Annotation of mandoc/mandocdb.c, Revision 1.49.2.6
1.49.2.6! schwarze 1: /* $Id: mandocdb.c,v 1.49.2.5 2013/09/18 00:54:20 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
4: * Copyright (c) 2011, 2012 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:
1.14 schwarze 22: #include <sys/types.h>
1.1 kristaps 23:
24: #include <assert.h>
1.43 kristaps 25: #include <ctype.h>
1.4 kristaps 26: #include <dirent.h>
1.45 kristaps 27: #include <errno.h>
1.1 kristaps 28: #include <fcntl.h>
29: #include <getopt.h>
1.49.2.1 schwarze 30: #include <limits.h>
1.1 kristaps 31: #include <stdio.h>
32: #include <stdint.h>
33: #include <stdlib.h>
34: #include <string.h>
1.17 schwarze 35: #include <unistd.h>
1.1 kristaps 36:
1.49.2.6! schwarze 37: #if defined(__linux__) || defined(__sun)
1.21 kristaps 38: # include <endian.h>
1.1 kristaps 39: # include <db_185.h>
1.21 kristaps 40: #elif defined(__APPLE__)
41: # include <libkern/OSByteOrder.h>
42: # include <db.h>
1.1 kristaps 43: #else
44: # include <db.h>
45: #endif
46:
1.49.2.6! schwarze 47: #if defined(__sun)
! 48: #include <sys/stat.h>
! 49: #endif
! 50:
1.1 kristaps 51: #include "man.h"
52: #include "mdoc.h"
53: #include "mandoc.h"
1.8 schwarze 54: #include "mandocdb.h"
1.10 kristaps 55: #include "manpath.h"
1.1 kristaps 56:
57: #define MANDOC_BUFSZ BUFSIZ
58: #define MANDOC_SLOP 1024
59:
1.14 schwarze 60: #define MANDOC_SRC 0x1
61: #define MANDOC_FORM 0x2
62:
1.38 schwarze 63: /* Access to the mandoc database on disk. */
64:
65: struct mdb {
1.49.2.1 schwarze 66: char idxn[PATH_MAX]; /* index db filename */
67: char dbn[PATH_MAX]; /* keyword db filename */
1.38 schwarze 68: DB *idx; /* index recno database */
69: DB *db; /* keyword btree database */
70: };
71:
72: /* Stack of temporarily unused index records. */
73:
74: struct recs {
75: recno_t *stack; /* pointer to a malloc'ed array */
76: size_t size; /* number of allocated slots */
77: size_t cur; /* current number of empty records */
78: recno_t last; /* last record number in the index */
79: };
80:
1.5 kristaps 81: /* Tiny list for files. No need to bring in QUEUE. */
82:
1.3 kristaps 83: struct of {
1.5 kristaps 84: char *fname; /* heap-allocated */
1.12 schwarze 85: char *sec;
86: char *arch;
87: char *title;
1.14 schwarze 88: int src_form;
1.5 kristaps 89: struct of *next; /* NULL for last one */
90: struct of *first; /* first in list */
1.3 kristaps 91: };
92:
1.1 kristaps 93: /* Buffer for storing growable data. */
94:
95: struct buf {
96: char *cp;
1.5 kristaps 97: size_t len; /* current length */
98: size_t size; /* total buffer size */
1.1 kristaps 99: };
100:
101: /* Operation we're going to perform. */
102:
103: enum op {
1.38 schwarze 104: OP_DEFAULT = 0, /* new dbs from dir list or default config */
105: OP_CONFFILE, /* new databases from custom config file */
1.5 kristaps 106: OP_UPDATE, /* delete/add entries in existing database */
1.38 schwarze 107: OP_DELETE, /* delete entries from existing database */
108: OP_TEST /* change no databases, report potential problems */
1.1 kristaps 109: };
110:
111: #define MAN_ARGS DB *hash, \
112: struct buf *buf, \
113: struct buf *dbuf, \
114: const struct man_node *n
115: #define MDOC_ARGS DB *hash, \
116: struct buf *buf, \
117: struct buf *dbuf, \
118: const struct mdoc_node *n, \
119: const struct mdoc_meta *m
120:
121: static void buf_appendmdoc(struct buf *,
122: const struct mdoc_node *, int);
123: static void buf_append(struct buf *, const char *);
124: static void buf_appendb(struct buf *,
125: const void *, size_t);
126: static void dbt_put(DB *, const char *, DBT *, DBT *);
1.9 kristaps 127: static void hash_put(DB *, const struct buf *, uint64_t);
1.1 kristaps 128: static void hash_reset(DB **);
1.3 kristaps 129: static void index_merge(const struct of *, struct mparse *,
1.16 schwarze 130: struct buf *, struct buf *, DB *,
1.49.2.2 schwarze 131: struct mdb *, struct recs *);
1.38 schwarze 132: static void index_prune(const struct of *, struct mdb *,
1.49.2.2 schwarze 133: struct recs *);
134: static void ofile_argbuild(int, char *[], struct of **,
135: const char *);
1.35 kristaps 136: static void ofile_dirbuild(const char *, const char *,
1.49.2.2 schwarze 137: const char *, int, struct of **);
1.4 kristaps 138: static void ofile_free(struct of *);
1.49.2.2 schwarze 139: static void pformatted(DB *, struct buf *,
140: struct buf *, const struct of *);
1.1 kristaps 141: static int pman_node(MAN_ARGS);
142: static void pmdoc_node(MDOC_ARGS);
1.25 schwarze 143: static int pmdoc_head(MDOC_ARGS);
144: static int pmdoc_body(MDOC_ARGS);
145: static int pmdoc_Fd(MDOC_ARGS);
146: static int pmdoc_In(MDOC_ARGS);
147: static int pmdoc_Fn(MDOC_ARGS);
148: static int pmdoc_Nd(MDOC_ARGS);
149: static int pmdoc_Nm(MDOC_ARGS);
150: static int pmdoc_Sh(MDOC_ARGS);
151: static int pmdoc_St(MDOC_ARGS);
152: static int pmdoc_Xr(MDOC_ARGS);
1.1 kristaps 153:
1.25 schwarze 154: #define MDOCF_CHILD 0x01 /* Automatically index child nodes. */
1.1 kristaps 155:
1.25 schwarze 156: struct mdoc_handler {
157: int (*fp)(MDOC_ARGS); /* Optional handler. */
158: uint64_t mask; /* Set unless handler returns 0. */
159: int flags; /* For use by pmdoc_node. */
160: };
161:
162: static const struct mdoc_handler mdocs[MDOC_MAX] = {
163: { NULL, 0, 0 }, /* Ap */
164: { NULL, 0, 0 }, /* Dd */
165: { NULL, 0, 0 }, /* Dt */
166: { NULL, 0, 0 }, /* Os */
167: { pmdoc_Sh, TYPE_Sh, MDOCF_CHILD }, /* Sh */
168: { pmdoc_head, TYPE_Ss, MDOCF_CHILD }, /* Ss */
169: { NULL, 0, 0 }, /* Pp */
170: { NULL, 0, 0 }, /* D1 */
171: { NULL, 0, 0 }, /* Dl */
172: { NULL, 0, 0 }, /* Bd */
173: { NULL, 0, 0 }, /* Ed */
174: { NULL, 0, 0 }, /* Bl */
175: { NULL, 0, 0 }, /* El */
176: { NULL, 0, 0 }, /* It */
177: { NULL, 0, 0 }, /* Ad */
178: { NULL, TYPE_An, MDOCF_CHILD }, /* An */
179: { NULL, TYPE_Ar, MDOCF_CHILD }, /* Ar */
180: { NULL, TYPE_Cd, MDOCF_CHILD }, /* Cd */
181: { NULL, TYPE_Cm, MDOCF_CHILD }, /* Cm */
182: { NULL, TYPE_Dv, MDOCF_CHILD }, /* Dv */
183: { NULL, TYPE_Er, MDOCF_CHILD }, /* Er */
184: { NULL, TYPE_Ev, MDOCF_CHILD }, /* Ev */
185: { NULL, 0, 0 }, /* Ex */
186: { NULL, TYPE_Fa, MDOCF_CHILD }, /* Fa */
187: { pmdoc_Fd, TYPE_In, 0 }, /* Fd */
188: { NULL, TYPE_Fl, MDOCF_CHILD }, /* Fl */
189: { pmdoc_Fn, 0, 0 }, /* Fn */
190: { NULL, TYPE_Ft, MDOCF_CHILD }, /* Ft */
191: { NULL, TYPE_Ic, MDOCF_CHILD }, /* Ic */
192: { pmdoc_In, TYPE_In, 0 }, /* In */
193: { NULL, TYPE_Li, MDOCF_CHILD }, /* Li */
194: { pmdoc_Nd, TYPE_Nd, MDOCF_CHILD }, /* Nd */
195: { pmdoc_Nm, TYPE_Nm, MDOCF_CHILD }, /* Nm */
196: { NULL, 0, 0 }, /* Op */
197: { NULL, 0, 0 }, /* Ot */
198: { NULL, TYPE_Pa, MDOCF_CHILD }, /* Pa */
199: { NULL, 0, 0 }, /* Rv */
200: { pmdoc_St, TYPE_St, 0 }, /* St */
201: { NULL, TYPE_Va, MDOCF_CHILD }, /* Va */
202: { pmdoc_body, TYPE_Va, MDOCF_CHILD }, /* Vt */
203: { pmdoc_Xr, TYPE_Xr, 0 }, /* Xr */
204: { NULL, 0, 0 }, /* %A */
205: { NULL, 0, 0 }, /* %B */
206: { NULL, 0, 0 }, /* %D */
207: { NULL, 0, 0 }, /* %I */
208: { NULL, 0, 0 }, /* %J */
209: { NULL, 0, 0 }, /* %N */
210: { NULL, 0, 0 }, /* %O */
211: { NULL, 0, 0 }, /* %P */
212: { NULL, 0, 0 }, /* %R */
213: { NULL, 0, 0 }, /* %T */
214: { NULL, 0, 0 }, /* %V */
215: { NULL, 0, 0 }, /* Ac */
216: { NULL, 0, 0 }, /* Ao */
217: { NULL, 0, 0 }, /* Aq */
218: { NULL, TYPE_At, MDOCF_CHILD }, /* At */
219: { NULL, 0, 0 }, /* Bc */
220: { NULL, 0, 0 }, /* Bf */
221: { NULL, 0, 0 }, /* Bo */
222: { NULL, 0, 0 }, /* Bq */
223: { NULL, TYPE_Bsx, MDOCF_CHILD }, /* Bsx */
224: { NULL, TYPE_Bx, MDOCF_CHILD }, /* Bx */
225: { NULL, 0, 0 }, /* Db */
226: { NULL, 0, 0 }, /* Dc */
227: { NULL, 0, 0 }, /* Do */
228: { NULL, 0, 0 }, /* Dq */
229: { NULL, 0, 0 }, /* Ec */
230: { NULL, 0, 0 }, /* Ef */
231: { NULL, TYPE_Em, MDOCF_CHILD }, /* Em */
232: { NULL, 0, 0 }, /* Eo */
233: { NULL, TYPE_Fx, MDOCF_CHILD }, /* Fx */
234: { NULL, TYPE_Ms, MDOCF_CHILD }, /* Ms */
235: { NULL, 0, 0 }, /* No */
236: { NULL, 0, 0 }, /* Ns */
237: { NULL, TYPE_Nx, MDOCF_CHILD }, /* Nx */
238: { NULL, TYPE_Ox, MDOCF_CHILD }, /* Ox */
239: { NULL, 0, 0 }, /* Pc */
240: { NULL, 0, 0 }, /* Pf */
241: { NULL, 0, 0 }, /* Po */
242: { NULL, 0, 0 }, /* Pq */
243: { NULL, 0, 0 }, /* Qc */
244: { NULL, 0, 0 }, /* Ql */
245: { NULL, 0, 0 }, /* Qo */
246: { NULL, 0, 0 }, /* Qq */
247: { NULL, 0, 0 }, /* Re */
248: { NULL, 0, 0 }, /* Rs */
249: { NULL, 0, 0 }, /* Sc */
250: { NULL, 0, 0 }, /* So */
251: { NULL, 0, 0 }, /* Sq */
252: { NULL, 0, 0 }, /* Sm */
253: { NULL, 0, 0 }, /* Sx */
254: { NULL, TYPE_Sy, MDOCF_CHILD }, /* Sy */
255: { NULL, TYPE_Tn, MDOCF_CHILD }, /* Tn */
256: { NULL, 0, 0 }, /* Ux */
257: { NULL, 0, 0 }, /* Xc */
258: { NULL, 0, 0 }, /* Xo */
259: { pmdoc_head, TYPE_Fn, 0 }, /* Fo */
260: { NULL, 0, 0 }, /* Fc */
261: { NULL, 0, 0 }, /* Oo */
262: { NULL, 0, 0 }, /* Oc */
263: { NULL, 0, 0 }, /* Bk */
264: { NULL, 0, 0 }, /* Ek */
265: { NULL, 0, 0 }, /* Bt */
266: { NULL, 0, 0 }, /* Hf */
267: { NULL, 0, 0 }, /* Fr */
268: { NULL, 0, 0 }, /* Ud */
269: { NULL, TYPE_Lb, MDOCF_CHILD }, /* Lb */
270: { NULL, 0, 0 }, /* Lp */
271: { NULL, TYPE_Lk, MDOCF_CHILD }, /* Lk */
272: { NULL, TYPE_Mt, MDOCF_CHILD }, /* Mt */
273: { NULL, 0, 0 }, /* Brq */
274: { NULL, 0, 0 }, /* Bro */
275: { NULL, 0, 0 }, /* Brc */
276: { NULL, 0, 0 }, /* %C */
277: { NULL, 0, 0 }, /* Es */
278: { NULL, 0, 0 }, /* En */
279: { NULL, TYPE_Dx, MDOCF_CHILD }, /* Dx */
280: { NULL, 0, 0 }, /* %Q */
281: { NULL, 0, 0 }, /* br */
282: { NULL, 0, 0 }, /* sp */
283: { NULL, 0, 0 }, /* %U */
284: { NULL, 0, 0 }, /* Ta */
1.1 kristaps 285: };
286:
287: static const char *progname;
1.16 schwarze 288: static int use_all; /* Use all directories and files. */
289: static int verb; /* Output verbosity level. */
1.38 schwarze 290: static int warnings; /* Potential problems in manuals. */
1.1 kristaps 291:
292: int
293: main(int argc, char *argv[])
294: {
295: struct mparse *mp; /* parse sequence */
1.10 kristaps 296: struct manpaths dirs;
1.38 schwarze 297: struct mdb mdb;
298: struct recs recs;
1.1 kristaps 299: enum op op; /* current operation */
1.5 kristaps 300: const char *dir;
1.49.2.2 schwarze 301: char *cp;
302: char pbuf[PATH_MAX];
1.16 schwarze 303: int ch, i, flags;
1.38 schwarze 304: DB *hash; /* temporary keyword hashtable */
1.1 kristaps 305: BTREEINFO info; /* btree configuration */
1.49.2.4 schwarze 306: size_t sz1, sz2, ipath;
1.1 kristaps 307: struct buf buf, /* keyword buffer */
308: dbuf; /* description buffer */
1.5 kristaps 309: struct of *of; /* list of files for processing */
1.1 kristaps 310: extern int optind;
311: extern char *optarg;
312:
313: progname = strrchr(argv[0], '/');
314: if (progname == NULL)
315: progname = argv[0];
316: else
317: ++progname;
318:
1.10 kristaps 319: memset(&dirs, 0, sizeof(struct manpaths));
1.38 schwarze 320: memset(&mdb, 0, sizeof(struct mdb));
321: memset(&recs, 0, sizeof(struct recs));
1.10 kristaps 322:
1.4 kristaps 323: of = NULL;
1.1 kristaps 324: mp = NULL;
325: hash = NULL;
1.38 schwarze 326: op = OP_DEFAULT;
1.5 kristaps 327: dir = NULL;
1.1 kristaps 328:
1.38 schwarze 329: while (-1 != (ch = getopt(argc, argv, "aC:d:tu:vW")))
1.1 kristaps 330: switch (ch) {
1.12 schwarze 331: case ('a'):
332: use_all = 1;
333: break;
1.34 schwarze 334: case ('C'):
1.38 schwarze 335: if (op) {
336: fprintf(stderr,
337: "-C: conflicting options\n");
338: goto usage;
339: }
340: dir = optarg;
341: op = OP_CONFFILE;
1.34 schwarze 342: break;
1.5 kristaps 343: case ('d'):
1.38 schwarze 344: if (op) {
345: fprintf(stderr,
346: "-d: conflicting options\n");
347: goto usage;
348: }
1.5 kristaps 349: dir = optarg;
350: op = OP_UPDATE;
351: break;
1.38 schwarze 352: case ('t'):
353: dup2(STDOUT_FILENO, STDERR_FILENO);
354: if (op) {
355: fprintf(stderr,
356: "-t: conflicting options\n");
357: goto usage;
358: }
359: op = OP_TEST;
360: use_all = 1;
361: warnings = 1;
362: break;
1.5 kristaps 363: case ('u'):
1.38 schwarze 364: if (op) {
365: fprintf(stderr,
366: "-u: conflicting options\n");
367: goto usage;
368: }
1.5 kristaps 369: dir = optarg;
370: op = OP_DELETE;
371: break;
372: case ('v'):
373: verb++;
374: break;
1.38 schwarze 375: case ('W'):
376: warnings = 1;
377: break;
1.1 kristaps 378: default:
1.38 schwarze 379: goto usage;
1.1 kristaps 380: }
381:
382: argc -= optind;
383: argv += optind;
384:
1.38 schwarze 385: if (OP_CONFFILE == op && argc > 0) {
386: fprintf(stderr, "-C: too many arguments\n");
387: goto usage;
388: }
389:
1.4 kristaps 390: memset(&info, 0, sizeof(BTREEINFO));
1.44 kristaps 391: info.lorder = 4321;
1.4 kristaps 392: info.flags = R_DUP;
1.1 kristaps 393:
1.49 schwarze 394: mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL, NULL);
1.1 kristaps 395:
1.5 kristaps 396: memset(&buf, 0, sizeof(struct buf));
397: memset(&dbuf, 0, sizeof(struct buf));
1.1 kristaps 398:
1.4 kristaps 399: buf.size = dbuf.size = MANDOC_BUFSZ;
1.1 kristaps 400:
1.4 kristaps 401: buf.cp = mandoc_malloc(buf.size);
402: dbuf.cp = mandoc_malloc(dbuf.size);
1.1 kristaps 403:
1.38 schwarze 404: if (OP_TEST == op) {
1.49.2.2 schwarze 405: ofile_argbuild(argc, argv, &of, NULL);
1.38 schwarze 406: if (NULL == of)
407: goto out;
1.49.2.2 schwarze 408: index_merge(of, mp, &dbuf, &buf, hash, &mdb, &recs);
1.38 schwarze 409: goto out;
410: }
1.5 kristaps 411:
412: if (OP_UPDATE == op || OP_DELETE == op) {
1.49.2.2 schwarze 413: if (NULL == realpath(dir, pbuf)) {
414: perror(dir);
415: exit((int)MANDOCLEVEL_BADARG);
416: }
417: if (strlcat(pbuf, "/", PATH_MAX) >= PATH_MAX) {
418: fprintf(stderr, "%s: path too long\n", pbuf);
419: exit((int)MANDOCLEVEL_BADARG);
420: }
421:
422: strlcat(mdb.dbn, pbuf, PATH_MAX);
1.49.2.1 schwarze 423: sz1 = strlcat(mdb.dbn, MANDOC_DB, PATH_MAX);
424:
1.49.2.2 schwarze 425: strlcat(mdb.idxn, pbuf, PATH_MAX);
1.49.2.1 schwarze 426: sz2 = strlcat(mdb.idxn, MANDOC_IDX, PATH_MAX);
1.5 kristaps 427:
1.49.2.1 schwarze 428: if (sz1 >= PATH_MAX || sz2 >= PATH_MAX) {
1.49.2.2 schwarze 429: fprintf(stderr, "%s: path too long\n", mdb.idxn);
1.5 kristaps 430: exit((int)MANDOCLEVEL_BADARG);
431: }
432:
1.44 kristaps 433: flags = O_CREAT | O_RDWR;
1.38 schwarze 434: mdb.db = dbopen(mdb.dbn, flags, 0644, DB_BTREE, &info);
435: mdb.idx = dbopen(mdb.idxn, flags, 0644, DB_RECNO, NULL);
1.5 kristaps 436:
1.38 schwarze 437: if (NULL == mdb.db) {
438: perror(mdb.dbn);
1.5 kristaps 439: exit((int)MANDOCLEVEL_SYSERR);
1.38 schwarze 440: } else if (NULL == mdb.idx) {
441: perror(mdb.idxn);
1.5 kristaps 442: exit((int)MANDOCLEVEL_SYSERR);
443: }
444:
1.49.2.2 schwarze 445: ofile_argbuild(argc, argv, &of, pbuf);
1.5 kristaps 446:
447: if (NULL == of)
448: goto out;
449:
1.49.2.2 schwarze 450: index_prune(of, &mdb, &recs);
1.5 kristaps 451:
1.17 schwarze 452: /*
1.35 kristaps 453: * Go to the root of the respective manual tree.
454: * This must work or no manuals may be found (they're
455: * indexed relative to the root).
1.17 schwarze 456: */
457:
458: if (OP_UPDATE == op) {
1.35 kristaps 459: if (-1 == chdir(dir)) {
460: perror(dir);
461: exit((int)MANDOCLEVEL_SYSERR);
462: }
1.13 schwarze 463: index_merge(of, mp, &dbuf, &buf, hash,
1.49.2.2 schwarze 464: &mdb, &recs);
1.17 schwarze 465: }
1.5 kristaps 466:
467: goto out;
468: }
469:
1.10 kristaps 470: /*
471: * Configure the directories we're going to scan.
472: * If we have command-line arguments, use them.
473: * If not, we use man(1)'s method (see mandocdb.8).
474: */
475:
476: if (argc > 0) {
1.26 kristaps 477: dirs.paths = mandoc_calloc(argc, sizeof(char *));
1.10 kristaps 478: dirs.sz = argc;
1.49.2.2 schwarze 479: for (i = 0; i < argc; i++) {
480: if (NULL == (cp = realpath(argv[i], pbuf))) {
481: perror(argv[i]);
482: goto out;
483: }
484: dirs.paths[i] = mandoc_strdup(cp);
485: }
1.10 kristaps 486: } else
1.38 schwarze 487: manpath_parse(&dirs, dir, NULL, NULL);
1.10 kristaps 488:
1.49.2.4 schwarze 489: for (ipath = 0; ipath < dirs.sz; ipath++) {
1.49.2.2 schwarze 490:
1.44 kristaps 491: /*
492: * Go to the root of the respective manual tree.
493: * This must work or no manuals may be found:
494: * They are indexed relative to the root.
495: */
1.1 kristaps 496:
1.49.2.4 schwarze 497: if (-1 == chdir(dirs.paths[ipath])) {
498: perror(dirs.paths[ipath]);
1.44 kristaps 499: exit((int)MANDOCLEVEL_SYSERR);
1.4 kristaps 500: }
1.3 kristaps 501:
1.49.2.3 schwarze 502: /* Create a new database in two temporary files. */
1.13 schwarze 503:
1.49.2.3 schwarze 504: flags = O_CREAT | O_EXCL | O_RDWR;
505: while (NULL == mdb.db) {
506: strlcpy(mdb.dbn, MANDOC_DB, PATH_MAX);
507: strlcat(mdb.dbn, ".XXXXXXXXXX", PATH_MAX);
508: if (NULL == mktemp(mdb.dbn)) {
509: perror(mdb.dbn);
510: exit((int)MANDOCLEVEL_SYSERR);
511: }
512: mdb.db = dbopen(mdb.dbn, flags, 0644,
513: DB_BTREE, &info);
514: if (NULL == mdb.db && EEXIST != errno) {
515: perror(mdb.dbn);
516: exit((int)MANDOCLEVEL_SYSERR);
517: }
518: }
519: while (NULL == mdb.idx) {
520: strlcpy(mdb.idxn, MANDOC_IDX, PATH_MAX);
521: strlcat(mdb.idxn, ".XXXXXXXXXX", PATH_MAX);
522: if (NULL == mktemp(mdb.idxn)) {
523: perror(mdb.idxn);
524: unlink(mdb.dbn);
525: exit((int)MANDOCLEVEL_SYSERR);
526: }
527: mdb.idx = dbopen(mdb.idxn, flags, 0644,
528: DB_RECNO, NULL);
529: if (NULL == mdb.idx && EEXIST != errno) {
530: perror(mdb.idxn);
531: unlink(mdb.dbn);
532: exit((int)MANDOCLEVEL_SYSERR);
533: }
1.5 kristaps 534: }
535:
1.44 kristaps 536: /*
537: * Search for manuals and fill the new database.
538: */
1.1 kristaps 539:
1.49.2.2 schwarze 540: ofile_dirbuild(".", "", "", 0, &of);
1.1 kristaps 541:
1.44 kristaps 542: if (NULL != of) {
543: index_merge(of, mp, &dbuf, &buf, hash,
1.49.2.2 schwarze 544: &mdb, &recs);
1.44 kristaps 545: ofile_free(of);
546: of = NULL;
1.35 kristaps 547: }
548:
1.44 kristaps 549: (*mdb.db->close)(mdb.db);
550: (*mdb.idx->close)(mdb.idx);
551: mdb.db = NULL;
552: mdb.idx = NULL;
1.49.2.3 schwarze 553:
554: /*
555: * Replace the old database with the new one.
556: * This is not perfectly atomic,
557: * but i cannot think of a better way.
558: */
559:
560: if (-1 == rename(mdb.dbn, MANDOC_DB)) {
561: perror(MANDOC_DB);
562: unlink(mdb.dbn);
563: unlink(mdb.idxn);
564: exit((int)MANDOCLEVEL_SYSERR);
565: }
566: if (-1 == rename(mdb.idxn, MANDOC_IDX)) {
567: perror(MANDOC_IDX);
568: unlink(MANDOC_DB);
569: unlink(MANDOC_IDX);
570: unlink(mdb.idxn);
571: exit((int)MANDOCLEVEL_SYSERR);
572: }
1.4 kristaps 573: }
1.3 kristaps 574:
1.5 kristaps 575: out:
1.38 schwarze 576: if (mdb.db)
577: (*mdb.db->close)(mdb.db);
578: if (mdb.idx)
579: (*mdb.idx->close)(mdb.idx);
1.3 kristaps 580: if (hash)
581: (*hash->close)(hash);
582: if (mp)
583: mparse_free(mp);
584:
1.10 kristaps 585: manpath_free(&dirs);
1.4 kristaps 586: ofile_free(of);
1.3 kristaps 587: free(buf.cp);
588: free(dbuf.cp);
1.38 schwarze 589: free(recs.stack);
1.3 kristaps 590:
1.5 kristaps 591: return(MANDOCLEVEL_OK);
1.38 schwarze 592:
593: usage:
594: fprintf(stderr,
1.49.2.2 schwarze 595: "usage: %s [-avvv] [-C file] | dir ... | -t file ...\n"
1.38 schwarze 596: " -d dir [file ...] | "
597: "-u dir [file ...]\n",
598: progname);
599:
600: return((int)MANDOCLEVEL_BADARG);
1.3 kristaps 601: }
602:
603: void
604: index_merge(const struct of *of, struct mparse *mp,
1.16 schwarze 605: struct buf *dbuf, struct buf *buf, DB *hash,
1.49.2.2 schwarze 606: struct mdb *mdb, struct recs *recs)
1.3 kristaps 607: {
608: recno_t rec;
1.38 schwarze 609: int ch, skip;
1.3 kristaps 610: DBT key, val;
1.44 kristaps 611: DB *files; /* temporary file name table */
1.3 kristaps 612: struct mdoc *mdoc;
613: struct man *man;
1.38 schwarze 614: const char *fn, *msec, *march, *mtitle;
1.49.2.2 schwarze 615: char *p;
1.37 schwarze 616: uint64_t mask;
1.3 kristaps 617: size_t sv;
618: unsigned seq;
1.39 schwarze 619: uint64_t vbuf[2];
1.36 kristaps 620: char type;
1.3 kristaps 621:
1.44 kristaps 622: if (warnings) {
623: files = NULL;
624: hash_reset(&files);
625: }
626:
1.38 schwarze 627: rec = 0;
628: for (of = of->first; of; of = of->next) {
1.3 kristaps 629: fn = of->fname;
1.14 schwarze 630:
631: /*
1.33 schwarze 632: * Try interpreting the file as mdoc(7) or man(7)
633: * source code, unless it is already known to be
634: * formatted. Fall back to formatted mode.
1.14 schwarze 635: */
636:
1.1 kristaps 637: mparse_reset(mp);
1.14 schwarze 638: mdoc = NULL;
639: man = NULL;
1.1 kristaps 640:
1.14 schwarze 641: if ((MANDOC_SRC & of->src_form ||
642: ! (MANDOC_FORM & of->src_form)) &&
643: MANDOCLEVEL_FATAL > mparse_readfd(mp, -1, fn))
644: mparse_result(mp, &mdoc, &man);
645:
646: if (NULL != mdoc) {
647: msec = mdoc_meta(mdoc)->msec;
1.38 schwarze 648: march = mdoc_meta(mdoc)->arch;
649: if (NULL == march)
650: march = "";
1.14 schwarze 651: mtitle = mdoc_meta(mdoc)->title;
652: } else if (NULL != man) {
653: msec = man_meta(man)->msec;
1.38 schwarze 654: march = "";
1.14 schwarze 655: mtitle = man_meta(man)->title;
656: } else {
657: msec = of->sec;
1.38 schwarze 658: march = of->arch;
1.14 schwarze 659: mtitle = of->title;
1.1 kristaps 660: }
661:
1.12 schwarze 662: /*
1.44 kristaps 663: * Check whether the manual section given in a file
664: * agrees with the directory where the file is located.
665: * Some manuals have suffixes like (3p) on their
666: * section number either inside the file or in the
667: * directory name, some are linked into more than one
668: * section, like encrypt(1) = makekey(8). Do not skip
669: * manuals for such reasons.
1.12 schwarze 670: */
671:
1.38 schwarze 672: skip = 0;
673: assert(of->sec);
674: assert(msec);
1.49.2.2 schwarze 675: if (warnings)
676: if (strcasecmp(msec, of->sec))
677: fprintf(stderr, "%s: "
678: "section \"%s\" manual "
679: "in \"%s\" directory\n",
680: fn, msec, of->sec);
681:
1.42 schwarze 682: /*
683: * Manual page directories exist for each kernel
684: * architecture as returned by machine(1).
685: * However, many manuals only depend on the
686: * application architecture as returned by arch(1).
687: * For example, some (2/ARM) manuals are shared
688: * across the "armish" and "zaurus" kernel
689: * architectures.
690: * A few manuals are even shared across completely
691: * different architectures, for example fdformat(1)
692: * on amd64, i386, sparc, and sparc64.
693: * Thus, warn about architecture mismatches,
694: * but don't skip manuals for this reason.
695: */
696:
1.38 schwarze 697: assert(of->arch);
698: assert(march);
1.49.2.2 schwarze 699: if (warnings)
700: if (strcasecmp(march, of->arch))
701: fprintf(stderr, "%s: "
702: "architecture \"%s\" manual "
703: "in \"%s\" directory\n",
704: fn, march, of->arch);
1.12 schwarze 705:
1.38 schwarze 706: /*
1.12 schwarze 707: * By default, skip a file if the title given
708: * in the file disagrees with the file name.
1.44 kristaps 709: * Do not warn, this happens for all MLINKs.
1.12 schwarze 710: */
711:
712: assert(of->title);
713: assert(mtitle);
1.44 kristaps 714: if (strcasecmp(mtitle, of->title))
1.38 schwarze 715: skip = 1;
1.44 kristaps 716:
717: /*
718: * Build a title string for the file. If it matches
719: * the location of the file, remember the title as
720: * found; else, remember it as missing.
721: */
722:
723: if (warnings) {
724: buf->len = 0;
725: buf_appendb(buf, mtitle, strlen(mtitle));
726: buf_appendb(buf, "(", 1);
727: buf_appendb(buf, msec, strlen(msec));
728: if ('\0' != *march) {
729: buf_appendb(buf, "/", 1);
730: buf_appendb(buf, march, strlen(march));
731: }
732: buf_appendb(buf, ")", 2);
733: for (p = buf->cp; '\0' != *p; p++)
734: *p = tolower(*p);
735: key.data = buf->cp;
736: key.size = buf->len;
737: val.data = NULL;
738: val.size = 0;
739: if (0 == skip)
1.49.2.2 schwarze 740: val.data = "";
1.44 kristaps 741: else {
742: ch = (*files->get)(files, &key, &val, 0);
743: if (ch < 0) {
744: perror("hash");
745: exit((int)MANDOCLEVEL_SYSERR);
746: } else if (ch > 0) {
747: val.data = (void *)fn;
748: val.size = strlen(fn) + 1;
749: } else
750: val.data = NULL;
751: }
752: if (NULL != val.data &&
753: (*files->put)(files, &key, &val, 0) < 0) {
754: perror("hash");
755: exit((int)MANDOCLEVEL_SYSERR);
756: }
757: }
1.12 schwarze 758:
1.38 schwarze 759: if (skip && !use_all)
1.12 schwarze 760: continue;
761:
1.38 schwarze 762: /*
1.1 kristaps 763: * The index record value consists of a nil-terminated
764: * filename, a nil-terminated manual section, and a
1.44 kristaps 765: * nil-terminated description. Use the actual
766: * location of the file, such that the user can find
767: * it with man(1). Since the description may not be
768: * set, we set a sentinel to see if we're going to
769: * write a nil byte in its place.
1.1 kristaps 770: */
771:
1.3 kristaps 772: dbuf->len = 0;
1.36 kristaps 773: type = mdoc ? 'd' : (man ? 'a' : 'c');
774: buf_appendb(dbuf, &type, 1);
1.3 kristaps 775: buf_appendb(dbuf, fn, strlen(fn) + 1);
1.44 kristaps 776: buf_appendb(dbuf, of->sec, strlen(of->sec) + 1);
777: buf_appendb(dbuf, of->title, strlen(of->title) + 1);
778: buf_appendb(dbuf, of->arch, strlen(of->arch) + 1);
1.1 kristaps 779:
1.3 kristaps 780: sv = dbuf->len;
1.1 kristaps 781:
1.33 schwarze 782: /*
783: * Collect keyword/mask pairs.
784: * Each pair will become a new btree node.
785: */
1.1 kristaps 786:
1.33 schwarze 787: hash_reset(&hash);
1.1 kristaps 788: if (mdoc)
1.3 kristaps 789: pmdoc_node(hash, buf, dbuf,
1.1 kristaps 790: mdoc_node(mdoc), mdoc_meta(mdoc));
1.14 schwarze 791: else if (man)
1.3 kristaps 792: pman_node(hash, buf, dbuf, man_node(man));
1.14 schwarze 793: else
1.49.2.2 schwarze 794: pformatted(hash, buf, dbuf, of);
1.1 kristaps 795:
1.38 schwarze 796: /* Test mode, do not access any database. */
797:
798: if (NULL == mdb->db || NULL == mdb->idx)
799: continue;
800:
1.1 kristaps 801: /*
1.44 kristaps 802: * Make sure the file name is always registered
803: * as an .Nm search key.
804: */
805: buf->len = 0;
806: buf_append(buf, of->title);
807: hash_put(hash, buf, TYPE_Nm);
808:
809: /*
1.33 schwarze 810: * Reclaim an empty index record, if available.
811: * Use its record number for all new btree nodes.
1.1 kristaps 812: */
813:
1.38 schwarze 814: if (recs->cur > 0) {
815: recs->cur--;
816: rec = recs->stack[(int)recs->cur];
817: } else if (recs->last > 0) {
818: rec = recs->last;
819: recs->last = 0;
1.33 schwarze 820: } else
821: rec++;
1.39 schwarze 822: vbuf[1] = htobe64(rec);
1.33 schwarze 823:
824: /*
825: * Copy from the in-memory hashtable of pending
826: * keyword/mask pairs into the database.
827: */
828:
1.1 kristaps 829: seq = R_FIRST;
830: while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) {
831: seq = R_NEXT;
1.37 schwarze 832: assert(sizeof(uint64_t) == val.size);
833: memcpy(&mask, val.data, val.size);
1.39 schwarze 834: vbuf[0] = htobe64(mask);
835: val.size = sizeof(vbuf);
1.9 kristaps 836: val.data = &vbuf;
1.38 schwarze 837: dbt_put(mdb->db, mdb->dbn, &key, &val);
1.1 kristaps 838: }
839: if (ch < 0) {
840: perror("hash");
1.49.2.2 schwarze 841: unlink(mdb->dbn);
842: unlink(mdb->idxn);
1.1 kristaps 843: exit((int)MANDOCLEVEL_SYSERR);
844: }
1.38 schwarze 845:
1.1 kristaps 846: /*
847: * Apply to the index. If we haven't had a description
848: * set, put an empty one in now.
849: */
850:
1.3 kristaps 851: if (dbuf->len == sv)
852: buf_appendb(dbuf, "", 1);
1.1 kristaps 853:
854: key.data = &rec;
855: key.size = sizeof(recno_t);
856:
1.3 kristaps 857: val.data = dbuf->cp;
858: val.size = dbuf->len;
1.1 kristaps 859:
1.5 kristaps 860: if (verb)
1.49.2.2 schwarze 861: printf("%s: adding to index\n", fn);
1.18 kristaps 862:
1.38 schwarze 863: dbt_put(mdb->idx, mdb->idxn, &key, &val);
1.3 kristaps 864: }
1.44 kristaps 865:
866: /*
867: * Iterate the remembered file titles and check that
868: * all files can be found by their main title.
869: */
870:
871: if (warnings) {
872: seq = R_FIRST;
873: while (0 == (*files->seq)(files, &key, &val, seq)) {
874: seq = R_NEXT;
875: if (val.size)
1.49.2.2 schwarze 876: fprintf(stderr, "%s: probably "
877: "unreachable, title is %s\n",
878: (char *)val.data, (char *)key.data);
1.44 kristaps 879: }
880: (*files->close)(files);
881: }
1.3 kristaps 882: }
883:
884: /*
885: * Scan through all entries in the index file `idx' and prune those
886: * entries in `ofile'.
887: * Pruning consists of removing from `db', then invalidating the entry
888: * in `idx' (zeroing its value size).
889: */
890: static void
1.49.2.2 schwarze 891: index_prune(const struct of *ofile, struct mdb *mdb, struct recs *recs)
1.3 kristaps 892: {
893: const struct of *of;
1.36 kristaps 894: const char *fn;
1.39 schwarze 895: uint64_t vbuf[2];
1.3 kristaps 896: unsigned seq, sseq;
897: DBT key, val;
898: int ch;
899:
1.38 schwarze 900: recs->cur = 0;
1.3 kristaps 901: seq = R_FIRST;
1.38 schwarze 902: while (0 == (ch = (*mdb->idx->seq)(mdb->idx, &key, &val, seq))) {
1.3 kristaps 903: seq = R_NEXT;
1.37 schwarze 904: assert(sizeof(recno_t) == key.size);
1.38 schwarze 905: memcpy(&recs->last, key.data, key.size);
1.18 kristaps 906:
907: /* Deleted records are zero-sized. Skip them. */
908:
909: if (0 == val.size)
910: goto cont;
911:
912: /*
913: * Make sure we're sane.
914: * Read past our mdoc/man/cat type to the next string,
915: * then make sure it's bounded by a NUL.
916: * Failing any of these, we go into our error handler.
917: */
918:
1.36 kristaps 919: fn = (char *)val.data + 1;
920: if (NULL == memchr(fn, '\0', val.size - 1))
1.18 kristaps 921: break;
922:
1.38 schwarze 923: /*
1.18 kristaps 924: * Search for the file in those we care about.
925: * XXX: build this into a tree. Too slow.
926: */
1.3 kristaps 927:
1.38 schwarze 928: for (of = ofile->first; of; of = of->next)
1.3 kristaps 929: if (0 == strcmp(fn, of->fname))
930: break;
931:
932: if (NULL == of)
933: continue;
934:
1.18 kristaps 935: /*
936: * Search through the keyword database, throwing out all
937: * references to our file.
938: */
939:
1.3 kristaps 940: sseq = R_FIRST;
1.38 schwarze 941: while (0 == (ch = (*mdb->db->seq)(mdb->db,
942: &key, &val, sseq))) {
1.3 kristaps 943: sseq = R_NEXT;
1.39 schwarze 944: if (sizeof(vbuf) != val.size)
1.18 kristaps 945: break;
946:
1.39 schwarze 947: memcpy(vbuf, val.data, val.size);
948: if (recs->last != betoh64(vbuf[1]))
1.3 kristaps 949: continue;
1.18 kristaps 950:
1.38 schwarze 951: if ((ch = (*mdb->db->del)(mdb->db,
952: &key, R_CURSOR)) < 0)
1.3 kristaps 953: break;
954: }
1.18 kristaps 955:
1.3 kristaps 956: if (ch < 0) {
1.38 schwarze 957: perror(mdb->dbn);
1.3 kristaps 958: exit((int)MANDOCLEVEL_SYSERR);
1.18 kristaps 959: } else if (1 != ch) {
1.38 schwarze 960: fprintf(stderr, "%s: corrupt database\n",
961: mdb->dbn);
1.18 kristaps 962: exit((int)MANDOCLEVEL_SYSERR);
1.3 kristaps 963: }
1.1 kristaps 964:
1.5 kristaps 965: if (verb)
1.49.2.2 schwarze 966: printf("%s: deleting from index\n", fn);
1.1 kristaps 967:
1.3 kristaps 968: val.size = 0;
1.38 schwarze 969: ch = (*mdb->idx->put)(mdb->idx, &key, &val, R_CURSOR);
1.1 kristaps 970:
1.18 kristaps 971: if (ch < 0)
972: break;
973: cont:
1.38 schwarze 974: if (recs->cur >= recs->size) {
975: recs->size += MANDOC_SLOP;
976: recs->stack = mandoc_realloc(recs->stack,
977: recs->size * sizeof(recno_t));
1.3 kristaps 978: }
1.1 kristaps 979:
1.38 schwarze 980: recs->stack[(int)recs->cur] = recs->last;
981: recs->cur++;
1.3 kristaps 982: }
1.18 kristaps 983:
984: if (ch < 0) {
1.38 schwarze 985: perror(mdb->idxn);
1.18 kristaps 986: exit((int)MANDOCLEVEL_SYSERR);
987: } else if (1 != ch) {
1.38 schwarze 988: fprintf(stderr, "%s: corrupt index\n", mdb->idxn);
1.18 kristaps 989: exit((int)MANDOCLEVEL_SYSERR);
990: }
991:
1.38 schwarze 992: recs->last++;
1.1 kristaps 993: }
994:
995: /*
996: * Grow the buffer (if necessary) and copy in a binary string.
997: */
998: static void
999: buf_appendb(struct buf *buf, const void *cp, size_t sz)
1000: {
1001:
1002: /* Overshoot by MANDOC_BUFSZ. */
1003:
1004: while (buf->len + sz >= buf->size) {
1005: buf->size = buf->len + sz + MANDOC_BUFSZ;
1006: buf->cp = mandoc_realloc(buf->cp, buf->size);
1007: }
1008:
1009: memcpy(buf->cp + (int)buf->len, cp, sz);
1010: buf->len += sz;
1011: }
1012:
1013: /*
1014: * Append a nil-terminated string to the buffer.
1015: * This can be invoked multiple times.
1016: * The buffer string will be nil-terminated.
1017: * If invoked multiple times, a space is put between strings.
1018: */
1019: static void
1020: buf_append(struct buf *buf, const char *cp)
1021: {
1022: size_t sz;
1023:
1024: if (0 == (sz = strlen(cp)))
1025: return;
1026:
1027: if (buf->len)
1028: buf->cp[(int)buf->len - 1] = ' ';
1029:
1030: buf_appendb(buf, cp, sz + 1);
1031: }
1032:
1033: /*
1034: * Recursively add all text from a given node.
1035: * This is optimised for general mdoc nodes in this context, which do
1036: * not consist of subexpressions and having a recursive call for n->next
1037: * would be wasteful.
1038: * The "f" variable should be 0 unless called from pmdoc_Nd for the
1039: * description buffer, which does not start at the beginning of the
1040: * buffer.
1041: */
1042: static void
1043: buf_appendmdoc(struct buf *buf, const struct mdoc_node *n, int f)
1044: {
1045:
1046: for ( ; n; n = n->next) {
1047: if (n->child)
1048: buf_appendmdoc(buf, n->child, f);
1049:
1050: if (MDOC_TEXT == n->type && f) {
1051: f = 0;
1052: buf_appendb(buf, n->string,
1053: strlen(n->string) + 1);
1054: } else if (MDOC_TEXT == n->type)
1055: buf_append(buf, n->string);
1056:
1057: }
1058: }
1059:
1060: static void
1061: hash_reset(DB **db)
1062: {
1063: DB *hash;
1064:
1065: if (NULL != (hash = *db))
1066: (*hash->close)(hash);
1067:
1.5 kristaps 1068: *db = dbopen(NULL, O_CREAT|O_RDWR, 0644, DB_HASH, NULL);
1.1 kristaps 1069: if (NULL == *db) {
1070: perror("hash");
1071: exit((int)MANDOCLEVEL_SYSERR);
1072: }
1073: }
1074:
1075: /* ARGSUSED */
1.25 schwarze 1076: static int
1077: pmdoc_head(MDOC_ARGS)
1078: {
1079:
1080: return(MDOC_HEAD == n->type);
1081: }
1082:
1083: /* ARGSUSED */
1084: static int
1085: pmdoc_body(MDOC_ARGS)
1086: {
1087:
1088: return(MDOC_BODY == n->type);
1089: }
1090:
1091: /* ARGSUSED */
1092: static int
1.1 kristaps 1093: pmdoc_Fd(MDOC_ARGS)
1094: {
1095: const char *start, *end;
1096: size_t sz;
1.25 schwarze 1097:
1.1 kristaps 1098: if (SEC_SYNOPSIS != n->sec)
1.25 schwarze 1099: return(0);
1.1 kristaps 1100: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.25 schwarze 1101: return(0);
1.1 kristaps 1102:
1103: /*
1104: * Only consider those `Fd' macro fields that begin with an
1105: * "inclusion" token (versus, e.g., #define).
1106: */
1107: if (strcmp("#include", n->string))
1.25 schwarze 1108: return(0);
1.1 kristaps 1109:
1110: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.25 schwarze 1111: return(0);
1.1 kristaps 1112:
1113: /*
1114: * Strip away the enclosing angle brackets and make sure we're
1115: * not zero-length.
1116: */
1117:
1118: start = n->string;
1119: if ('<' == *start || '"' == *start)
1120: start++;
1121:
1122: if (0 == (sz = strlen(start)))
1.25 schwarze 1123: return(0);
1.1 kristaps 1124:
1125: end = &start[(int)sz - 1];
1126: if ('>' == *end || '"' == *end)
1127: end--;
1128:
1129: assert(end >= start);
1130:
1131: buf_appendb(buf, start, (size_t)(end - start + 1));
1132: buf_appendb(buf, "", 1);
1.25 schwarze 1133: return(1);
1.1 kristaps 1134: }
1135:
1136: /* ARGSUSED */
1.25 schwarze 1137: static int
1138: pmdoc_In(MDOC_ARGS)
1.1 kristaps 1139: {
1140:
1141: if (NULL == n->child || MDOC_TEXT != n->child->type)
1.25 schwarze 1142: return(0);
1.1 kristaps 1143:
1144: buf_append(buf, n->child->string);
1.25 schwarze 1145: return(1);
1.1 kristaps 1146: }
1147:
1148: /* ARGSUSED */
1.25 schwarze 1149: static int
1.1 kristaps 1150: pmdoc_Fn(MDOC_ARGS)
1151: {
1.25 schwarze 1152: struct mdoc_node *nn;
1.1 kristaps 1153: const char *cp;
1154:
1.25 schwarze 1155: nn = n->child;
1156:
1157: if (NULL == nn || MDOC_TEXT != nn->type)
1158: return(0);
1159:
1160: /* .Fn "struct type *name" "char *arg" */
1.1 kristaps 1161:
1.25 schwarze 1162: cp = strrchr(nn->string, ' ');
1.1 kristaps 1163: if (NULL == cp)
1.25 schwarze 1164: cp = nn->string;
1.1 kristaps 1165:
1166: /* Strip away pointer symbol. */
1167:
1168: while ('*' == *cp)
1169: cp++;
1170:
1.25 schwarze 1171: /* Store the function name. */
1172:
1.1 kristaps 1173: buf_append(buf, cp);
1.8 schwarze 1174: hash_put(hash, buf, TYPE_Fn);
1.25 schwarze 1175:
1176: /* Store the function type. */
1177:
1178: if (nn->string < cp) {
1179: buf->len = 0;
1180: buf_appendb(buf, nn->string, cp - nn->string);
1181: buf_appendb(buf, "", 1);
1182: hash_put(hash, buf, TYPE_Ft);
1183: }
1184:
1185: /* Store the arguments. */
1186:
1187: for (nn = nn->next; nn; nn = nn->next) {
1188: if (MDOC_TEXT != nn->type)
1189: continue;
1190: buf->len = 0;
1191: buf_append(buf, nn->string);
1192: hash_put(hash, buf, TYPE_Fa);
1193: }
1194:
1195: return(0);
1.1 kristaps 1196: }
1197:
1198: /* ARGSUSED */
1.25 schwarze 1199: static int
1.1 kristaps 1200: pmdoc_St(MDOC_ARGS)
1201: {
1.25 schwarze 1202:
1.1 kristaps 1203: if (NULL == n->child || MDOC_TEXT != n->child->type)
1.25 schwarze 1204: return(0);
1.1 kristaps 1205:
1206: buf_append(buf, n->child->string);
1.25 schwarze 1207: return(1);
1.1 kristaps 1208: }
1209:
1210: /* ARGSUSED */
1.25 schwarze 1211: static int
1.1 kristaps 1212: pmdoc_Xr(MDOC_ARGS)
1213: {
1214:
1215: if (NULL == (n = n->child))
1.25 schwarze 1216: return(0);
1.1 kristaps 1217:
1218: buf_appendb(buf, n->string, strlen(n->string));
1219:
1220: if (NULL != (n = n->next)) {
1221: buf_appendb(buf, ".", 1);
1222: buf_appendb(buf, n->string, strlen(n->string) + 1);
1223: } else
1224: buf_appendb(buf, ".", 2);
1225:
1.25 schwarze 1226: return(1);
1.1 kristaps 1227: }
1228:
1229: /* ARGSUSED */
1.25 schwarze 1230: static int
1.1 kristaps 1231: pmdoc_Nd(MDOC_ARGS)
1232: {
1233:
1234: if (MDOC_BODY != n->type)
1.25 schwarze 1235: return(0);
1.1 kristaps 1236:
1237: buf_appendmdoc(dbuf, n->child, 1);
1.25 schwarze 1238: return(1);
1.1 kristaps 1239: }
1240:
1241: /* ARGSUSED */
1.25 schwarze 1242: static int
1243: pmdoc_Nm(MDOC_ARGS)
1.1 kristaps 1244: {
1245:
1.25 schwarze 1246: if (SEC_NAME == n->sec)
1247: return(1);
1248: else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
1249: return(0);
1.1 kristaps 1250:
1.25 schwarze 1251: if (NULL == n->child)
1252: buf_append(buf, m->name);
1.1 kristaps 1253:
1.25 schwarze 1254: return(1);
1.1 kristaps 1255: }
1256:
1257: /* ARGSUSED */
1.25 schwarze 1258: static int
1259: pmdoc_Sh(MDOC_ARGS)
1.1 kristaps 1260: {
1261:
1.25 schwarze 1262: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 kristaps 1263: }
1264:
1265: static void
1.9 kristaps 1266: hash_put(DB *db, const struct buf *buf, uint64_t mask)
1.1 kristaps 1267: {
1.37 schwarze 1268: uint64_t oldmask;
1.1 kristaps 1269: DBT key, val;
1270: int rc;
1271:
1272: if (buf->len < 2)
1273: return;
1274:
1275: key.data = buf->cp;
1276: key.size = buf->len;
1277:
1278: if ((rc = (*db->get)(db, &key, &val, 0)) < 0) {
1279: perror("hash");
1280: exit((int)MANDOCLEVEL_SYSERR);
1.37 schwarze 1281: } else if (0 == rc) {
1282: assert(sizeof(uint64_t) == val.size);
1283: memcpy(&oldmask, val.data, val.size);
1284: mask |= oldmask;
1285: }
1.1 kristaps 1286:
1287: val.data = &mask;
1.9 kristaps 1288: val.size = sizeof(uint64_t);
1.1 kristaps 1289:
1290: if ((rc = (*db->put)(db, &key, &val, 0)) < 0) {
1291: perror("hash");
1292: exit((int)MANDOCLEVEL_SYSERR);
1293: }
1294: }
1295:
1296: static void
1297: dbt_put(DB *db, const char *dbn, DBT *key, DBT *val)
1298: {
1299:
1300: assert(key->size);
1301: assert(val->size);
1302:
1303: if (0 == (*db->put)(db, key, val, 0))
1304: return;
1305:
1306: perror(dbn);
1307: exit((int)MANDOCLEVEL_SYSERR);
1308: /* NOTREACHED */
1309: }
1310:
1311: /*
1312: * Call out to per-macro handlers after clearing the persistent database
1313: * key. If the macro sets the database key, flush it to the database.
1314: */
1315: static void
1316: pmdoc_node(MDOC_ARGS)
1317: {
1318:
1319: if (NULL == n)
1320: return;
1321:
1322: switch (n->type) {
1323: case (MDOC_HEAD):
1324: /* FALLTHROUGH */
1325: case (MDOC_BODY):
1326: /* FALLTHROUGH */
1327: case (MDOC_TAIL):
1328: /* FALLTHROUGH */
1329: case (MDOC_BLOCK):
1330: /* FALLTHROUGH */
1331: case (MDOC_ELEM):
1.25 schwarze 1332: buf->len = 0;
1333:
1334: /*
1335: * Both NULL handlers and handlers returning true
1336: * request using the data. Only skip the element
1337: * when the handler returns false.
1338: */
1339:
1340: if (NULL != mdocs[n->tok].fp &&
1341: 0 == (*mdocs[n->tok].fp)(hash, buf, dbuf, n, m))
1.1 kristaps 1342: break;
1343:
1.25 schwarze 1344: /*
1345: * For many macros, use the text from all children.
1346: * Set zero flags for macros not needing this.
1347: * In that case, the handler must fill the buffer.
1348: */
1349:
1350: if (MDOCF_CHILD & mdocs[n->tok].flags)
1351: buf_appendmdoc(buf, n->child, 0);
1352:
1353: /*
1354: * Cover the most common case:
1355: * Automatically stage one string per element.
1356: * Set a zero mask for macros not needing this.
1357: * Additional staging can be done in the handler.
1358: */
1359:
1360: if (mdocs[n->tok].mask)
1361: hash_put(hash, buf, mdocs[n->tok].mask);
1.1 kristaps 1362: break;
1363: default:
1364: break;
1365: }
1366:
1367: pmdoc_node(hash, buf, dbuf, n->child, m);
1368: pmdoc_node(hash, buf, dbuf, n->next, m);
1369: }
1370:
1371: static int
1372: pman_node(MAN_ARGS)
1373: {
1374: const struct man_node *head, *body;
1.46 kristaps 1375: char *start, *sv, *title;
1376: size_t sz, titlesz;
1.1 kristaps 1377:
1378: if (NULL == n)
1379: return(0);
1380:
1381: /*
1382: * We're only searching for one thing: the first text child in
1383: * the BODY of a NAME section. Since we don't keep track of
1384: * sections in -man, run some hoops to find out whether we're in
1385: * the correct section or not.
1386: */
1387:
1388: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1389: body = n;
1390: assert(body->parent);
1391: if (NULL != (head = body->parent->head) &&
1392: 1 == head->nchild &&
1393: NULL != (head = (head->child)) &&
1394: MAN_TEXT == head->type &&
1395: 0 == strcmp(head->string, "NAME") &&
1396: NULL != (body = body->child) &&
1397: MAN_TEXT == body->type) {
1398:
1.46 kristaps 1399: title = NULL;
1400: titlesz = 0;
1401: /*
1402: * Suck the entire NAME section into memory.
1403: * Yes, we might run away.
1404: * But too many manuals have big, spread-out
1405: * NAME sections over many lines.
1406: */
1407: for ( ; NULL != body; body = body->next) {
1408: if (MAN_TEXT != body->type)
1409: break;
1410: if (0 == (sz = strlen(body->string)))
1411: continue;
1412: title = mandoc_realloc
1413: (title, titlesz + sz + 1);
1414: memcpy(title + titlesz, body->string, sz);
1415: titlesz += sz + 1;
1416: title[(int)titlesz - 1] = ' ';
1417: }
1418: if (NULL == title)
1419: return(0);
1420:
1421: title = mandoc_realloc(title, titlesz + 1);
1422: title[(int)titlesz] = '\0';
1423:
1424: /* Skip leading space. */
1425:
1426: sv = title;
1427: while (isspace((unsigned char)*sv))
1428: sv++;
1429:
1430: if (0 == (sz = strlen(sv))) {
1431: free(title);
1432: return(0);
1433: }
1434:
1435: /* Erase trailing space. */
1436:
1437: start = &sv[sz - 1];
1438: while (start > sv && isspace((unsigned char)*start))
1439: *start-- = '\0';
1440:
1441: if (start == sv) {
1442: free(title);
1443: return(0);
1444: }
1445:
1446: start = sv;
1.1 kristaps 1447:
1448: /*
1449: * Go through a special heuristic dance here.
1450: * This is why -man manuals are great!
1451: * (I'm being sarcastic: my eyes are bleeding.)
1452: * Conventionally, one or more manual names are
1453: * comma-specified prior to a whitespace, then a
1454: * dash, then a description. Try to puzzle out
1455: * the name parts here.
1456: */
1457:
1458: for ( ;; ) {
1459: sz = strcspn(start, " ,");
1460: if ('\0' == start[(int)sz])
1461: break;
1462:
1463: buf->len = 0;
1464: buf_appendb(buf, start, sz);
1465: buf_appendb(buf, "", 1);
1466:
1.8 schwarze 1467: hash_put(hash, buf, TYPE_Nm);
1.1 kristaps 1468:
1469: if (' ' == start[(int)sz]) {
1470: start += (int)sz + 1;
1471: break;
1472: }
1473:
1474: assert(',' == start[(int)sz]);
1475: start += (int)sz + 1;
1476: while (' ' == *start)
1477: start++;
1478: }
1479:
1480: buf->len = 0;
1481:
1482: if (sv == start) {
1483: buf_append(buf, start);
1.46 kristaps 1484: free(title);
1.1 kristaps 1485: return(1);
1486: }
1487:
1.46 kristaps 1488: while (isspace((unsigned char)*start))
1.1 kristaps 1489: start++;
1490:
1491: if (0 == strncmp(start, "-", 1))
1492: start += 1;
1.43 kristaps 1493: else if (0 == strncmp(start, "\\-\\-", 4))
1494: start += 4;
1.1 kristaps 1495: else if (0 == strncmp(start, "\\-", 2))
1496: start += 2;
1497: else if (0 == strncmp(start, "\\(en", 4))
1498: start += 4;
1499: else if (0 == strncmp(start, "\\(em", 4))
1500: start += 4;
1501:
1502: while (' ' == *start)
1503: start++;
1504:
1505: sz = strlen(start) + 1;
1506: buf_appendb(dbuf, start, sz);
1507: buf_appendb(buf, start, sz);
1508:
1.8 schwarze 1509: hash_put(hash, buf, TYPE_Nd);
1.46 kristaps 1510: free(title);
1.1 kristaps 1511: }
1512: }
1513:
1.7 schwarze 1514: for (n = n->child; n; n = n->next)
1515: if (pman_node(hash, buf, dbuf, n))
1516: return(1);
1.1 kristaps 1517:
1518: return(0);
1519: }
1520:
1.14 schwarze 1521: /*
1522: * Parse a formatted manual page.
1523: * By necessity, this involves rather crude guesswork.
1524: */
1525: static void
1.49.2.2 schwarze 1526: pformatted(DB *hash, struct buf *buf,
1527: struct buf *dbuf, const struct of *of)
1.14 schwarze 1528: {
1529: FILE *stream;
1.43 kristaps 1530: char *line, *p, *title;
1531: size_t len, plen, titlesz;
1.14 schwarze 1532:
1533: if (NULL == (stream = fopen(of->fname, "r"))) {
1.49.2.2 schwarze 1534: if (warnings)
1535: perror(of->fname);
1.14 schwarze 1536: return;
1537: }
1538:
1539: /*
1540: * Always use the title derived from the filename up front,
1541: * do not even try to find it in the file. This also makes
1542: * sure we don't end up with an orphan index record, even if
1543: * the file content turns out to be completely unintelligible.
1544: */
1545:
1546: buf->len = 0;
1547: buf_append(buf, of->title);
1548: hash_put(hash, buf, TYPE_Nm);
1549:
1.31 schwarze 1550: /* Skip to first blank line. */
1.14 schwarze 1551:
1.28 kristaps 1552: while (NULL != (line = fgetln(stream, &len)))
1.31 schwarze 1553: if ('\n' == *line)
1.28 kristaps 1554: break;
1555:
1.31 schwarze 1556: /*
1557: * Assume the first line that is not indented
1558: * is the first section header. Skip to it.
1.28 kristaps 1559: */
1560:
1561: while (NULL != (line = fgetln(stream, &len)))
1.31 schwarze 1562: if ('\n' != *line && ' ' != *line)
1.28 kristaps 1563: break;
1.43 kristaps 1564:
1565: /*
1566: * Read up until the next section into a buffer.
1567: * Strip the leading and trailing newline from each read line,
1568: * appending a trailing space.
1569: * Ignore empty (whitespace-only) lines.
1570: */
1571:
1572: titlesz = 0;
1573: title = NULL;
1574:
1575: while (NULL != (line = fgetln(stream, &len))) {
1576: if (' ' != *line || '\n' != line[(int)len - 1])
1577: break;
1578: while (len > 0 && isspace((unsigned char)*line)) {
1579: line++;
1580: len--;
1581: }
1582: if (1 == len)
1583: continue;
1584: title = mandoc_realloc(title, titlesz + len);
1585: memcpy(title + titlesz, line, len);
1586: titlesz += len;
1587: title[(int)titlesz - 1] = ' ';
1588: }
1589:
1.49.2.2 schwarze 1590:
1.14 schwarze 1591: /*
1.31 schwarze 1592: * If no page content can be found, or the input line
1593: * is already the next section header, or there is no
1594: * trailing newline, reuse the page title as the page
1595: * description.
1.14 schwarze 1596: */
1597:
1.43 kristaps 1598: if (NULL == title || '\0' == *title) {
1.49.2.2 schwarze 1599: if (warnings)
1600: fprintf(stderr, "%s: cannot find NAME section\n",
1601: of->fname);
1.14 schwarze 1602: buf_appendb(dbuf, buf->cp, buf->size);
1603: hash_put(hash, buf, TYPE_Nd);
1604: fclose(stream);
1.43 kristaps 1605: free(title);
1.14 schwarze 1606: return;
1607: }
1608:
1.43 kristaps 1609: title = mandoc_realloc(title, titlesz + 1);
1610: title[(int)titlesz] = '\0';
1.28 kristaps 1611:
1.31 schwarze 1612: /*
1613: * Skip to the first dash.
1.28 kristaps 1614: * Use the remaining line as the description (no more than 70
1615: * bytes).
1.14 schwarze 1616: */
1617:
1.43 kristaps 1618: if (NULL != (p = strstr(title, "- "))) {
1.30 kristaps 1619: for (p += 2; ' ' == *p || '\b' == *p; p++)
1.28 kristaps 1620: /* Skip to next word. */ ;
1.38 schwarze 1621: } else {
1.49.2.2 schwarze 1622: if (warnings)
1623: fprintf(stderr, "%s: no dash in title line\n",
1624: of->fname);
1.43 kristaps 1625: p = title;
1.38 schwarze 1626: }
1.28 kristaps 1627:
1.43 kristaps 1628: plen = strlen(p);
1.29 kristaps 1629:
1630: /* Strip backspace-encoding from line. */
1631:
1632: while (NULL != (line = memchr(p, '\b', plen))) {
1633: len = line - p;
1634: if (0 == len) {
1635: memmove(line, line + 1, plen--);
1636: continue;
1637: }
1638: memmove(line - 1, line + 1, plen - len);
1639: plen -= 2;
1.14 schwarze 1640: }
1641:
1.28 kristaps 1642: buf_appendb(dbuf, p, plen + 1);
1.14 schwarze 1643: buf->len = 0;
1.28 kristaps 1644: buf_appendb(buf, p, plen + 1);
1.14 schwarze 1645: hash_put(hash, buf, TYPE_Nd);
1.28 kristaps 1646: fclose(stream);
1.43 kristaps 1647: free(title);
1.14 schwarze 1648: }
1649:
1.5 kristaps 1650: static void
1.49.2.2 schwarze 1651: ofile_argbuild(int argc, char *argv[], struct of **of,
1652: const char *basedir)
1.5 kristaps 1653: {
1.49.2.1 schwarze 1654: char buf[PATH_MAX];
1.49.2.2 schwarze 1655: char pbuf[PATH_MAX];
1.41 kristaps 1656: const char *sec, *arch, *title;
1.49.2.2 schwarze 1657: char *relpath, *p;
1.14 schwarze 1658: int i, src_form;
1.5 kristaps 1659: struct of *nof;
1660:
1661: for (i = 0; i < argc; i++) {
1.49.2.2 schwarze 1662: if (NULL == (relpath = realpath(argv[i], pbuf))) {
1663: perror(argv[i]);
1664: continue;
1665: }
1666: if (NULL != basedir) {
1667: if (strstr(pbuf, basedir) != pbuf) {
1668: fprintf(stderr, "%s: file outside "
1669: "base directory %s\n",
1670: pbuf, basedir);
1671: continue;
1672: }
1673: relpath = pbuf + strlen(basedir);
1674: }
1.12 schwarze 1675:
1676: /*
1677: * Try to infer the manual section, architecture and
1678: * page title from the path, assuming it looks like
1.14 schwarze 1679: * man*[/<arch>]/<title>.<section> or
1680: * cat<section>[/<arch>]/<title>.0
1.12 schwarze 1681: */
1682:
1.49.2.2 schwarze 1683: if (strlcpy(buf, relpath, sizeof(buf)) >= sizeof(buf)) {
1684: fprintf(stderr, "%s: path too long\n", relpath);
1.12 schwarze 1685: continue;
1686: }
1.38 schwarze 1687: sec = arch = title = "";
1.14 schwarze 1688: src_form = 0;
1.12 schwarze 1689: p = strrchr(buf, '\0');
1690: while (p-- > buf) {
1.38 schwarze 1691: if ('\0' == *sec && '.' == *p) {
1.12 schwarze 1692: sec = p + 1;
1693: *p = '\0';
1.14 schwarze 1694: if ('0' == *sec)
1695: src_form |= MANDOC_FORM;
1696: else if ('1' <= *sec && '9' >= *sec)
1697: src_form |= MANDOC_SRC;
1.12 schwarze 1698: continue;
1699: }
1700: if ('/' != *p)
1701: continue;
1.38 schwarze 1702: if ('\0' == *title) {
1.12 schwarze 1703: title = p + 1;
1704: *p = '\0';
1705: continue;
1706: }
1.24 schwarze 1707: if (0 == strncmp("man", p + 1, 3))
1.14 schwarze 1708: src_form |= MANDOC_SRC;
1.24 schwarze 1709: else if (0 == strncmp("cat", p + 1, 3))
1.14 schwarze 1710: src_form |= MANDOC_FORM;
1.24 schwarze 1711: else
1.12 schwarze 1712: arch = p + 1;
1713: break;
1714: }
1.38 schwarze 1715: if ('\0' == *title) {
1.49.2.2 schwarze 1716: if (warnings)
1717: fprintf(stderr,
1718: "%s: cannot deduce title "
1719: "from filename\n",
1720: relpath);
1.12 schwarze 1721: title = buf;
1.38 schwarze 1722: }
1.12 schwarze 1723:
1724: /*
1725: * Build the file structure.
1726: */
1727:
1.5 kristaps 1728: nof = mandoc_calloc(1, sizeof(struct of));
1.49.2.2 schwarze 1729: nof->fname = mandoc_strdup(relpath);
1.38 schwarze 1730: nof->sec = mandoc_strdup(sec);
1731: nof->arch = mandoc_strdup(arch);
1.12 schwarze 1732: nof->title = mandoc_strdup(title);
1.14 schwarze 1733: nof->src_form = src_form;
1.12 schwarze 1734:
1735: /*
1736: * Add the structure to the list.
1737: */
1738:
1.5 kristaps 1739: if (NULL == *of) {
1740: *of = nof;
1741: (*of)->first = nof;
1742: } else {
1743: nof->first = (*of)->first;
1744: (*of)->next = nof;
1745: *of = nof;
1746: }
1747: }
1748: }
1749:
1.4 kristaps 1750: /*
1751: * Recursively build up a list of files to parse.
1752: * We use this instead of ftw() and so on because I don't want global
1753: * variables hanging around.
1.49.2.5 schwarze 1754: * This ignores the mandoc.db and mandoc.index files, but assumes that
1.4 kristaps 1755: * everything else is a manual.
1756: * Pass in a pointer to a NULL structure for the first invocation.
1757: */
1.35 kristaps 1758: static void
1.12 schwarze 1759: ofile_dirbuild(const char *dir, const char* psec, const char *parch,
1.49.2.2 schwarze 1760: int p_src_form, struct of **of)
1.4 kristaps 1761: {
1.49.2.1 schwarze 1762: char buf[PATH_MAX];
1.49.2.6! schwarze 1763: #if defined(__sun)
! 1764: struct stat sb;
! 1765: #endif
1.5 kristaps 1766: size_t sz;
1.4 kristaps 1767: DIR *d;
1.12 schwarze 1768: const char *fn, *sec, *arch;
1.14 schwarze 1769: char *p, *q, *suffix;
1.4 kristaps 1770: struct of *nof;
1771: struct dirent *dp;
1.14 schwarze 1772: int src_form;
1.4 kristaps 1773:
1774: if (NULL == (d = opendir(dir))) {
1.49.2.2 schwarze 1775: if (warnings)
1776: perror(dir);
1.38 schwarze 1777: return;
1.4 kristaps 1778: }
1779:
1780: while (NULL != (dp = readdir(d))) {
1781: fn = dp->d_name;
1.12 schwarze 1782:
1783: if ('.' == *fn)
1784: continue;
1785:
1.14 schwarze 1786: src_form = p_src_form;
1787:
1.49.2.6! schwarze 1788: #if defined(__sun)
! 1789: stat(dp->d_name, &sb);
! 1790: if (S_IFDIR & sb.st_mode) {
! 1791: #else
1.4 kristaps 1792: if (DT_DIR == dp->d_type) {
1.49.2.6! schwarze 1793: #endif
1.12 schwarze 1794: sec = psec;
1795: arch = parch;
1796:
1797: /*
1798: * By default, only use directories called:
1.14 schwarze 1799: * man<section>/[<arch>/] or
1800: * cat<section>/[<arch>/]
1.12 schwarze 1801: */
1802:
1.38 schwarze 1803: if ('\0' == *sec) {
1.14 schwarze 1804: if(0 == strncmp("man", fn, 3)) {
1805: src_form |= MANDOC_SRC;
1.12 schwarze 1806: sec = fn + 3;
1.14 schwarze 1807: } else if (0 == strncmp("cat", fn, 3)) {
1808: src_form |= MANDOC_FORM;
1809: sec = fn + 3;
1.38 schwarze 1810: } else {
1.49.2.2 schwarze 1811: if (warnings) fprintf(stderr,
1812: "%s/%s: bad section\n",
1813: dir, fn);
1.38 schwarze 1814: if (use_all)
1815: sec = fn;
1816: else
1817: continue;
1818: }
1819: } else if ('\0' == *arch) {
1820: if (NULL != strchr(fn, '.')) {
1.49.2.2 schwarze 1821: if (warnings) fprintf(stderr,
1822: "%s/%s: bad architecture\n",
1823: dir, fn);
1.38 schwarze 1824: if (0 == use_all)
1825: continue;
1826: }
1827: arch = fn;
1828: } else {
1.49.2.2 schwarze 1829: if (warnings) fprintf(stderr, "%s/%s: "
1830: "excessive subdirectory\n", dir, fn);
1.38 schwarze 1831: if (0 == use_all)
1.12 schwarze 1832: continue;
1.38 schwarze 1833: }
1.5 kristaps 1834:
1835: buf[0] = '\0';
1.49.2.1 schwarze 1836: strlcat(buf, dir, PATH_MAX);
1837: strlcat(buf, "/", PATH_MAX);
1838: sz = strlcat(buf, fn, PATH_MAX);
1.5 kristaps 1839:
1.49.2.1 schwarze 1840: if (PATH_MAX <= sz) {
1.49.2.2 schwarze 1841: if (warnings) fprintf(stderr, "%s/%s: "
1842: "path too long\n", dir, fn);
1.38 schwarze 1843: continue;
1.12 schwarze 1844: }
1.38 schwarze 1845:
1.49.2.2 schwarze 1846: ofile_dirbuild(buf, sec, arch, src_form, of);
1.38 schwarze 1847: continue;
1.35 kristaps 1848: }
1.12 schwarze 1849:
1.49.2.6! schwarze 1850: #if defined(__sun)
! 1851: if (0 == S_IFREG & sb.st_mode) {
! 1852: #else
1.38 schwarze 1853: if (DT_REG != dp->d_type) {
1.49.2.6! schwarze 1854: #endif
1.49.2.2 schwarze 1855: if (warnings)
1856: fprintf(stderr,
1857: "%s/%s: not a regular file\n",
1858: dir, fn);
1.38 schwarze 1859: continue;
1860: }
1861: if (!strcmp(MANDOC_DB, fn) || !strcmp(MANDOC_IDX, fn))
1.12 schwarze 1862: continue;
1.38 schwarze 1863: if ('\0' == *psec) {
1.49.2.2 schwarze 1864: if (warnings)
1865: fprintf(stderr,
1866: "%s/%s: file outside section\n",
1867: dir, fn);
1.38 schwarze 1868: if (0 == use_all)
1869: continue;
1870: }
1.12 schwarze 1871:
1872: /*
1873: * By default, skip files where the file name suffix
1874: * does not agree with the section directory
1875: * they are located in.
1876: */
1877:
1878: suffix = strrchr(fn, '.');
1.38 schwarze 1879: if (NULL == suffix) {
1.49.2.2 schwarze 1880: if (warnings)
1881: fprintf(stderr,
1882: "%s/%s: no filename suffix\n",
1883: dir, fn);
1.38 schwarze 1884: if (0 == use_all)
1.5 kristaps 1885: continue;
1.38 schwarze 1886: } else if ((MANDOC_SRC & src_form &&
1887: strcmp(suffix + 1, psec)) ||
1.14 schwarze 1888: (MANDOC_FORM & src_form &&
1.38 schwarze 1889: strcmp(suffix + 1, "0"))) {
1.49.2.2 schwarze 1890: if (warnings)
1891: fprintf(stderr,
1892: "%s/%s: wrong filename suffix\n",
1893: dir, fn);
1.38 schwarze 1894: if (0 == use_all)
1895: continue;
1.14 schwarze 1896: if ('0' == suffix[1])
1897: src_form |= MANDOC_FORM;
1898: else if ('1' <= suffix[1] && '9' >= suffix[1])
1899: src_form |= MANDOC_SRC;
1900: }
1901:
1902: /*
1903: * Skip formatted manuals if a source version is
1904: * available. Ignore the age: it is very unlikely
1905: * that people install newer formatted base manuals
1906: * when they used to have source manuals before,
1907: * and in ports, old manuals get removed on update.
1908: */
1909: if (0 == use_all && MANDOC_FORM & src_form &&
1.38 schwarze 1910: '\0' != *psec) {
1.14 schwarze 1911: buf[0] = '\0';
1.49.2.1 schwarze 1912: strlcat(buf, dir, PATH_MAX);
1.14 schwarze 1913: p = strrchr(buf, '/');
1.38 schwarze 1914: if ('\0' != *parch && NULL != p)
1.32 schwarze 1915: for (p--; p > buf; p--)
1916: if ('/' == *p)
1917: break;
1.14 schwarze 1918: if (NULL == p)
1919: p = buf;
1920: else
1921: p++;
1922: if (0 == strncmp("cat", p, 3))
1923: memcpy(p, "man", 3);
1.49.2.1 schwarze 1924: strlcat(buf, "/", PATH_MAX);
1925: sz = strlcat(buf, fn, PATH_MAX);
1926: if (sz >= PATH_MAX) {
1.49.2.2 schwarze 1927: if (warnings) fprintf(stderr,
1928: "%s/%s: path too long\n",
1929: dir, fn);
1.5 kristaps 1930: continue;
1.14 schwarze 1931: }
1932: q = strrchr(buf, '.');
1933: if (NULL != q && p < q++) {
1934: *q = '\0';
1.49.2.1 schwarze 1935: sz = strlcat(buf, psec, PATH_MAX);
1936: if (sz >= PATH_MAX) {
1.49.2.2 schwarze 1937: if (warnings) fprintf(stderr,
1938: "%s/%s: path too long\n",
1939: dir, fn);
1.14 schwarze 1940: continue;
1941: }
1.35 kristaps 1942: if (0 == access(buf, R_OK))
1.14 schwarze 1943: continue;
1944: }
1.5 kristaps 1945: }
1.4 kristaps 1946:
1.38 schwarze 1947: buf[0] = '\0';
1.35 kristaps 1948: assert('.' == dir[0]);
1.38 schwarze 1949: if ('/' == dir[1]) {
1.49.2.1 schwarze 1950: strlcat(buf, dir + 2, PATH_MAX);
1951: strlcat(buf, "/", PATH_MAX);
1.38 schwarze 1952: }
1.49.2.1 schwarze 1953: sz = strlcat(buf, fn, PATH_MAX);
1954: if (sz >= PATH_MAX) {
1.49.2.2 schwarze 1955: if (warnings) fprintf(stderr,
1956: "%s/%s: path too long\n", dir, fn);
1.14 schwarze 1957: continue;
1.5 kristaps 1958: }
1959:
1.4 kristaps 1960: nof = mandoc_calloc(1, sizeof(struct of));
1.5 kristaps 1961: nof->fname = mandoc_strdup(buf);
1.38 schwarze 1962: nof->sec = mandoc_strdup(psec);
1963: nof->arch = mandoc_strdup(parch);
1.14 schwarze 1964: nof->src_form = src_form;
1.12 schwarze 1965:
1966: /*
1967: * Remember the file name without the extension,
1968: * to be used as the page title in the database.
1969: */
1970:
1971: if (NULL != suffix)
1972: *suffix = '\0';
1973: nof->title = mandoc_strdup(fn);
1.5 kristaps 1974:
1.14 schwarze 1975: /*
1976: * Add the structure to the list.
1977: */
1.41 kristaps 1978:
1.4 kristaps 1979: if (NULL == *of) {
1980: *of = nof;
1981: (*of)->first = nof;
1982: } else {
1.5 kristaps 1983: nof->first = (*of)->first;
1.4 kristaps 1984: (*of)->next = nof;
1985: *of = nof;
1986: }
1987: }
1988:
1.7 schwarze 1989: closedir(d);
1.4 kristaps 1990: }
1991:
1992: static void
1993: ofile_free(struct of *of)
1994: {
1995: struct of *nof;
1996:
1.41 kristaps 1997: if (NULL != of)
1998: of = of->first;
1999:
2000: while (NULL != of) {
1.4 kristaps 2001: nof = of->next;
2002: free(of->fname);
1.12 schwarze 2003: free(of->sec);
2004: free(of->arch);
2005: free(of->title);
1.4 kristaps 2006: free(of);
2007: of = nof;
2008: }
1.1 kristaps 2009: }
CVSweb