf));
133 * Loop over the directories (containing databases) for us to
135 * Don't let missing/bad databases/directories phase us.
136 * In each, try to open the resident database and, if it opens,
137 * scan it for our match expression.
141 for (i = 0; i < paths->sz; i++) {
142 if (chdir_status && paths->paths[i][0] != '/') {
143 if ( ! getcwd_status) {
144 warnx("%s: getcwd: %s", paths->paths[i], buf);
146 } else if (chdir(buf) == -1) {
151 if (chdir(paths->paths[i]) == -1) {
152 warn("%s", paths->paths[i]);
157 if (dbm_open(MANDOC_DB) == -1) {
158 warn("%s/%s", paths->paths[i], MANDOC_DB);
162 if ((htab = manmerge(e, NULL)) == NULL) {
167 for (rp = ohash_first(htab, &slot); rp != NULL;
168 rp = ohash_next(htab, &slot)) {
169 page = dbm_page_get(rp->page);
171 if (lstmatch(search->sec, page->sect) == 0 ||
172 lstmatch(search->arch, page->arch) == 0)
175 if (cur + 1 > maxres) {
177 *res = mandoc_reallocarray(*res,
178 maxres, sizeof(**res));
181 mandoc_asprintf(&mpage->file, "%s/%s",
182 paths->paths[i], page->file + 1);
183 mpage->names = buildnames(page);
184 mpage->output = (int)outkey == KEY_Nd ?
185 mandoc_strdup(page->desc) :
186 buildoutput(outkey, page->addr);
188 mpage->bits = rp->bits;
189 mpage->sec = *page->sect - '0';
190 if (mpage->sec < 0 || mpage->sec > 9)
192 mpage->form = *page->file;
201 * In man(1) mode, prefer matches in earlier trees
202 * over matches in later trees.
205 if (cur && search->firstmatch)
208 qsort(*res, cur, sizeof(struct manpage), manpage_compare);
209 if (chdir_status && getcwd_status && chdir(buf) == -1)
217 * Merge the results for the expression tree rooted at e
218 * into the the result list htab.
220 static struct ohash *
221 manmerge(struct expr *e, struct ohash *htab)
225 return manmerge_term(e, htab);
227 return manmerge_or(e->child, htab);
229 return manmerge_and(e->child, htab);
235 static struct ohash *
236 manmerge_term(struct expr *e, struct ohash *htab)
238 struct dbm_res res, *rp;
244 htab = mandoc_malloc(sizeof(*htab));
245 mandoc_ohash_init(htab, 4, offsetof(struct dbm_res, page));
248 for (im = 0, ib = 1; im < KEY_MAX; im++, ib <<= 1) {
249 if ((e->bits & ib) == 0)
254 dbm_page_byarch(&e->match);
257 dbm_page_bysect(&e->match);
260 dbm_page_byname(&e->match);
263 dbm_page_bydesc(&e->match);
266 dbm_page_bymacro(im - 2, &e->match);
271 * When hashing for deduplication, use the unique
272 * page ID itself instead of a hash function;
273 * that is quite efficient.
277 res = dbm_page_next();
280 slot = ohash_lookup_memory(htab,
281 (char *)&res, sizeof(res.page), res.page);
282 if ((rp = ohash_find(htab, slot)) != NULL) {
283 rp->bits |= res.bits;
286 rp = mandoc_malloc(sizeof(*rp));
288 ohash_insert(htab, slot, rp);
294 static struct ohash *
295 manmerge_or(struct expr *e, struct ohash *htab)
298 htab = manmerge(e, htab);
304 static struct ohash *
305 manmerge_and(struct expr *e, struct ohash *htab)
307 struct ohash *hand, *h1, *h2;
309 unsigned int slot1, slot2;
311 /* Evaluate the first term of the AND clause. */
313 hand = manmerge(e, NULL);
315 while ((e = e->next) != NULL) {
317 /* Evaluate the next term and prepare for ANDing. */
319 h2 = manmerge(e, NULL);
320 if (ohash_entries(h2) < ohash_entries(hand)) {
325 hand = mandoc_malloc(sizeof(*hand));
326 mandoc_ohash_init(hand, 4, offsetof(struct dbm_res, page));
328 /* Keep all pages that are in both result sets. */
330 for (res = ohash_first(h1, &slot1); res != NULL;
331 res = ohash_next(h1, &slot1)) {
332 if (ohash_find(h2, ohash_lookup_memory(h2,
333 (char *)res, sizeof(res->page),
337 ohash_insert(hand, ohash_lookup_memory(hand,
338 (char *)res, sizeof(res->page),
342 /* Discard the merged results. */
344 for (res = ohash_first(h2, &slot2); res != NULL;
345 res = ohash_next(h2, &slot2))
353 /* Merge the result of the AND into htab. */
358 for (res = ohash_first(hand, &slot1); res != NULL;
359 res = ohash_next(hand, &slot1)) {
360 slot2 = ohash_lookup_memory(htab,
361 (char *)res, sizeof(res->page), res->page);
362 if (ohash_find(htab, slot2) == NULL)
363 ohash_insert(htab, slot2, res);
368 /* Discard the merged result. */
376 mansearch_free(struct manpage *res, size_t sz)
380 for (i = 0; i < sz; i++) {
389 manpage_compare(const void *vp1, const void *vp2)
391 const struct manpage *mp1, *mp2;
396 return (diff = mp2->bits - mp1->bits) ? diff :
397 (diff = mp1->sec - mp2->sec) ? diff :
398 strcasecmp(mp1->names, mp2->names);
402 buildnames(const struct dbm_page *page)
407 sz = lstlen(page->name) + 1 + lstlen(page->sect) +
408 (page->arch == NULL ? 0 : 1 + lstlen(page->arch)) + 2;
409 buf = mandoc_malloc(sz);
411 lstcat(buf, &i, page->name);
413 lstcat(buf, &i, page->sect);
414 if (page->arch != NULL) {
416 lstcat(buf, &i, page->arch);
425 * Count the buffer space needed to print the NUL-terminated
426 * list of NUL-terminated strings, when printing two separator
427 * characters between strings.
430 lstlen(const char *cp)
434 for (sz = 0;; sz++) {
439 } else if (cp[0] < ' ')
447 * Print the NUL-terminated list of NUL-terminated strings
448 * into the buffer, seperating strings with a comma and a blank.
451 lstcat(char *buf, size_t *i, const char *cp)
459 } else if (cp[0] >= ' ')
466 * Return 1 if the string *want occurs in any of the strings
467 * in the NUL-terminated string list *have, or 0 otherwise.
468 * If either argument is NULL or empty, assume no filtering
469 * is desired and return 1.
472 lstmatch(const char *want, const char *have)
474 if (want == NULL || have == NULL || *have == '\0')
476 while (*have != '\0') {
477 if (strcasestr(have, want) != NULL)
479 have = strchr(have, '\0') + 1;
485 * Build a list of values taken by the macro im
486 * in the manual page with big-endian address addr.
489 buildoutput(size_t im, int32_t addr)
491 const char *oldoutput, *sep;
492 char *output, *newoutput, *value;
495 dbm_macro_bypage(im - 2, addr);
496 while ((value = dbm_macro_next()) != NULL) {
497 if (output == NULL) {
504 mandoc_asprintf(&newoutput, "%s%s%s", oldoutput, sep, value);
512 * Compile a set of string tokens into an expression.
513 * Tokens in "argv" are assumed to be individual expression atoms (e.g.,
514 * "(", "foo=bar", etc.).
517 exprcomp(const struct mansearch *search, int argc, char *argv[], int *argi)
519 struct expr *parent, *child;
520 int needterm, nested;
522 if ((nested = *argi) == argc)
525 parent = child = NULL;
526 while (*argi < argc) {
527 if (strcmp(")", argv[*argi]) == 0) {
529 warnx("missing term "
530 "before closing parenthesis");
534 warnx("ignoring unmatched right parenthesis");
538 if (strcmp("-o", argv[*argi]) == 0) {
541 warnx("ignoring -o after %s",
544 warnx("ignoring initial -o");
552 child = expr_and(search, argc, argv, argi);
555 if (parent == NULL) {
556 parent = mandoc_calloc(1, sizeof(*parent));
557 parent->type = EXPR_OR;
559 parent->child = child;
561 child->next = expr_and(search, argc, argv, argi);
564 if (needterm && *argi)
565 warnx("ignoring trailing %s", argv[*argi - 1]);
566 return parent == NULL ? child : parent;
570 expr_and(const struct mansearch *search, int argc, char *argv[], int *argi)
572 struct expr *parent, *child;
576 parent = child = NULL;
577 while (*argi < argc) {
578 if (strcmp(")", argv[*argi]) == 0) {
580 warnx("missing term "
581 "before closing parenthesis");
585 if (strcmp("-o", argv[*argi]) == 0)
587 if (strcmp("-a", argv[*argi]) == 0) {
590 warnx("ignoring -a after %s",
593 warnx("ignoring initial -a");
602 child = exprterm(search, argc, argv, argi);
608 if (parent == NULL) {
609 parent = mandoc_calloc(1, sizeof(*parent));
610 parent->type = EXPR_AND;
612 parent->child = child;
614 child->next = exprterm(search, argc, argv, argi);
615 if (child->next != NULL) {
620 if (needterm && *argi)
621 warnx("ignoring trailing %s", argv[*argi - 1]);
622 return parent == NULL ? child : parent;
626 exprterm(const struct mansearch *search, int argc, char *argv[], int *argi)
634 if (strcmp("(", argv[*argi]) == 0) {
636 e = exprcomp(search, argc, argv, argi);
638 assert(strcmp(")", argv[*argi]) == 0);
641 warnx("unclosed parenthesis");
645 e = mandoc_calloc(1, sizeof(*e));
651 if (search->argmode == ARG_NAME) {
653 e->match.type = DBM_EXACT;
654 e->match.str = argv[(*argi)++];
659 * Separate macro keys from search string.
660 * If needed, request regular expression handling.
663 if (search->argmode == ARG_WORD) {
665 e->match.type = DBM_REGEX;
667 mandoc_asprintf(&val, "[[:<:]]%s[[:>:]]", argv[*argi]);
669 mandoc_asprintf(&val, "\\<%s\\>", argv[*argi]);
671 mandoc_asprintf(&val,
672 "(^|[^a-zA-Z01-9_])%s([^a-zA-Z01-9_]|$)", argv[*argi]);
675 } else if ((val = strpbrk(argv[*argi], "=~")) == NULL) {
676 e->bits = TYPE_Nm | TYPE_Nd;
677 e->match.type = DBM_SUB;
678 e->match.str = argv[*argi];
680 if (val == argv[*argi])
681 e->bits = TYPE_Nm | TYPE_Nd;
683 e->match.type = DBM_SUB;
684 e->match.str = val + 1;
686 e->match.type = DBM_REGEX;
688 if (strstr(argv[*argi], "arch") != NULL)
692 /* Compile regular expressions. */
694 if (e->match.type == DBM_REGEX) {
695 e->match.re = mandoc_malloc(sizeof(*e->match.re));
696 irc = regcomp(e->match.re, val,
697 REG_EXTENDED | REG_NOSUB | (cs ? 0 : REG_ICASE));
699 regerror(irc, e->match.re, errbuf, sizeof(errbuf));
700 warnx("regcomp /%s/: %s", val, errbuf);
702 if (search->argmode == ARG_WORD)
718 * Parse out all possible fields.
719 * If the field doesn't resolve, bail.
722 while (NULL != (key = strsep(&argv[*argi], ","))) {
725 for (i = 0, iterbit = 1; i < KEY_MAX; i++, iterbit <<= 1) {
726 if (0 == strcasecmp(key, mansearch_keynames[i])) {
732 if (strcasecmp(key, "any"))
733 warnx("treating unknown key "
734 "\"%s\" as \"any\"", key);
744 exprfree(struct expr *e)
748 if (e->child != NULL)