Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.71
1.71 ! schwarze 1: /* $Id: mandocdb.c,v 1.70 2014/01/19 22:48:00 schwarze Exp $ */
1.1 schwarze 2: /*
1.47 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.52 schwarze 4: * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 5: *
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.
9: *
10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR 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.
17: */
1.47 schwarze 18: #include <sys/stat.h>
1.1 schwarze 19:
20: #include <assert.h>
1.33 schwarze 21: #include <ctype.h>
1.34 schwarze 22: #include <errno.h>
1.1 schwarze 23: #include <fcntl.h>
1.47 schwarze 24: #include <fts.h>
1.1 schwarze 25: #include <getopt.h>
1.44 schwarze 26: #include <limits.h>
1.47 schwarze 27: #include <stddef.h>
1.1 schwarze 28: #include <stdio.h>
29: #include <stdint.h>
30: #include <stdlib.h>
31: #include <string.h>
1.14 schwarze 32: #include <unistd.h>
1.1 schwarze 33:
1.47 schwarze 34: #include <ohash.h>
35: #include <sqlite3.h>
36:
37: #include "mdoc.h"
1.1 schwarze 38: #include "man.h"
39: #include "mandoc.h"
1.10 schwarze 40: #include "manpath.h"
1.47 schwarze 41: #include "mansearch.h"
1.1 schwarze 42:
1.68 schwarze 43: extern int mansearch_keymax;
44: extern const char *const mansearch_keynames[];
45:
1.47 schwarze 46: #define SQL_EXEC(_v) \
47: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
48: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
49: #define SQL_BIND_TEXT(_s, _i, _v) \
50: if (SQLITE_OK != sqlite3_bind_text \
51: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
52: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
53: #define SQL_BIND_INT(_s, _i, _v) \
54: if (SQLITE_OK != sqlite3_bind_int \
55: ((_s), (_i)++, (_v))) \
56: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
57: #define SQL_BIND_INT64(_s, _i, _v) \
58: if (SQLITE_OK != sqlite3_bind_int64 \
59: ((_s), (_i)++, (_v))) \
60: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
61: #define SQL_STEP(_s) \
62: if (SQLITE_DONE != sqlite3_step((_s))) \
63: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
1.1 schwarze 64:
1.47 schwarze 65: enum op {
66: OP_DEFAULT = 0, /* new dbs from dir list or default config */
67: OP_CONFFILE, /* new databases from custom config file */
68: OP_UPDATE, /* delete/add entries in existing database */
69: OP_DELETE, /* delete entries from existing database */
70: OP_TEST /* change no databases, report potential problems */
71: };
1.11 schwarze 72:
1.47 schwarze 73: enum form {
74: FORM_NONE, /* format is unknown */
75: FORM_SRC, /* format is -man or -mdoc */
76: FORM_CAT /* format is cat */
77: };
1.28 schwarze 78:
1.47 schwarze 79: struct str {
1.53 schwarze 80: char *rendered; /* key in UTF-8 or ASCII form */
1.47 schwarze 81: const struct mpage *mpage; /* if set, the owning parse */
82: uint64_t mask; /* bitmask in sequence */
1.53 schwarze 83: char key[]; /* may contain escape sequences */
1.28 schwarze 84: };
85:
1.47 schwarze 86: struct inodev {
87: ino_t st_ino;
88: dev_t st_dev;
89: };
1.28 schwarze 90:
1.47 schwarze 91: struct mpage {
92: struct inodev inodev; /* used for hashing routine */
93: enum form form; /* format from file content */
94: char *sec; /* section from file content */
95: char *arch; /* architecture from file content */
96: char *title; /* title from file content */
97: char *desc; /* description from file content */
98: struct mlink *mlinks; /* singly linked list */
1.28 schwarze 99: };
100:
1.47 schwarze 101: struct mlink {
102: char file[PATH_MAX]; /* filename rel. to manpath */
103: enum form dform; /* format from directory */
104: enum form fform; /* format from file name suffix */
105: char *dsec; /* section from directory */
106: char *arch; /* architecture from directory */
107: char *name; /* name from file name (not empty) */
108: char *fsec; /* section from file name suffix */
109: struct mlink *next; /* singly linked list */
1.2 schwarze 110: };
111:
1.47 schwarze 112: enum stmt {
113: STMT_DELETE_PAGE = 0, /* delete mpage */
114: STMT_INSERT_PAGE, /* insert mpage */
115: STMT_INSERT_LINK, /* insert mlink */
116: STMT_INSERT_KEY, /* insert parsed key */
117: STMT__MAX
1.1 schwarze 118: };
119:
1.47 schwarze 120: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
1.1 schwarze 121:
1.19 schwarze 122: struct mdoc_handler {
1.47 schwarze 123: mdoc_fp fp; /* optional handler */
124: uint64_t mask; /* set unless handler returns 0 */
1.19 schwarze 125: };
126:
1.47 schwarze 127: static void dbclose(int);
1.62 schwarze 128: static void dbadd(const struct mpage *, struct mchars *);
1.47 schwarze 129: static int dbopen(int);
130: static void dbprune(void);
131: static void filescan(const char *);
132: static void *hash_alloc(size_t, void *);
133: static void hash_free(void *, size_t, void *);
134: static void *hash_halloc(size_t, void *);
135: static void mlink_add(struct mlink *, const struct stat *);
1.50 schwarze 136: static int mlink_check(struct mpage *, struct mlink *);
1.47 schwarze 137: static void mlink_free(struct mlink *);
138: static void mlinks_undupe(struct mpage *);
139: static void mpages_free(void);
1.58 schwarze 140: static void mpages_merge(struct mchars *, struct mparse *);
1.47 schwarze 141: static void parse_cat(struct mpage *);
142: static void parse_man(struct mpage *, const struct man_node *);
143: static void parse_mdoc(struct mpage *, const struct mdoc_node *);
144: static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
145: static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
146: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
147: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
148: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
149: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
150: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
151: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
1.69 schwarze 152: static void putkey(const struct mpage *, char *, uint64_t);
1.47 schwarze 153: static void putkeys(const struct mpage *,
154: const char *, size_t, uint64_t);
155: static void putmdockey(const struct mpage *,
156: const struct mdoc_node *, uint64_t);
1.53 schwarze 157: static void render_key(struct mchars *, struct str *);
1.47 schwarze 158: static void say(const char *, const char *, ...);
159: static int set_basedir(const char *);
160: static int treescan(void);
161: static size_t utf8(unsigned int, char [7]);
162:
163: static char *progname;
1.59 schwarze 164: static int nodb; /* no database changes */
165: static int quick; /* abort the parse early */
1.47 schwarze 166: static int use_all; /* use all found files */
167: static int verb; /* print what we're doing */
168: static int warnings; /* warn about crap */
1.52 schwarze 169: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 170: static int exitcode; /* to be returned by main */
171: static enum op op; /* operational mode */
172: static char basedir[PATH_MAX]; /* current base directory */
173: static struct ohash mpages; /* table of distinct manual pages */
174: static struct ohash mlinks; /* table of directory entries */
175: static struct ohash strings; /* table of all strings */
176: static sqlite3 *db = NULL; /* current database */
177: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
178:
1.19 schwarze 179: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 180: { NULL, 0 }, /* Ap */
181: { NULL, 0 }, /* Dd */
182: { NULL, 0 }, /* Dt */
183: { NULL, 0 }, /* Os */
184: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
185: { parse_mdoc_head, TYPE_Ss }, /* Ss */
186: { NULL, 0 }, /* Pp */
187: { NULL, 0 }, /* D1 */
188: { NULL, 0 }, /* Dl */
189: { NULL, 0 }, /* Bd */
190: { NULL, 0 }, /* Ed */
191: { NULL, 0 }, /* Bl */
192: { NULL, 0 }, /* El */
193: { NULL, 0 }, /* It */
194: { NULL, 0 }, /* Ad */
195: { NULL, TYPE_An }, /* An */
196: { NULL, TYPE_Ar }, /* Ar */
197: { NULL, TYPE_Cd }, /* Cd */
198: { NULL, TYPE_Cm }, /* Cm */
199: { NULL, TYPE_Dv }, /* Dv */
200: { NULL, TYPE_Er }, /* Er */
201: { NULL, TYPE_Ev }, /* Ev */
202: { NULL, 0 }, /* Ex */
203: { NULL, TYPE_Fa }, /* Fa */
204: { parse_mdoc_Fd, 0 }, /* Fd */
205: { NULL, TYPE_Fl }, /* Fl */
206: { parse_mdoc_Fn, 0 }, /* Fn */
207: { NULL, TYPE_Ft }, /* Ft */
208: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 209: { NULL, TYPE_In }, /* In */
1.47 schwarze 210: { NULL, TYPE_Li }, /* Li */
211: { parse_mdoc_Nd, TYPE_Nd }, /* Nd */
212: { parse_mdoc_Nm, TYPE_Nm }, /* Nm */
213: { NULL, 0 }, /* Op */
214: { NULL, 0 }, /* Ot */
215: { NULL, TYPE_Pa }, /* Pa */
216: { NULL, 0 }, /* Rv */
1.49 schwarze 217: { NULL, TYPE_St }, /* St */
1.47 schwarze 218: { NULL, TYPE_Va }, /* Va */
219: { parse_mdoc_body, TYPE_Va }, /* Vt */
220: { parse_mdoc_Xr, 0 }, /* Xr */
221: { NULL, 0 }, /* %A */
222: { NULL, 0 }, /* %B */
223: { NULL, 0 }, /* %D */
224: { NULL, 0 }, /* %I */
225: { NULL, 0 }, /* %J */
226: { NULL, 0 }, /* %N */
227: { NULL, 0 }, /* %O */
228: { NULL, 0 }, /* %P */
229: { NULL, 0 }, /* %R */
230: { NULL, 0 }, /* %T */
231: { NULL, 0 }, /* %V */
232: { NULL, 0 }, /* Ac */
233: { NULL, 0 }, /* Ao */
234: { NULL, 0 }, /* Aq */
235: { NULL, TYPE_At }, /* At */
236: { NULL, 0 }, /* Bc */
237: { NULL, 0 }, /* Bf */
238: { NULL, 0 }, /* Bo */
239: { NULL, 0 }, /* Bq */
240: { NULL, TYPE_Bsx }, /* Bsx */
241: { NULL, TYPE_Bx }, /* Bx */
242: { NULL, 0 }, /* Db */
243: { NULL, 0 }, /* Dc */
244: { NULL, 0 }, /* Do */
245: { NULL, 0 }, /* Dq */
246: { NULL, 0 }, /* Ec */
247: { NULL, 0 }, /* Ef */
248: { NULL, TYPE_Em }, /* Em */
249: { NULL, 0 }, /* Eo */
250: { NULL, TYPE_Fx }, /* Fx */
251: { NULL, TYPE_Ms }, /* Ms */
252: { NULL, 0 }, /* No */
253: { NULL, 0 }, /* Ns */
254: { NULL, TYPE_Nx }, /* Nx */
255: { NULL, TYPE_Ox }, /* Ox */
256: { NULL, 0 }, /* Pc */
257: { NULL, 0 }, /* Pf */
258: { NULL, 0 }, /* Po */
259: { NULL, 0 }, /* Pq */
260: { NULL, 0 }, /* Qc */
261: { NULL, 0 }, /* Ql */
262: { NULL, 0 }, /* Qo */
263: { NULL, 0 }, /* Qq */
264: { NULL, 0 }, /* Re */
265: { NULL, 0 }, /* Rs */
266: { NULL, 0 }, /* Sc */
267: { NULL, 0 }, /* So */
268: { NULL, 0 }, /* Sq */
269: { NULL, 0 }, /* Sm */
270: { NULL, 0 }, /* Sx */
271: { NULL, TYPE_Sy }, /* Sy */
272: { NULL, TYPE_Tn }, /* Tn */
273: { NULL, 0 }, /* Ux */
274: { NULL, 0 }, /* Xc */
275: { NULL, 0 }, /* Xo */
276: { parse_mdoc_head, 0 }, /* Fo */
277: { NULL, 0 }, /* Fc */
278: { NULL, 0 }, /* Oo */
279: { NULL, 0 }, /* Oc */
280: { NULL, 0 }, /* Bk */
281: { NULL, 0 }, /* Ek */
282: { NULL, 0 }, /* Bt */
283: { NULL, 0 }, /* Hf */
284: { NULL, 0 }, /* Fr */
285: { NULL, 0 }, /* Ud */
286: { NULL, TYPE_Lb }, /* Lb */
287: { NULL, 0 }, /* Lp */
288: { NULL, TYPE_Lk }, /* Lk */
289: { NULL, TYPE_Mt }, /* Mt */
290: { NULL, 0 }, /* Brq */
291: { NULL, 0 }, /* Bro */
292: { NULL, 0 }, /* Brc */
293: { NULL, 0 }, /* %C */
294: { NULL, 0 }, /* Es */
295: { NULL, 0 }, /* En */
296: { NULL, TYPE_Dx }, /* Dx */
297: { NULL, 0 }, /* %Q */
298: { NULL, 0 }, /* br */
299: { NULL, 0 }, /* sp */
300: { NULL, 0 }, /* %U */
301: { NULL, 0 }, /* Ta */
1.1 schwarze 302: };
303:
304: int
1.3 schwarze 305: mandocdb(int argc, char *argv[])
1.1 schwarze 306: {
1.47 schwarze 307: int ch, i;
308: size_t j, sz;
309: const char *path_arg;
310: struct mchars *mc;
311: struct manpaths dirs;
312: struct mparse *mp;
313: struct ohash_info mpages_info, mlinks_info;
314:
315: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
316: memset(&dirs, 0, sizeof(struct manpaths));
317:
318: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
319: mpages_info.halloc = mlinks_info.halloc = hash_halloc;
320: mpages_info.hfree = mlinks_info.hfree = hash_free;
321:
322: mpages_info.key_offset = offsetof(struct mpage, inodev);
323: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 324:
325: progname = strrchr(argv[0], '/');
326: if (progname == NULL)
327: progname = argv[0];
328: else
329: ++progname;
330:
1.47 schwarze 331: /*
332: * We accept a few different invocations.
333: * The CHECKOP macro makes sure that invocation styles don't
334: * clobber each other.
335: */
336: #define CHECKOP(_op, _ch) do \
337: if (OP_DEFAULT != (_op)) { \
338: fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
339: goto usage; \
340: } while (/*CONSTCOND*/0)
1.10 schwarze 341:
1.47 schwarze 342: path_arg = NULL;
1.28 schwarze 343: op = OP_DEFAULT;
1.1 schwarze 344:
1.59 schwarze 345: while (-1 != (ch = getopt(argc, argv, "aC:d:nQT:tu:vW")))
1.1 schwarze 346: switch (ch) {
1.6 schwarze 347: case ('a'):
348: use_all = 1;
349: break;
1.25 schwarze 350: case ('C'):
1.47 schwarze 351: CHECKOP(op, ch);
352: path_arg = optarg;
1.28 schwarze 353: op = OP_CONFFILE;
1.25 schwarze 354: break;
1.1 schwarze 355: case ('d'):
1.47 schwarze 356: CHECKOP(op, ch);
357: path_arg = optarg;
1.2 schwarze 358: op = OP_UPDATE;
1.1 schwarze 359: break;
1.47 schwarze 360: case ('n'):
361: nodb = 1;
362: break;
1.59 schwarze 363: case ('Q'):
364: quick = 1;
365: break;
1.52 schwarze 366: case ('T'):
367: if (strcmp(optarg, "utf8")) {
368: fprintf(stderr, "-T%s: Unsupported "
369: "output format\n", optarg);
370: goto usage;
371: }
372: write_utf8 = 1;
373: break;
1.28 schwarze 374: case ('t'):
1.47 schwarze 375: CHECKOP(op, ch);
1.28 schwarze 376: dup2(STDOUT_FILENO, STDERR_FILENO);
377: op = OP_TEST;
1.47 schwarze 378: nodb = warnings = 1;
1.28 schwarze 379: break;
1.2 schwarze 380: case ('u'):
1.47 schwarze 381: CHECKOP(op, ch);
382: path_arg = optarg;
1.1 schwarze 383: op = OP_DELETE;
384: break;
385: case ('v'):
386: verb++;
387: break;
1.28 schwarze 388: case ('W'):
389: warnings = 1;
390: break;
1.1 schwarze 391: default:
1.28 schwarze 392: goto usage;
1.1 schwarze 393: }
394:
395: argc -= optind;
396: argv += optind;
397:
1.28 schwarze 398: if (OP_CONFFILE == op && argc > 0) {
1.47 schwarze 399: fprintf(stderr, "-C: Too many arguments\n");
1.28 schwarze 400: goto usage;
401: }
402:
1.47 schwarze 403: exitcode = (int)MANDOCLEVEL_OK;
404: mp = mparse_alloc(MPARSE_AUTO,
1.59 schwarze 405: MANDOCLEVEL_FATAL, NULL, NULL, quick);
1.47 schwarze 406: mc = mchars_alloc();
1.2 schwarze 407:
1.47 schwarze 408: ohash_init(&mpages, 6, &mpages_info);
409: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 410:
1.47 schwarze 411: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
412: /*
413: * Force processing all files.
414: */
415: use_all = 1;
1.2 schwarze 416:
1.47 schwarze 417: /*
418: * All of these deal with a specific directory.
419: * Jump into that directory then collect files specified
420: * on the command-line.
421: */
422: if (0 == set_basedir(path_arg))
423: goto out;
424: for (i = 0; i < argc; i++)
425: filescan(argv[i]);
426: if (0 == dbopen(1))
427: goto out;
428: if (OP_TEST != op)
429: dbprune();
430: if (OP_DELETE != op)
1.58 schwarze 431: mpages_merge(mc, mp);
1.47 schwarze 432: dbclose(1);
433: } else {
434: /*
435: * If we have arguments, use them as our manpaths.
436: * If we don't, grok from manpath(1) or however else
437: * manpath_parse() wants to do it.
438: */
439: if (argc > 0) {
440: dirs.paths = mandoc_calloc
441: (argc, sizeof(char *));
442: dirs.sz = (size_t)argc;
443: for (i = 0; i < argc; i++)
444: dirs.paths[i] = mandoc_strdup(argv[i]);
445: } else
446: manpath_parse(&dirs, path_arg, NULL, NULL);
1.2 schwarze 447:
1.47 schwarze 448: /*
449: * First scan the tree rooted at a base directory, then
450: * build a new database and finally move it into place.
451: * Ignore zero-length directories and strip trailing
452: * slashes.
453: */
454: for (j = 0; j < dirs.sz; j++) {
455: sz = strlen(dirs.paths[j]);
456: if (sz && '/' == dirs.paths[j][sz - 1])
457: dirs.paths[j][--sz] = '\0';
458: if (0 == sz)
459: continue;
1.2 schwarze 460:
1.47 schwarze 461: if (j) {
462: ohash_init(&mpages, 6, &mpages_info);
463: ohash_init(&mlinks, 6, &mlinks_info);
464: }
1.1 schwarze 465:
1.47 schwarze 466: if (0 == set_basedir(dirs.paths[j]))
467: goto out;
468: if (0 == treescan())
469: goto out;
470: if (0 == set_basedir(dirs.paths[j]))
471: goto out;
472: if (0 == dbopen(0))
473: goto out;
1.41 deraadt 474:
1.58 schwarze 475: mpages_merge(mc, mp);
1.47 schwarze 476: dbclose(0);
1.28 schwarze 477:
1.47 schwarze 478: if (j + 1 < dirs.sz) {
479: mpages_free();
480: ohash_delete(&mpages);
481: ohash_delete(&mlinks);
482: }
1.2 schwarze 483: }
1.47 schwarze 484: }
485: out:
486: set_basedir(NULL);
487: manpath_free(&dirs);
488: mchars_free(mc);
489: mparse_free(mp);
490: mpages_free();
491: ohash_delete(&mpages);
492: ohash_delete(&mlinks);
493: return(exitcode);
494: usage:
1.59 schwarze 495: fprintf(stderr, "usage: %s [-anQvW] [-C file] [-Tutf8]\n"
496: " %s [-anQvW] [-Tutf8] dir ...\n"
497: " %s [-nQvW] [-Tutf8] -d dir [file ...]\n"
1.47 schwarze 498: " %s [-nvW] -u dir [file ...]\n"
1.59 schwarze 499: " %s [-Q] -t file ...\n",
1.47 schwarze 500: progname, progname, progname,
501: progname, progname);
1.1 schwarze 502:
1.47 schwarze 503: return((int)MANDOCLEVEL_BADARG);
504: }
1.1 schwarze 505:
1.47 schwarze 506: /*
507: * Scan a directory tree rooted at "basedir" for manpages.
508: * We use fts(), scanning directory parts along the way for clues to our
509: * section and architecture.
510: *
511: * If use_all has been specified, grok all files.
512: * If not, sanitise paths to the following:
513: *
514: * [./]man*[/<arch>]/<name>.<section>
515: * or
516: * [./]cat<section>[/<arch>]/<name>.0
517: *
518: * TODO: accomodate for multi-language directories.
519: */
520: static int
521: treescan(void)
522: {
523: FTS *f;
524: FTSENT *ff;
525: struct mlink *mlink;
526: int dform;
1.51 schwarze 527: char *dsec, *arch, *fsec, *cp;
528: const char *path;
1.47 schwarze 529: const char *argv[2];
1.1 schwarze 530:
1.47 schwarze 531: argv[0] = ".";
532: argv[1] = (char *)NULL;
1.1 schwarze 533:
1.47 schwarze 534: /*
535: * Walk through all components under the directory, using the
536: * logical descent of files.
537: */
538: f = fts_open((char * const *)argv, FTS_LOGICAL, NULL);
539: if (NULL == f) {
540: exitcode = (int)MANDOCLEVEL_SYSERR;
541: say("", NULL);
542: return(0);
543: }
1.2 schwarze 544:
1.47 schwarze 545: dsec = arch = NULL;
546: dform = FORM_NONE;
1.2 schwarze 547:
1.47 schwarze 548: while (NULL != (ff = fts_read(f))) {
549: path = ff->fts_path + 2;
1.15 schwarze 550: /*
1.47 schwarze 551: * If we're a regular file, add an mlink by using the
552: * stored directory data and handling the filename.
1.15 schwarze 553: */
1.47 schwarze 554: if (FTS_F == ff->fts_info) {
555: if (0 == strcmp(path, MANDOC_DB))
556: continue;
557: if ( ! use_all && ff->fts_level < 2) {
558: if (warnings)
559: say(path, "Extraneous file");
560: continue;
561: } else if (NULL == (fsec =
562: strrchr(ff->fts_name, '.'))) {
563: if ( ! use_all) {
564: if (warnings)
565: say(path,
566: "No filename suffix");
567: continue;
568: }
569: } else if (0 == strcmp(++fsec, "html")) {
570: if (warnings)
571: say(path, "Skip html");
572: continue;
573: } else if (0 == strcmp(fsec, "gz")) {
574: if (warnings)
575: say(path, "Skip gz");
576: continue;
577: } else if (0 == strcmp(fsec, "ps")) {
578: if (warnings)
579: say(path, "Skip ps");
580: continue;
581: } else if (0 == strcmp(fsec, "pdf")) {
582: if (warnings)
583: say(path, "Skip pdf");
584: continue;
585: } else if ( ! use_all &&
586: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
587: (FORM_CAT == dform && strcmp(fsec, "0")))) {
588: if (warnings)
589: say(path, "Wrong filename suffix");
590: continue;
591: } else
592: fsec[-1] = '\0';
1.51 schwarze 593:
1.47 schwarze 594: mlink = mandoc_calloc(1, sizeof(struct mlink));
595: strlcpy(mlink->file, path, sizeof(mlink->file));
596: mlink->dform = dform;
1.51 schwarze 597: mlink->dsec = dsec;
598: mlink->arch = arch;
599: mlink->name = ff->fts_name;
600: mlink->fsec = fsec;
1.47 schwarze 601: mlink_add(mlink, ff->fts_statp);
602: continue;
603: } else if (FTS_D != ff->fts_info &&
604: FTS_DP != ff->fts_info) {
605: if (warnings)
606: say(path, "Not a regular file");
607: continue;
608: }
609:
610: switch (ff->fts_level) {
611: case (0):
612: /* Ignore the root directory. */
613: break;
614: case (1):
615: /*
616: * This might contain manX/ or catX/.
617: * Try to infer this from the name.
618: * If we're not in use_all, enforce it.
619: */
620: cp = ff->fts_name;
621: if (FTS_DP == ff->fts_info)
622: break;
1.15 schwarze 623:
1.47 schwarze 624: if (0 == strncmp(cp, "man", 3)) {
625: dform = FORM_SRC;
626: dsec = cp + 3;
627: } else if (0 == strncmp(cp, "cat", 3)) {
628: dform = FORM_CAT;
629: dsec = cp + 3;
1.51 schwarze 630: } else {
631: dform = FORM_NONE;
632: dsec = NULL;
1.26 schwarze 633: }
1.47 schwarze 634:
635: if (NULL != dsec || use_all)
636: break;
637:
638: if (warnings)
639: say(path, "Unknown directory part");
640: fts_set(f, ff, FTS_SKIP);
641: break;
642: case (2):
643: /*
644: * Possibly our architecture.
645: * If we're descending, keep tabs on it.
646: */
647: if (FTS_DP != ff->fts_info && NULL != dsec)
648: arch = ff->fts_name;
1.51 schwarze 649: else
650: arch = NULL;
1.47 schwarze 651: break;
652: default:
653: if (FTS_DP == ff->fts_info || use_all)
654: break;
655: if (warnings)
656: say(path, "Extraneous directory part");
657: fts_set(f, ff, FTS_SKIP);
658: break;
1.14 schwarze 659: }
1.47 schwarze 660: }
661:
662: fts_close(f);
663: return(1);
664: }
1.1 schwarze 665:
1.47 schwarze 666: /*
667: * Add a file to the mlinks table.
668: * Do not verify that it's a "valid" looking manpage (we'll do that
669: * later).
670: *
671: * Try to infer the manual section, architecture, and page name from the
672: * path, assuming it looks like
673: *
674: * [./]man*[/<arch>]/<name>.<section>
675: * or
676: * [./]cat<section>[/<arch>]/<name>.0
677: *
678: * See treescan() for the fts(3) version of this.
679: */
680: static void
681: filescan(const char *file)
682: {
683: char buf[PATH_MAX];
684: struct stat st;
685: struct mlink *mlink;
686: char *p, *start;
687:
688: assert(use_all);
689:
690: if (0 == strncmp(file, "./", 2))
691: file += 2;
692:
693: if (NULL == realpath(file, buf)) {
694: exitcode = (int)MANDOCLEVEL_BADARG;
695: say(file, NULL);
696: return;
1.63 schwarze 697: }
698:
699: if (strstr(buf, basedir) == buf)
700: start = buf + strlen(basedir) + 1;
701: else if (OP_TEST == op)
702: start = buf;
703: else {
1.47 schwarze 704: exitcode = (int)MANDOCLEVEL_BADARG;
705: say("", "%s: outside base directory", buf);
706: return;
1.63 schwarze 707: }
708:
709: if (-1 == stat(buf, &st)) {
1.47 schwarze 710: exitcode = (int)MANDOCLEVEL_BADARG;
711: say(file, NULL);
712: return;
713: } else if ( ! (S_IFREG & st.st_mode)) {
714: exitcode = (int)MANDOCLEVEL_BADARG;
715: say(file, "Not a regular file");
716: return;
1.1 schwarze 717: }
1.63 schwarze 718:
1.47 schwarze 719: mlink = mandoc_calloc(1, sizeof(struct mlink));
720: strlcpy(mlink->file, start, sizeof(mlink->file));
1.1 schwarze 721:
1.10 schwarze 722: /*
1.47 schwarze 723: * First try to guess our directory structure.
724: * If we find a separator, try to look for man* or cat*.
725: * If we find one of these and what's underneath is a directory,
726: * assume it's an architecture.
1.10 schwarze 727: */
1.47 schwarze 728: if (NULL != (p = strchr(start, '/'))) {
729: *p++ = '\0';
730: if (0 == strncmp(start, "man", 3)) {
731: mlink->dform = FORM_SRC;
1.51 schwarze 732: mlink->dsec = start + 3;
1.47 schwarze 733: } else if (0 == strncmp(start, "cat", 3)) {
734: mlink->dform = FORM_CAT;
1.51 schwarze 735: mlink->dsec = start + 3;
1.47 schwarze 736: }
1.10 schwarze 737:
1.47 schwarze 738: start = p;
739: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
740: *p++ = '\0';
1.51 schwarze 741: mlink->arch = start;
1.47 schwarze 742: start = p;
1.41 deraadt 743: }
1.47 schwarze 744: }
1.7 schwarze 745:
1.47 schwarze 746: /*
747: * Now check the file suffix.
748: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
749: */
750: p = strrchr(start, '\0');
751: while (p-- > start && '/' != *p && '.' != *p)
752: /* Loop. */ ;
753:
754: if ('.' == *p) {
755: *p++ = '\0';
1.51 schwarze 756: mlink->fsec = p;
1.47 schwarze 757: }
1.41 deraadt 758:
1.47 schwarze 759: /*
760: * Now try to parse the name.
761: * Use the filename portion of the path.
762: */
763: mlink->name = start;
764: if (NULL != (p = strrchr(start, '/'))) {
765: mlink->name = p + 1;
766: *p = '\0';
767: }
768: mlink_add(mlink, &st);
769: }
1.2 schwarze 770:
1.47 schwarze 771: static void
772: mlink_add(struct mlink *mlink, const struct stat *st)
773: {
774: struct inodev inodev;
775: struct mpage *mpage;
776: unsigned int slot;
777:
778: assert(NULL != mlink->file);
779:
1.51 schwarze 780: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
781: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
782: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
783: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 784:
785: if ('0' == *mlink->fsec) {
786: free(mlink->fsec);
787: mlink->fsec = mandoc_strdup(mlink->dsec);
788: mlink->fform = FORM_CAT;
789: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
790: mlink->fform = FORM_SRC;
791: else
792: mlink->fform = FORM_NONE;
1.1 schwarze 793:
1.47 schwarze 794: slot = ohash_qlookup(&mlinks, mlink->file);
795: assert(NULL == ohash_find(&mlinks, slot));
796: ohash_insert(&mlinks, slot, mlink);
797:
798: inodev.st_ino = st->st_ino;
799: inodev.st_dev = st->st_dev;
800: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
801: sizeof(struct inodev), inodev.st_ino);
802: mpage = ohash_find(&mpages, slot);
803: if (NULL == mpage) {
804: mpage = mandoc_calloc(1, sizeof(struct mpage));
805: mpage->inodev.st_ino = inodev.st_ino;
806: mpage->inodev.st_dev = inodev.st_dev;
807: ohash_insert(&mpages, slot, mpage);
808: } else
809: mlink->next = mpage->mlinks;
810: mpage->mlinks = mlink;
811: }
1.1 schwarze 812:
1.47 schwarze 813: static void
814: mlink_free(struct mlink *mlink)
815: {
1.1 schwarze 816:
1.47 schwarze 817: free(mlink->dsec);
818: free(mlink->arch);
819: free(mlink->name);
820: free(mlink->fsec);
821: free(mlink);
822: }
1.1 schwarze 823:
1.47 schwarze 824: static void
825: mpages_free(void)
826: {
827: struct mpage *mpage;
828: struct mlink *mlink;
829: unsigned int slot;
830:
831: mpage = ohash_first(&mpages, &slot);
832: while (NULL != mpage) {
833: while (NULL != (mlink = mpage->mlinks)) {
834: mpage->mlinks = mlink->next;
835: mlink_free(mlink);
836: }
837: free(mpage->sec);
838: free(mpage->arch);
839: free(mpage->title);
840: free(mpage->desc);
841: free(mpage);
842: mpage = ohash_next(&mpages, &slot);
843: }
844: }
1.1 schwarze 845:
1.47 schwarze 846: /*
847: * For each mlink to the mpage, check whether the path looks like
848: * it is formatted, and if it does, check whether a source manual
849: * exists by the same name, ignoring the suffix.
850: * If both conditions hold, drop the mlink.
851: */
852: static void
853: mlinks_undupe(struct mpage *mpage)
854: {
855: char buf[PATH_MAX];
856: struct mlink **prev;
857: struct mlink *mlink;
858: char *bufp;
859:
860: mpage->form = FORM_CAT;
861: prev = &mpage->mlinks;
862: while (NULL != (mlink = *prev)) {
863: if (FORM_CAT != mlink->dform) {
864: mpage->form = FORM_NONE;
865: goto nextlink;
866: }
867: if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
868: if (warnings)
869: say(mlink->file, "Filename too long");
870: goto nextlink;
1.26 schwarze 871: }
1.47 schwarze 872: bufp = strstr(buf, "cat");
873: assert(NULL != bufp);
874: memcpy(bufp, "man", 3);
875: if (NULL != (bufp = strrchr(buf, '.')))
876: *++bufp = '\0';
877: strlcat(buf, mlink->dsec, PATH_MAX);
878: if (NULL == ohash_find(&mlinks,
879: ohash_qlookup(&mlinks, buf)))
880: goto nextlink;
881: if (warnings)
882: say(mlink->file, "Man source exists: %s", buf);
883: if (use_all)
884: goto nextlink;
885: *prev = mlink->next;
886: mlink_free(mlink);
887: continue;
888: nextlink:
889: prev = &(*prev)->next;
1.1 schwarze 890: }
1.47 schwarze 891: }
1.1 schwarze 892:
1.50 schwarze 893: static int
894: mlink_check(struct mpage *mpage, struct mlink *mlink)
895: {
896: int match;
897:
898: match = 1;
899:
900: /*
901: * Check whether the manual section given in a file
902: * agrees with the directory where the file is located.
903: * Some manuals have suffixes like (3p) on their
904: * section number either inside the file or in the
905: * directory name, some are linked into more than one
906: * section, like encrypt(1) = makekey(8).
907: */
908:
909: if (FORM_SRC == mpage->form &&
910: strcasecmp(mpage->sec, mlink->dsec)) {
911: match = 0;
912: say(mlink->file, "Section \"%s\" manual in %s directory",
913: mpage->sec, mlink->dsec);
914: }
915:
916: /*
917: * Manual page directories exist for each kernel
918: * architecture as returned by machine(1).
919: * However, many manuals only depend on the
920: * application architecture as returned by arch(1).
921: * For example, some (2/ARM) manuals are shared
922: * across the "armish" and "zaurus" kernel
923: * architectures.
924: * A few manuals are even shared across completely
925: * different architectures, for example fdformat(1)
926: * on amd64, i386, sparc, and sparc64.
927: */
928:
929: if (strcasecmp(mpage->arch, mlink->arch)) {
930: match = 0;
931: say(mlink->file, "Architecture \"%s\" manual in "
932: "\"%s\" directory", mpage->arch, mlink->arch);
933: }
934:
935: if (strcasecmp(mpage->title, mlink->name))
936: match = 0;
937:
938: return(match);
939: }
940:
1.47 schwarze 941: /*
942: * Run through the files in the global vector "mpages"
943: * and add them to the database specified in "basedir".
944: *
945: * This handles the parsing scheme itself, using the cues of directory
946: * and filename to determine whether the file is parsable or not.
947: */
948: static void
1.58 schwarze 949: mpages_merge(struct mchars *mc, struct mparse *mp)
1.47 schwarze 950: {
1.70 schwarze 951: char any[] = "any";
1.58 schwarze 952: struct ohash_info str_info;
1.47 schwarze 953: struct mpage *mpage;
1.50 schwarze 954: struct mlink *mlink;
1.47 schwarze 955: struct mdoc *mdoc;
956: struct man *man;
1.69 schwarze 957: char *cp;
1.47 schwarze 958: int match;
1.58 schwarze 959: unsigned int pslot;
1.47 schwarze 960: enum mandoclevel lvl;
961:
962: str_info.alloc = hash_alloc;
963: str_info.halloc = hash_halloc;
964: str_info.hfree = hash_free;
965: str_info.key_offset = offsetof(struct str, key);
966:
1.64 schwarze 967: if (0 == nodb)
968: SQL_EXEC("BEGIN TRANSACTION");
969:
1.47 schwarze 970: mpage = ohash_first(&mpages, &pslot);
971: while (NULL != mpage) {
972: mlinks_undupe(mpage);
973: if (NULL == mpage->mlinks) {
974: mpage = ohash_next(&mpages, &pslot);
975: continue;
976: }
1.1 schwarze 977:
1.47 schwarze 978: ohash_init(&strings, 6, &str_info);
979: mparse_reset(mp);
980: mdoc = NULL;
981: man = NULL;
1.11 schwarze 982:
983: /*
1.24 schwarze 984: * Try interpreting the file as mdoc(7) or man(7)
985: * source code, unless it is already known to be
986: * formatted. Fall back to formatted mode.
1.11 schwarze 987: */
1.47 schwarze 988: if (FORM_CAT != mpage->mlinks->dform ||
989: FORM_CAT != mpage->mlinks->fform) {
990: lvl = mparse_readfd(mp, -1, mpage->mlinks->file);
991: if (lvl < MANDOCLEVEL_FATAL)
992: mparse_result(mp, &mdoc, &man);
993: }
1.11 schwarze 994:
995: if (NULL != mdoc) {
1.47 schwarze 996: mpage->form = FORM_SRC;
997: mpage->sec =
998: mandoc_strdup(mdoc_meta(mdoc)->msec);
999: mpage->arch = mdoc_meta(mdoc)->arch;
1000: mpage->arch = mandoc_strdup(
1001: NULL == mpage->arch ? "" : mpage->arch);
1002: mpage->title =
1003: mandoc_strdup(mdoc_meta(mdoc)->title);
1.11 schwarze 1004: } else if (NULL != man) {
1.47 schwarze 1005: mpage->form = FORM_SRC;
1006: mpage->sec =
1007: mandoc_strdup(man_meta(man)->msec);
1008: mpage->arch =
1009: mandoc_strdup(mpage->mlinks->arch);
1010: mpage->title =
1011: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1012: } else {
1.47 schwarze 1013: mpage->form = FORM_CAT;
1014: mpage->sec =
1015: mandoc_strdup(mpage->mlinks->dsec);
1016: mpage->arch =
1017: mandoc_strdup(mpage->mlinks->arch);
1018: mpage->title =
1019: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1020: }
1.54 schwarze 1021: putkey(mpage, mpage->sec, TYPE_sec);
1.55 schwarze 1022: putkey(mpage, '\0' == *mpage->arch ?
1.70 schwarze 1023: any : mpage->arch, TYPE_arch);
1.1 schwarze 1024:
1.54 schwarze 1025: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1026: if ('\0' != *mlink->dsec)
1027: putkey(mpage, mlink->dsec, TYPE_sec);
1028: if ('\0' != *mlink->fsec)
1029: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1030: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1031: any : mlink->arch, TYPE_arch);
1.50 schwarze 1032: putkey(mpage, mlink->name, TYPE_Nm);
1.54 schwarze 1033: }
1.41 deraadt 1034:
1.50 schwarze 1035: if (warnings && !use_all) {
1.47 schwarze 1036: match = 0;
1.50 schwarze 1037: for (mlink = mpage->mlinks; mlink;
1038: mlink = mlink->next)
1039: if (mlink_check(mpage, mlink))
1040: match = 1;
1041: } else
1042: match = 1;
1.6 schwarze 1043:
1.47 schwarze 1044: if (NULL != mdoc) {
1045: if (NULL != (cp = mdoc_meta(mdoc)->name))
1046: putkey(mpage, cp, TYPE_Nm);
1047: assert(NULL == mpage->desc);
1048: parse_mdoc(mpage, mdoc_node(mdoc));
1049: putkey(mpage, NULL != mpage->desc ?
1050: mpage->desc : mpage->mlinks->name, TYPE_Nd);
1051: } else if (NULL != man)
1052: parse_man(mpage, man_node(man));
1053: else
1054: parse_cat(mpage);
1.6 schwarze 1055:
1.62 schwarze 1056: dbadd(mpage, mc);
1.47 schwarze 1057: ohash_delete(&strings);
1058: mpage = ohash_next(&mpages, &pslot);
1059: }
1.64 schwarze 1060:
1061: if (0 == nodb)
1062: SQL_EXEC("END TRANSACTION");
1.47 schwarze 1063: }
1.6 schwarze 1064:
1.47 schwarze 1065: static void
1066: parse_cat(struct mpage *mpage)
1067: {
1068: FILE *stream;
1069: char *line, *p, *title;
1070: size_t len, plen, titlesz;
1.1 schwarze 1071:
1.47 schwarze 1072: if (NULL == (stream = fopen(mpage->mlinks->file, "r"))) {
1073: if (warnings)
1074: say(mpage->mlinks->file, NULL);
1075: return;
1076: }
1.1 schwarze 1077:
1.47 schwarze 1078: /* Skip to first blank line. */
1.1 schwarze 1079:
1.47 schwarze 1080: while (NULL != (line = fgetln(stream, &len)))
1081: if ('\n' == *line)
1082: break;
1.1 schwarze 1083:
1.47 schwarze 1084: /*
1085: * Assume the first line that is not indented
1086: * is the first section header. Skip to it.
1087: */
1.1 schwarze 1088:
1.47 schwarze 1089: while (NULL != (line = fgetln(stream, &len)))
1090: if ('\n' != *line && ' ' != *line)
1091: break;
1092:
1093: /*
1094: * Read up until the next section into a buffer.
1095: * Strip the leading and trailing newline from each read line,
1096: * appending a trailing space.
1097: * Ignore empty (whitespace-only) lines.
1098: */
1.28 schwarze 1099:
1.47 schwarze 1100: titlesz = 0;
1101: title = NULL;
1.38 schwarze 1102:
1.47 schwarze 1103: while (NULL != (line = fgetln(stream, &len))) {
1104: if (' ' != *line || '\n' != line[len - 1])
1105: break;
1106: while (len > 0 && isspace((unsigned char)*line)) {
1107: line++;
1108: len--;
1109: }
1110: if (1 == len)
1111: continue;
1112: title = mandoc_realloc(title, titlesz + len);
1113: memcpy(title + titlesz, line, len);
1114: titlesz += len;
1115: title[titlesz - 1] = ' ';
1116: }
1.28 schwarze 1117:
1.47 schwarze 1118: /*
1119: * If no page content can be found, or the input line
1120: * is already the next section header, or there is no
1121: * trailing newline, reuse the page title as the page
1122: * description.
1123: */
1.1 schwarze 1124:
1.47 schwarze 1125: if (NULL == title || '\0' == *title) {
1126: if (warnings)
1127: say(mpage->mlinks->file,
1128: "Cannot find NAME section");
1129: assert(NULL == mpage->desc);
1130: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1131: putkey(mpage, mpage->mlinks->name, TYPE_Nd);
1132: fclose(stream);
1133: free(title);
1134: return;
1135: }
1.24 schwarze 1136:
1.47 schwarze 1137: title = mandoc_realloc(title, titlesz + 1);
1138: title[titlesz] = '\0';
1.24 schwarze 1139:
1.47 schwarze 1140: /*
1141: * Skip to the first dash.
1142: * Use the remaining line as the description (no more than 70
1143: * bytes).
1144: */
1.28 schwarze 1145:
1.47 schwarze 1146: if (NULL != (p = strstr(title, "- "))) {
1147: for (p += 2; ' ' == *p || '\b' == *p; p++)
1148: /* Skip to next word. */ ;
1149: } else {
1150: if (warnings)
1151: say(mpage->mlinks->file,
1152: "No dash in title line");
1153: p = title;
1154: }
1.1 schwarze 1155:
1.47 schwarze 1156: plen = strlen(p);
1.1 schwarze 1157:
1.47 schwarze 1158: /* Strip backspace-encoding from line. */
1.1 schwarze 1159:
1.47 schwarze 1160: while (NULL != (line = memchr(p, '\b', plen))) {
1161: len = line - p;
1162: if (0 == len) {
1163: memmove(line, line + 1, plen--);
1164: continue;
1165: }
1166: memmove(line - 1, line + 1, plen - len);
1167: plen -= 2;
1168: }
1.1 schwarze 1169:
1.47 schwarze 1170: assert(NULL == mpage->desc);
1171: mpage->desc = mandoc_strdup(p);
1172: putkey(mpage, mpage->desc, TYPE_Nd);
1173: fclose(stream);
1174: free(title);
1175: }
1.16 schwarze 1176:
1.47 schwarze 1177: /*
1178: * Put a type/word pair into the word database for this particular file.
1179: */
1180: static void
1.69 schwarze 1181: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1182: {
1.69 schwarze 1183: char *cp;
1.37 schwarze 1184:
1.47 schwarze 1185: assert(NULL != value);
1.69 schwarze 1186: if (TYPE_arch == type)
1187: for (cp = value; *cp; cp++)
1188: if (isupper((unsigned char)*cp))
1189: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1190: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1191: }
1192:
1193: /*
1.47 schwarze 1194: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1195: */
1196: static void
1.47 schwarze 1197: putmdockey(const struct mpage *mpage,
1198: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1199: {
1.16 schwarze 1200:
1.47 schwarze 1201: for ( ; NULL != n; n = n->next) {
1202: if (NULL != n->child)
1203: putmdockey(mpage, n->child, m);
1204: if (MDOC_TEXT == n->type)
1205: putkey(mpage, n->string, m);
1206: }
1207: }
1.16 schwarze 1208:
1.47 schwarze 1209: static void
1210: parse_man(struct mpage *mpage, const struct man_node *n)
1211: {
1212: const struct man_node *head, *body;
1213: char *start, *sv, *title;
1214: char byte;
1215: size_t sz, titlesz;
1.16 schwarze 1216:
1.47 schwarze 1217: if (NULL == n)
1218: return;
1.16 schwarze 1219:
1.47 schwarze 1220: /*
1221: * We're only searching for one thing: the first text child in
1222: * the BODY of a NAME section. Since we don't keep track of
1223: * sections in -man, run some hoops to find out whether we're in
1224: * the correct section or not.
1225: */
1.16 schwarze 1226:
1.47 schwarze 1227: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1228: body = n;
1229: assert(body->parent);
1230: if (NULL != (head = body->parent->head) &&
1231: 1 == head->nchild &&
1232: NULL != (head = (head->child)) &&
1233: MAN_TEXT == head->type &&
1234: 0 == strcmp(head->string, "NAME") &&
1235: NULL != (body = body->child) &&
1236: MAN_TEXT == body->type) {
1.2 schwarze 1237:
1.47 schwarze 1238: title = NULL;
1239: titlesz = 0;
1.2 schwarze 1240:
1.47 schwarze 1241: /*
1242: * Suck the entire NAME section into memory.
1243: * Yes, we might run away.
1244: * But too many manuals have big, spread-out
1245: * NAME sections over many lines.
1246: */
1.2 schwarze 1247:
1.47 schwarze 1248: for ( ; NULL != body; body = body->next) {
1249: if (MAN_TEXT != body->type)
1250: break;
1251: if (0 == (sz = strlen(body->string)))
1252: continue;
1253: title = mandoc_realloc
1254: (title, titlesz + sz + 1);
1255: memcpy(title + titlesz, body->string, sz);
1256: titlesz += sz + 1;
1257: title[titlesz - 1] = ' ';
1258: }
1259: if (NULL == title)
1260: return;
1.16 schwarze 1261:
1.47 schwarze 1262: title = mandoc_realloc(title, titlesz + 1);
1263: title[titlesz] = '\0';
1.16 schwarze 1264:
1.47 schwarze 1265: /* Skip leading space. */
1.16 schwarze 1266:
1.47 schwarze 1267: sv = title;
1268: while (isspace((unsigned char)*sv))
1269: sv++;
1.16 schwarze 1270:
1.47 schwarze 1271: if (0 == (sz = strlen(sv))) {
1272: free(title);
1273: return;
1274: }
1.1 schwarze 1275:
1.47 schwarze 1276: /* Erase trailing space. */
1.1 schwarze 1277:
1.47 schwarze 1278: start = &sv[sz - 1];
1279: while (start > sv && isspace((unsigned char)*start))
1280: *start-- = '\0';
1.1 schwarze 1281:
1.47 schwarze 1282: if (start == sv) {
1283: free(title);
1284: return;
1285: }
1.1 schwarze 1286:
1.47 schwarze 1287: start = sv;
1.16 schwarze 1288:
1.47 schwarze 1289: /*
1290: * Go through a special heuristic dance here.
1291: * Conventionally, one or more manual names are
1292: * comma-specified prior to a whitespace, then a
1293: * dash, then a description. Try to puzzle out
1294: * the name parts here.
1295: */
1.16 schwarze 1296:
1.47 schwarze 1297: for ( ;; ) {
1298: sz = strcspn(start, " ,");
1299: if ('\0' == start[sz])
1300: break;
1.1 schwarze 1301:
1.47 schwarze 1302: byte = start[sz];
1303: start[sz] = '\0';
1.67 schwarze 1304:
1305: /*
1306: * Assume a stray trailing comma in the
1307: * name list if a name begins with a dash.
1308: */
1309:
1310: if ('-' == start[0] ||
1311: ('\\' == start[0] && '-' == start[1]))
1312: break;
1.1 schwarze 1313:
1.47 schwarze 1314: putkey(mpage, start, TYPE_Nm);
1.1 schwarze 1315:
1.47 schwarze 1316: if (' ' == byte) {
1317: start += sz + 1;
1318: break;
1319: }
1.1 schwarze 1320:
1.47 schwarze 1321: assert(',' == byte);
1322: start += sz + 1;
1323: while (' ' == *start)
1324: start++;
1325: }
1.1 schwarze 1326:
1.47 schwarze 1327: if (sv == start) {
1328: putkey(mpage, start, TYPE_Nm);
1329: free(title);
1330: return;
1331: }
1.1 schwarze 1332:
1.47 schwarze 1333: while (isspace((unsigned char)*start))
1334: start++;
1.1 schwarze 1335:
1.47 schwarze 1336: if (0 == strncmp(start, "-", 1))
1337: start += 1;
1338: else if (0 == strncmp(start, "\\-\\-", 4))
1339: start += 4;
1340: else if (0 == strncmp(start, "\\-", 2))
1341: start += 2;
1342: else if (0 == strncmp(start, "\\(en", 4))
1343: start += 4;
1344: else if (0 == strncmp(start, "\\(em", 4))
1345: start += 4;
1.1 schwarze 1346:
1.47 schwarze 1347: while (' ' == *start)
1348: start++;
1.1 schwarze 1349:
1.47 schwarze 1350: assert(NULL == mpage->desc);
1351: mpage->desc = mandoc_strdup(start);
1352: putkey(mpage, mpage->desc, TYPE_Nd);
1353: free(title);
1354: return;
1355: }
1356: }
1.1 schwarze 1357:
1.47 schwarze 1358: for (n = n->child; n; n = n->next) {
1359: if (NULL != mpage->desc)
1360: break;
1361: parse_man(mpage, n);
1.1 schwarze 1362: }
1363: }
1364:
1365: static void
1.47 schwarze 1366: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1367: {
1368:
1.47 schwarze 1369: assert(NULL != n);
1370: for (n = n->child; NULL != n; n = n->next) {
1371: switch (n->type) {
1372: case (MDOC_ELEM):
1373: /* FALLTHROUGH */
1374: case (MDOC_BLOCK):
1375: /* FALLTHROUGH */
1376: case (MDOC_HEAD):
1377: /* FALLTHROUGH */
1378: case (MDOC_BODY):
1379: /* FALLTHROUGH */
1380: case (MDOC_TAIL):
1381: if (NULL != mdocs[n->tok].fp)
1382: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1383: break;
1384: if (mdocs[n->tok].mask)
1385: putmdockey(mpage, n->child,
1386: mdocs[n->tok].mask);
1387: break;
1388: default:
1389: assert(MDOC_ROOT != n->type);
1390: continue;
1391: }
1392: if (NULL != n->child)
1393: parse_mdoc(mpage, n);
1.1 schwarze 1394: }
1395: }
1396:
1.19 schwarze 1397: static int
1.47 schwarze 1398: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1399: {
1400: const char *start, *end;
1401: size_t sz;
1.19 schwarze 1402:
1.47 schwarze 1403: if (SEC_SYNOPSIS != n->sec ||
1404: NULL == (n = n->child) ||
1405: MDOC_TEXT != n->type)
1.19 schwarze 1406: return(0);
1.1 schwarze 1407:
1408: /*
1409: * Only consider those `Fd' macro fields that begin with an
1410: * "inclusion" token (versus, e.g., #define).
1411: */
1.47 schwarze 1412:
1.1 schwarze 1413: if (strcmp("#include", n->string))
1.19 schwarze 1414: return(0);
1.1 schwarze 1415:
1416: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1417: return(0);
1.1 schwarze 1418:
1419: /*
1420: * Strip away the enclosing angle brackets and make sure we're
1421: * not zero-length.
1422: */
1423:
1424: start = n->string;
1425: if ('<' == *start || '"' == *start)
1426: start++;
1427:
1428: if (0 == (sz = strlen(start)))
1.19 schwarze 1429: return(0);
1.1 schwarze 1430:
1431: end = &start[(int)sz - 1];
1432: if ('>' == *end || '"' == *end)
1433: end--;
1434:
1.47 schwarze 1435: if (end > start)
1436: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1437: return(0);
1.1 schwarze 1438: }
1439:
1.19 schwarze 1440: static int
1.47 schwarze 1441: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1442: {
1.69 schwarze 1443: char *cp;
1.1 schwarze 1444:
1.47 schwarze 1445: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.19 schwarze 1446: return(0);
1447:
1.47 schwarze 1448: /*
1449: * Parse: .Fn "struct type *name" "char *arg".
1450: * First strip away pointer symbol.
1451: * Then store the function name, then type.
1452: * Finally, store the arguments.
1453: */
1.1 schwarze 1454:
1.47 schwarze 1455: if (NULL == (cp = strrchr(n->string, ' ')))
1456: cp = n->string;
1.1 schwarze 1457:
1458: while ('*' == *cp)
1459: cp++;
1460:
1.47 schwarze 1461: putkey(mpage, cp, TYPE_Fn);
1.19 schwarze 1462:
1.47 schwarze 1463: if (n->string < cp)
1464: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.19 schwarze 1465:
1.47 schwarze 1466: for (n = n->next; NULL != n; n = n->next)
1467: if (MDOC_TEXT == n->type)
1468: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1469:
1470: return(0);
1.1 schwarze 1471: }
1472:
1.19 schwarze 1473: static int
1.47 schwarze 1474: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1475: {
1.47 schwarze 1476: char *cp;
1.1 schwarze 1477:
1478: if (NULL == (n = n->child))
1.19 schwarze 1479: return(0);
1.1 schwarze 1480:
1.47 schwarze 1481: if (NULL == n->next) {
1482: putkey(mpage, n->string, TYPE_Xr);
1483: return(0);
1484: }
1.1 schwarze 1485:
1.47 schwarze 1486: if (-1 == asprintf(&cp, "%s(%s)", n->string, n->next->string)) {
1487: perror(NULL);
1488: exit((int)MANDOCLEVEL_SYSERR);
1489: }
1490: putkey(mpage, cp, TYPE_Xr);
1491: free(cp);
1492: return(0);
1.1 schwarze 1493: }
1494:
1.19 schwarze 1495: static int
1.47 schwarze 1496: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1497: {
1.47 schwarze 1498: size_t sz;
1.1 schwarze 1499:
1500: if (MDOC_BODY != n->type)
1.19 schwarze 1501: return(0);
1.1 schwarze 1502:
1.47 schwarze 1503: /*
1504: * Special-case the `Nd' because we need to put the description
1505: * into the document table.
1506: */
1507:
1508: for (n = n->child; NULL != n; n = n->next) {
1509: if (MDOC_TEXT == n->type) {
1510: if (NULL != mpage->desc) {
1511: sz = strlen(mpage->desc) +
1512: strlen(n->string) + 2;
1513: mpage->desc = mandoc_realloc(
1514: mpage->desc, sz);
1515: strlcat(mpage->desc, " ", sz);
1516: strlcat(mpage->desc, n->string, sz);
1517: } else
1518: mpage->desc = mandoc_strdup(n->string);
1519: }
1520: if (NULL != n->child)
1521: parse_mdoc_Nd(mpage, n);
1522: }
1.19 schwarze 1523: return(1);
1.1 schwarze 1524: }
1525:
1.19 schwarze 1526: static int
1.47 schwarze 1527: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1528: {
1529:
1.49 schwarze 1530: return(SEC_NAME == n->sec ||
1531: (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type));
1.1 schwarze 1532: }
1533:
1.19 schwarze 1534: static int
1.47 schwarze 1535: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1536: {
1537:
1.19 schwarze 1538: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1539: }
1540:
1.47 schwarze 1541: static int
1542: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1543: {
1544:
1.47 schwarze 1545: return(MDOC_HEAD == n->type);
1546: }
1.1 schwarze 1547:
1.47 schwarze 1548: static int
1549: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1550: {
1.1 schwarze 1551:
1.47 schwarze 1552: return(MDOC_BODY == n->type);
1.1 schwarze 1553: }
1554:
1.47 schwarze 1555: /*
1556: * Add a string to the hash table for the current manual.
1557: * Each string has a bitmask telling which macros it belongs to.
1558: * When we finish the manual, we'll dump the table.
1559: */
1.1 schwarze 1560: static void
1.47 schwarze 1561: putkeys(const struct mpage *mpage,
1562: const char *cp, size_t sz, uint64_t v)
1.1 schwarze 1563: {
1.47 schwarze 1564: struct str *s;
1.68 schwarze 1565: const char *end;
1566: uint64_t mask;
1.47 schwarze 1567: unsigned int slot;
1.68 schwarze 1568: int i;
1.1 schwarze 1569:
1.47 schwarze 1570: if (0 == sz)
1571: return;
1.68 schwarze 1572:
1573: if (verb > 1) {
1574: for (i = 0, mask = 1;
1575: i < mansearch_keymax;
1576: i++, mask <<= 1)
1577: if (mask & v)
1578: break;
1579: say(mpage->mlinks->file, "Adding key %s=%*s",
1580: mansearch_keynames[i], sz, cp);
1581: }
1.47 schwarze 1582:
1583: end = cp + sz;
1584: slot = ohash_qlookupi(&strings, cp, &end);
1585: s = ohash_find(&strings, slot);
1.1 schwarze 1586:
1.47 schwarze 1587: if (NULL != s && mpage == s->mpage) {
1588: s->mask |= v;
1.1 schwarze 1589: return;
1.47 schwarze 1590: } else if (NULL == s) {
1591: s = mandoc_calloc(sizeof(struct str) + sz + 1, 1);
1592: memcpy(s->key, cp, sz);
1593: ohash_insert(&strings, slot, s);
1594: }
1595: s->mpage = mpage;
1596: s->mask = v;
1.1 schwarze 1597: }
1598:
1599: /*
1.47 schwarze 1600: * Take a Unicode codepoint and produce its UTF-8 encoding.
1601: * This isn't the best way to do this, but it works.
1602: * The magic numbers are from the UTF-8 packaging.
1603: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1604: */
1.47 schwarze 1605: static size_t
1606: utf8(unsigned int cp, char out[7])
1.1 schwarze 1607: {
1.47 schwarze 1608: size_t rc;
1.1 schwarze 1609:
1.47 schwarze 1610: rc = 0;
1611: if (cp <= 0x0000007F) {
1612: rc = 1;
1613: out[0] = (char)cp;
1614: } else if (cp <= 0x000007FF) {
1615: rc = 2;
1616: out[0] = (cp >> 6 & 31) | 192;
1617: out[1] = (cp & 63) | 128;
1618: } else if (cp <= 0x0000FFFF) {
1619: rc = 3;
1620: out[0] = (cp >> 12 & 15) | 224;
1621: out[1] = (cp >> 6 & 63) | 128;
1622: out[2] = (cp & 63) | 128;
1623: } else if (cp <= 0x001FFFFF) {
1624: rc = 4;
1625: out[0] = (cp >> 18 & 7) | 240;
1626: out[1] = (cp >> 12 & 63) | 128;
1627: out[2] = (cp >> 6 & 63) | 128;
1628: out[3] = (cp & 63) | 128;
1629: } else if (cp <= 0x03FFFFFF) {
1630: rc = 5;
1631: out[0] = (cp >> 24 & 3) | 248;
1632: out[1] = (cp >> 18 & 63) | 128;
1633: out[2] = (cp >> 12 & 63) | 128;
1634: out[3] = (cp >> 6 & 63) | 128;
1635: out[4] = (cp & 63) | 128;
1636: } else if (cp <= 0x7FFFFFFF) {
1637: rc = 6;
1638: out[0] = (cp >> 30 & 1) | 252;
1639: out[1] = (cp >> 24 & 63) | 128;
1640: out[2] = (cp >> 18 & 63) | 128;
1641: out[3] = (cp >> 12 & 63) | 128;
1642: out[4] = (cp >> 6 & 63) | 128;
1643: out[5] = (cp & 63) | 128;
1644: } else
1645: return(0);
1.19 schwarze 1646:
1.47 schwarze 1647: out[rc] = '\0';
1648: return(rc);
1.1 schwarze 1649: }
1650:
1.47 schwarze 1651: /*
1.53 schwarze 1652: * Store the rendered version of a key, or alias the pointer
1653: * if the key contains no escape sequences.
1.47 schwarze 1654: */
1655: static void
1.53 schwarze 1656: render_key(struct mchars *mc, struct str *key)
1.1 schwarze 1657: {
1.47 schwarze 1658: size_t sz, bsz, pos;
1.71 ! schwarze 1659: char utfbuf[7], res[6];
1.47 schwarze 1660: char *buf;
1661: const char *seq, *cpp, *val;
1662: int len, u;
1663: enum mandoc_esc esc;
1664:
1.53 schwarze 1665: assert(NULL == key->rendered);
1.47 schwarze 1666:
1667: res[0] = '\\';
1668: res[1] = '\t';
1669: res[2] = ASCII_NBRSP;
1670: res[3] = ASCII_HYPH;
1.71 ! schwarze 1671: res[4] = ASCII_BREAK;
! 1672: res[5] = '\0';
1.1 schwarze 1673:
1.47 schwarze 1674: val = key->key;
1675: bsz = strlen(val);
1.1 schwarze 1676:
1677: /*
1.47 schwarze 1678: * Pre-check: if we have no stop-characters, then set the
1679: * pointer as ourselvse and get out of here.
1.1 schwarze 1680: */
1.47 schwarze 1681: if (strcspn(val, res) == bsz) {
1.53 schwarze 1682: key->rendered = key->key;
1.47 schwarze 1683: return;
1684: }
1.1 schwarze 1685:
1.47 schwarze 1686: /* Pre-allocate by the length of the input */
1.39 schwarze 1687:
1.47 schwarze 1688: buf = mandoc_malloc(++bsz);
1689: pos = 0;
1.39 schwarze 1690:
1.47 schwarze 1691: while ('\0' != *val) {
1692: /*
1693: * Halt on the first escape sequence.
1694: * This also halts on the end of string, in which case
1695: * we just copy, fallthrough, and exit the loop.
1696: */
1697: if ((sz = strcspn(val, res)) > 0) {
1698: memcpy(&buf[pos], val, sz);
1699: pos += sz;
1700: val += sz;
1701: }
1.39 schwarze 1702:
1.71 ! schwarze 1703: switch (*val) {
! 1704: case (ASCII_HYPH):
1.47 schwarze 1705: buf[pos++] = '-';
1706: val++;
1707: continue;
1.71 ! schwarze 1708: case ('\t'):
! 1709: /* FALLTHROUGH */
! 1710: case (ASCII_NBRSP):
1.47 schwarze 1711: buf[pos++] = ' ';
1712: val++;
1.71 ! schwarze 1713: /* FALLTHROUGH */
! 1714: case (ASCII_BREAK):
1.47 schwarze 1715: continue;
1.71 ! schwarze 1716: default:
! 1717: break;
! 1718: }
! 1719: if ('\\' != *val)
1.47 schwarze 1720: break;
1.39 schwarze 1721:
1.47 schwarze 1722: /* Read past the slash. */
1.39 schwarze 1723:
1.47 schwarze 1724: val++;
1.39 schwarze 1725:
1.47 schwarze 1726: /*
1727: * Parse the escape sequence and see if it's a
1728: * predefined character or special character.
1729: */
1.52 schwarze 1730:
1.47 schwarze 1731: esc = mandoc_escape
1732: ((const char **)&val, &seq, &len);
1733: if (ESCAPE_ERROR == esc)
1734: break;
1735: if (ESCAPE_SPECIAL != esc)
1736: continue;
1.39 schwarze 1737:
1.47 schwarze 1738: /*
1.52 schwarze 1739: * Render the special character
1740: * as either UTF-8 or ASCII.
1.47 schwarze 1741: */
1.52 schwarze 1742:
1743: if (write_utf8) {
1744: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1745: continue;
1746: cpp = utfbuf;
1747: if (0 == (sz = utf8(u, utfbuf)))
1748: continue;
1749: sz = strlen(cpp);
1750: } else {
1751: cpp = mchars_spec2str(mc, seq, len, &sz);
1752: if (NULL == cpp)
1753: continue;
1754: if (ASCII_NBRSP == *cpp) {
1755: cpp = " ";
1756: sz = 1;
1757: }
1758: }
1.1 schwarze 1759:
1.47 schwarze 1760: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1761:
1.47 schwarze 1762: bsz += sz;
1763: buf = mandoc_realloc(buf, bsz);
1764: memcpy(&buf[pos], cpp, sz);
1765: pos += sz;
1.1 schwarze 1766: }
1767:
1.47 schwarze 1768: buf[pos] = '\0';
1.53 schwarze 1769: key->rendered = buf;
1.1 schwarze 1770: }
1771:
1.11 schwarze 1772: /*
1.47 schwarze 1773: * Flush the current page's terms (and their bits) into the database.
1774: * Wrap the entire set of additions in a transaction to make sqlite be a
1775: * little faster.
1.53 schwarze 1776: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 1777: */
1778: static void
1.62 schwarze 1779: dbadd(const struct mpage *mpage, struct mchars *mc)
1.11 schwarze 1780: {
1.47 schwarze 1781: struct mlink *mlink;
1782: struct str *key;
1783: int64_t recno;
1784: size_t i;
1785: unsigned int slot;
1786:
1787: if (verb)
1.62 schwarze 1788: say(mpage->mlinks->file, "Adding to database");
1.11 schwarze 1789:
1.47 schwarze 1790: if (nodb)
1.11 schwarze 1791: return;
1.47 schwarze 1792:
1793: i = 1;
1794: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
1795: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1796: recno = sqlite3_last_insert_rowid(db);
1797: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1798:
1799: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1800: i = 1;
1801: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1802: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1803: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1804: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, recno);
1805: SQL_STEP(stmts[STMT_INSERT_LINK]);
1806: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1807: }
1808:
1809: for (key = ohash_first(&strings, &slot); NULL != key;
1810: key = ohash_next(&strings, &slot)) {
1811: assert(key->mpage == mpage);
1.53 schwarze 1812: if (NULL == key->rendered)
1813: render_key(mc, key);
1.47 schwarze 1814: i = 1;
1815: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.53 schwarze 1816: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.47 schwarze 1817: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno);
1818: SQL_STEP(stmts[STMT_INSERT_KEY]);
1819: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.53 schwarze 1820: if (key->rendered != key->key)
1821: free(key->rendered);
1.47 schwarze 1822: free(key);
1.33 schwarze 1823: }
1.47 schwarze 1824: }
1.41 deraadt 1825:
1.47 schwarze 1826: static void
1827: dbprune(void)
1828: {
1829: struct mpage *mpage;
1830: struct mlink *mlink;
1831: size_t i;
1832: unsigned int slot;
1.11 schwarze 1833:
1.63 schwarze 1834: if (0 == nodb)
1835: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 1836:
1.63 schwarze 1837: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
1838: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 1839: mlink = mpage->mlinks;
1840: if (verb)
1.63 schwarze 1841: say(mlink->file, "Deleting from database");
1842: if (nodb)
1843: continue;
1844: for ( ; NULL != mlink; mlink = mlink->next) {
1845: i = 1;
1846: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
1847: i, mlink->dsec);
1848: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
1849: i, mlink->arch);
1850: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
1851: i, mlink->name);
1852: SQL_STEP(stmts[STMT_DELETE_PAGE]);
1853: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
1854: }
1.11 schwarze 1855: }
1.63 schwarze 1856:
1857: if (0 == nodb)
1858: SQL_EXEC("END TRANSACTION");
1.47 schwarze 1859: }
1.22 schwarze 1860:
1.47 schwarze 1861: /*
1862: * Close an existing database and its prepared statements.
1863: * If "real" is not set, rename the temporary file into the real one.
1864: */
1865: static void
1866: dbclose(int real)
1867: {
1868: size_t i;
1.11 schwarze 1869:
1.47 schwarze 1870: if (nodb)
1871: return;
1.11 schwarze 1872:
1.47 schwarze 1873: for (i = 0; i < STMT__MAX; i++) {
1874: sqlite3_finalize(stmts[i]);
1875: stmts[i] = NULL;
1.28 schwarze 1876: }
1.22 schwarze 1877:
1.47 schwarze 1878: sqlite3_close(db);
1879: db = NULL;
1.11 schwarze 1880:
1.47 schwarze 1881: if (real)
1882: return;
1.22 schwarze 1883:
1.47 schwarze 1884: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
1885: exitcode = (int)MANDOCLEVEL_SYSERR;
1886: say(MANDOC_DB, NULL);
1.22 schwarze 1887: }
1.11 schwarze 1888: }
1889:
1.47 schwarze 1890: /*
1891: * This is straightforward stuff.
1892: * Open a database connection to a "temporary" database, then open a set
1893: * of prepared statements we'll use over and over again.
1894: * If "real" is set, we use the existing database; if not, we truncate a
1895: * temporary one.
1896: * Must be matched by dbclose().
1897: */
1898: static int
1899: dbopen(int real)
1.2 schwarze 1900: {
1.47 schwarze 1901: const char *file, *sql;
1902: int rc, ofl;
1.6 schwarze 1903:
1.47 schwarze 1904: if (nodb)
1905: return(1);
1.6 schwarze 1906:
1.47 schwarze 1907: ofl = SQLITE_OPEN_READWRITE;
1908: if (0 == real) {
1909: file = MANDOC_DB "~";
1910: if (-1 == remove(file) && ENOENT != errno) {
1911: exitcode = (int)MANDOCLEVEL_SYSERR;
1912: say(file, NULL);
1913: return(0);
1.28 schwarze 1914: }
1.47 schwarze 1915: ofl |= SQLITE_OPEN_EXCLUSIVE;
1916: } else
1917: file = MANDOC_DB;
1.6 schwarze 1918:
1.47 schwarze 1919: rc = sqlite3_open_v2(file, &db, ofl, NULL);
1920: if (SQLITE_OK == rc)
1921: goto prepare_statements;
1922: if (SQLITE_CANTOPEN != rc) {
1923: exitcode = (int)MANDOCLEVEL_SYSERR;
1924: say(file, NULL);
1925: return(0);
1926: }
1.6 schwarze 1927:
1.47 schwarze 1928: sqlite3_close(db);
1929: db = NULL;
1.6 schwarze 1930:
1.47 schwarze 1931: if (SQLITE_OK != (rc = sqlite3_open(file, &db))) {
1932: exitcode = (int)MANDOCLEVEL_SYSERR;
1933: say(file, NULL);
1934: return(0);
1.2 schwarze 1935: }
1936:
1.47 schwarze 1937: sql = "CREATE TABLE \"mpages\" (\n"
1938: " \"form\" INTEGER NOT NULL,\n"
1939: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1940: ");\n"
1941: "\n"
1942: "CREATE TABLE \"mlinks\" (\n"
1943: " \"sec\" TEXT NOT NULL,\n"
1944: " \"arch\" TEXT NOT NULL,\n"
1945: " \"name\" TEXT NOT NULL,\n"
1946: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1.66 schwarze 1947: "ON DELETE CASCADE\n"
1.47 schwarze 1948: ");\n"
1949: "\n"
1950: "CREATE TABLE \"keys\" (\n"
1951: " \"bits\" INTEGER NOT NULL,\n"
1952: " \"key\" TEXT NOT NULL,\n"
1953: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1.66 schwarze 1954: "ON DELETE CASCADE\n"
1.65 schwarze 1955: ");\n";
1.47 schwarze 1956:
1957: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1958: exitcode = (int)MANDOCLEVEL_SYSERR;
1959: say(file, "%s", sqlite3_errmsg(db));
1960: return(0);
1.2 schwarze 1961: }
1962:
1.47 schwarze 1963: prepare_statements:
1964: SQL_EXEC("PRAGMA foreign_keys = ON");
1.63 schwarze 1965: sql = "DELETE FROM mpages WHERE id IN "
1966: "(SELECT pageid FROM mlinks WHERE "
1967: "sec=? AND arch=? AND name=?)";
1.47 schwarze 1968: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
1969: sql = "INSERT INTO mpages "
1.60 schwarze 1970: "(form) VALUES (?)";
1.47 schwarze 1971: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
1972: sql = "INSERT INTO mlinks "
1.61 schwarze 1973: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 1974: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1975: sql = "INSERT INTO keys "
1976: "(bits,key,pageid) VALUES (?,?,?)";
1977: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 1978:
1.47 schwarze 1979: /*
1980: * When opening a new database, we can turn off
1981: * synchronous mode for much better performance.
1982: */
1.6 schwarze 1983:
1.47 schwarze 1984: if (real)
1985: SQL_EXEC("PRAGMA synchronous = OFF");
1.11 schwarze 1986:
1.47 schwarze 1987: return(1);
1988: }
1.6 schwarze 1989:
1.47 schwarze 1990: static void *
1991: hash_halloc(size_t sz, void *arg)
1992: {
1.6 schwarze 1993:
1.47 schwarze 1994: return(mandoc_calloc(sz, 1));
1995: }
1.2 schwarze 1996:
1.47 schwarze 1997: static void *
1998: hash_alloc(size_t sz, void *arg)
1999: {
1.28 schwarze 2000:
1.47 schwarze 2001: return(mandoc_malloc(sz));
2002: }
1.6 schwarze 2003:
1.47 schwarze 2004: static void
2005: hash_free(void *p, size_t sz, void *arg)
2006: {
1.26 schwarze 2007:
1.47 schwarze 2008: free(p);
2009: }
1.6 schwarze 2010:
1.47 schwarze 2011: static int
2012: set_basedir(const char *targetdir)
2013: {
2014: static char startdir[PATH_MAX];
2015: static int fd;
1.6 schwarze 2016:
1.47 schwarze 2017: /*
2018: * Remember where we started by keeping a fd open to the origin
2019: * path component: throughout this utility, we chdir() a lot to
2020: * handle relative paths, and by doing this, we can return to
2021: * the starting point.
2022: */
2023: if ('\0' == *startdir) {
2024: if (NULL == getcwd(startdir, PATH_MAX)) {
2025: exitcode = (int)MANDOCLEVEL_SYSERR;
2026: if (NULL != targetdir)
2027: say(".", NULL);
2028: return(0);
2029: }
2030: if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
2031: exitcode = (int)MANDOCLEVEL_SYSERR;
2032: say(startdir, NULL);
2033: return(0);
1.11 schwarze 2034: }
1.47 schwarze 2035: if (NULL == targetdir)
2036: targetdir = startdir;
2037: } else {
2038: if (-1 == fd)
2039: return(0);
2040: if (-1 == fchdir(fd)) {
2041: close(fd);
2042: basedir[0] = '\0';
2043: exitcode = (int)MANDOCLEVEL_SYSERR;
2044: say(startdir, NULL);
2045: return(0);
1.2 schwarze 2046: }
1.47 schwarze 2047: if (NULL == targetdir) {
2048: close(fd);
2049: return(1);
1.2 schwarze 2050: }
2051: }
1.47 schwarze 2052: if (NULL == realpath(targetdir, basedir)) {
2053: basedir[0] = '\0';
2054: exitcode = (int)MANDOCLEVEL_BADARG;
2055: say(targetdir, NULL);
2056: return(0);
2057: } else if (-1 == chdir(basedir)) {
2058: exitcode = (int)MANDOCLEVEL_BADARG;
2059: say("", NULL);
2060: return(0);
2061: }
2062: return(1);
1.2 schwarze 2063: }
2064:
2065: static void
1.47 schwarze 2066: say(const char *file, const char *format, ...)
1.2 schwarze 2067: {
1.47 schwarze 2068: va_list ap;
1.2 schwarze 2069:
1.47 schwarze 2070: if ('\0' != *basedir)
2071: fprintf(stderr, "%s", basedir);
2072: if ('\0' != *basedir && '\0' != *file)
2073: fputs("//", stderr);
2074: if ('\0' != *file)
2075: fprintf(stderr, "%s", file);
2076: fputs(": ", stderr);
1.31 schwarze 2077:
1.47 schwarze 2078: if (NULL == format) {
2079: perror(NULL);
2080: return;
1.2 schwarze 2081: }
1.47 schwarze 2082:
2083: va_start(ap, format);
2084: vfprintf(stderr, format, ap);
2085: va_end(ap);
2086:
2087: fputc('\n', stderr);
1.1 schwarze 2088: }
![[BACK]](/icons/back.gif){kind=icon}
Return to mandocdb.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc