Annotation of mandoc/mandocdb.c, Revision 1.94
1.94 ! schwarze 1: /* $Id: mandocdb.c,v 1.93 2014/01/02 18:52:15 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.56 schwarze 4: * Copyright (c) 2011, 2012, 2013 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 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: */
18: #ifdef HAVE_CONFIG_H
19: #include "config.h"
20: #endif
21:
1.50 kristaps 22: #include <sys/stat.h>
1.1 kristaps 23:
24: #include <assert.h>
1.43 kristaps 25: #include <ctype.h>
1.63 schwarze 26: #include <errno.h>
1.1 kristaps 27: #include <fcntl.h>
1.50 kristaps 28: #include <fts.h>
1.1 kristaps 29: #include <getopt.h>
1.58 schwarze 30: #include <limits.h>
1.50 kristaps 31: #include <stddef.h>
1.59 schwarze 32: #include <stdio.h>
1.1 kristaps 33: #include <stdint.h>
34: #include <stdlib.h>
35: #include <string.h>
1.17 schwarze 36: #include <unistd.h>
1.1 kristaps 37:
1.53 kristaps 38: #ifdef HAVE_OHASH
1.50 kristaps 39: #include <ohash.h>
1.53 kristaps 40: #else
41: #include "compat_ohash.h"
42: #endif
1.50 kristaps 43: #include <sqlite3.h>
1.1 kristaps 44:
1.50 kristaps 45: #include "mdoc.h"
1.1 kristaps 46: #include "man.h"
47: #include "mandoc.h"
1.10 kristaps 48: #include "manpath.h"
1.55 kristaps 49: #include "mansearch.h"
1.1 kristaps 50:
1.52 kristaps 51: #define SQL_EXEC(_v) \
52: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
53: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
54: #define SQL_BIND_TEXT(_s, _i, _v) \
55: if (SQLITE_OK != sqlite3_bind_text \
56: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
57: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
58: #define SQL_BIND_INT(_s, _i, _v) \
59: if (SQLITE_OK != sqlite3_bind_int \
60: ((_s), (_i)++, (_v))) \
61: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
62: #define SQL_BIND_INT64(_s, _i, _v) \
63: if (SQLITE_OK != sqlite3_bind_int64 \
64: ((_s), (_i)++, (_v))) \
65: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
66: #define SQL_STEP(_s) \
67: if (SQLITE_DONE != sqlite3_step((_s))) \
68: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
69:
1.50 kristaps 70: enum op {
71: OP_DEFAULT = 0, /* new dbs from dir list or default config */
72: OP_CONFFILE, /* new databases from custom config file */
73: OP_UPDATE, /* delete/add entries in existing database */
74: OP_DELETE, /* delete entries from existing database */
75: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 76: };
77:
1.50 kristaps 78: enum form {
1.84 schwarze 79: FORM_NONE, /* format is unknown */
80: FORM_SRC, /* format is -man or -mdoc */
81: FORM_CAT /* format is cat */
1.50 kristaps 82: };
1.38 schwarze 83:
1.50 kristaps 84: struct str {
85: char *utf8; /* key in UTF-8 form */
1.78 schwarze 86: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 87: uint64_t mask; /* bitmask in sequence */
1.51 kristaps 88: char key[]; /* the string itself */
1.38 schwarze 89: };
90:
1.79 schwarze 91: struct inodev {
92: ino_t st_ino;
93: dev_t st_dev;
1.50 kristaps 94: };
1.5 kristaps 95:
1.78 schwarze 96: struct mpage {
1.79 schwarze 97: struct inodev inodev; /* used for hashing routine */
1.82 schwarze 98: enum form form; /* format from file content */
99: char *sec; /* section from file content */
100: char *arch; /* architecture from file content */
101: char *title; /* title from file content */
102: char *desc; /* description from file content */
103: struct mlink *mlinks; /* singly linked list */
104: };
105:
106: struct mlink {
1.58 schwarze 107: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 108: enum form dform; /* format from directory */
1.84 schwarze 109: enum form fform; /* format from file name suffix */
1.82 schwarze 110: char *dsec; /* section from directory */
111: char *arch; /* architecture from directory */
112: char *name; /* name from file name (not empty) */
113: char *fsec; /* section from file name suffix */
1.85 schwarze 114: struct mlink *next; /* singly linked list */
1.3 kristaps 115: };
116:
1.69 schwarze 117: struct title {
118: char *title; /* name(sec/arch) given inside the file */
119: char *file; /* file name in case of mismatch */
120: };
121:
1.50 kristaps 122: enum stmt {
1.81 schwarze 123: STMT_DELETE_PAGE = 0, /* delete mpage */
124: STMT_INSERT_PAGE, /* insert mpage */
125: STMT_INSERT_LINK, /* insert mlink */
126: STMT_INSERT_KEY, /* insert parsed key */
1.50 kristaps 127: STMT__MAX
1.1 kristaps 128: };
129:
1.78 schwarze 130: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
1.1 kristaps 131:
1.50 kristaps 132: struct mdoc_handler {
133: mdoc_fp fp; /* optional handler */
134: uint64_t mask; /* set unless handler returns 0 */
1.1 kristaps 135: };
136:
1.59 schwarze 137: static void dbclose(int);
1.82 schwarze 138: static void dbindex(const struct mpage *, struct mchars *);
1.59 schwarze 139: static int dbopen(int);
140: static void dbprune(void);
141: static void filescan(const char *);
1.50 kristaps 142: static void *hash_alloc(size_t, void *);
143: static void hash_free(void *, size_t, void *);
144: static void *hash_halloc(size_t, void *);
1.84 schwarze 145: static void mlink_add(struct mlink *, const struct stat *);
1.93 schwarze 146: static int mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 147: static void mlink_free(struct mlink *);
1.89 schwarze 148: static void mlinks_undupe(struct mpage *);
1.78 schwarze 149: static void mpages_free(void);
150: static void mpages_merge(struct mchars *, struct mparse *, int);
151: static void parse_cat(struct mpage *);
152: static void parse_man(struct mpage *, const struct man_node *);
153: static void parse_mdoc(struct mpage *, const struct mdoc_node *);
154: static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
155: static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
156: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
157: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
158: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
159: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
160: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
161: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
162: static void putkey(const struct mpage *,
1.50 kristaps 163: const char *, uint64_t);
1.78 schwarze 164: static void putkeys(const struct mpage *,
1.64 schwarze 165: const char *, size_t, uint64_t);
1.78 schwarze 166: static void putmdockey(const struct mpage *,
1.50 kristaps 167: const struct mdoc_node *, uint64_t);
1.59 schwarze 168: static void say(const char *, const char *, ...);
1.80 schwarze 169: static int set_basedir(const char *);
1.59 schwarze 170: static int treescan(void);
1.50 kristaps 171: static size_t utf8(unsigned int, char [7]);
172: static void utf8key(struct mchars *, struct str *);
173:
174: static char *progname;
175: static int use_all; /* use all found files */
176: static int nodb; /* no database changes */
177: static int verb; /* print what we're doing */
178: static int warnings; /* warn about crap */
1.59 schwarze 179: static int exitcode; /* to be returned by main */
1.50 kristaps 180: static enum op op; /* operational mode */
1.59 schwarze 181: static char basedir[PATH_MAX]; /* current base directory */
1.78 schwarze 182: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 183: static struct ohash mlinks; /* table of directory entries */
1.50 kristaps 184: static struct ohash strings; /* table of all strings */
185: static sqlite3 *db = NULL; /* current database */
186: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.25 schwarze 187:
188: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 189: { NULL, 0 }, /* Ap */
190: { NULL, 0 }, /* Dd */
191: { NULL, 0 }, /* Dt */
192: { NULL, 0 }, /* Os */
193: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
194: { parse_mdoc_head, TYPE_Ss }, /* Ss */
195: { NULL, 0 }, /* Pp */
196: { NULL, 0 }, /* D1 */
197: { NULL, 0 }, /* Dl */
198: { NULL, 0 }, /* Bd */
199: { NULL, 0 }, /* Ed */
200: { NULL, 0 }, /* Bl */
201: { NULL, 0 }, /* El */
202: { NULL, 0 }, /* It */
203: { NULL, 0 }, /* Ad */
204: { NULL, TYPE_An }, /* An */
205: { NULL, TYPE_Ar }, /* Ar */
206: { NULL, TYPE_Cd }, /* Cd */
207: { NULL, TYPE_Cm }, /* Cm */
208: { NULL, TYPE_Dv }, /* Dv */
209: { NULL, TYPE_Er }, /* Er */
210: { NULL, TYPE_Ev }, /* Ev */
211: { NULL, 0 }, /* Ex */
212: { NULL, TYPE_Fa }, /* Fa */
213: { parse_mdoc_Fd, 0 }, /* Fd */
214: { NULL, TYPE_Fl }, /* Fl */
215: { parse_mdoc_Fn, 0 }, /* Fn */
216: { NULL, TYPE_Ft }, /* Ft */
217: { NULL, TYPE_Ic }, /* Ic */
1.92 schwarze 218: { NULL, TYPE_In }, /* In */
1.68 schwarze 219: { NULL, TYPE_Li }, /* Li */
220: { parse_mdoc_Nd, TYPE_Nd }, /* Nd */
221: { parse_mdoc_Nm, TYPE_Nm }, /* Nm */
222: { NULL, 0 }, /* Op */
223: { NULL, 0 }, /* Ot */
224: { NULL, TYPE_Pa }, /* Pa */
225: { NULL, 0 }, /* Rv */
1.92 schwarze 226: { NULL, TYPE_St }, /* St */
1.68 schwarze 227: { NULL, TYPE_Va }, /* Va */
228: { parse_mdoc_body, TYPE_Va }, /* Vt */
229: { parse_mdoc_Xr, 0 }, /* Xr */
230: { NULL, 0 }, /* %A */
231: { NULL, 0 }, /* %B */
232: { NULL, 0 }, /* %D */
233: { NULL, 0 }, /* %I */
234: { NULL, 0 }, /* %J */
235: { NULL, 0 }, /* %N */
236: { NULL, 0 }, /* %O */
237: { NULL, 0 }, /* %P */
238: { NULL, 0 }, /* %R */
239: { NULL, 0 }, /* %T */
240: { NULL, 0 }, /* %V */
241: { NULL, 0 }, /* Ac */
242: { NULL, 0 }, /* Ao */
243: { NULL, 0 }, /* Aq */
244: { NULL, TYPE_At }, /* At */
245: { NULL, 0 }, /* Bc */
246: { NULL, 0 }, /* Bf */
247: { NULL, 0 }, /* Bo */
248: { NULL, 0 }, /* Bq */
249: { NULL, TYPE_Bsx }, /* Bsx */
250: { NULL, TYPE_Bx }, /* Bx */
251: { NULL, 0 }, /* Db */
252: { NULL, 0 }, /* Dc */
253: { NULL, 0 }, /* Do */
254: { NULL, 0 }, /* Dq */
255: { NULL, 0 }, /* Ec */
256: { NULL, 0 }, /* Ef */
257: { NULL, TYPE_Em }, /* Em */
258: { NULL, 0 }, /* Eo */
259: { NULL, TYPE_Fx }, /* Fx */
260: { NULL, TYPE_Ms }, /* Ms */
261: { NULL, 0 }, /* No */
262: { NULL, 0 }, /* Ns */
263: { NULL, TYPE_Nx }, /* Nx */
264: { NULL, TYPE_Ox }, /* Ox */
265: { NULL, 0 }, /* Pc */
266: { NULL, 0 }, /* Pf */
267: { NULL, 0 }, /* Po */
268: { NULL, 0 }, /* Pq */
269: { NULL, 0 }, /* Qc */
270: { NULL, 0 }, /* Ql */
271: { NULL, 0 }, /* Qo */
272: { NULL, 0 }, /* Qq */
273: { NULL, 0 }, /* Re */
274: { NULL, 0 }, /* Rs */
275: { NULL, 0 }, /* Sc */
276: { NULL, 0 }, /* So */
277: { NULL, 0 }, /* Sq */
278: { NULL, 0 }, /* Sm */
279: { NULL, 0 }, /* Sx */
280: { NULL, TYPE_Sy }, /* Sy */
281: { NULL, TYPE_Tn }, /* Tn */
282: { NULL, 0 }, /* Ux */
283: { NULL, 0 }, /* Xc */
284: { NULL, 0 }, /* Xo */
285: { parse_mdoc_head, 0 }, /* Fo */
286: { NULL, 0 }, /* Fc */
287: { NULL, 0 }, /* Oo */
288: { NULL, 0 }, /* Oc */
289: { NULL, 0 }, /* Bk */
290: { NULL, 0 }, /* Ek */
291: { NULL, 0 }, /* Bt */
292: { NULL, 0 }, /* Hf */
293: { NULL, 0 }, /* Fr */
294: { NULL, 0 }, /* Ud */
295: { NULL, TYPE_Lb }, /* Lb */
296: { NULL, 0 }, /* Lp */
297: { NULL, TYPE_Lk }, /* Lk */
298: { NULL, TYPE_Mt }, /* Mt */
299: { NULL, 0 }, /* Brq */
300: { NULL, 0 }, /* Bro */
301: { NULL, 0 }, /* Brc */
302: { NULL, 0 }, /* %C */
303: { NULL, 0 }, /* Es */
304: { NULL, 0 }, /* En */
305: { NULL, TYPE_Dx }, /* Dx */
306: { NULL, 0 }, /* %Q */
307: { NULL, 0 }, /* br */
308: { NULL, 0 }, /* sp */
309: { NULL, 0 }, /* %U */
310: { NULL, 0 }, /* Ta */
1.1 kristaps 311: };
312:
313: int
314: main(int argc, char *argv[])
315: {
1.59 schwarze 316: int ch, i;
1.50 kristaps 317: size_t j, sz;
1.59 schwarze 318: const char *path_arg;
1.50 kristaps 319: struct mchars *mc;
320: struct manpaths dirs;
321: struct mparse *mp;
1.83 schwarze 322: struct ohash_info mpages_info, mlinks_info;
1.50 kristaps 323:
324: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
325: memset(&dirs, 0, sizeof(struct manpaths));
326:
1.83 schwarze 327: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
328: mpages_info.halloc = mlinks_info.halloc = hash_halloc;
329: mpages_info.hfree = mlinks_info.hfree = hash_free;
1.50 kristaps 330:
1.79 schwarze 331: mpages_info.key_offset = offsetof(struct mpage, inodev);
1.83 schwarze 332: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 kristaps 333:
334: progname = strrchr(argv[0], '/');
335: if (progname == NULL)
336: progname = argv[0];
337: else
338: ++progname;
339:
1.50 kristaps 340: /*
341: * We accept a few different invocations.
342: * The CHECKOP macro makes sure that invocation styles don't
343: * clobber each other.
344: */
345: #define CHECKOP(_op, _ch) do \
346: if (OP_DEFAULT != (_op)) { \
347: fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
348: goto usage; \
349: } while (/*CONSTCOND*/0)
1.10 kristaps 350:
1.59 schwarze 351: path_arg = NULL;
1.38 schwarze 352: op = OP_DEFAULT;
1.1 kristaps 353:
1.50 kristaps 354: while (-1 != (ch = getopt(argc, argv, "aC:d:ntu:vW")))
1.1 kristaps 355: switch (ch) {
1.12 schwarze 356: case ('a'):
357: use_all = 1;
358: break;
1.34 schwarze 359: case ('C'):
1.50 kristaps 360: CHECKOP(op, ch);
1.59 schwarze 361: path_arg = optarg;
1.38 schwarze 362: op = OP_CONFFILE;
1.34 schwarze 363: break;
1.5 kristaps 364: case ('d'):
1.50 kristaps 365: CHECKOP(op, ch);
1.59 schwarze 366: path_arg = optarg;
1.5 kristaps 367: op = OP_UPDATE;
368: break;
1.50 kristaps 369: case ('n'):
370: nodb = 1;
371: break;
1.38 schwarze 372: case ('t'):
1.50 kristaps 373: CHECKOP(op, ch);
1.38 schwarze 374: dup2(STDOUT_FILENO, STDERR_FILENO);
375: op = OP_TEST;
1.50 kristaps 376: nodb = warnings = 1;
1.38 schwarze 377: break;
1.5 kristaps 378: case ('u'):
1.50 kristaps 379: CHECKOP(op, ch);
1.59 schwarze 380: path_arg = optarg;
1.5 kristaps 381: op = OP_DELETE;
382: break;
383: case ('v'):
384: verb++;
385: break;
1.38 schwarze 386: case ('W'):
387: warnings = 1;
388: break;
1.1 kristaps 389: default:
1.38 schwarze 390: goto usage;
1.1 kristaps 391: }
392:
393: argc -= optind;
394: argv += optind;
395:
1.38 schwarze 396: if (OP_CONFFILE == op && argc > 0) {
1.50 kristaps 397: fprintf(stderr, "-C: Too many arguments\n");
1.38 schwarze 398: goto usage;
399: }
400:
1.59 schwarze 401: exitcode = (int)MANDOCLEVEL_OK;
1.50 kristaps 402: mp = mparse_alloc(MPARSE_AUTO,
403: MANDOCLEVEL_FATAL, NULL, NULL, NULL);
404: mc = mchars_alloc();
405:
1.78 schwarze 406: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 407: ohash_init(&mlinks, 6, &mlinks_info);
1.50 kristaps 408:
409: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
410: /*
411: * Force processing all files.
412: */
413: use_all = 1;
1.59 schwarze 414:
1.50 kristaps 415: /*
416: * All of these deal with a specific directory.
417: * Jump into that directory then collect files specified
418: * on the command-line.
419: */
1.59 schwarze 420: if (0 == set_basedir(path_arg))
1.50 kristaps 421: goto out;
422: for (i = 0; i < argc; i++)
1.59 schwarze 423: filescan(argv[i]);
424: if (0 == dbopen(1))
1.50 kristaps 425: goto out;
426: if (OP_TEST != op)
1.59 schwarze 427: dbprune();
1.50 kristaps 428: if (OP_DELETE != op)
1.78 schwarze 429: mpages_merge(mc, mp, 0);
1.59 schwarze 430: dbclose(1);
1.50 kristaps 431: } else {
432: /*
433: * If we have arguments, use them as our manpaths.
434: * If we don't, grok from manpath(1) or however else
435: * manpath_parse() wants to do it.
436: */
437: if (argc > 0) {
438: dirs.paths = mandoc_calloc
439: (argc, sizeof(char *));
440: dirs.sz = (size_t)argc;
441: for (i = 0; i < argc; i++)
442: dirs.paths[i] = mandoc_strdup(argv[i]);
443: } else
1.59 schwarze 444: manpath_parse(&dirs, path_arg, NULL, NULL);
1.50 kristaps 445:
446: /*
1.69 schwarze 447: * First scan the tree rooted at a base directory, then
448: * build a new database and finally move it into place.
1.50 kristaps 449: * Ignore zero-length directories and strip trailing
450: * slashes.
451: */
452: for (j = 0; j < dirs.sz; j++) {
453: sz = strlen(dirs.paths[j]);
454: if (sz && '/' == dirs.paths[j][sz - 1])
455: dirs.paths[j][--sz] = '\0';
456: if (0 == sz)
457: continue;
1.66 schwarze 458:
459: if (j) {
1.78 schwarze 460: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 461: ohash_init(&mlinks, 6, &mlinks_info);
1.66 schwarze 462: }
463:
1.59 schwarze 464: if (0 == set_basedir(dirs.paths[j]))
1.50 kristaps 465: goto out;
1.59 schwarze 466: if (0 == treescan())
1.50 kristaps 467: goto out;
1.59 schwarze 468: if (0 == set_basedir(dirs.paths[j]))
1.50 kristaps 469: goto out;
1.59 schwarze 470: if (0 == dbopen(0))
1.50 kristaps 471: goto out;
1.52 kristaps 472:
1.78 schwarze 473: mpages_merge(mc, mp, warnings && !use_all);
1.59 schwarze 474: dbclose(0);
1.66 schwarze 475:
476: if (j + 1 < dirs.sz) {
1.80 schwarze 477: mpages_free();
1.78 schwarze 478: ohash_delete(&mpages);
1.83 schwarze 479: ohash_delete(&mlinks);
1.66 schwarze 480: }
1.50 kristaps 481: }
482: }
483: out:
1.59 schwarze 484: set_basedir(NULL);
1.50 kristaps 485: manpath_free(&dirs);
486: mchars_free(mc);
487: mparse_free(mp);
1.80 schwarze 488: mpages_free();
1.78 schwarze 489: ohash_delete(&mpages);
1.83 schwarze 490: ohash_delete(&mlinks);
1.59 schwarze 491: return(exitcode);
1.50 kristaps 492: usage:
493: fprintf(stderr, "usage: %s [-anvW] [-C file]\n"
494: " %s [-anvW] dir ...\n"
495: " %s [-nvW] -d dir [file ...]\n"
496: " %s [-nvW] -u dir [file ...]\n"
497: " %s -t file ...\n",
498: progname, progname, progname,
499: progname, progname);
500:
1.59 schwarze 501: return((int)MANDOCLEVEL_BADARG);
1.50 kristaps 502: }
503:
504: /*
1.59 schwarze 505: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 506: * We use fts(), scanning directory parts along the way for clues to our
507: * section and architecture.
508: *
509: * If use_all has been specified, grok all files.
510: * If not, sanitise paths to the following:
511: *
512: * [./]man*[/<arch>]/<name>.<section>
513: * or
514: * [./]cat<section>[/<arch>]/<name>.0
515: *
516: * TODO: accomodate for multi-language directories.
517: */
518: static int
1.59 schwarze 519: treescan(void)
1.50 kristaps 520: {
521: FTS *f;
522: FTSENT *ff;
1.84 schwarze 523: struct mlink *mlink;
1.50 kristaps 524: int dform;
1.94 ! schwarze 525: char *dsec, *arch, *fsec, *cp;
! 526: const char *path;
1.50 kristaps 527: const char *argv[2];
528:
529: argv[0] = ".";
530: argv[1] = (char *)NULL;
531:
532: /*
533: * Walk through all components under the directory, using the
534: * logical descent of files.
535: */
536: f = fts_open((char * const *)argv, FTS_LOGICAL, NULL);
537: if (NULL == f) {
1.59 schwarze 538: exitcode = (int)MANDOCLEVEL_SYSERR;
539: say("", NULL);
1.50 kristaps 540: return(0);
541: }
542:
543: dsec = arch = NULL;
544: dform = FORM_NONE;
545:
546: while (NULL != (ff = fts_read(f))) {
547: path = ff->fts_path + 2;
548: /*
1.83 schwarze 549: * If we're a regular file, add an mlink by using the
1.50 kristaps 550: * stored directory data and handling the filename.
551: */
552: if (FTS_F == ff->fts_info) {
1.60 schwarze 553: if (0 == strcmp(path, MANDOC_DB))
554: continue;
1.50 kristaps 555: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 556: if (warnings)
1.59 schwarze 557: say(path, "Extraneous file");
1.50 kristaps 558: continue;
1.84 schwarze 559: } else if (NULL == (fsec =
1.60 schwarze 560: strrchr(ff->fts_name, '.'))) {
561: if ( ! use_all) {
1.56 schwarze 562: if (warnings)
1.60 schwarze 563: say(path,
564: "No filename suffix");
1.50 kristaps 565: continue;
566: }
1.84 schwarze 567: } else if (0 == strcmp(++fsec, "html")) {
1.60 schwarze 568: if (warnings)
569: say(path, "Skip html");
570: continue;
1.84 schwarze 571: } else if (0 == strcmp(fsec, "gz")) {
1.60 schwarze 572: if (warnings)
573: say(path, "Skip gz");
574: continue;
1.84 schwarze 575: } else if (0 == strcmp(fsec, "ps")) {
1.60 schwarze 576: if (warnings)
577: say(path, "Skip ps");
578: continue;
1.84 schwarze 579: } else if (0 == strcmp(fsec, "pdf")) {
1.60 schwarze 580: if (warnings)
581: say(path, "Skip pdf");
582: continue;
583: } else if ( ! use_all &&
1.84 schwarze 584: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
585: (FORM_CAT == dform && strcmp(fsec, "0")))) {
1.60 schwarze 586: if (warnings)
587: say(path, "Wrong filename suffix");
588: continue;
1.66 schwarze 589: } else
1.84 schwarze 590: fsec[-1] = '\0';
1.94 ! schwarze 591:
1.84 schwarze 592: mlink = mandoc_calloc(1, sizeof(struct mlink));
593: strlcpy(mlink->file, path, sizeof(mlink->file));
594: mlink->dform = dform;
1.94 ! schwarze 595: mlink->dsec = dsec;
! 596: mlink->arch = arch;
! 597: mlink->name = ff->fts_name;
! 598: mlink->fsec = fsec;
1.84 schwarze 599: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 600: continue;
1.84 schwarze 601: } else if (FTS_D != ff->fts_info &&
1.60 schwarze 602: FTS_DP != ff->fts_info) {
603: if (warnings)
604: say(path, "Not a regular file");
1.50 kristaps 605: continue;
1.60 schwarze 606: }
1.1 kristaps 607:
1.50 kristaps 608: switch (ff->fts_level) {
609: case (0):
610: /* Ignore the root directory. */
611: break;
612: case (1):
613: /*
614: * This might contain manX/ or catX/.
615: * Try to infer this from the name.
616: * If we're not in use_all, enforce it.
617: */
618: cp = ff->fts_name;
619: if (FTS_DP == ff->fts_info)
620: break;
1.1 kristaps 621:
1.50 kristaps 622: if (0 == strncmp(cp, "man", 3)) {
623: dform = FORM_SRC;
1.66 schwarze 624: dsec = cp + 3;
1.50 kristaps 625: } else if (0 == strncmp(cp, "cat", 3)) {
626: dform = FORM_CAT;
1.66 schwarze 627: dsec = cp + 3;
1.94 ! schwarze 628: } else {
! 629: dform = FORM_NONE;
! 630: dsec = NULL;
1.50 kristaps 631: }
1.1 kristaps 632:
1.50 kristaps 633: if (NULL != dsec || use_all)
634: break;
1.1 kristaps 635:
1.56 schwarze 636: if (warnings)
1.59 schwarze 637: say(path, "Unknown directory part");
1.50 kristaps 638: fts_set(f, ff, FTS_SKIP);
639: break;
640: case (2):
641: /*
642: * Possibly our architecture.
643: * If we're descending, keep tabs on it.
644: */
645: if (FTS_DP != ff->fts_info && NULL != dsec)
1.66 schwarze 646: arch = ff->fts_name;
1.94 ! schwarze 647: else
! 648: arch = NULL;
1.50 kristaps 649: break;
650: default:
651: if (FTS_DP == ff->fts_info || use_all)
652: break;
1.56 schwarze 653: if (warnings)
1.59 schwarze 654: say(path, "Extraneous directory part");
1.50 kristaps 655: fts_set(f, ff, FTS_SKIP);
656: break;
1.5 kristaps 657: }
1.50 kristaps 658: }
659:
660: fts_close(f);
661: return(1);
662: }
1.5 kristaps 663:
1.50 kristaps 664: /*
1.89 schwarze 665: * Add a file to the mlinks table.
1.50 kristaps 666: * Do not verify that it's a "valid" looking manpage (we'll do that
667: * later).
668: *
669: * Try to infer the manual section, architecture, and page name from the
670: * path, assuming it looks like
671: *
672: * [./]man*[/<arch>]/<name>.<section>
673: * or
674: * [./]cat<section>[/<arch>]/<name>.0
675: *
676: * See treescan() for the fts(3) version of this.
677: */
678: static void
1.59 schwarze 679: filescan(const char *file)
1.50 kristaps 680: {
1.59 schwarze 681: char buf[PATH_MAX];
1.84 schwarze 682: struct stat st;
683: struct mlink *mlink;
1.59 schwarze 684: char *p, *start;
1.5 kristaps 685:
1.50 kristaps 686: assert(use_all);
1.5 kristaps 687:
1.50 kristaps 688: if (0 == strncmp(file, "./", 2))
689: file += 2;
1.5 kristaps 690:
1.59 schwarze 691: if (NULL == realpath(file, buf)) {
692: exitcode = (int)MANDOCLEVEL_BADARG;
693: say(file, NULL);
694: return;
1.72 schwarze 695: } else if (OP_TEST != op && strstr(buf, basedir) != buf) {
1.59 schwarze 696: exitcode = (int)MANDOCLEVEL_BADARG;
697: say("", "%s: outside base directory", buf);
698: return;
699: } else if (-1 == stat(buf, &st)) {
700: exitcode = (int)MANDOCLEVEL_BADARG;
701: say(file, NULL);
1.50 kristaps 702: return;
703: } else if ( ! (S_IFREG & st.st_mode)) {
1.59 schwarze 704: exitcode = (int)MANDOCLEVEL_BADARG;
705: say(file, "Not a regular file");
1.50 kristaps 706: return;
707: }
1.59 schwarze 708: start = buf + strlen(basedir);
1.84 schwarze 709: mlink = mandoc_calloc(1, sizeof(struct mlink));
710: strlcpy(mlink->file, start, sizeof(mlink->file));
1.17 schwarze 711:
1.50 kristaps 712: /*
713: * First try to guess our directory structure.
714: * If we find a separator, try to look for man* or cat*.
715: * If we find one of these and what's underneath is a directory,
716: * assume it's an architecture.
717: */
718: if (NULL != (p = strchr(start, '/'))) {
719: *p++ = '\0';
720: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 721: mlink->dform = FORM_SRC;
1.94 ! schwarze 722: mlink->dsec = start + 3;
1.50 kristaps 723: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 724: mlink->dform = FORM_CAT;
1.94 ! schwarze 725: mlink->dsec = start + 3;
1.17 schwarze 726: }
1.5 kristaps 727:
1.50 kristaps 728: start = p;
1.84 schwarze 729: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 730: *p++ = '\0';
1.94 ! schwarze 731: mlink->arch = start;
1.50 kristaps 732: start = p;
1.84 schwarze 733: }
1.50 kristaps 734: }
735:
736: /*
737: * Now check the file suffix.
738: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
739: */
740: p = strrchr(start, '\0');
741: while (p-- > start && '/' != *p && '.' != *p)
742: /* Loop. */ ;
743:
744: if ('.' == *p) {
745: *p++ = '\0';
1.94 ! schwarze 746: mlink->fsec = p;
1.5 kristaps 747: }
748:
1.10 kristaps 749: /*
1.50 kristaps 750: * Now try to parse the name.
751: * Use the filename portion of the path.
1.10 kristaps 752: */
1.84 schwarze 753: mlink->name = start;
1.50 kristaps 754: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 755: mlink->name = p + 1;
1.50 kristaps 756: *p = '\0';
1.84 schwarze 757: }
758: mlink_add(mlink, &st);
1.50 kristaps 759: }
1.5 kristaps 760:
1.50 kristaps 761: static void
1.84 schwarze 762: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 763: {
1.83 schwarze 764: struct inodev inodev;
1.78 schwarze 765: struct mpage *mpage;
1.83 schwarze 766: unsigned int slot;
1.50 kristaps 767:
1.84 schwarze 768: assert(NULL != mlink->file);
1.50 kristaps 769:
1.94 ! schwarze 770: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
! 771: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
! 772: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
! 773: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 774:
775: if ('0' == *mlink->fsec) {
776: free(mlink->fsec);
777: mlink->fsec = mandoc_strdup(mlink->dsec);
778: mlink->fform = FORM_CAT;
779: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
780: mlink->fform = FORM_SRC;
1.74 schwarze 781: else
1.84 schwarze 782: mlink->fform = FORM_NONE;
1.82 schwarze 783:
1.83 schwarze 784: slot = ohash_qlookup(&mlinks, mlink->file);
785: assert(NULL == ohash_find(&mlinks, slot));
786: ohash_insert(&mlinks, slot, mlink);
787:
788: inodev.st_ino = st->st_ino;
789: inodev.st_dev = st->st_dev;
790: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
791: sizeof(struct inodev), inodev.st_ino);
792: mpage = ohash_find(&mpages, slot);
793: if (NULL == mpage) {
794: mpage = mandoc_calloc(1, sizeof(struct mpage));
795: mpage->inodev.st_ino = inodev.st_ino;
796: mpage->inodev.st_dev = inodev.st_dev;
797: ohash_insert(&mpages, slot, mpage);
798: } else
1.85 schwarze 799: mlink->next = mpage->mlinks;
1.82 schwarze 800: mpage->mlinks = mlink;
801: }
802:
803: static void
804: mlink_free(struct mlink *mlink)
805: {
806:
807: free(mlink->dsec);
808: free(mlink->arch);
809: free(mlink->name);
810: free(mlink->fsec);
811: free(mlink);
1.50 kristaps 812: }
1.3 kristaps 813:
1.50 kristaps 814: static void
1.78 schwarze 815: mpages_free(void)
1.50 kristaps 816: {
1.78 schwarze 817: struct mpage *mpage;
1.82 schwarze 818: struct mlink *mlink;
1.80 schwarze 819: unsigned int slot;
1.3 kristaps 820:
1.80 schwarze 821: mpage = ohash_first(&mpages, &slot);
822: while (NULL != mpage) {
1.82 schwarze 823: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 824: mpage->mlinks = mlink->next;
1.82 schwarze 825: mlink_free(mlink);
826: }
1.78 schwarze 827: free(mpage->sec);
828: free(mpage->arch);
1.82 schwarze 829: free(mpage->title);
830: free(mpage->desc);
1.78 schwarze 831: free(mpage);
1.80 schwarze 832: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 833: }
834: }
1.38 schwarze 835:
1.50 kristaps 836: /*
1.89 schwarze 837: * For each mlink to the mpage, check whether the path looks like
838: * it is formatted, and if it does, check whether a source manual
839: * exists by the same name, ignoring the suffix.
840: * If both conditions hold, drop the mlink.
841: */
842: static void
843: mlinks_undupe(struct mpage *mpage)
844: {
845: char buf[PATH_MAX];
846: struct mlink **prev;
847: struct mlink *mlink;
848: char *bufp;
849:
850: mpage->form = FORM_CAT;
1.90 schwarze 851: prev = &mpage->mlinks;
852: while (NULL != (mlink = *prev)) {
1.89 schwarze 853: if (FORM_CAT != mlink->dform) {
854: mpage->form = FORM_NONE;
1.90 schwarze 855: goto nextlink;
1.89 schwarze 856: }
857: if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
858: if (warnings)
859: say(mlink->file, "Filename too long");
1.90 schwarze 860: goto nextlink;
1.89 schwarze 861: }
862: bufp = strstr(buf, "cat");
863: assert(NULL != bufp);
864: memcpy(bufp, "man", 3);
865: if (NULL != (bufp = strrchr(buf, '.')))
866: *++bufp = '\0';
867: strlcat(buf, mlink->dsec, PATH_MAX);
868: if (NULL == ohash_find(&mlinks,
869: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 870: goto nextlink;
1.89 schwarze 871: if (warnings)
872: say(mlink->file, "Man source exists: %s", buf);
873: if (use_all)
1.90 schwarze 874: goto nextlink;
1.89 schwarze 875: *prev = mlink->next;
876: mlink_free(mlink);
1.90 schwarze 877: continue;
878: nextlink:
879: prev = &(*prev)->next;
1.89 schwarze 880: }
881: }
882:
1.93 schwarze 883: static int
884: mlink_check(struct mpage *mpage, struct mlink *mlink)
885: {
886: int match;
887:
888: match = 1;
889:
890: /*
891: * Check whether the manual section given in a file
892: * agrees with the directory where the file is located.
893: * Some manuals have suffixes like (3p) on their
894: * section number either inside the file or in the
895: * directory name, some are linked into more than one
896: * section, like encrypt(1) = makekey(8).
897: */
898:
899: if (FORM_SRC == mpage->form &&
900: strcasecmp(mpage->sec, mlink->dsec)) {
901: match = 0;
902: say(mlink->file, "Section \"%s\" manual in %s directory",
903: mpage->sec, mlink->dsec);
904: }
905:
906: /*
907: * Manual page directories exist for each kernel
908: * architecture as returned by machine(1).
909: * However, many manuals only depend on the
910: * application architecture as returned by arch(1).
911: * For example, some (2/ARM) manuals are shared
912: * across the "armish" and "zaurus" kernel
913: * architectures.
914: * A few manuals are even shared across completely
915: * different architectures, for example fdformat(1)
916: * on amd64, i386, sparc, and sparc64.
917: */
918:
919: if (strcasecmp(mpage->arch, mlink->arch)) {
920: match = 0;
921: say(mlink->file, "Architecture \"%s\" manual in "
922: "\"%s\" directory", mpage->arch, mlink->arch);
923: }
924:
925: if (strcasecmp(mpage->title, mlink->name))
926: match = 0;
927:
928: return(match);
929: }
930:
1.89 schwarze 931: /*
1.80 schwarze 932: * Run through the files in the global vector "mpages"
933: * and add them to the database specified in "basedir".
1.50 kristaps 934: *
935: * This handles the parsing scheme itself, using the cues of directory
936: * and filename to determine whether the file is parsable or not.
937: */
1.59 schwarze 938: static void
1.78 schwarze 939: mpages_merge(struct mchars *mc, struct mparse *mp, int check_reachable)
1.50 kristaps 940: {
1.69 schwarze 941: struct ohash title_table;
1.71 schwarze 942: struct ohash_info title_info, str_info;
1.78 schwarze 943: struct mpage *mpage;
1.93 schwarze 944: struct mlink *mlink;
1.69 schwarze 945: struct mdoc *mdoc;
946: struct man *man;
947: struct title *title_entry;
1.89 schwarze 948: char *title_str;
1.82 schwarze 949: const char *cp;
1.69 schwarze 950: int match;
1.80 schwarze 951: unsigned int pslot, tslot;
1.69 schwarze 952: enum mandoclevel lvl;
953:
1.71 schwarze 954: str_info.alloc = hash_alloc;
955: str_info.halloc = hash_halloc;
956: str_info.hfree = hash_free;
957: str_info.key_offset = offsetof(struct str, key);
958:
1.69 schwarze 959: if (check_reachable) {
960: title_info.alloc = hash_alloc;
961: title_info.halloc = hash_halloc;
962: title_info.hfree = hash_free;
963: title_info.key_offset = offsetof(struct title, title);
964: ohash_init(&title_table, 6, &title_info);
965: }
1.50 kristaps 966:
1.80 schwarze 967: mpage = ohash_first(&mpages, &pslot);
968: while (NULL != mpage) {
1.89 schwarze 969: mlinks_undupe(mpage);
970: if (NULL == mpage->mlinks) {
971: mpage = ohash_next(&mpages, &pslot);
972: continue;
1.50 kristaps 973: }
974:
1.71 schwarze 975: ohash_init(&strings, 6, &str_info);
1.50 kristaps 976: mparse_reset(mp);
977: mdoc = NULL;
978: man = NULL;
1.14 schwarze 979:
980: /*
1.33 schwarze 981: * Try interpreting the file as mdoc(7) or man(7)
982: * source code, unless it is already known to be
983: * formatted. Fall back to formatted mode.
1.14 schwarze 984: */
1.82 schwarze 985: if (FORM_CAT != mpage->mlinks->dform ||
1.84 schwarze 986: FORM_CAT != mpage->mlinks->fform) {
1.82 schwarze 987: lvl = mparse_readfd(mp, -1, mpage->mlinks->file);
1.50 kristaps 988: if (lvl < MANDOCLEVEL_FATAL)
989: mparse_result(mp, &mdoc, &man);
1.76 schwarze 990: }
1.14 schwarze 991:
992: if (NULL != mdoc) {
1.82 schwarze 993: mpage->form = FORM_SRC;
994: mpage->sec =
995: mandoc_strdup(mdoc_meta(mdoc)->msec);
996: mpage->arch = mdoc_meta(mdoc)->arch;
1.84 schwarze 997: mpage->arch = mandoc_strdup(
998: NULL == mpage->arch ? "" : mpage->arch);
1.82 schwarze 999: mpage->title =
1000: mandoc_strdup(mdoc_meta(mdoc)->title);
1.14 schwarze 1001: } else if (NULL != man) {
1.82 schwarze 1002: mpage->form = FORM_SRC;
1003: mpage->sec =
1004: mandoc_strdup(man_meta(man)->msec);
1005: mpage->arch =
1006: mandoc_strdup(mpage->mlinks->arch);
1007: mpage->title =
1008: mandoc_strdup(man_meta(man)->title);
1.76 schwarze 1009: } else {
1.82 schwarze 1010: mpage->form = FORM_CAT;
1011: mpage->sec =
1012: mandoc_strdup(mpage->mlinks->dsec);
1013: mpage->arch =
1014: mandoc_strdup(mpage->mlinks->arch);
1015: mpage->title =
1016: mandoc_strdup(mpage->mlinks->name);
1.76 schwarze 1017: }
1.50 kristaps 1018:
1.93 schwarze 1019: for (mlink = mpage->mlinks; mlink; mlink = mlink->next)
1020: putkey(mpage, mlink->name, TYPE_Nm);
1.12 schwarze 1021:
1.93 schwarze 1022: if (warnings && !use_all) {
1.69 schwarze 1023: match = 0;
1.93 schwarze 1024: for (mlink = mpage->mlinks; mlink;
1025: mlink = mlink->next)
1026: if (mlink_check(mpage, mlink))
1027: match = 1;
1028: } else
1029: match = 1;
1.12 schwarze 1030:
1.50 kristaps 1031: if (NULL != mdoc) {
1032: if (NULL != (cp = mdoc_meta(mdoc)->name))
1.78 schwarze 1033: putkey(mpage, cp, TYPE_Nm);
1034: assert(NULL == mpage->desc);
1035: parse_mdoc(mpage, mdoc_node(mdoc));
1036: putkey(mpage, NULL != mpage->desc ?
1.82 schwarze 1037: mpage->desc : mpage->mlinks->name, TYPE_Nd);
1.50 kristaps 1038: } else if (NULL != man)
1.78 schwarze 1039: parse_man(mpage, man_node(man));
1.50 kristaps 1040: else
1.78 schwarze 1041: parse_cat(mpage);
1.44 kristaps 1042:
1.69 schwarze 1043: /*
1044: * Build a title string for the file. If it matches
1045: * the location of the file, remember the title as
1046: * found; else, remember it as missing.
1047: */
1048:
1049: if (check_reachable) {
1.82 schwarze 1050: if (-1 == asprintf(&title_str, "%s(%s%s%s)",
1051: mpage->title, mpage->sec,
1052: '\0' == *mpage->arch ? "" : "/",
1053: mpage->arch)) {
1.69 schwarze 1054: perror(NULL);
1055: exit((int)MANDOCLEVEL_SYSERR);
1056: }
1.80 schwarze 1057: tslot = ohash_qlookup(&title_table, title_str);
1058: title_entry = ohash_find(&title_table, tslot);
1.69 schwarze 1059: if (NULL == title_entry) {
1060: title_entry = mandoc_malloc(
1061: sizeof(struct title));
1062: title_entry->title = title_str;
1063: title_entry->file = mandoc_strdup(
1.82 schwarze 1064: match ? "" : mpage->mlinks->file);
1.80 schwarze 1065: ohash_insert(&title_table, tslot,
1.69 schwarze 1066: title_entry);
1067: } else {
1068: if (match)
1069: *title_entry->file = '\0';
1070: free(title_str);
1071: }
1072: }
1073:
1.82 schwarze 1074: dbindex(mpage, mc);
1.66 schwarze 1075: ohash_delete(&strings);
1.80 schwarze 1076: mpage = ohash_next(&mpages, &pslot);
1.69 schwarze 1077: }
1078:
1079: if (check_reachable) {
1.80 schwarze 1080: title_entry = ohash_first(&title_table, &tslot);
1.69 schwarze 1081: while (NULL != title_entry) {
1082: if ('\0' != *title_entry->file)
1083: say(title_entry->file,
1084: "Probably unreachable, title is %s",
1085: title_entry->title);
1086: free(title_entry->title);
1087: free(title_entry->file);
1088: free(title_entry);
1.80 schwarze 1089: title_entry = ohash_next(&title_table, &tslot);
1.69 schwarze 1090: }
1091: ohash_delete(&title_table);
1.50 kristaps 1092: }
1093: }
1.12 schwarze 1094:
1.50 kristaps 1095: static void
1.78 schwarze 1096: parse_cat(struct mpage *mpage)
1.50 kristaps 1097: {
1098: FILE *stream;
1099: char *line, *p, *title;
1100: size_t len, plen, titlesz;
1.12 schwarze 1101:
1.82 schwarze 1102: if (NULL == (stream = fopen(mpage->mlinks->file, "r"))) {
1.56 schwarze 1103: if (warnings)
1.82 schwarze 1104: say(mpage->mlinks->file, NULL);
1.50 kristaps 1105: return;
1106: }
1.1 kristaps 1107:
1.50 kristaps 1108: /* Skip to first blank line. */
1.1 kristaps 1109:
1.50 kristaps 1110: while (NULL != (line = fgetln(stream, &len)))
1111: if ('\n' == *line)
1112: break;
1.1 kristaps 1113:
1.50 kristaps 1114: /*
1115: * Assume the first line that is not indented
1116: * is the first section header. Skip to it.
1117: */
1.1 kristaps 1118:
1.50 kristaps 1119: while (NULL != (line = fgetln(stream, &len)))
1120: if ('\n' != *line && ' ' != *line)
1121: break;
1122:
1123: /*
1124: * Read up until the next section into a buffer.
1125: * Strip the leading and trailing newline from each read line,
1126: * appending a trailing space.
1127: * Ignore empty (whitespace-only) lines.
1128: */
1.1 kristaps 1129:
1.50 kristaps 1130: titlesz = 0;
1131: title = NULL;
1.38 schwarze 1132:
1.50 kristaps 1133: while (NULL != (line = fgetln(stream, &len))) {
1134: if (' ' != *line || '\n' != line[len - 1])
1135: break;
1136: while (len > 0 && isspace((unsigned char)*line)) {
1137: line++;
1138: len--;
1139: }
1140: if (1 == len)
1.38 schwarze 1141: continue;
1.50 kristaps 1142: title = mandoc_realloc(title, titlesz + len);
1143: memcpy(title + titlesz, line, len);
1144: titlesz += len;
1145: title[titlesz - 1] = ' ';
1146: }
1.38 schwarze 1147:
1.50 kristaps 1148: /*
1149: * If no page content can be found, or the input line
1150: * is already the next section header, or there is no
1151: * trailing newline, reuse the page title as the page
1152: * description.
1153: */
1.44 kristaps 1154:
1.50 kristaps 1155: if (NULL == title || '\0' == *title) {
1.56 schwarze 1156: if (warnings)
1.82 schwarze 1157: say(mpage->mlinks->file,
1158: "Cannot find NAME section");
1.78 schwarze 1159: assert(NULL == mpage->desc);
1.82 schwarze 1160: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1161: putkey(mpage, mpage->mlinks->name, TYPE_Nd);
1.50 kristaps 1162: fclose(stream);
1163: free(title);
1164: return;
1165: }
1.1 kristaps 1166:
1.50 kristaps 1167: title = mandoc_realloc(title, titlesz + 1);
1168: title[titlesz] = '\0';
1.33 schwarze 1169:
1.50 kristaps 1170: /*
1171: * Skip to the first dash.
1172: * Use the remaining line as the description (no more than 70
1173: * bytes).
1174: */
1.33 schwarze 1175:
1.50 kristaps 1176: if (NULL != (p = strstr(title, "- "))) {
1177: for (p += 2; ' ' == *p || '\b' == *p; p++)
1178: /* Skip to next word. */ ;
1179: } else {
1.56 schwarze 1180: if (warnings)
1.82 schwarze 1181: say(mpage->mlinks->file,
1182: "No dash in title line");
1.50 kristaps 1183: p = title;
1184: }
1.38 schwarze 1185:
1.50 kristaps 1186: plen = strlen(p);
1.1 kristaps 1187:
1.50 kristaps 1188: /* Strip backspace-encoding from line. */
1.1 kristaps 1189:
1.50 kristaps 1190: while (NULL != (line = memchr(p, '\b', plen))) {
1191: len = line - p;
1192: if (0 == len) {
1193: memmove(line, line + 1, plen--);
1194: continue;
1195: }
1196: memmove(line - 1, line + 1, plen - len);
1197: plen -= 2;
1198: }
1.1 kristaps 1199:
1.78 schwarze 1200: assert(NULL == mpage->desc);
1201: mpage->desc = mandoc_strdup(p);
1202: putkey(mpage, mpage->desc, TYPE_Nd);
1.50 kristaps 1203: fclose(stream);
1204: free(title);
1205: }
1.1 kristaps 1206:
1.50 kristaps 1207: /*
1208: * Put a type/word pair into the word database for this particular file.
1209: */
1210: static void
1.78 schwarze 1211: putkey(const struct mpage *mpage, const char *value, uint64_t type)
1.50 kristaps 1212: {
1.18 kristaps 1213:
1.50 kristaps 1214: assert(NULL != value);
1.78 schwarze 1215: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1216: }
1217:
1218: /*
1.50 kristaps 1219: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1220: */
1221: static void
1.78 schwarze 1222: putmdockey(const struct mpage *mpage,
1223: const struct mdoc_node *n, uint64_t m)
1.3 kristaps 1224: {
1.18 kristaps 1225:
1.50 kristaps 1226: for ( ; NULL != n; n = n->next) {
1227: if (NULL != n->child)
1.78 schwarze 1228: putmdockey(mpage, n->child, m);
1.50 kristaps 1229: if (MDOC_TEXT == n->type)
1.78 schwarze 1230: putkey(mpage, n->string, m);
1.50 kristaps 1231: }
1232: }
1.18 kristaps 1233:
1.61 schwarze 1234: static void
1.78 schwarze 1235: parse_man(struct mpage *mpage, const struct man_node *n)
1.50 kristaps 1236: {
1237: const struct man_node *head, *body;
1238: char *start, *sv, *title;
1239: char byte;
1240: size_t sz, titlesz;
1.18 kristaps 1241:
1.50 kristaps 1242: if (NULL == n)
1.61 schwarze 1243: return;
1.18 kristaps 1244:
1.50 kristaps 1245: /*
1246: * We're only searching for one thing: the first text child in
1247: * the BODY of a NAME section. Since we don't keep track of
1248: * sections in -man, run some hoops to find out whether we're in
1249: * the correct section or not.
1250: */
1.18 kristaps 1251:
1.50 kristaps 1252: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1253: body = n;
1254: assert(body->parent);
1255: if (NULL != (head = body->parent->head) &&
1256: 1 == head->nchild &&
1257: NULL != (head = (head->child)) &&
1258: MAN_TEXT == head->type &&
1259: 0 == strcmp(head->string, "NAME") &&
1260: NULL != (body = body->child) &&
1261: MAN_TEXT == body->type) {
1.3 kristaps 1262:
1.50 kristaps 1263: title = NULL;
1264: titlesz = 0;
1.3 kristaps 1265:
1.50 kristaps 1266: /*
1267: * Suck the entire NAME section into memory.
1268: * Yes, we might run away.
1269: * But too many manuals have big, spread-out
1270: * NAME sections over many lines.
1271: */
1.3 kristaps 1272:
1.50 kristaps 1273: for ( ; NULL != body; body = body->next) {
1274: if (MAN_TEXT != body->type)
1275: break;
1276: if (0 == (sz = strlen(body->string)))
1277: continue;
1278: title = mandoc_realloc
1279: (title, titlesz + sz + 1);
1280: memcpy(title + titlesz, body->string, sz);
1281: titlesz += sz + 1;
1282: title[titlesz - 1] = ' ';
1283: }
1284: if (NULL == title)
1.61 schwarze 1285: return;
1.18 kristaps 1286:
1.50 kristaps 1287: title = mandoc_realloc(title, titlesz + 1);
1288: title[titlesz] = '\0';
1.18 kristaps 1289:
1.50 kristaps 1290: /* Skip leading space. */
1.18 kristaps 1291:
1.50 kristaps 1292: sv = title;
1293: while (isspace((unsigned char)*sv))
1294: sv++;
1.18 kristaps 1295:
1.50 kristaps 1296: if (0 == (sz = strlen(sv))) {
1297: free(title);
1.61 schwarze 1298: return;
1.50 kristaps 1299: }
1.1 kristaps 1300:
1.50 kristaps 1301: /* Erase trailing space. */
1.1 kristaps 1302:
1.50 kristaps 1303: start = &sv[sz - 1];
1304: while (start > sv && isspace((unsigned char)*start))
1305: *start-- = '\0';
1.1 kristaps 1306:
1.50 kristaps 1307: if (start == sv) {
1308: free(title);
1.61 schwarze 1309: return;
1.50 kristaps 1310: }
1.1 kristaps 1311:
1.50 kristaps 1312: start = sv;
1.18 kristaps 1313:
1.50 kristaps 1314: /*
1315: * Go through a special heuristic dance here.
1316: * Conventionally, one or more manual names are
1317: * comma-specified prior to a whitespace, then a
1318: * dash, then a description. Try to puzzle out
1319: * the name parts here.
1320: */
1.18 kristaps 1321:
1.50 kristaps 1322: for ( ;; ) {
1323: sz = strcspn(start, " ,");
1324: if ('\0' == start[sz])
1325: break;
1.1 kristaps 1326:
1.50 kristaps 1327: byte = start[sz];
1328: start[sz] = '\0';
1.1 kristaps 1329:
1.78 schwarze 1330: putkey(mpage, start, TYPE_Nm);
1.1 kristaps 1331:
1.50 kristaps 1332: if (' ' == byte) {
1333: start += sz + 1;
1334: break;
1335: }
1.1 kristaps 1336:
1.50 kristaps 1337: assert(',' == byte);
1338: start += sz + 1;
1339: while (' ' == *start)
1340: start++;
1341: }
1.1 kristaps 1342:
1.50 kristaps 1343: if (sv == start) {
1.78 schwarze 1344: putkey(mpage, start, TYPE_Nm);
1.50 kristaps 1345: free(title);
1.61 schwarze 1346: return;
1.50 kristaps 1347: }
1.1 kristaps 1348:
1.50 kristaps 1349: while (isspace((unsigned char)*start))
1350: start++;
1.1 kristaps 1351:
1.50 kristaps 1352: if (0 == strncmp(start, "-", 1))
1353: start += 1;
1354: else if (0 == strncmp(start, "\\-\\-", 4))
1355: start += 4;
1356: else if (0 == strncmp(start, "\\-", 2))
1357: start += 2;
1358: else if (0 == strncmp(start, "\\(en", 4))
1359: start += 4;
1360: else if (0 == strncmp(start, "\\(em", 4))
1361: start += 4;
1.1 kristaps 1362:
1.50 kristaps 1363: while (' ' == *start)
1364: start++;
1.1 kristaps 1365:
1.78 schwarze 1366: assert(NULL == mpage->desc);
1367: mpage->desc = mandoc_strdup(start);
1368: putkey(mpage, mpage->desc, TYPE_Nd);
1.50 kristaps 1369: free(title);
1.61 schwarze 1370: return;
1.50 kristaps 1371: }
1372: }
1.1 kristaps 1373:
1.77 schwarze 1374: for (n = n->child; n; n = n->next) {
1.78 schwarze 1375: if (NULL != mpage->desc)
1.77 schwarze 1376: break;
1.78 schwarze 1377: parse_man(mpage, n);
1.77 schwarze 1378: }
1.1 kristaps 1379: }
1380:
1381: static void
1.78 schwarze 1382: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1383: {
1384:
1.50 kristaps 1385: assert(NULL != n);
1386: for (n = n->child; NULL != n; n = n->next) {
1387: switch (n->type) {
1388: case (MDOC_ELEM):
1389: /* FALLTHROUGH */
1390: case (MDOC_BLOCK):
1391: /* FALLTHROUGH */
1392: case (MDOC_HEAD):
1393: /* FALLTHROUGH */
1394: case (MDOC_BODY):
1395: /* FALLTHROUGH */
1396: case (MDOC_TAIL):
1397: if (NULL != mdocs[n->tok].fp)
1.78 schwarze 1398: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1.50 kristaps 1399: break;
1.68 schwarze 1400: if (mdocs[n->tok].mask)
1.78 schwarze 1401: putmdockey(mpage, n->child,
1402: mdocs[n->tok].mask);
1.50 kristaps 1403: break;
1404: default:
1405: assert(MDOC_ROOT != n->type);
1406: continue;
1407: }
1408: if (NULL != n->child)
1.78 schwarze 1409: parse_mdoc(mpage, n);
1.1 kristaps 1410: }
1411: }
1412:
1.25 schwarze 1413: static int
1.78 schwarze 1414: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1415: {
1416: const char *start, *end;
1417: size_t sz;
1.25 schwarze 1418:
1.50 kristaps 1419: if (SEC_SYNOPSIS != n->sec ||
1420: NULL == (n = n->child) ||
1421: MDOC_TEXT != n->type)
1.25 schwarze 1422: return(0);
1.1 kristaps 1423:
1424: /*
1425: * Only consider those `Fd' macro fields that begin with an
1426: * "inclusion" token (versus, e.g., #define).
1427: */
1.50 kristaps 1428:
1.1 kristaps 1429: if (strcmp("#include", n->string))
1.25 schwarze 1430: return(0);
1.1 kristaps 1431:
1432: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.25 schwarze 1433: return(0);
1.1 kristaps 1434:
1435: /*
1436: * Strip away the enclosing angle brackets and make sure we're
1437: * not zero-length.
1438: */
1439:
1440: start = n->string;
1441: if ('<' == *start || '"' == *start)
1442: start++;
1443:
1444: if (0 == (sz = strlen(start)))
1.25 schwarze 1445: return(0);
1.1 kristaps 1446:
1447: end = &start[(int)sz - 1];
1448: if ('>' == *end || '"' == *end)
1449: end--;
1450:
1.50 kristaps 1451: if (end > start)
1.78 schwarze 1452: putkeys(mpage, start, end - start + 1, TYPE_In);
1.92 schwarze 1453: return(0);
1.1 kristaps 1454: }
1455:
1.25 schwarze 1456: static int
1.78 schwarze 1457: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1458: {
1459: const char *cp;
1460:
1.50 kristaps 1461: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.25 schwarze 1462: return(0);
1463:
1.50 kristaps 1464: /*
1465: * Parse: .Fn "struct type *name" "char *arg".
1466: * First strip away pointer symbol.
1467: * Then store the function name, then type.
1468: * Finally, store the arguments.
1469: */
1.1 kristaps 1470:
1.50 kristaps 1471: if (NULL == (cp = strrchr(n->string, ' ')))
1472: cp = n->string;
1.1 kristaps 1473:
1474: while ('*' == *cp)
1475: cp++;
1476:
1.78 schwarze 1477: putkey(mpage, cp, TYPE_Fn);
1.25 schwarze 1478:
1.50 kristaps 1479: if (n->string < cp)
1.78 schwarze 1480: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.25 schwarze 1481:
1.50 kristaps 1482: for (n = n->next; NULL != n; n = n->next)
1483: if (MDOC_TEXT == n->type)
1.78 schwarze 1484: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1485:
1486: return(0);
1.1 kristaps 1487: }
1488:
1.25 schwarze 1489: static int
1.78 schwarze 1490: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1491: {
1.67 schwarze 1492: char *cp;
1.1 kristaps 1493:
1494: if (NULL == (n = n->child))
1.25 schwarze 1495: return(0);
1.1 kristaps 1496:
1.67 schwarze 1497: if (NULL == n->next) {
1.78 schwarze 1498: putkey(mpage, n->string, TYPE_Xr);
1.67 schwarze 1499: return(0);
1500: }
1501:
1502: if (-1 == asprintf(&cp, "%s(%s)", n->string, n->next->string)) {
1503: perror(NULL);
1504: exit((int)MANDOCLEVEL_SYSERR);
1505: }
1.78 schwarze 1506: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1507: free(cp);
1508: return(0);
1.1 kristaps 1509: }
1510:
1.25 schwarze 1511: static int
1.78 schwarze 1512: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1513: {
1.50 kristaps 1514: size_t sz;
1.1 kristaps 1515:
1516: if (MDOC_BODY != n->type)
1.25 schwarze 1517: return(0);
1.1 kristaps 1518:
1.50 kristaps 1519: /*
1520: * Special-case the `Nd' because we need to put the description
1521: * into the document table.
1522: */
1523:
1524: for (n = n->child; NULL != n; n = n->next) {
1525: if (MDOC_TEXT == n->type) {
1.78 schwarze 1526: if (NULL != mpage->desc) {
1527: sz = strlen(mpage->desc) +
1.66 schwarze 1528: strlen(n->string) + 2;
1.78 schwarze 1529: mpage->desc = mandoc_realloc(
1530: mpage->desc, sz);
1531: strlcat(mpage->desc, " ", sz);
1532: strlcat(mpage->desc, n->string, sz);
1.66 schwarze 1533: } else
1.78 schwarze 1534: mpage->desc = mandoc_strdup(n->string);
1.50 kristaps 1535: }
1536: if (NULL != n->child)
1.78 schwarze 1537: parse_mdoc_Nd(mpage, n);
1.50 kristaps 1538: }
1.25 schwarze 1539: return(1);
1.1 kristaps 1540: }
1541:
1.25 schwarze 1542: static int
1.78 schwarze 1543: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1544: {
1545:
1.92 schwarze 1546: return(SEC_NAME == n->sec ||
1547: (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type));
1.1 kristaps 1548: }
1549:
1.25 schwarze 1550: static int
1.78 schwarze 1551: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1552: {
1553:
1.25 schwarze 1554: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 kristaps 1555: }
1556:
1.50 kristaps 1557: static int
1.78 schwarze 1558: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1559: {
1560:
1.50 kristaps 1561: return(MDOC_HEAD == n->type);
1562: }
1.1 kristaps 1563:
1.50 kristaps 1564: static int
1.78 schwarze 1565: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1.50 kristaps 1566: {
1.1 kristaps 1567:
1.50 kristaps 1568: return(MDOC_BODY == n->type);
1.1 kristaps 1569: }
1570:
1.50 kristaps 1571: /*
1.66 schwarze 1572: * Add a string to the hash table for the current manual.
1573: * Each string has a bitmask telling which macros it belongs to.
1574: * When we finish the manual, we'll dump the table.
1.50 kristaps 1575: */
1576: static void
1.78 schwarze 1577: putkeys(const struct mpage *mpage,
1578: const char *cp, size_t sz, uint64_t v)
1.50 kristaps 1579: {
1580: struct str *s;
1.65 schwarze 1581: unsigned int slot;
1.50 kristaps 1582: const char *end;
1.25 schwarze 1583:
1.50 kristaps 1584: if (0 == sz)
1585: return;
1.25 schwarze 1586:
1.65 schwarze 1587: end = cp + sz;
1588: slot = ohash_qlookupi(&strings, cp, &end);
1589: s = ohash_find(&strings, slot);
1.25 schwarze 1590:
1.78 schwarze 1591: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1592: s->mask |= v;
1593: return;
1594: } else if (NULL == s) {
1.51 kristaps 1595: s = mandoc_calloc(sizeof(struct str) + sz + 1, 1);
1.50 kristaps 1596: memcpy(s->key, cp, sz);
1.65 schwarze 1597: ohash_insert(&strings, slot, s);
1.1 kristaps 1598: }
1.78 schwarze 1599: s->mpage = mpage;
1.50 kristaps 1600: s->mask = v;
1.1 kristaps 1601: }
1602:
1.50 kristaps 1603: /*
1604: * Take a Unicode codepoint and produce its UTF-8 encoding.
1605: * This isn't the best way to do this, but it works.
1606: * The magic numbers are from the UTF-8 packaging.
1607: * They're not as scary as they seem: read the UTF-8 spec for details.
1608: */
1609: static size_t
1610: utf8(unsigned int cp, char out[7])
1.1 kristaps 1611: {
1.50 kristaps 1612: size_t rc;
1.1 kristaps 1613:
1.50 kristaps 1614: rc = 0;
1615: if (cp <= 0x0000007F) {
1616: rc = 1;
1617: out[0] = (char)cp;
1618: } else if (cp <= 0x000007FF) {
1619: rc = 2;
1620: out[0] = (cp >> 6 & 31) | 192;
1621: out[1] = (cp & 63) | 128;
1622: } else if (cp <= 0x0000FFFF) {
1623: rc = 3;
1624: out[0] = (cp >> 12 & 15) | 224;
1625: out[1] = (cp >> 6 & 63) | 128;
1626: out[2] = (cp & 63) | 128;
1627: } else if (cp <= 0x001FFFFF) {
1628: rc = 4;
1629: out[0] = (cp >> 18 & 7) | 240;
1630: out[1] = (cp >> 12 & 63) | 128;
1631: out[2] = (cp >> 6 & 63) | 128;
1632: out[3] = (cp & 63) | 128;
1633: } else if (cp <= 0x03FFFFFF) {
1634: rc = 5;
1635: out[0] = (cp >> 24 & 3) | 248;
1636: out[1] = (cp >> 18 & 63) | 128;
1637: out[2] = (cp >> 12 & 63) | 128;
1638: out[3] = (cp >> 6 & 63) | 128;
1639: out[4] = (cp & 63) | 128;
1640: } else if (cp <= 0x7FFFFFFF) {
1641: rc = 6;
1642: out[0] = (cp >> 30 & 1) | 252;
1643: out[1] = (cp >> 24 & 63) | 128;
1644: out[2] = (cp >> 18 & 63) | 128;
1645: out[3] = (cp >> 12 & 63) | 128;
1646: out[4] = (cp >> 6 & 63) | 128;
1647: out[5] = (cp & 63) | 128;
1648: } else
1.1 kristaps 1649: return(0);
1650:
1.50 kristaps 1651: out[rc] = '\0';
1652: return(rc);
1653: }
1.1 kristaps 1654:
1.50 kristaps 1655: /*
1656: * Store the UTF-8 version of a key, or alias the pointer if the key has
1657: * no UTF-8 transcription marks in it.
1658: */
1659: static void
1660: utf8key(struct mchars *mc, struct str *key)
1661: {
1662: size_t sz, bsz, pos;
1663: char utfbuf[7], res[5];
1664: char *buf;
1665: const char *seq, *cpp, *val;
1666: int len, u;
1667: enum mandoc_esc esc;
1668:
1669: assert(NULL == key->utf8);
1670:
1671: res[0] = '\\';
1672: res[1] = '\t';
1673: res[2] = ASCII_NBRSP;
1674: res[3] = ASCII_HYPH;
1675: res[4] = '\0';
1.1 kristaps 1676:
1.50 kristaps 1677: val = key->key;
1678: bsz = strlen(val);
1.46 kristaps 1679:
1.50 kristaps 1680: /*
1681: * Pre-check: if we have no stop-characters, then set the
1682: * pointer as ourselvse and get out of here.
1683: */
1684: if (strcspn(val, res) == bsz) {
1685: key->utf8 = key->key;
1686: return;
1687: }
1.46 kristaps 1688:
1.50 kristaps 1689: /* Pre-allocate by the length of the input */
1.46 kristaps 1690:
1.50 kristaps 1691: buf = mandoc_malloc(++bsz);
1692: pos = 0;
1.46 kristaps 1693:
1.50 kristaps 1694: while ('\0' != *val) {
1695: /*
1696: * Halt on the first escape sequence.
1697: * This also halts on the end of string, in which case
1698: * we just copy, fallthrough, and exit the loop.
1699: */
1700: if ((sz = strcspn(val, res)) > 0) {
1701: memcpy(&buf[pos], val, sz);
1702: pos += sz;
1703: val += sz;
1704: }
1.46 kristaps 1705:
1.50 kristaps 1706: if (ASCII_HYPH == *val) {
1707: buf[pos++] = '-';
1708: val++;
1709: continue;
1710: } else if ('\t' == *val || ASCII_NBRSP == *val) {
1711: buf[pos++] = ' ';
1712: val++;
1713: continue;
1714: } else if ('\\' != *val)
1715: break;
1.46 kristaps 1716:
1.50 kristaps 1717: /* Read past the slash. */
1.46 kristaps 1718:
1.50 kristaps 1719: val++;
1.46 kristaps 1720:
1.50 kristaps 1721: /*
1722: * Parse the escape sequence and see if it's a
1723: * predefined character or special character.
1724: */
1725: esc = mandoc_escape
1726: ((const char **)&val, &seq, &len);
1727: if (ESCAPE_ERROR == esc)
1728: break;
1.1 kristaps 1729:
1.50 kristaps 1730: if (ESCAPE_SPECIAL != esc)
1731: continue;
1732: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1733: continue;
1.1 kristaps 1734:
1.50 kristaps 1735: /*
1736: * If we have a Unicode codepoint, try to convert that
1737: * to a UTF-8 byte string.
1738: */
1739: cpp = utfbuf;
1740: if (0 == (sz = utf8(u, utfbuf)))
1741: continue;
1.1 kristaps 1742:
1.50 kristaps 1743: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1744:
1.50 kristaps 1745: sz = strlen(cpp);
1746: bsz += sz;
1.1 kristaps 1747:
1.50 kristaps 1748: buf = mandoc_realloc(buf, bsz);
1.1 kristaps 1749:
1.50 kristaps 1750: memcpy(&buf[pos], cpp, sz);
1751: pos += sz;
1.1 kristaps 1752: }
1753:
1.50 kristaps 1754: buf[pos] = '\0';
1755: key->utf8 = buf;
1.1 kristaps 1756: }
1757:
1.14 schwarze 1758: /*
1.50 kristaps 1759: * Flush the current page's terms (and their bits) into the database.
1760: * Wrap the entire set of additions in a transaction to make sqlite be a
1761: * little faster.
1762: * Also, UTF-8-encode the description at the last possible moment.
1.14 schwarze 1763: */
1764: static void
1.82 schwarze 1765: dbindex(const struct mpage *mpage, struct mchars *mc)
1.14 schwarze 1766: {
1.87 schwarze 1767: struct mlink *mlink;
1.50 kristaps 1768: struct str *key;
1769: const char *desc;
1770: int64_t recno;
1.52 kristaps 1771: size_t i;
1.66 schwarze 1772: unsigned int slot;
1.14 schwarze 1773:
1.56 schwarze 1774: if (verb)
1.82 schwarze 1775: say(mpage->mlinks->file, "Adding to index");
1.43 kristaps 1776:
1.50 kristaps 1777: if (nodb)
1.14 schwarze 1778: return;
1.28 kristaps 1779:
1.50 kristaps 1780: desc = "";
1.78 schwarze 1781: if (NULL != mpage->desc && '\0' != *mpage->desc) {
1.65 schwarze 1782: key = ohash_find(&strings,
1.78 schwarze 1783: ohash_qlookup(&strings, mpage->desc));
1.50 kristaps 1784: assert(NULL != key);
1785: if (NULL == key->utf8)
1786: utf8key(mc, key);
1787: desc = key->utf8;
1788: }
1789:
1.52 kristaps 1790: SQL_EXEC("BEGIN TRANSACTION");
1.50 kristaps 1791:
1.52 kristaps 1792: i = 1;
1.81 schwarze 1793: /*
1794: * XXX The following three lines are obsolete
1795: * and only kept for backward compatibility
1796: * until apropos(1) and friends have caught up.
1797: */
1.82 schwarze 1798: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->mlinks->file);
1799: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->mlinks->dsec);
1800: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->mlinks->arch);
1.81 schwarze 1801: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, desc);
1.82 schwarze 1802: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
1.81 schwarze 1803: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.50 kristaps 1804: recno = sqlite3_last_insert_rowid(db);
1.81 schwarze 1805: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1806:
1.87 schwarze 1807: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1808: i = 1;
1.88 schwarze 1809: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->file);
1.87 schwarze 1810: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1811: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1.88 schwarze 1812: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.87 schwarze 1813: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, recno);
1814: SQL_STEP(stmts[STMT_INSERT_LINK]);
1815: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1816: }
1.50 kristaps 1817:
1.66 schwarze 1818: for (key = ohash_first(&strings, &slot); NULL != key;
1819: key = ohash_next(&strings, &slot)) {
1.78 schwarze 1820: assert(key->mpage == mpage);
1.50 kristaps 1821: if (NULL == key->utf8)
1822: utf8key(mc, key);
1.52 kristaps 1823: i = 1;
1824: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1825: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->utf8);
1826: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno);
1827: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 1828: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 1829: if (key->utf8 != key->key)
1830: free(key->utf8);
1831: free(key);
1.38 schwarze 1832: }
1.28 kristaps 1833:
1.52 kristaps 1834: SQL_EXEC("END TRANSACTION");
1.14 schwarze 1835: }
1836:
1.5 kristaps 1837: static void
1.59 schwarze 1838: dbprune(void)
1.5 kristaps 1839: {
1.78 schwarze 1840: struct mpage *mpage;
1.82 schwarze 1841: struct mlink *mlink;
1.52 kristaps 1842: size_t i;
1.80 schwarze 1843: unsigned int slot;
1.5 kristaps 1844:
1.50 kristaps 1845: if (nodb)
1846: return;
1.12 schwarze 1847:
1.80 schwarze 1848: mpage = ohash_first(&mpages, &slot);
1849: while (NULL != mpage) {
1.82 schwarze 1850: mlink = mpage->mlinks;
1.52 kristaps 1851: i = 1;
1.82 schwarze 1852: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], i, mlink->file);
1.81 schwarze 1853: SQL_STEP(stmts[STMT_DELETE_PAGE]);
1854: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
1.56 schwarze 1855: if (verb)
1.82 schwarze 1856: say(mlink->file, "Deleted from index");
1.80 schwarze 1857: mpage = ohash_next(&mpages, &slot);
1.5 kristaps 1858: }
1859: }
1860:
1.4 kristaps 1861: /*
1.50 kristaps 1862: * Close an existing database and its prepared statements.
1863: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 1864: */
1.35 kristaps 1865: static void
1.59 schwarze 1866: dbclose(int real)
1.4 kristaps 1867: {
1.50 kristaps 1868: size_t i;
1.4 kristaps 1869:
1.50 kristaps 1870: if (nodb)
1.38 schwarze 1871: return;
1.50 kristaps 1872:
1873: for (i = 0; i < STMT__MAX; i++) {
1874: sqlite3_finalize(stmts[i]);
1875: stmts[i] = NULL;
1.4 kristaps 1876: }
1877:
1.50 kristaps 1878: sqlite3_close(db);
1879: db = NULL;
1.12 schwarze 1880:
1.50 kristaps 1881: if (real)
1882: return;
1.12 schwarze 1883:
1.63 schwarze 1884: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
1.59 schwarze 1885: exitcode = (int)MANDOCLEVEL_SYSERR;
1886: say(MANDOC_DB, NULL);
1887: }
1.50 kristaps 1888: }
1.14 schwarze 1889:
1.50 kristaps 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
1.59 schwarze 1899: dbopen(int real)
1.50 kristaps 1900: {
1.63 schwarze 1901: const char *file, *sql;
1.50 kristaps 1902: int rc, ofl;
1.12 schwarze 1903:
1.50 kristaps 1904: if (nodb)
1905: return(1);
1.12 schwarze 1906:
1.63 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);
1914: }
1915: ofl |= SQLITE_OPEN_EXCLUSIVE;
1916: } else
1917: file = MANDOC_DB;
1.45 kristaps 1918:
1.50 kristaps 1919: rc = sqlite3_open_v2(file, &db, ofl, NULL);
1920: if (SQLITE_OK == rc)
1.57 schwarze 1921: goto prepare_statements;
1.50 kristaps 1922: if (SQLITE_CANTOPEN != rc) {
1.59 schwarze 1923: exitcode = (int)MANDOCLEVEL_SYSERR;
1924: say(file, NULL);
1.50 kristaps 1925: return(0);
1926: }
1.12 schwarze 1927:
1.50 kristaps 1928: sqlite3_close(db);
1929: db = NULL;
1.12 schwarze 1930:
1.50 kristaps 1931: if (SQLITE_OK != (rc = sqlite3_open(file, &db))) {
1.59 schwarze 1932: exitcode = (int)MANDOCLEVEL_SYSERR;
1933: say(file, NULL);
1.50 kristaps 1934: return(0);
1935: }
1.12 schwarze 1936:
1.81 schwarze 1937: /*
1938: * XXX The first three columns in table mpages are obsolete
1939: * and only kept for backward compatibility
1940: * until apropos(1) and friends have caught up.
1941: */
1942: sql = "CREATE TABLE \"mpages\" (\n"
1.50 kristaps 1943: " \"file\" TEXT NOT NULL,\n"
1944: " \"sec\" TEXT NOT NULL,\n"
1945: " \"arch\" TEXT NOT NULL,\n"
1946: " \"desc\" TEXT NOT NULL,\n"
1947: " \"form\" INTEGER NOT NULL,\n"
1948: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1949: ");\n"
1950: "\n"
1.81 schwarze 1951: "CREATE TABLE \"mlinks\" (\n"
1.88 schwarze 1952: " \"file\" TEXT NOT NULL,\n"
1.81 schwarze 1953: " \"sec\" TEXT NOT NULL,\n"
1954: " \"arch\" TEXT NOT NULL,\n"
1955: " \"name\" TEXT NOT NULL,\n"
1956: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1957: "ON DELETE CASCADE,\n"
1958: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1959: ");\n"
1960: "\n"
1.50 kristaps 1961: "CREATE TABLE \"keys\" (\n"
1962: " \"bits\" INTEGER NOT NULL,\n"
1963: " \"key\" TEXT NOT NULL,\n"
1.81 schwarze 1964: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1965: "ON DELETE CASCADE,\n"
1.50 kristaps 1966: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1967: ");\n"
1968: "\n"
1969: "CREATE INDEX \"key_index\" ON keys (key);\n";
1.14 schwarze 1970:
1.50 kristaps 1971: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 1972: exitcode = (int)MANDOCLEVEL_SYSERR;
1973: say(file, "%s", sqlite3_errmsg(db));
1.50 kristaps 1974: return(0);
1975: }
1.4 kristaps 1976:
1.57 schwarze 1977: prepare_statements:
1978: SQL_EXEC("PRAGMA foreign_keys = ON");
1.81 schwarze 1979: sql = "DELETE FROM mpages where file=?";
1980: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
1981: sql = "INSERT INTO mpages "
1.50 kristaps 1982: "(file,sec,arch,desc,form) VALUES (?,?,?,?,?)";
1.81 schwarze 1983: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
1984: sql = "INSERT INTO mlinks "
1.88 schwarze 1985: "(file,sec,arch,name,pageid) VALUES (?,?,?,?,?)";
1.81 schwarze 1986: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.50 kristaps 1987: sql = "INSERT INTO keys "
1.81 schwarze 1988: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 1989: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 1990:
1991: #ifndef __APPLE__
1992: /*
1993: * When opening a new database, we can turn off
1994: * synchronous mode for much better performance.
1995: */
1996:
1997: if (real)
1998: SQL_EXEC("PRAGMA synchronous = OFF");
1999: #endif
2000:
1.50 kristaps 2001: return(1);
2002: }
1.5 kristaps 2003:
1.50 kristaps 2004: static void *
2005: hash_halloc(size_t sz, void *arg)
2006: {
1.12 schwarze 2007:
1.50 kristaps 2008: return(mandoc_calloc(sz, 1));
2009: }
1.12 schwarze 2010:
1.50 kristaps 2011: static void *
2012: hash_alloc(size_t sz, void *arg)
2013: {
1.5 kristaps 2014:
1.50 kristaps 2015: return(mandoc_malloc(sz));
2016: }
1.41 kristaps 2017:
1.50 kristaps 2018: static void
2019: hash_free(void *p, size_t sz, void *arg)
2020: {
1.4 kristaps 2021:
1.50 kristaps 2022: free(p);
1.4 kristaps 2023: }
2024:
1.50 kristaps 2025: static int
1.59 schwarze 2026: set_basedir(const char *targetdir)
1.4 kristaps 2027: {
1.59 schwarze 2028: static char startdir[PATH_MAX];
2029: static int fd;
1.4 kristaps 2030:
1.59 schwarze 2031: /*
2032: * Remember where we started by keeping a fd open to the origin
2033: * path component: throughout this utility, we chdir() a lot to
2034: * handle relative paths, and by doing this, we can return to
2035: * the starting point.
2036: */
2037: if ('\0' == *startdir) {
2038: if (NULL == getcwd(startdir, PATH_MAX)) {
2039: exitcode = (int)MANDOCLEVEL_SYSERR;
2040: if (NULL != targetdir)
2041: say(".", NULL);
2042: return(0);
2043: }
2044: if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
2045: exitcode = (int)MANDOCLEVEL_SYSERR;
2046: say(startdir, NULL);
2047: return(0);
2048: }
2049: if (NULL == targetdir)
2050: targetdir = startdir;
2051: } else {
2052: if (-1 == fd)
2053: return(0);
2054: if (-1 == fchdir(fd)) {
2055: close(fd);
2056: basedir[0] = '\0';
2057: exitcode = (int)MANDOCLEVEL_SYSERR;
2058: say(startdir, NULL);
2059: return(0);
2060: }
2061: if (NULL == targetdir) {
2062: close(fd);
2063: return(1);
2064: }
2065: }
2066: if (NULL == realpath(targetdir, basedir)) {
2067: basedir[0] = '\0';
2068: exitcode = (int)MANDOCLEVEL_BADARG;
2069: say(targetdir, NULL);
1.50 kristaps 2070: return(0);
1.59 schwarze 2071: } else if (-1 == chdir(basedir)) {
2072: exitcode = (int)MANDOCLEVEL_BADARG;
2073: say("", NULL);
1.50 kristaps 2074: return(0);
1.4 kristaps 2075: }
1.50 kristaps 2076: return(1);
1.56 schwarze 2077: }
2078:
2079: static void
1.59 schwarze 2080: say(const char *file, const char *format, ...)
1.56 schwarze 2081: {
2082: va_list ap;
2083:
1.59 schwarze 2084: if ('\0' != *basedir)
2085: fprintf(stderr, "%s", basedir);
2086: if ('\0' != *basedir && '\0' != *file)
2087: fputs("//", stderr);
1.56 schwarze 2088: if ('\0' != *file)
1.59 schwarze 2089: fprintf(stderr, "%s", file);
1.56 schwarze 2090: fputs(": ", stderr);
1.59 schwarze 2091:
2092: if (NULL == format) {
2093: perror(NULL);
2094: return;
2095: }
1.56 schwarze 2096:
2097: va_start(ap, format);
2098: vfprintf(stderr, format, ap);
2099: va_end(ap);
2100:
2101: fputc('\n', stderr);
1.1 kristaps 2102: }
CVSweb
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to mandocdb.c CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [cvsweb.bsd.lv] / mandoc