Annotation of mandoc/mandocdb.c, Revision 1.15
1.15 ! schwarze 1: /* $Id: mandocdb.c,v 1.14 2011/11/27 22:57:53 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.12 schwarze 4: * Copyright (c) 2011 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:
22: #include <sys/param.h>
1.14 schwarze 23: #include <sys/types.h>
24: #include <sys/stat.h>
1.1 kristaps 25:
26: #include <assert.h>
1.4 kristaps 27: #include <dirent.h>
1.1 kristaps 28: #include <fcntl.h>
29: #include <getopt.h>
30: #include <stdio.h>
31: #include <stdint.h>
32: #include <stdlib.h>
33: #include <string.h>
34:
35: #ifdef __linux__
36: # include <db_185.h>
37: #else
38: # include <db.h>
39: #endif
40:
41: #include "man.h"
42: #include "mdoc.h"
43: #include "mandoc.h"
1.8 schwarze 44: #include "mandocdb.h"
1.10 kristaps 45: #include "manpath.h"
1.1 kristaps 46:
47: #define MANDOC_BUFSZ BUFSIZ
48: #define MANDOC_SLOP 1024
49:
1.14 schwarze 50: #define MANDOC_SRC 0x1
51: #define MANDOC_FORM 0x2
52:
1.5 kristaps 53: /* Tiny list for files. No need to bring in QUEUE. */
54:
1.3 kristaps 55: struct of {
1.5 kristaps 56: char *fname; /* heap-allocated */
1.12 schwarze 57: char *sec;
58: char *arch;
59: char *title;
1.14 schwarze 60: int src_form;
1.5 kristaps 61: struct of *next; /* NULL for last one */
62: struct of *first; /* first in list */
1.3 kristaps 63: };
64:
1.1 kristaps 65: /* Buffer for storing growable data. */
66:
67: struct buf {
68: char *cp;
1.5 kristaps 69: size_t len; /* current length */
70: size_t size; /* total buffer size */
1.1 kristaps 71: };
72:
73: /* Operation we're going to perform. */
74:
75: enum op {
76: OP_NEW = 0, /* new database */
1.5 kristaps 77: OP_UPDATE, /* delete/add entries in existing database */
1.1 kristaps 78: OP_DELETE /* delete entries from existing database */
79: };
80:
81: #define MAN_ARGS DB *hash, \
82: struct buf *buf, \
83: struct buf *dbuf, \
84: const struct man_node *n
85: #define MDOC_ARGS DB *hash, \
86: struct buf *buf, \
87: struct buf *dbuf, \
88: const struct mdoc_node *n, \
89: const struct mdoc_meta *m
90:
91: static void buf_appendmdoc(struct buf *,
92: const struct mdoc_node *, int);
93: static void buf_append(struct buf *, const char *);
94: static void buf_appendb(struct buf *,
95: const void *, size_t);
96: static void dbt_put(DB *, const char *, DBT *, DBT *);
1.9 kristaps 97: static void hash_put(DB *, const struct buf *, uint64_t);
1.1 kristaps 98: static void hash_reset(DB **);
1.3 kristaps 99: static void index_merge(const struct of *, struct mparse *,
100: struct buf *, struct buf *,
101: DB *, DB *, const char *,
1.12 schwarze 102: DB *, const char *, int, int,
1.3 kristaps 103: recno_t, const recno_t *, size_t);
104: static void index_prune(const struct of *, DB *,
105: const char *, DB *, const char *,
1.5 kristaps 106: int, recno_t *, recno_t **, size_t *);
1.12 schwarze 107: static void ofile_argbuild(char *[], int, int, int,
108: struct of **);
109: static int ofile_dirbuild(const char *, const char *,
1.14 schwarze 110: const char *, int, int, int,
111: struct of **);
1.4 kristaps 112: static void ofile_free(struct of *);
1.14 schwarze 113: static void pformatted(DB *, struct buf *, struct buf *,
114: const struct of *);
1.1 kristaps 115: static int pman_node(MAN_ARGS);
116: static void pmdoc_node(MDOC_ARGS);
117: static void pmdoc_An(MDOC_ARGS);
118: static void pmdoc_Cd(MDOC_ARGS);
119: static void pmdoc_Er(MDOC_ARGS);
120: static void pmdoc_Ev(MDOC_ARGS);
121: static void pmdoc_Fd(MDOC_ARGS);
122: static void pmdoc_In(MDOC_ARGS);
123: static void pmdoc_Fn(MDOC_ARGS);
124: static void pmdoc_Fo(MDOC_ARGS);
125: static void pmdoc_Nd(MDOC_ARGS);
126: static void pmdoc_Nm(MDOC_ARGS);
127: static void pmdoc_Pa(MDOC_ARGS);
128: static void pmdoc_St(MDOC_ARGS);
129: static void pmdoc_Vt(MDOC_ARGS);
130: static void pmdoc_Xr(MDOC_ARGS);
131: static void usage(void);
132:
133: typedef void (*pmdoc_nf)(MDOC_ARGS);
134:
135: static const pmdoc_nf mdocs[MDOC_MAX] = {
136: NULL, /* Ap */
137: NULL, /* Dd */
138: NULL, /* Dt */
139: NULL, /* Os */
140: NULL, /* Sh */
141: NULL, /* Ss */
142: NULL, /* Pp */
143: NULL, /* D1 */
144: NULL, /* Dl */
145: NULL, /* Bd */
146: NULL, /* Ed */
147: NULL, /* Bl */
148: NULL, /* El */
149: NULL, /* It */
150: NULL, /* Ad */
151: pmdoc_An, /* An */
152: NULL, /* Ar */
153: pmdoc_Cd, /* Cd */
154: NULL, /* Cm */
155: NULL, /* Dv */
156: pmdoc_Er, /* Er */
157: pmdoc_Ev, /* Ev */
158: NULL, /* Ex */
159: NULL, /* Fa */
160: pmdoc_Fd, /* Fd */
161: NULL, /* Fl */
162: pmdoc_Fn, /* Fn */
163: NULL, /* Ft */
164: NULL, /* Ic */
165: pmdoc_In, /* In */
166: NULL, /* Li */
167: pmdoc_Nd, /* Nd */
168: pmdoc_Nm, /* Nm */
169: NULL, /* Op */
170: NULL, /* Ot */
171: pmdoc_Pa, /* Pa */
172: NULL, /* Rv */
173: pmdoc_St, /* St */
174: pmdoc_Vt, /* Va */
175: pmdoc_Vt, /* Vt */
176: pmdoc_Xr, /* Xr */
177: NULL, /* %A */
178: NULL, /* %B */
179: NULL, /* %D */
180: NULL, /* %I */
181: NULL, /* %J */
182: NULL, /* %N */
183: NULL, /* %O */
184: NULL, /* %P */
185: NULL, /* %R */
186: NULL, /* %T */
187: NULL, /* %V */
188: NULL, /* Ac */
189: NULL, /* Ao */
190: NULL, /* Aq */
191: NULL, /* At */
192: NULL, /* Bc */
193: NULL, /* Bf */
194: NULL, /* Bo */
195: NULL, /* Bq */
196: NULL, /* Bsx */
197: NULL, /* Bx */
198: NULL, /* Db */
199: NULL, /* Dc */
200: NULL, /* Do */
201: NULL, /* Dq */
202: NULL, /* Ec */
203: NULL, /* Ef */
204: NULL, /* Em */
205: NULL, /* Eo */
206: NULL, /* Fx */
207: NULL, /* Ms */
208: NULL, /* No */
209: NULL, /* Ns */
210: NULL, /* Nx */
211: NULL, /* Ox */
212: NULL, /* Pc */
213: NULL, /* Pf */
214: NULL, /* Po */
215: NULL, /* Pq */
216: NULL, /* Qc */
217: NULL, /* Ql */
218: NULL, /* Qo */
219: NULL, /* Qq */
220: NULL, /* Re */
221: NULL, /* Rs */
222: NULL, /* Sc */
223: NULL, /* So */
224: NULL, /* Sq */
225: NULL, /* Sm */
226: NULL, /* Sx */
227: NULL, /* Sy */
228: NULL, /* Tn */
229: NULL, /* Ux */
230: NULL, /* Xc */
231: NULL, /* Xo */
232: pmdoc_Fo, /* Fo */
233: NULL, /* Fc */
234: NULL, /* Oo */
235: NULL, /* Oc */
236: NULL, /* Bk */
237: NULL, /* Ek */
238: NULL, /* Bt */
239: NULL, /* Hf */
240: NULL, /* Fr */
241: NULL, /* Ud */
242: NULL, /* Lb */
243: NULL, /* Lp */
244: NULL, /* Lk */
245: NULL, /* Mt */
246: NULL, /* Brq */
247: NULL, /* Bro */
248: NULL, /* Brc */
249: NULL, /* %C */
250: NULL, /* Es */
251: NULL, /* En */
252: NULL, /* Dx */
253: NULL, /* %Q */
254: NULL, /* br */
255: NULL, /* sp */
256: NULL, /* %U */
257: NULL, /* Ta */
258: };
259:
260: static const char *progname;
261:
262: int
263: main(int argc, char *argv[])
264: {
265: struct mparse *mp; /* parse sequence */
1.10 kristaps 266: struct manpaths dirs;
1.1 kristaps 267: enum op op; /* current operation */
1.5 kristaps 268: const char *dir;
1.1 kristaps 269: char ibuf[MAXPATHLEN], /* index fname */
1.3 kristaps 270: fbuf[MAXPATHLEN]; /* btree fname */
1.5 kristaps 271: int verb, /* output verbosity */
1.12 schwarze 272: use_all, /* use all directories and files */
1.5 kristaps 273: ch, i, flags;
1.1 kristaps 274: DB *idx, /* index database */
275: *db, /* keyword database */
276: *hash; /* temporary keyword hashtable */
277: BTREEINFO info; /* btree configuration */
1.12 schwarze 278: recno_t maxrec; /* last record number in the index */
279: recno_t *recs; /* the numbers of all empty records */
1.5 kristaps 280: size_t sz1, sz2,
1.12 schwarze 281: recsz, /* number of allocated slots in recs */
282: reccur; /* current number of empty records */
1.1 kristaps 283: struct buf buf, /* keyword buffer */
284: dbuf; /* description buffer */
1.5 kristaps 285: struct of *of; /* list of files for processing */
1.1 kristaps 286: extern int optind;
287: extern char *optarg;
288:
289: progname = strrchr(argv[0], '/');
290: if (progname == NULL)
291: progname = argv[0];
292: else
293: ++progname;
294:
1.10 kristaps 295: memset(&dirs, 0, sizeof(struct manpaths));
296:
1.5 kristaps 297: verb = 0;
1.12 schwarze 298: use_all = 0;
1.4 kristaps 299: of = NULL;
1.1 kristaps 300: db = idx = NULL;
301: mp = NULL;
302: hash = NULL;
303: recs = NULL;
304: recsz = reccur = 0;
305: maxrec = 0;
306: op = OP_NEW;
1.5 kristaps 307: dir = NULL;
1.1 kristaps 308:
1.12 schwarze 309: while (-1 != (ch = getopt(argc, argv, "ad:u:v")))
1.1 kristaps 310: switch (ch) {
1.12 schwarze 311: case ('a'):
312: use_all = 1;
313: break;
1.5 kristaps 314: case ('d'):
315: dir = optarg;
316: op = OP_UPDATE;
317: break;
318: case ('u'):
319: dir = optarg;
320: op = OP_DELETE;
321: break;
322: case ('v'):
323: verb++;
324: break;
1.1 kristaps 325: default:
326: usage();
327: return((int)MANDOCLEVEL_BADARG);
328: }
329:
330: argc -= optind;
331: argv += optind;
332:
1.4 kristaps 333: memset(&info, 0, sizeof(BTREEINFO));
334: info.flags = R_DUP;
1.1 kristaps 335:
1.4 kristaps 336: mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL);
1.1 kristaps 337:
1.5 kristaps 338: memset(&buf, 0, sizeof(struct buf));
339: memset(&dbuf, 0, sizeof(struct buf));
1.1 kristaps 340:
1.4 kristaps 341: buf.size = dbuf.size = MANDOC_BUFSZ;
1.1 kristaps 342:
1.4 kristaps 343: buf.cp = mandoc_malloc(buf.size);
344: dbuf.cp = mandoc_malloc(dbuf.size);
1.1 kristaps 345:
1.5 kristaps 346: flags = OP_NEW == op ? O_CREAT|O_TRUNC|O_RDWR : O_CREAT|O_RDWR;
347:
348: if (OP_UPDATE == op || OP_DELETE == op) {
349: ibuf[0] = fbuf[0] = '\0';
350:
351: strlcat(fbuf, dir, MAXPATHLEN);
352: strlcat(fbuf, "/", MAXPATHLEN);
353: sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
354:
355: strlcat(ibuf, dir, MAXPATHLEN);
356: strlcat(ibuf, "/", MAXPATHLEN);
357: sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
358:
359: if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) {
360: fprintf(stderr, "%s: Path too long\n", dir);
361: exit((int)MANDOCLEVEL_BADARG);
362: }
363:
364: db = dbopen(fbuf, flags, 0644, DB_BTREE, &info);
365: idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL);
366:
367: if (NULL == db) {
368: perror(fbuf);
369: exit((int)MANDOCLEVEL_SYSERR);
1.12 schwarze 370: } else if (NULL == idx) {
1.5 kristaps 371: perror(ibuf);
372: exit((int)MANDOCLEVEL_SYSERR);
373: }
374:
375: if (verb > 2) {
376: printf("%s: Opened\n", fbuf);
377: printf("%s: Opened\n", ibuf);
378: }
379:
1.12 schwarze 380: ofile_argbuild(argv, argc, use_all, verb, &of);
1.5 kristaps 381: if (NULL == of)
382: goto out;
383:
384: of = of->first;
385:
386: index_prune(of, db, fbuf, idx, ibuf, verb,
387: &maxrec, &recs, &recsz);
388:
389: if (OP_UPDATE == op)
1.13 schwarze 390: index_merge(of, mp, &dbuf, &buf, hash,
1.12 schwarze 391: db, fbuf, idx, ibuf, use_all,
392: verb, maxrec, recs, reccur);
1.5 kristaps 393:
394: goto out;
395: }
396:
1.10 kristaps 397: /*
398: * Configure the directories we're going to scan.
399: * If we have command-line arguments, use them.
400: * If not, we use man(1)'s method (see mandocdb.8).
401: */
402:
403: if (argc > 0) {
404: dirs.paths = mandoc_malloc(argc * sizeof(char *));
405: dirs.sz = argc;
406: for (i = 0; i < argc; i++)
407: dirs.paths[i] = mandoc_strdup(argv[i]);
408: } else
1.11 kristaps 409: manpath_parse(&dirs, NULL, NULL);
1.10 kristaps 410:
411: for (i = 0; i < dirs.sz; i++) {
1.5 kristaps 412: ibuf[0] = fbuf[0] = '\0';
1.1 kristaps 413:
1.10 kristaps 414: strlcat(fbuf, dirs.paths[i], MAXPATHLEN);
1.5 kristaps 415: strlcat(fbuf, "/", MAXPATHLEN);
416: sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
1.1 kristaps 417:
1.10 kristaps 418: strlcat(ibuf, dirs.paths[i], MAXPATHLEN);
1.5 kristaps 419: strlcat(ibuf, "/", MAXPATHLEN);
420: sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
1.1 kristaps 421:
1.5 kristaps 422: if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) {
1.13 schwarze 423: fprintf(stderr, "%s: Path too long\n",
1.10 kristaps 424: dirs.paths[i]);
1.5 kristaps 425: exit((int)MANDOCLEVEL_BADARG);
1.4 kristaps 426: }
1.3 kristaps 427:
1.13 schwarze 428: if (db)
429: (*db->close)(db);
430: if (idx)
431: (*idx->close)(idx);
432:
1.4 kristaps 433: db = dbopen(fbuf, flags, 0644, DB_BTREE, &info);
434: idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL);
1.3 kristaps 435:
1.4 kristaps 436: if (NULL == db) {
437: perror(fbuf);
1.5 kristaps 438: exit((int)MANDOCLEVEL_SYSERR);
1.12 schwarze 439: } else if (NULL == idx) {
1.4 kristaps 440: perror(ibuf);
1.5 kristaps 441: exit((int)MANDOCLEVEL_SYSERR);
442: }
443:
444: if (verb > 2) {
445: printf("%s: Truncated\n", fbuf);
446: printf("%s: Truncated\n", ibuf);
1.4 kristaps 447: }
1.1 kristaps 448:
1.4 kristaps 449: ofile_free(of);
450: of = NULL;
1.1 kristaps 451:
1.12 schwarze 452: if ( ! ofile_dirbuild(dirs.paths[i], NULL, NULL,
1.14 schwarze 453: 0, use_all, verb, &of))
1.5 kristaps 454: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 455:
1.5 kristaps 456: if (NULL == of)
457: continue;
1.1 kristaps 458:
1.5 kristaps 459: of = of->first;
1.1 kristaps 460:
1.13 schwarze 461: index_merge(of, mp, &dbuf, &buf, hash, db, fbuf,
1.12 schwarze 462: idx, ibuf, use_all, verb,
463: maxrec, recs, reccur);
1.4 kristaps 464: }
1.3 kristaps 465:
1.5 kristaps 466: out:
1.3 kristaps 467: if (db)
468: (*db->close)(db);
469: if (idx)
470: (*idx->close)(idx);
471: if (hash)
472: (*hash->close)(hash);
473: if (mp)
474: mparse_free(mp);
475:
1.10 kristaps 476: manpath_free(&dirs);
1.4 kristaps 477: ofile_free(of);
1.3 kristaps 478: free(buf.cp);
479: free(dbuf.cp);
480: free(recs);
481:
1.5 kristaps 482: return(MANDOCLEVEL_OK);
1.3 kristaps 483: }
484:
485: void
486: index_merge(const struct of *of, struct mparse *mp,
487: struct buf *dbuf, struct buf *buf,
1.13 schwarze 488: DB *hash, DB *db, const char *dbf,
1.12 schwarze 489: DB *idx, const char *idxf, int use_all, int verb,
1.3 kristaps 490: recno_t maxrec, const recno_t *recs, size_t reccur)
491: {
492: recno_t rec;
493: int ch;
494: DBT key, val;
495: struct mdoc *mdoc;
496: struct man *man;
497: const char *fn, *msec, *mtitle, *arch;
498: size_t sv;
499: unsigned seq;
1.9 kristaps 500: struct db_val vbuf;
1.3 kristaps 501:
502: for (rec = 0; of; of = of->next) {
503: fn = of->fname;
1.14 schwarze 504:
505: /*
506: * Reclaim an empty index record, if available.
507: */
508:
1.3 kristaps 509: if (reccur > 0) {
510: --reccur;
511: rec = recs[(int)reccur];
512: } else if (maxrec > 0) {
513: rec = maxrec;
514: maxrec = 0;
1.1 kristaps 515: } else
516: rec++;
517:
518: mparse_reset(mp);
519: hash_reset(&hash);
1.14 schwarze 520: mdoc = NULL;
521: man = NULL;
1.1 kristaps 522:
1.14 schwarze 523: /*
524: * Try interpreting the file as mdoc(7) or man(7)
525: * source code, unless it is already known to be
526: * formatted. Fall back to formatted mode.
527: */
528:
529: if ((MANDOC_SRC & of->src_form ||
530: ! (MANDOC_FORM & of->src_form)) &&
531: MANDOCLEVEL_FATAL > mparse_readfd(mp, -1, fn))
532: mparse_result(mp, &mdoc, &man);
533:
534: if (NULL != mdoc) {
535: msec = mdoc_meta(mdoc)->msec;
536: arch = mdoc_meta(mdoc)->arch;
537: mtitle = mdoc_meta(mdoc)->title;
538: } else if (NULL != man) {
539: msec = man_meta(man)->msec;
540: arch = NULL;
541: mtitle = man_meta(man)->title;
542: } else {
543: msec = of->sec;
544: arch = of->arch;
545: mtitle = of->title;
1.1 kristaps 546: }
547:
1.12 schwarze 548: /*
549: * By default, skip a file if the manual section
550: * and architecture given in the file disagree
551: * with the directory where the file is located.
552: */
553:
554: if (0 == use_all) {
555: assert(of->sec);
556: assert(msec);
557: if (strcmp(msec, of->sec))
558: continue;
559:
560: if (NULL == arch) {
561: if (NULL != of->arch)
562: continue;
563: } else if (NULL == of->arch ||
564: strcmp(arch, of->arch))
565: continue;
566: }
567:
1.1 kristaps 568: if (NULL == arch)
569: arch = "";
570:
571: /*
1.12 schwarze 572: * By default, skip a file if the title given
573: * in the file disagrees with the file name.
574: * If both agree, use the file name as the title,
575: * because the one in the file usually is all caps.
576: */
577:
578: assert(of->title);
579: assert(mtitle);
580:
581: if (0 == strcasecmp(mtitle, of->title))
582: mtitle = of->title;
583: else if (0 == use_all)
584: continue;
585:
586: /*
1.1 kristaps 587: * The index record value consists of a nil-terminated
588: * filename, a nil-terminated manual section, and a
589: * nil-terminated description. Since the description
590: * may not be set, we set a sentinel to see if we're
591: * going to write a nil byte in its place.
592: */
593:
1.3 kristaps 594: dbuf->len = 0;
1.15 ! schwarze 595: buf_append(dbuf, mdoc ? "mdoc" : (man ? "man" : "cat"));
1.3 kristaps 596: buf_appendb(dbuf, fn, strlen(fn) + 1);
597: buf_appendb(dbuf, msec, strlen(msec) + 1);
598: buf_appendb(dbuf, mtitle, strlen(mtitle) + 1);
599: buf_appendb(dbuf, arch, strlen(arch) + 1);
1.1 kristaps 600:
1.3 kristaps 601: sv = dbuf->len;
1.1 kristaps 602:
603: /* Fix the record number in the btree value. */
604:
605: if (mdoc)
1.3 kristaps 606: pmdoc_node(hash, buf, dbuf,
1.1 kristaps 607: mdoc_node(mdoc), mdoc_meta(mdoc));
1.14 schwarze 608: else if (man)
1.3 kristaps 609: pman_node(hash, buf, dbuf, man_node(man));
1.14 schwarze 610: else
611: pformatted(hash, buf, dbuf, of);
1.1 kristaps 612:
613: /*
614: * Copy from the in-memory hashtable of pending keywords
615: * into the database.
616: */
617:
1.9 kristaps 618: vbuf.rec = rec;
1.1 kristaps 619: seq = R_FIRST;
620: while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) {
621: seq = R_NEXT;
622:
1.9 kristaps 623: vbuf.mask = *(uint64_t *)val.data;
624: val.size = sizeof(struct db_val);
625: val.data = &vbuf;
1.1 kristaps 626:
1.5 kristaps 627: if (verb > 1)
628: printf("%s: Added keyword: %s\n",
629: fn, (char *)key.data);
1.3 kristaps 630: dbt_put(db, dbf, &key, &val);
1.1 kristaps 631: }
632: if (ch < 0) {
633: perror("hash");
634: exit((int)MANDOCLEVEL_SYSERR);
635: }
636:
637: /*
638: * Apply to the index. If we haven't had a description
639: * set, put an empty one in now.
640: */
641:
1.3 kristaps 642: if (dbuf->len == sv)
643: buf_appendb(dbuf, "", 1);
1.1 kristaps 644:
645: key.data = &rec;
646: key.size = sizeof(recno_t);
647:
1.3 kristaps 648: val.data = dbuf->cp;
649: val.size = dbuf->len;
1.1 kristaps 650:
1.5 kristaps 651: if (verb)
652: printf("%s: Added index\n", fn);
1.3 kristaps 653: dbt_put(idx, idxf, &key, &val);
654: }
655: }
656:
657: /*
658: * Scan through all entries in the index file `idx' and prune those
659: * entries in `ofile'.
660: * Pruning consists of removing from `db', then invalidating the entry
661: * in `idx' (zeroing its value size).
662: */
663: static void
664: index_prune(const struct of *ofile, DB *db, const char *dbf,
1.5 kristaps 665: DB *idx, const char *idxf, int verb,
1.3 kristaps 666: recno_t *maxrec, recno_t **recs, size_t *recsz)
667: {
668: const struct of *of;
669: const char *fn;
1.9 kristaps 670: struct db_val *vbuf;
1.3 kristaps 671: unsigned seq, sseq;
672: DBT key, val;
673: size_t reccur;
674: int ch;
675:
676: reccur = 0;
677: seq = R_FIRST;
678: while (0 == (ch = (*idx->seq)(idx, &key, &val, seq))) {
679: seq = R_NEXT;
680: *maxrec = *(recno_t *)key.data;
681: if (0 == val.size) {
682: if (reccur >= *recsz) {
683: *recsz += MANDOC_SLOP;
684: *recs = mandoc_realloc(*recs,
685: *recsz * sizeof(recno_t));
686: }
687: (*recs)[(int)reccur] = *maxrec;
688: reccur++;
689: continue;
690: }
691:
692: fn = (char *)val.data;
693: for (of = ofile; of; of = of->next)
694: if (0 == strcmp(fn, of->fname))
695: break;
696:
697: if (NULL == of)
698: continue;
699:
700: sseq = R_FIRST;
701: while (0 == (ch = (*db->seq)(db, &key, &val, sseq))) {
702: sseq = R_NEXT;
1.9 kristaps 703: assert(sizeof(struct db_val) == val.size);
704: vbuf = val.data;
705: if (*maxrec != vbuf->rec)
1.3 kristaps 706: continue;
1.5 kristaps 707: if (verb)
708: printf("%s: Deleted keyword: %s\n",
709: fn, (char *)key.data);
1.3 kristaps 710: ch = (*db->del)(db, &key, R_CURSOR);
711: if (ch < 0)
712: break;
713: }
714: if (ch < 0) {
715: perror(dbf);
716: exit((int)MANDOCLEVEL_SYSERR);
717: }
1.1 kristaps 718:
1.5 kristaps 719: if (verb)
720: printf("%s: Deleted index\n", fn);
1.1 kristaps 721:
1.3 kristaps 722: val.size = 0;
723: ch = (*idx->put)(idx, &key, &val, R_CURSOR);
724: if (ch < 0) {
725: perror(idxf);
726: exit((int)MANDOCLEVEL_SYSERR);
727: }
1.1 kristaps 728:
1.3 kristaps 729: if (reccur >= *recsz) {
730: *recsz += MANDOC_SLOP;
731: *recs = mandoc_realloc
732: (*recs, *recsz * sizeof(recno_t));
733: }
1.1 kristaps 734:
1.3 kristaps 735: (*recs)[(int)reccur] = *maxrec;
736: reccur++;
737: }
738: (*maxrec)++;
1.1 kristaps 739: }
740:
741: /*
742: * Grow the buffer (if necessary) and copy in a binary string.
743: */
744: static void
745: buf_appendb(struct buf *buf, const void *cp, size_t sz)
746: {
747:
748: /* Overshoot by MANDOC_BUFSZ. */
749:
750: while (buf->len + sz >= buf->size) {
751: buf->size = buf->len + sz + MANDOC_BUFSZ;
752: buf->cp = mandoc_realloc(buf->cp, buf->size);
753: }
754:
755: memcpy(buf->cp + (int)buf->len, cp, sz);
756: buf->len += sz;
757: }
758:
759: /*
760: * Append a nil-terminated string to the buffer.
761: * This can be invoked multiple times.
762: * The buffer string will be nil-terminated.
763: * If invoked multiple times, a space is put between strings.
764: */
765: static void
766: buf_append(struct buf *buf, const char *cp)
767: {
768: size_t sz;
769:
770: if (0 == (sz = strlen(cp)))
771: return;
772:
773: if (buf->len)
774: buf->cp[(int)buf->len - 1] = ' ';
775:
776: buf_appendb(buf, cp, sz + 1);
777: }
778:
779: /*
780: * Recursively add all text from a given node.
781: * This is optimised for general mdoc nodes in this context, which do
782: * not consist of subexpressions and having a recursive call for n->next
783: * would be wasteful.
784: * The "f" variable should be 0 unless called from pmdoc_Nd for the
785: * description buffer, which does not start at the beginning of the
786: * buffer.
787: */
788: static void
789: buf_appendmdoc(struct buf *buf, const struct mdoc_node *n, int f)
790: {
791:
792: for ( ; n; n = n->next) {
793: if (n->child)
794: buf_appendmdoc(buf, n->child, f);
795:
796: if (MDOC_TEXT == n->type && f) {
797: f = 0;
798: buf_appendb(buf, n->string,
799: strlen(n->string) + 1);
800: } else if (MDOC_TEXT == n->type)
801: buf_append(buf, n->string);
802:
803: }
804: }
805:
806: /* ARGSUSED */
807: static void
808: pmdoc_An(MDOC_ARGS)
809: {
810:
811: if (SEC_AUTHORS != n->sec)
812: return;
813:
814: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 815: hash_put(hash, buf, TYPE_An);
1.1 kristaps 816: }
817:
818: static void
819: hash_reset(DB **db)
820: {
821: DB *hash;
822:
823: if (NULL != (hash = *db))
824: (*hash->close)(hash);
825:
1.5 kristaps 826: *db = dbopen(NULL, O_CREAT|O_RDWR, 0644, DB_HASH, NULL);
1.1 kristaps 827: if (NULL == *db) {
828: perror("hash");
829: exit((int)MANDOCLEVEL_SYSERR);
830: }
831: }
832:
833: /* ARGSUSED */
834: static void
835: pmdoc_Fd(MDOC_ARGS)
836: {
837: const char *start, *end;
838: size_t sz;
839:
840: if (SEC_SYNOPSIS != n->sec)
841: return;
842: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
843: return;
844:
845: /*
846: * Only consider those `Fd' macro fields that begin with an
847: * "inclusion" token (versus, e.g., #define).
848: */
849: if (strcmp("#include", n->string))
850: return;
851:
852: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
853: return;
854:
855: /*
856: * Strip away the enclosing angle brackets and make sure we're
857: * not zero-length.
858: */
859:
860: start = n->string;
861: if ('<' == *start || '"' == *start)
862: start++;
863:
864: if (0 == (sz = strlen(start)))
865: return;
866:
867: end = &start[(int)sz - 1];
868: if ('>' == *end || '"' == *end)
869: end--;
870:
871: assert(end >= start);
872:
873: buf_appendb(buf, start, (size_t)(end - start + 1));
874: buf_appendb(buf, "", 1);
875:
1.8 schwarze 876: hash_put(hash, buf, TYPE_In);
1.1 kristaps 877: }
878:
879: /* ARGSUSED */
880: static void
881: pmdoc_Cd(MDOC_ARGS)
882: {
883:
884: if (SEC_SYNOPSIS != n->sec)
885: return;
886:
887: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 888: hash_put(hash, buf, TYPE_Cd);
1.1 kristaps 889: }
890:
891: /* ARGSUSED */
892: static void
893: pmdoc_In(MDOC_ARGS)
894: {
895:
896: if (SEC_SYNOPSIS != n->sec)
897: return;
898: if (NULL == n->child || MDOC_TEXT != n->child->type)
899: return;
900:
901: buf_append(buf, n->child->string);
1.8 schwarze 902: hash_put(hash, buf, TYPE_In);
1.1 kristaps 903: }
904:
905: /* ARGSUSED */
906: static void
907: pmdoc_Fn(MDOC_ARGS)
908: {
909: const char *cp;
910:
911: if (SEC_SYNOPSIS != n->sec)
912: return;
913: if (NULL == n->child || MDOC_TEXT != n->child->type)
914: return;
915:
916: /* .Fn "struct type *arg" "foo" */
917:
918: cp = strrchr(n->child->string, ' ');
919: if (NULL == cp)
920: cp = n->child->string;
921:
922: /* Strip away pointer symbol. */
923:
924: while ('*' == *cp)
925: cp++;
926:
927: buf_append(buf, cp);
1.8 schwarze 928: hash_put(hash, buf, TYPE_Fn);
1.1 kristaps 929: }
930:
931: /* ARGSUSED */
932: static void
933: pmdoc_St(MDOC_ARGS)
934: {
935:
936: if (SEC_STANDARDS != n->sec)
937: return;
938: if (NULL == n->child || MDOC_TEXT != n->child->type)
939: return;
940:
941: buf_append(buf, n->child->string);
1.8 schwarze 942: hash_put(hash, buf, TYPE_St);
1.1 kristaps 943: }
944:
945: /* ARGSUSED */
946: static void
947: pmdoc_Xr(MDOC_ARGS)
948: {
949:
950: if (NULL == (n = n->child))
951: return;
952:
953: buf_appendb(buf, n->string, strlen(n->string));
954:
955: if (NULL != (n = n->next)) {
956: buf_appendb(buf, ".", 1);
957: buf_appendb(buf, n->string, strlen(n->string) + 1);
958: } else
959: buf_appendb(buf, ".", 2);
960:
1.8 schwarze 961: hash_put(hash, buf, TYPE_Xr);
1.1 kristaps 962: }
963:
964: /* ARGSUSED */
965: static void
966: pmdoc_Vt(MDOC_ARGS)
967: {
968: const char *start;
969: size_t sz;
970:
971: if (SEC_SYNOPSIS != n->sec)
972: return;
973: if (MDOC_Vt == n->tok && MDOC_BODY != n->type)
974: return;
975: if (NULL == n->last || MDOC_TEXT != n->last->type)
976: return;
977:
978: /*
979: * Strip away leading pointer symbol '*' and trailing ';'.
980: */
981:
982: start = n->last->string;
983:
984: while ('*' == *start)
985: start++;
986:
987: if (0 == (sz = strlen(start)))
988: return;
989:
990: if (';' == start[(int)sz - 1])
991: sz--;
992:
993: if (0 == sz)
994: return;
995:
996: buf_appendb(buf, start, sz);
997: buf_appendb(buf, "", 1);
1.8 schwarze 998: hash_put(hash, buf, TYPE_Va);
1.1 kristaps 999: }
1000:
1001: /* ARGSUSED */
1002: static void
1003: pmdoc_Fo(MDOC_ARGS)
1004: {
1005:
1006: if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
1007: return;
1008: if (NULL == n->child || MDOC_TEXT != n->child->type)
1009: return;
1010:
1011: buf_append(buf, n->child->string);
1.8 schwarze 1012: hash_put(hash, buf, TYPE_Fn);
1.1 kristaps 1013: }
1014:
1015:
1016: /* ARGSUSED */
1017: static void
1018: pmdoc_Nd(MDOC_ARGS)
1019: {
1020:
1021: if (MDOC_BODY != n->type)
1022: return;
1023:
1024: buf_appendmdoc(dbuf, n->child, 1);
1025: buf_appendmdoc(buf, n->child, 0);
1026:
1.8 schwarze 1027: hash_put(hash, buf, TYPE_Nd);
1.1 kristaps 1028: }
1029:
1030: /* ARGSUSED */
1031: static void
1032: pmdoc_Er(MDOC_ARGS)
1033: {
1034:
1035: if (SEC_ERRORS != n->sec)
1036: return;
1037:
1038: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 1039: hash_put(hash, buf, TYPE_Er);
1.1 kristaps 1040: }
1041:
1042: /* ARGSUSED */
1043: static void
1044: pmdoc_Ev(MDOC_ARGS)
1045: {
1046:
1047: if (SEC_ENVIRONMENT != n->sec)
1048: return;
1049:
1050: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 1051: hash_put(hash, buf, TYPE_Ev);
1.1 kristaps 1052: }
1053:
1054: /* ARGSUSED */
1055: static void
1056: pmdoc_Pa(MDOC_ARGS)
1057: {
1058:
1059: if (SEC_FILES != n->sec)
1060: return;
1061:
1062: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 1063: hash_put(hash, buf, TYPE_Pa);
1.1 kristaps 1064: }
1065:
1066: /* ARGSUSED */
1067: static void
1068: pmdoc_Nm(MDOC_ARGS)
1069: {
1070:
1071: if (SEC_NAME == n->sec) {
1072: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 1073: hash_put(hash, buf, TYPE_Nm);
1.1 kristaps 1074: return;
1075: } else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
1076: return;
1077:
1078: if (NULL == n->child)
1079: buf_append(buf, m->name);
1080:
1081: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 1082: hash_put(hash, buf, TYPE_Nm);
1.1 kristaps 1083: }
1084:
1085: static void
1.9 kristaps 1086: hash_put(DB *db, const struct buf *buf, uint64_t mask)
1.1 kristaps 1087: {
1088: DBT key, val;
1089: int rc;
1090:
1091: if (buf->len < 2)
1092: return;
1093:
1094: key.data = buf->cp;
1095: key.size = buf->len;
1096:
1097: if ((rc = (*db->get)(db, &key, &val, 0)) < 0) {
1098: perror("hash");
1099: exit((int)MANDOCLEVEL_SYSERR);
1100: } else if (0 == rc)
1.9 kristaps 1101: mask |= *(uint64_t *)val.data;
1.1 kristaps 1102:
1103: val.data = &mask;
1.9 kristaps 1104: val.size = sizeof(uint64_t);
1.1 kristaps 1105:
1106: if ((rc = (*db->put)(db, &key, &val, 0)) < 0) {
1107: perror("hash");
1108: exit((int)MANDOCLEVEL_SYSERR);
1109: }
1110: }
1111:
1112: static void
1113: dbt_put(DB *db, const char *dbn, DBT *key, DBT *val)
1114: {
1115:
1116: assert(key->size);
1117: assert(val->size);
1118:
1119: if (0 == (*db->put)(db, key, val, 0))
1120: return;
1121:
1122: perror(dbn);
1123: exit((int)MANDOCLEVEL_SYSERR);
1124: /* NOTREACHED */
1125: }
1126:
1127: /*
1128: * Call out to per-macro handlers after clearing the persistent database
1129: * key. If the macro sets the database key, flush it to the database.
1130: */
1131: static void
1132: pmdoc_node(MDOC_ARGS)
1133: {
1134:
1135: if (NULL == n)
1136: return;
1137:
1138: switch (n->type) {
1139: case (MDOC_HEAD):
1140: /* FALLTHROUGH */
1141: case (MDOC_BODY):
1142: /* FALLTHROUGH */
1143: case (MDOC_TAIL):
1144: /* FALLTHROUGH */
1145: case (MDOC_BLOCK):
1146: /* FALLTHROUGH */
1147: case (MDOC_ELEM):
1148: if (NULL == mdocs[n->tok])
1149: break;
1150:
1151: buf->len = 0;
1152: (*mdocs[n->tok])(hash, buf, dbuf, n, m);
1153: break;
1154: default:
1155: break;
1156: }
1157:
1158: pmdoc_node(hash, buf, dbuf, n->child, m);
1159: pmdoc_node(hash, buf, dbuf, n->next, m);
1160: }
1161:
1162: static int
1163: pman_node(MAN_ARGS)
1164: {
1165: const struct man_node *head, *body;
1166: const char *start, *sv;
1167: size_t sz;
1168:
1169: if (NULL == n)
1170: return(0);
1171:
1172: /*
1173: * We're only searching for one thing: the first text child in
1174: * the BODY of a NAME section. Since we don't keep track of
1175: * sections in -man, run some hoops to find out whether we're in
1176: * the correct section or not.
1177: */
1178:
1179: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1180: body = n;
1181: assert(body->parent);
1182: if (NULL != (head = body->parent->head) &&
1183: 1 == head->nchild &&
1184: NULL != (head = (head->child)) &&
1185: MAN_TEXT == head->type &&
1186: 0 == strcmp(head->string, "NAME") &&
1187: NULL != (body = body->child) &&
1188: MAN_TEXT == body->type) {
1189:
1190: assert(body->string);
1191: start = sv = body->string;
1192:
1193: /*
1194: * Go through a special heuristic dance here.
1195: * This is why -man manuals are great!
1196: * (I'm being sarcastic: my eyes are bleeding.)
1197: * Conventionally, one or more manual names are
1198: * comma-specified prior to a whitespace, then a
1199: * dash, then a description. Try to puzzle out
1200: * the name parts here.
1201: */
1202:
1203: for ( ;; ) {
1204: sz = strcspn(start, " ,");
1205: if ('\0' == start[(int)sz])
1206: break;
1207:
1208: buf->len = 0;
1209: buf_appendb(buf, start, sz);
1210: buf_appendb(buf, "", 1);
1211:
1.8 schwarze 1212: hash_put(hash, buf, TYPE_Nm);
1.1 kristaps 1213:
1214: if (' ' == start[(int)sz]) {
1215: start += (int)sz + 1;
1216: break;
1217: }
1218:
1219: assert(',' == start[(int)sz]);
1220: start += (int)sz + 1;
1221: while (' ' == *start)
1222: start++;
1223: }
1224:
1225: buf->len = 0;
1226:
1227: if (sv == start) {
1228: buf_append(buf, start);
1229: return(1);
1230: }
1231:
1232: while (' ' == *start)
1233: start++;
1234:
1235: if (0 == strncmp(start, "-", 1))
1236: start += 1;
1237: else if (0 == strncmp(start, "\\-", 2))
1238: start += 2;
1239: else if (0 == strncmp(start, "\\(en", 4))
1240: start += 4;
1241: else if (0 == strncmp(start, "\\(em", 4))
1242: start += 4;
1243:
1244: while (' ' == *start)
1245: start++;
1246:
1247: sz = strlen(start) + 1;
1248: buf_appendb(dbuf, start, sz);
1249: buf_appendb(buf, start, sz);
1250:
1.8 schwarze 1251: hash_put(hash, buf, TYPE_Nd);
1.1 kristaps 1252: }
1253: }
1254:
1.7 schwarze 1255: for (n = n->child; n; n = n->next)
1256: if (pman_node(hash, buf, dbuf, n))
1257: return(1);
1.1 kristaps 1258:
1259: return(0);
1260: }
1261:
1.14 schwarze 1262: /*
1263: * Parse a formatted manual page.
1264: * By necessity, this involves rather crude guesswork.
1265: */
1266: static void
1267: pformatted(DB *hash, struct buf *buf, struct buf *dbuf,
1268: const struct of *of)
1269: {
1270: FILE *stream;
1271: char *line, *p;
1272: size_t len, plen;
1273:
1274: if (NULL == (stream = fopen(of->fname, "r"))) {
1275: perror(of->fname);
1276: return;
1277: }
1278:
1279: /*
1280: * Always use the title derived from the filename up front,
1281: * do not even try to find it in the file. This also makes
1282: * sure we don't end up with an orphan index record, even if
1283: * the file content turns out to be completely unintelligible.
1284: */
1285:
1286: buf->len = 0;
1287: buf_append(buf, of->title);
1288: hash_put(hash, buf, TYPE_Nm);
1289:
1290: while (NULL != (line = fgetln(stream, &len)) && '\n' != *line)
1291: /* Skip to first blank line. */ ;
1292:
1293: while (NULL != (line = fgetln(stream, &len)) &&
1294: ('\n' == *line || ' ' == *line))
1295: /* Skip to first section header. */ ;
1296:
1297: /*
1298: * If no page content can be found,
1299: * reuse the page title as the page description.
1300: */
1301:
1302: if (NULL == (line = fgetln(stream, &len))) {
1303: buf_appendb(dbuf, buf->cp, buf->size);
1304: hash_put(hash, buf, TYPE_Nd);
1305: fclose(stream);
1306: return;
1307: }
1308: fclose(stream);
1309:
1310: /*
1311: * If there is a dash, skip to the text following it.
1312: */
1313:
1314: for (p = line, plen = len; plen; p++, plen--)
1315: if ('-' == *p)
1316: break;
1317: for ( ; plen; p++, plen--)
1318: if ('-' != *p && ' ' != *p && 8 != *p)
1319: break;
1320: if (0 == plen) {
1321: p = line;
1322: plen = len;
1323: }
1324:
1325: /*
1326: * Copy the rest of the line, but no more than 70 bytes.
1327: */
1328:
1329: if (70 < plen)
1330: plen = 70;
1331: p[plen-1] = '\0';
1332: buf_appendb(dbuf, p, plen);
1333: buf->len = 0;
1334: buf_appendb(buf, p, plen);
1335: hash_put(hash, buf, TYPE_Nd);
1336: }
1337:
1.5 kristaps 1338: static void
1.12 schwarze 1339: ofile_argbuild(char *argv[], int argc, int use_all, int verb,
1340: struct of **of)
1.5 kristaps 1341: {
1.12 schwarze 1342: char buf[MAXPATHLEN];
1343: char *sec, *arch, *title, *p;
1.14 schwarze 1344: int i, src_form;
1.5 kristaps 1345: struct of *nof;
1346:
1347: for (i = 0; i < argc; i++) {
1.12 schwarze 1348:
1349: /*
1350: * Try to infer the manual section, architecture and
1351: * page title from the path, assuming it looks like
1.14 schwarze 1352: * man*[/<arch>]/<title>.<section> or
1353: * cat<section>[/<arch>]/<title>.0
1.12 schwarze 1354: */
1355:
1356: if (strlcpy(buf, argv[i], sizeof(buf)) >= sizeof(buf)) {
1357: fprintf(stderr, "%s: Path too long\n", argv[i]);
1358: continue;
1359: }
1360: sec = arch = title = NULL;
1.14 schwarze 1361: src_form = 0;
1.12 schwarze 1362: p = strrchr(buf, '\0');
1363: while (p-- > buf) {
1364: if (NULL == sec && '.' == *p) {
1365: sec = p + 1;
1366: *p = '\0';
1.14 schwarze 1367: if ('0' == *sec)
1368: src_form |= MANDOC_FORM;
1369: else if ('1' <= *sec && '9' >= *sec)
1370: src_form |= MANDOC_SRC;
1.12 schwarze 1371: continue;
1372: }
1373: if ('/' != *p)
1374: continue;
1375: if (NULL == title) {
1376: title = p + 1;
1377: *p = '\0';
1378: continue;
1379: }
1.14 schwarze 1380: if (strncmp("man", p + 1, 3)) {
1381: src_form |= MANDOC_SRC;
1382: arch = p + 1;
1383: } else if (strncmp("cat", p + 1, 3)) {
1384: src_form |= MANDOC_FORM;
1.12 schwarze 1385: arch = p + 1;
1.14 schwarze 1386: }
1.12 schwarze 1387: break;
1388: }
1389: if (NULL == title)
1390: title = buf;
1391:
1392: /*
1393: * Build the file structure.
1394: */
1395:
1.5 kristaps 1396: nof = mandoc_calloc(1, sizeof(struct of));
1.12 schwarze 1397: nof->fname = mandoc_strdup(argv[i]);
1398: if (NULL != sec)
1399: nof->sec = mandoc_strdup(sec);
1400: if (NULL != arch)
1401: nof->arch = mandoc_strdup(arch);
1402: nof->title = mandoc_strdup(title);
1.14 schwarze 1403: nof->src_form = src_form;
1.12 schwarze 1404:
1405: /*
1406: * Add the structure to the list.
1407: */
1408:
1.5 kristaps 1409: if (verb > 2)
1410: printf("%s: Scheduling\n", argv[i]);
1411: if (NULL == *of) {
1412: *of = nof;
1413: (*of)->first = nof;
1414: } else {
1415: nof->first = (*of)->first;
1416: (*of)->next = nof;
1417: *of = nof;
1418: }
1419: }
1420: }
1421:
1.4 kristaps 1422: /*
1423: * Recursively build up a list of files to parse.
1424: * We use this instead of ftw() and so on because I don't want global
1425: * variables hanging around.
1426: * This ignores the mandoc.db and mandoc.index files, but assumes that
1427: * everything else is a manual.
1428: * Pass in a pointer to a NULL structure for the first invocation.
1429: */
1430: static int
1.12 schwarze 1431: ofile_dirbuild(const char *dir, const char* psec, const char *parch,
1.14 schwarze 1432: int p_src_form, int use_all, int verb, struct of **of)
1.4 kristaps 1433: {
1.5 kristaps 1434: char buf[MAXPATHLEN];
1.14 schwarze 1435: struct stat sb;
1.5 kristaps 1436: size_t sz;
1.4 kristaps 1437: DIR *d;
1.12 schwarze 1438: const char *fn, *sec, *arch;
1.14 schwarze 1439: char *p, *q, *suffix;
1.4 kristaps 1440: struct of *nof;
1441: struct dirent *dp;
1.14 schwarze 1442: int src_form;
1.4 kristaps 1443:
1444: if (NULL == (d = opendir(dir))) {
1445: perror(dir);
1446: return(0);
1447: }
1448:
1449: while (NULL != (dp = readdir(d))) {
1450: fn = dp->d_name;
1.12 schwarze 1451:
1452: if ('.' == *fn)
1453: continue;
1454:
1.14 schwarze 1455: src_form = p_src_form;
1456:
1.4 kristaps 1457: if (DT_DIR == dp->d_type) {
1.12 schwarze 1458: sec = psec;
1459: arch = parch;
1460:
1461: /*
1462: * By default, only use directories called:
1.14 schwarze 1463: * man<section>/[<arch>/] or
1464: * cat<section>/[<arch>/]
1.12 schwarze 1465: */
1466:
1467: if (NULL == sec) {
1.14 schwarze 1468: if(0 == strncmp("man", fn, 3)) {
1469: src_form |= MANDOC_SRC;
1.12 schwarze 1470: sec = fn + 3;
1.14 schwarze 1471: } else if (0 == strncmp("cat", fn, 3)) {
1472: src_form |= MANDOC_FORM;
1473: sec = fn + 3;
1474: } else if (use_all)
1.12 schwarze 1475: sec = fn;
1476: else
1477: continue;
1478: } else if (NULL == arch && (use_all ||
1479: NULL == strchr(fn, '.')))
1480: arch = fn;
1481: else if (0 == use_all)
1.5 kristaps 1482: continue;
1483:
1484: buf[0] = '\0';
1485: strlcat(buf, dir, MAXPATHLEN);
1486: strlcat(buf, "/", MAXPATHLEN);
1487: sz = strlcat(buf, fn, MAXPATHLEN);
1488:
1.12 schwarze 1489: if (MAXPATHLEN <= sz) {
1490: fprintf(stderr, "%s: Path too long\n", dir);
1491: return(0);
1492: }
1493:
1494: if (verb > 2)
1495: printf("%s: Scanning\n", buf);
1496:
1497: if ( ! ofile_dirbuild(buf, sec, arch,
1.14 schwarze 1498: src_form, use_all, verb, of))
1.12 schwarze 1499: return(0);
1500: }
1501: if (DT_REG != dp->d_type ||
1502: (NULL == psec && !use_all) ||
1503: !strcmp(MANDOC_DB, fn) ||
1504: !strcmp(MANDOC_IDX, fn))
1505: continue;
1506:
1507: /*
1508: * By default, skip files where the file name suffix
1509: * does not agree with the section directory
1510: * they are located in.
1511: */
1512:
1513: suffix = strrchr(fn, '.');
1514: if (0 == use_all) {
1515: if (NULL == suffix)
1.5 kristaps 1516: continue;
1.14 schwarze 1517: if ((MANDOC_SRC & src_form &&
1518: strcmp(suffix + 1, psec)) ||
1519: (MANDOC_FORM & src_form &&
1520: strcmp(suffix + 1, "0")))
1521: continue;
1522: }
1523: if (NULL != suffix) {
1524: if ('0' == suffix[1])
1525: src_form |= MANDOC_FORM;
1526: else if ('1' <= suffix[1] && '9' >= suffix[1])
1527: src_form |= MANDOC_SRC;
1528: }
1529:
1530:
1531: /*
1532: * Skip formatted manuals if a source version is
1533: * available. Ignore the age: it is very unlikely
1534: * that people install newer formatted base manuals
1535: * when they used to have source manuals before,
1536: * and in ports, old manuals get removed on update.
1537: */
1538: if (0 == use_all && MANDOC_FORM & src_form &&
1539: NULL != psec) {
1540: buf[0] = '\0';
1541: strlcat(buf, dir, MAXPATHLEN);
1542: p = strrchr(buf, '/');
1543: if (NULL == p)
1544: p = buf;
1545: else
1546: p++;
1547: if (0 == strncmp("cat", p, 3))
1548: memcpy(p, "man", 3);
1549: strlcat(buf, "/", MAXPATHLEN);
1550: sz = strlcat(buf, fn, MAXPATHLEN);
1551: if (sz >= MAXPATHLEN) {
1552: fprintf(stderr, "%s: Path too long\n", buf);
1.5 kristaps 1553: continue;
1.14 schwarze 1554: }
1555: q = strrchr(buf, '.');
1556: if (NULL != q && p < q++) {
1557: *q = '\0';
1558: sz = strlcat(buf, psec, MAXPATHLEN);
1559: if (sz >= MAXPATHLEN) {
1560: fprintf(stderr,
1561: "%s: Path too long\n", buf);
1562: continue;
1563: }
1564: if (0 == stat(buf, &sb))
1565: continue;
1566: }
1.5 kristaps 1567: }
1.4 kristaps 1568:
1.5 kristaps 1569: buf[0] = '\0';
1570: strlcat(buf, dir, MAXPATHLEN);
1571: strlcat(buf, "/", MAXPATHLEN);
1.6 schwarze 1572: sz = strlcat(buf, fn, MAXPATHLEN);
1.5 kristaps 1573: if (sz >= MAXPATHLEN) {
1574: fprintf(stderr, "%s: Path too long\n", dir);
1.14 schwarze 1575: continue;
1.5 kristaps 1576: }
1577:
1.4 kristaps 1578: nof = mandoc_calloc(1, sizeof(struct of));
1.5 kristaps 1579: nof->fname = mandoc_strdup(buf);
1.12 schwarze 1580: if (NULL != psec)
1581: nof->sec = mandoc_strdup(psec);
1582: if (NULL != parch)
1583: nof->arch = mandoc_strdup(parch);
1.14 schwarze 1584: nof->src_form = src_form;
1.12 schwarze 1585:
1586: /*
1587: * Remember the file name without the extension,
1588: * to be used as the page title in the database.
1589: */
1590:
1591: if (NULL != suffix)
1592: *suffix = '\0';
1593: nof->title = mandoc_strdup(fn);
1.5 kristaps 1594:
1.14 schwarze 1595: /*
1596: * Add the structure to the list.
1597: */
1598:
1.5 kristaps 1599: if (verb > 2)
1600: printf("%s: Scheduling\n", buf);
1.4 kristaps 1601: if (NULL == *of) {
1602: *of = nof;
1603: (*of)->first = nof;
1604: } else {
1.5 kristaps 1605: nof->first = (*of)->first;
1.4 kristaps 1606: (*of)->next = nof;
1607: *of = nof;
1608: }
1609: }
1610:
1.7 schwarze 1611: closedir(d);
1.4 kristaps 1612: return(1);
1613: }
1614:
1615: static void
1616: ofile_free(struct of *of)
1617: {
1618: struct of *nof;
1619:
1620: while (of) {
1621: nof = of->next;
1622: free(of->fname);
1.12 schwarze 1623: free(of->sec);
1624: free(of->arch);
1625: free(of->title);
1.4 kristaps 1626: free(of);
1627: of = nof;
1628: }
1629: }
1630:
1.1 kristaps 1631: static void
1632: usage(void)
1633: {
1634:
1.5 kristaps 1635: fprintf(stderr, "usage: %s [-v] "
1636: "[-d dir [files...] |"
1637: " -u dir [files...] |"
1638: " dir...]\n", progname);
1.1 kristaps 1639: }
CVSweb