Annotation of mandoc/mandocdb.c, Revision 1.16
1.16 ! schwarze 1: /* $Id: mandocdb.c,v 1.15 2011/11/27 23:11:37 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 *,
1.16 ! schwarze 100: struct buf *, struct buf *, DB *,
! 101: DB *, const char *, DB *, const char *,
1.3 kristaps 102: recno_t, const recno_t *, size_t);
103: static void index_prune(const struct of *, DB *,
104: const char *, DB *, const char *,
1.16 ! schwarze 105: recno_t *, recno_t **, size_t *);
! 106: static void ofile_argbuild(int, char *[], struct of **);
1.12 schwarze 107: static int ofile_dirbuild(const char *, const char *,
1.16 ! schwarze 108: const char *, int, struct of **);
1.4 kristaps 109: static void ofile_free(struct of *);
1.14 schwarze 110: static void pformatted(DB *, struct buf *, struct buf *,
111: const struct of *);
1.1 kristaps 112: static int pman_node(MAN_ARGS);
113: static void pmdoc_node(MDOC_ARGS);
114: static void pmdoc_An(MDOC_ARGS);
115: static void pmdoc_Cd(MDOC_ARGS);
116: static void pmdoc_Er(MDOC_ARGS);
117: static void pmdoc_Ev(MDOC_ARGS);
118: static void pmdoc_Fd(MDOC_ARGS);
119: static void pmdoc_In(MDOC_ARGS);
120: static void pmdoc_Fn(MDOC_ARGS);
121: static void pmdoc_Fo(MDOC_ARGS);
122: static void pmdoc_Nd(MDOC_ARGS);
123: static void pmdoc_Nm(MDOC_ARGS);
124: static void pmdoc_Pa(MDOC_ARGS);
125: static void pmdoc_St(MDOC_ARGS);
126: static void pmdoc_Vt(MDOC_ARGS);
127: static void pmdoc_Xr(MDOC_ARGS);
128: static void usage(void);
129:
130: typedef void (*pmdoc_nf)(MDOC_ARGS);
131:
132: static const pmdoc_nf mdocs[MDOC_MAX] = {
133: NULL, /* Ap */
134: NULL, /* Dd */
135: NULL, /* Dt */
136: NULL, /* Os */
137: NULL, /* Sh */
138: NULL, /* Ss */
139: NULL, /* Pp */
140: NULL, /* D1 */
141: NULL, /* Dl */
142: NULL, /* Bd */
143: NULL, /* Ed */
144: NULL, /* Bl */
145: NULL, /* El */
146: NULL, /* It */
147: NULL, /* Ad */
148: pmdoc_An, /* An */
149: NULL, /* Ar */
150: pmdoc_Cd, /* Cd */
151: NULL, /* Cm */
152: NULL, /* Dv */
153: pmdoc_Er, /* Er */
154: pmdoc_Ev, /* Ev */
155: NULL, /* Ex */
156: NULL, /* Fa */
157: pmdoc_Fd, /* Fd */
158: NULL, /* Fl */
159: pmdoc_Fn, /* Fn */
160: NULL, /* Ft */
161: NULL, /* Ic */
162: pmdoc_In, /* In */
163: NULL, /* Li */
164: pmdoc_Nd, /* Nd */
165: pmdoc_Nm, /* Nm */
166: NULL, /* Op */
167: NULL, /* Ot */
168: pmdoc_Pa, /* Pa */
169: NULL, /* Rv */
170: pmdoc_St, /* St */
171: pmdoc_Vt, /* Va */
172: pmdoc_Vt, /* Vt */
173: pmdoc_Xr, /* Xr */
174: NULL, /* %A */
175: NULL, /* %B */
176: NULL, /* %D */
177: NULL, /* %I */
178: NULL, /* %J */
179: NULL, /* %N */
180: NULL, /* %O */
181: NULL, /* %P */
182: NULL, /* %R */
183: NULL, /* %T */
184: NULL, /* %V */
185: NULL, /* Ac */
186: NULL, /* Ao */
187: NULL, /* Aq */
188: NULL, /* At */
189: NULL, /* Bc */
190: NULL, /* Bf */
191: NULL, /* Bo */
192: NULL, /* Bq */
193: NULL, /* Bsx */
194: NULL, /* Bx */
195: NULL, /* Db */
196: NULL, /* Dc */
197: NULL, /* Do */
198: NULL, /* Dq */
199: NULL, /* Ec */
200: NULL, /* Ef */
201: NULL, /* Em */
202: NULL, /* Eo */
203: NULL, /* Fx */
204: NULL, /* Ms */
205: NULL, /* No */
206: NULL, /* Ns */
207: NULL, /* Nx */
208: NULL, /* Ox */
209: NULL, /* Pc */
210: NULL, /* Pf */
211: NULL, /* Po */
212: NULL, /* Pq */
213: NULL, /* Qc */
214: NULL, /* Ql */
215: NULL, /* Qo */
216: NULL, /* Qq */
217: NULL, /* Re */
218: NULL, /* Rs */
219: NULL, /* Sc */
220: NULL, /* So */
221: NULL, /* Sq */
222: NULL, /* Sm */
223: NULL, /* Sx */
224: NULL, /* Sy */
225: NULL, /* Tn */
226: NULL, /* Ux */
227: NULL, /* Xc */
228: NULL, /* Xo */
229: pmdoc_Fo, /* Fo */
230: NULL, /* Fc */
231: NULL, /* Oo */
232: NULL, /* Oc */
233: NULL, /* Bk */
234: NULL, /* Ek */
235: NULL, /* Bt */
236: NULL, /* Hf */
237: NULL, /* Fr */
238: NULL, /* Ud */
239: NULL, /* Lb */
240: NULL, /* Lp */
241: NULL, /* Lk */
242: NULL, /* Mt */
243: NULL, /* Brq */
244: NULL, /* Bro */
245: NULL, /* Brc */
246: NULL, /* %C */
247: NULL, /* Es */
248: NULL, /* En */
249: NULL, /* Dx */
250: NULL, /* %Q */
251: NULL, /* br */
252: NULL, /* sp */
253: NULL, /* %U */
254: NULL, /* Ta */
255: };
256:
257: static const char *progname;
1.16 ! schwarze 258: static int use_all; /* Use all directories and files. */
! 259: static int verb; /* Output verbosity level. */
1.1 kristaps 260:
261: int
262: main(int argc, char *argv[])
263: {
264: struct mparse *mp; /* parse sequence */
1.10 kristaps 265: struct manpaths dirs;
1.1 kristaps 266: enum op op; /* current operation */
1.5 kristaps 267: const char *dir;
1.1 kristaps 268: char ibuf[MAXPATHLEN], /* index fname */
1.3 kristaps 269: fbuf[MAXPATHLEN]; /* btree fname */
1.16 ! schwarze 270: int ch, i, flags;
1.1 kristaps 271: DB *idx, /* index database */
272: *db, /* keyword database */
273: *hash; /* temporary keyword hashtable */
274: BTREEINFO info; /* btree configuration */
1.12 schwarze 275: recno_t maxrec; /* last record number in the index */
276: recno_t *recs; /* the numbers of all empty records */
1.5 kristaps 277: size_t sz1, sz2,
1.12 schwarze 278: recsz, /* number of allocated slots in recs */
279: reccur; /* current number of empty records */
1.1 kristaps 280: struct buf buf, /* keyword buffer */
281: dbuf; /* description buffer */
1.5 kristaps 282: struct of *of; /* list of files for processing */
1.1 kristaps 283: extern int optind;
284: extern char *optarg;
285:
286: progname = strrchr(argv[0], '/');
287: if (progname == NULL)
288: progname = argv[0];
289: else
290: ++progname;
291:
1.10 kristaps 292: memset(&dirs, 0, sizeof(struct manpaths));
293:
1.5 kristaps 294: verb = 0;
1.12 schwarze 295: use_all = 0;
1.4 kristaps 296: of = NULL;
1.1 kristaps 297: db = idx = NULL;
298: mp = NULL;
299: hash = NULL;
300: recs = NULL;
301: recsz = reccur = 0;
302: maxrec = 0;
303: op = OP_NEW;
1.5 kristaps 304: dir = NULL;
1.1 kristaps 305:
1.12 schwarze 306: while (-1 != (ch = getopt(argc, argv, "ad:u:v")))
1.1 kristaps 307: switch (ch) {
1.12 schwarze 308: case ('a'):
309: use_all = 1;
310: break;
1.5 kristaps 311: case ('d'):
312: dir = optarg;
313: op = OP_UPDATE;
314: break;
315: case ('u'):
316: dir = optarg;
317: op = OP_DELETE;
318: break;
319: case ('v'):
320: verb++;
321: break;
1.1 kristaps 322: default:
323: usage();
324: return((int)MANDOCLEVEL_BADARG);
325: }
326:
327: argc -= optind;
328: argv += optind;
329:
1.4 kristaps 330: memset(&info, 0, sizeof(BTREEINFO));
331: info.flags = R_DUP;
1.1 kristaps 332:
1.4 kristaps 333: mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL);
1.1 kristaps 334:
1.5 kristaps 335: memset(&buf, 0, sizeof(struct buf));
336: memset(&dbuf, 0, sizeof(struct buf));
1.1 kristaps 337:
1.4 kristaps 338: buf.size = dbuf.size = MANDOC_BUFSZ;
1.1 kristaps 339:
1.4 kristaps 340: buf.cp = mandoc_malloc(buf.size);
341: dbuf.cp = mandoc_malloc(dbuf.size);
1.1 kristaps 342:
1.5 kristaps 343: flags = OP_NEW == op ? O_CREAT|O_TRUNC|O_RDWR : O_CREAT|O_RDWR;
344:
345: if (OP_UPDATE == op || OP_DELETE == op) {
346: ibuf[0] = fbuf[0] = '\0';
347:
348: strlcat(fbuf, dir, MAXPATHLEN);
349: strlcat(fbuf, "/", MAXPATHLEN);
350: sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
351:
352: strlcat(ibuf, dir, MAXPATHLEN);
353: strlcat(ibuf, "/", MAXPATHLEN);
354: sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
355:
356: if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) {
357: fprintf(stderr, "%s: Path too long\n", dir);
358: exit((int)MANDOCLEVEL_BADARG);
359: }
360:
361: db = dbopen(fbuf, flags, 0644, DB_BTREE, &info);
362: idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL);
363:
364: if (NULL == db) {
365: perror(fbuf);
366: exit((int)MANDOCLEVEL_SYSERR);
1.12 schwarze 367: } else if (NULL == idx) {
1.5 kristaps 368: perror(ibuf);
369: exit((int)MANDOCLEVEL_SYSERR);
370: }
371:
372: if (verb > 2) {
373: printf("%s: Opened\n", fbuf);
374: printf("%s: Opened\n", ibuf);
375: }
376:
1.16 ! schwarze 377: ofile_argbuild(argc, argv, &of);
1.5 kristaps 378: if (NULL == of)
379: goto out;
380:
381: of = of->first;
382:
1.16 ! schwarze 383: index_prune(of, db, fbuf, idx, ibuf,
1.5 kristaps 384: &maxrec, &recs, &recsz);
385:
386: if (OP_UPDATE == op)
1.13 schwarze 387: index_merge(of, mp, &dbuf, &buf, hash,
1.16 ! schwarze 388: db, fbuf, idx, ibuf,
! 389: maxrec, recs, reccur);
1.5 kristaps 390:
391: goto out;
392: }
393:
1.10 kristaps 394: /*
395: * Configure the directories we're going to scan.
396: * If we have command-line arguments, use them.
397: * If not, we use man(1)'s method (see mandocdb.8).
398: */
399:
400: if (argc > 0) {
401: dirs.paths = mandoc_malloc(argc * sizeof(char *));
402: dirs.sz = argc;
403: for (i = 0; i < argc; i++)
404: dirs.paths[i] = mandoc_strdup(argv[i]);
405: } else
1.11 kristaps 406: manpath_parse(&dirs, NULL, NULL);
1.10 kristaps 407:
408: for (i = 0; i < dirs.sz; i++) {
1.5 kristaps 409: ibuf[0] = fbuf[0] = '\0';
1.1 kristaps 410:
1.10 kristaps 411: strlcat(fbuf, dirs.paths[i], MAXPATHLEN);
1.5 kristaps 412: strlcat(fbuf, "/", MAXPATHLEN);
413: sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
1.1 kristaps 414:
1.10 kristaps 415: strlcat(ibuf, dirs.paths[i], MAXPATHLEN);
1.5 kristaps 416: strlcat(ibuf, "/", MAXPATHLEN);
417: sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
1.1 kristaps 418:
1.5 kristaps 419: if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) {
1.13 schwarze 420: fprintf(stderr, "%s: Path too long\n",
1.10 kristaps 421: dirs.paths[i]);
1.5 kristaps 422: exit((int)MANDOCLEVEL_BADARG);
1.4 kristaps 423: }
1.3 kristaps 424:
1.13 schwarze 425: if (db)
426: (*db->close)(db);
427: if (idx)
428: (*idx->close)(idx);
429:
1.4 kristaps 430: db = dbopen(fbuf, flags, 0644, DB_BTREE, &info);
431: idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL);
1.3 kristaps 432:
1.4 kristaps 433: if (NULL == db) {
434: perror(fbuf);
1.5 kristaps 435: exit((int)MANDOCLEVEL_SYSERR);
1.12 schwarze 436: } else if (NULL == idx) {
1.4 kristaps 437: perror(ibuf);
1.5 kristaps 438: exit((int)MANDOCLEVEL_SYSERR);
439: }
440:
441: if (verb > 2) {
442: printf("%s: Truncated\n", fbuf);
443: printf("%s: Truncated\n", ibuf);
1.4 kristaps 444: }
1.1 kristaps 445:
1.4 kristaps 446: ofile_free(of);
447: of = NULL;
1.1 kristaps 448:
1.12 schwarze 449: if ( ! ofile_dirbuild(dirs.paths[i], NULL, NULL,
1.16 ! schwarze 450: 0, &of))
1.5 kristaps 451: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 452:
1.5 kristaps 453: if (NULL == of)
454: continue;
1.1 kristaps 455:
1.5 kristaps 456: of = of->first;
1.1 kristaps 457:
1.13 schwarze 458: index_merge(of, mp, &dbuf, &buf, hash, db, fbuf,
1.16 ! schwarze 459: idx, ibuf, maxrec, recs, reccur);
1.4 kristaps 460: }
1.3 kristaps 461:
1.5 kristaps 462: out:
1.3 kristaps 463: if (db)
464: (*db->close)(db);
465: if (idx)
466: (*idx->close)(idx);
467: if (hash)
468: (*hash->close)(hash);
469: if (mp)
470: mparse_free(mp);
471:
1.10 kristaps 472: manpath_free(&dirs);
1.4 kristaps 473: ofile_free(of);
1.3 kristaps 474: free(buf.cp);
475: free(dbuf.cp);
476: free(recs);
477:
1.5 kristaps 478: return(MANDOCLEVEL_OK);
1.3 kristaps 479: }
480:
481: void
482: index_merge(const struct of *of, struct mparse *mp,
1.16 ! schwarze 483: struct buf *dbuf, struct buf *buf, DB *hash,
! 484: DB *db, const char *dbf, DB *idx, const char *idxf,
1.3 kristaps 485: recno_t maxrec, const recno_t *recs, size_t reccur)
486: {
487: recno_t rec;
488: int ch;
489: DBT key, val;
490: struct mdoc *mdoc;
491: struct man *man;
492: const char *fn, *msec, *mtitle, *arch;
493: size_t sv;
494: unsigned seq;
1.9 kristaps 495: struct db_val vbuf;
1.3 kristaps 496:
497: for (rec = 0; of; of = of->next) {
498: fn = of->fname;
1.14 schwarze 499:
500: /*
501: * Reclaim an empty index record, if available.
502: */
503:
1.3 kristaps 504: if (reccur > 0) {
505: --reccur;
506: rec = recs[(int)reccur];
507: } else if (maxrec > 0) {
508: rec = maxrec;
509: maxrec = 0;
1.1 kristaps 510: } else
511: rec++;
512:
513: mparse_reset(mp);
514: hash_reset(&hash);
1.14 schwarze 515: mdoc = NULL;
516: man = NULL;
1.1 kristaps 517:
1.14 schwarze 518: /*
519: * Try interpreting the file as mdoc(7) or man(7)
520: * source code, unless it is already known to be
521: * formatted. Fall back to formatted mode.
522: */
523:
524: if ((MANDOC_SRC & of->src_form ||
525: ! (MANDOC_FORM & of->src_form)) &&
526: MANDOCLEVEL_FATAL > mparse_readfd(mp, -1, fn))
527: mparse_result(mp, &mdoc, &man);
528:
529: if (NULL != mdoc) {
530: msec = mdoc_meta(mdoc)->msec;
531: arch = mdoc_meta(mdoc)->arch;
532: mtitle = mdoc_meta(mdoc)->title;
533: } else if (NULL != man) {
534: msec = man_meta(man)->msec;
535: arch = NULL;
536: mtitle = man_meta(man)->title;
537: } else {
538: msec = of->sec;
539: arch = of->arch;
540: mtitle = of->title;
1.1 kristaps 541: }
542:
1.12 schwarze 543: /*
544: * By default, skip a file if the manual section
545: * and architecture given in the file disagree
546: * with the directory where the file is located.
547: */
548:
549: if (0 == use_all) {
550: assert(of->sec);
551: assert(msec);
552: if (strcmp(msec, of->sec))
553: continue;
554:
555: if (NULL == arch) {
556: if (NULL != of->arch)
557: continue;
558: } else if (NULL == of->arch ||
559: strcmp(arch, of->arch))
560: continue;
561: }
562:
1.1 kristaps 563: if (NULL == arch)
564: arch = "";
565:
566: /*
1.12 schwarze 567: * By default, skip a file if the title given
568: * in the file disagrees with the file name.
569: * If both agree, use the file name as the title,
570: * because the one in the file usually is all caps.
571: */
572:
573: assert(of->title);
574: assert(mtitle);
575:
576: if (0 == strcasecmp(mtitle, of->title))
577: mtitle = of->title;
578: else if (0 == use_all)
579: continue;
580:
581: /*
1.1 kristaps 582: * The index record value consists of a nil-terminated
583: * filename, a nil-terminated manual section, and a
584: * nil-terminated description. Since the description
585: * may not be set, we set a sentinel to see if we're
586: * going to write a nil byte in its place.
587: */
588:
1.3 kristaps 589: dbuf->len = 0;
1.15 schwarze 590: buf_append(dbuf, mdoc ? "mdoc" : (man ? "man" : "cat"));
1.3 kristaps 591: buf_appendb(dbuf, fn, strlen(fn) + 1);
592: buf_appendb(dbuf, msec, strlen(msec) + 1);
593: buf_appendb(dbuf, mtitle, strlen(mtitle) + 1);
594: buf_appendb(dbuf, arch, strlen(arch) + 1);
1.1 kristaps 595:
1.3 kristaps 596: sv = dbuf->len;
1.1 kristaps 597:
598: /* Fix the record number in the btree value. */
599:
600: if (mdoc)
1.3 kristaps 601: pmdoc_node(hash, buf, dbuf,
1.1 kristaps 602: mdoc_node(mdoc), mdoc_meta(mdoc));
1.14 schwarze 603: else if (man)
1.3 kristaps 604: pman_node(hash, buf, dbuf, man_node(man));
1.14 schwarze 605: else
606: pformatted(hash, buf, dbuf, of);
1.1 kristaps 607:
608: /*
609: * Copy from the in-memory hashtable of pending keywords
610: * into the database.
611: */
612:
1.9 kristaps 613: vbuf.rec = rec;
1.1 kristaps 614: seq = R_FIRST;
615: while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) {
616: seq = R_NEXT;
617:
1.9 kristaps 618: vbuf.mask = *(uint64_t *)val.data;
619: val.size = sizeof(struct db_val);
620: val.data = &vbuf;
1.1 kristaps 621:
1.5 kristaps 622: if (verb > 1)
623: printf("%s: Added keyword: %s\n",
624: fn, (char *)key.data);
1.3 kristaps 625: dbt_put(db, dbf, &key, &val);
1.1 kristaps 626: }
627: if (ch < 0) {
628: perror("hash");
629: exit((int)MANDOCLEVEL_SYSERR);
630: }
631:
632: /*
633: * Apply to the index. If we haven't had a description
634: * set, put an empty one in now.
635: */
636:
1.3 kristaps 637: if (dbuf->len == sv)
638: buf_appendb(dbuf, "", 1);
1.1 kristaps 639:
640: key.data = &rec;
641: key.size = sizeof(recno_t);
642:
1.3 kristaps 643: val.data = dbuf->cp;
644: val.size = dbuf->len;
1.1 kristaps 645:
1.5 kristaps 646: if (verb)
647: printf("%s: Added index\n", fn);
1.3 kristaps 648: dbt_put(idx, idxf, &key, &val);
649: }
650: }
651:
652: /*
653: * Scan through all entries in the index file `idx' and prune those
654: * entries in `ofile'.
655: * Pruning consists of removing from `db', then invalidating the entry
656: * in `idx' (zeroing its value size).
657: */
658: static void
659: index_prune(const struct of *ofile, DB *db, const char *dbf,
1.16 ! schwarze 660: DB *idx, const char *idxf,
1.3 kristaps 661: recno_t *maxrec, recno_t **recs, size_t *recsz)
662: {
663: const struct of *of;
664: const char *fn;
1.9 kristaps 665: struct db_val *vbuf;
1.3 kristaps 666: unsigned seq, sseq;
667: DBT key, val;
668: size_t reccur;
669: int ch;
670:
671: reccur = 0;
672: seq = R_FIRST;
673: while (0 == (ch = (*idx->seq)(idx, &key, &val, seq))) {
674: seq = R_NEXT;
675: *maxrec = *(recno_t *)key.data;
676: if (0 == val.size) {
677: if (reccur >= *recsz) {
678: *recsz += MANDOC_SLOP;
679: *recs = mandoc_realloc(*recs,
680: *recsz * sizeof(recno_t));
681: }
682: (*recs)[(int)reccur] = *maxrec;
683: reccur++;
684: continue;
685: }
686:
687: fn = (char *)val.data;
688: for (of = ofile; of; of = of->next)
689: if (0 == strcmp(fn, of->fname))
690: break;
691:
692: if (NULL == of)
693: continue;
694:
695: sseq = R_FIRST;
696: while (0 == (ch = (*db->seq)(db, &key, &val, sseq))) {
697: sseq = R_NEXT;
1.9 kristaps 698: assert(sizeof(struct db_val) == val.size);
699: vbuf = val.data;
700: if (*maxrec != vbuf->rec)
1.3 kristaps 701: continue;
1.5 kristaps 702: if (verb)
703: printf("%s: Deleted keyword: %s\n",
704: fn, (char *)key.data);
1.3 kristaps 705: ch = (*db->del)(db, &key, R_CURSOR);
706: if (ch < 0)
707: break;
708: }
709: if (ch < 0) {
710: perror(dbf);
711: exit((int)MANDOCLEVEL_SYSERR);
712: }
1.1 kristaps 713:
1.5 kristaps 714: if (verb)
715: printf("%s: Deleted index\n", fn);
1.1 kristaps 716:
1.3 kristaps 717: val.size = 0;
718: ch = (*idx->put)(idx, &key, &val, R_CURSOR);
719: if (ch < 0) {
720: perror(idxf);
721: exit((int)MANDOCLEVEL_SYSERR);
722: }
1.1 kristaps 723:
1.3 kristaps 724: if (reccur >= *recsz) {
725: *recsz += MANDOC_SLOP;
726: *recs = mandoc_realloc
727: (*recs, *recsz * sizeof(recno_t));
728: }
1.1 kristaps 729:
1.3 kristaps 730: (*recs)[(int)reccur] = *maxrec;
731: reccur++;
732: }
733: (*maxrec)++;
1.1 kristaps 734: }
735:
736: /*
737: * Grow the buffer (if necessary) and copy in a binary string.
738: */
739: static void
740: buf_appendb(struct buf *buf, const void *cp, size_t sz)
741: {
742:
743: /* Overshoot by MANDOC_BUFSZ. */
744:
745: while (buf->len + sz >= buf->size) {
746: buf->size = buf->len + sz + MANDOC_BUFSZ;
747: buf->cp = mandoc_realloc(buf->cp, buf->size);
748: }
749:
750: memcpy(buf->cp + (int)buf->len, cp, sz);
751: buf->len += sz;
752: }
753:
754: /*
755: * Append a nil-terminated string to the buffer.
756: * This can be invoked multiple times.
757: * The buffer string will be nil-terminated.
758: * If invoked multiple times, a space is put between strings.
759: */
760: static void
761: buf_append(struct buf *buf, const char *cp)
762: {
763: size_t sz;
764:
765: if (0 == (sz = strlen(cp)))
766: return;
767:
768: if (buf->len)
769: buf->cp[(int)buf->len - 1] = ' ';
770:
771: buf_appendb(buf, cp, sz + 1);
772: }
773:
774: /*
775: * Recursively add all text from a given node.
776: * This is optimised for general mdoc nodes in this context, which do
777: * not consist of subexpressions and having a recursive call for n->next
778: * would be wasteful.
779: * The "f" variable should be 0 unless called from pmdoc_Nd for the
780: * description buffer, which does not start at the beginning of the
781: * buffer.
782: */
783: static void
784: buf_appendmdoc(struct buf *buf, const struct mdoc_node *n, int f)
785: {
786:
787: for ( ; n; n = n->next) {
788: if (n->child)
789: buf_appendmdoc(buf, n->child, f);
790:
791: if (MDOC_TEXT == n->type && f) {
792: f = 0;
793: buf_appendb(buf, n->string,
794: strlen(n->string) + 1);
795: } else if (MDOC_TEXT == n->type)
796: buf_append(buf, n->string);
797:
798: }
799: }
800:
801: /* ARGSUSED */
802: static void
803: pmdoc_An(MDOC_ARGS)
804: {
805:
806: if (SEC_AUTHORS != n->sec)
807: return;
808:
809: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 810: hash_put(hash, buf, TYPE_An);
1.1 kristaps 811: }
812:
813: static void
814: hash_reset(DB **db)
815: {
816: DB *hash;
817:
818: if (NULL != (hash = *db))
819: (*hash->close)(hash);
820:
1.5 kristaps 821: *db = dbopen(NULL, O_CREAT|O_RDWR, 0644, DB_HASH, NULL);
1.1 kristaps 822: if (NULL == *db) {
823: perror("hash");
824: exit((int)MANDOCLEVEL_SYSERR);
825: }
826: }
827:
828: /* ARGSUSED */
829: static void
830: pmdoc_Fd(MDOC_ARGS)
831: {
832: const char *start, *end;
833: size_t sz;
834:
835: if (SEC_SYNOPSIS != n->sec)
836: return;
837: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
838: return;
839:
840: /*
841: * Only consider those `Fd' macro fields that begin with an
842: * "inclusion" token (versus, e.g., #define).
843: */
844: if (strcmp("#include", n->string))
845: return;
846:
847: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
848: return;
849:
850: /*
851: * Strip away the enclosing angle brackets and make sure we're
852: * not zero-length.
853: */
854:
855: start = n->string;
856: if ('<' == *start || '"' == *start)
857: start++;
858:
859: if (0 == (sz = strlen(start)))
860: return;
861:
862: end = &start[(int)sz - 1];
863: if ('>' == *end || '"' == *end)
864: end--;
865:
866: assert(end >= start);
867:
868: buf_appendb(buf, start, (size_t)(end - start + 1));
869: buf_appendb(buf, "", 1);
870:
1.8 schwarze 871: hash_put(hash, buf, TYPE_In);
1.1 kristaps 872: }
873:
874: /* ARGSUSED */
875: static void
876: pmdoc_Cd(MDOC_ARGS)
877: {
878:
879: if (SEC_SYNOPSIS != n->sec)
880: return;
881:
882: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 883: hash_put(hash, buf, TYPE_Cd);
1.1 kristaps 884: }
885:
886: /* ARGSUSED */
887: static void
888: pmdoc_In(MDOC_ARGS)
889: {
890:
891: if (SEC_SYNOPSIS != n->sec)
892: return;
893: if (NULL == n->child || MDOC_TEXT != n->child->type)
894: return;
895:
896: buf_append(buf, n->child->string);
1.8 schwarze 897: hash_put(hash, buf, TYPE_In);
1.1 kristaps 898: }
899:
900: /* ARGSUSED */
901: static void
902: pmdoc_Fn(MDOC_ARGS)
903: {
904: const char *cp;
905:
906: if (SEC_SYNOPSIS != n->sec)
907: return;
908: if (NULL == n->child || MDOC_TEXT != n->child->type)
909: return;
910:
911: /* .Fn "struct type *arg" "foo" */
912:
913: cp = strrchr(n->child->string, ' ');
914: if (NULL == cp)
915: cp = n->child->string;
916:
917: /* Strip away pointer symbol. */
918:
919: while ('*' == *cp)
920: cp++;
921:
922: buf_append(buf, cp);
1.8 schwarze 923: hash_put(hash, buf, TYPE_Fn);
1.1 kristaps 924: }
925:
926: /* ARGSUSED */
927: static void
928: pmdoc_St(MDOC_ARGS)
929: {
930:
931: if (SEC_STANDARDS != n->sec)
932: return;
933: if (NULL == n->child || MDOC_TEXT != n->child->type)
934: return;
935:
936: buf_append(buf, n->child->string);
1.8 schwarze 937: hash_put(hash, buf, TYPE_St);
1.1 kristaps 938: }
939:
940: /* ARGSUSED */
941: static void
942: pmdoc_Xr(MDOC_ARGS)
943: {
944:
945: if (NULL == (n = n->child))
946: return;
947:
948: buf_appendb(buf, n->string, strlen(n->string));
949:
950: if (NULL != (n = n->next)) {
951: buf_appendb(buf, ".", 1);
952: buf_appendb(buf, n->string, strlen(n->string) + 1);
953: } else
954: buf_appendb(buf, ".", 2);
955:
1.8 schwarze 956: hash_put(hash, buf, TYPE_Xr);
1.1 kristaps 957: }
958:
959: /* ARGSUSED */
960: static void
961: pmdoc_Vt(MDOC_ARGS)
962: {
963: const char *start;
964: size_t sz;
965:
966: if (SEC_SYNOPSIS != n->sec)
967: return;
968: if (MDOC_Vt == n->tok && MDOC_BODY != n->type)
969: return;
970: if (NULL == n->last || MDOC_TEXT != n->last->type)
971: return;
972:
973: /*
974: * Strip away leading pointer symbol '*' and trailing ';'.
975: */
976:
977: start = n->last->string;
978:
979: while ('*' == *start)
980: start++;
981:
982: if (0 == (sz = strlen(start)))
983: return;
984:
985: if (';' == start[(int)sz - 1])
986: sz--;
987:
988: if (0 == sz)
989: return;
990:
991: buf_appendb(buf, start, sz);
992: buf_appendb(buf, "", 1);
1.8 schwarze 993: hash_put(hash, buf, TYPE_Va);
1.1 kristaps 994: }
995:
996: /* ARGSUSED */
997: static void
998: pmdoc_Fo(MDOC_ARGS)
999: {
1000:
1001: if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
1002: return;
1003: if (NULL == n->child || MDOC_TEXT != n->child->type)
1004: return;
1005:
1006: buf_append(buf, n->child->string);
1.8 schwarze 1007: hash_put(hash, buf, TYPE_Fn);
1.1 kristaps 1008: }
1009:
1010:
1011: /* ARGSUSED */
1012: static void
1013: pmdoc_Nd(MDOC_ARGS)
1014: {
1015:
1016: if (MDOC_BODY != n->type)
1017: return;
1018:
1019: buf_appendmdoc(dbuf, n->child, 1);
1020: buf_appendmdoc(buf, n->child, 0);
1021:
1.8 schwarze 1022: hash_put(hash, buf, TYPE_Nd);
1.1 kristaps 1023: }
1024:
1025: /* ARGSUSED */
1026: static void
1027: pmdoc_Er(MDOC_ARGS)
1028: {
1029:
1030: if (SEC_ERRORS != n->sec)
1031: return;
1032:
1033: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 1034: hash_put(hash, buf, TYPE_Er);
1.1 kristaps 1035: }
1036:
1037: /* ARGSUSED */
1038: static void
1039: pmdoc_Ev(MDOC_ARGS)
1040: {
1041:
1042: if (SEC_ENVIRONMENT != n->sec)
1043: return;
1044:
1045: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 1046: hash_put(hash, buf, TYPE_Ev);
1.1 kristaps 1047: }
1048:
1049: /* ARGSUSED */
1050: static void
1051: pmdoc_Pa(MDOC_ARGS)
1052: {
1053:
1054: if (SEC_FILES != n->sec)
1055: return;
1056:
1057: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 1058: hash_put(hash, buf, TYPE_Pa);
1.1 kristaps 1059: }
1060:
1061: /* ARGSUSED */
1062: static void
1063: pmdoc_Nm(MDOC_ARGS)
1064: {
1065:
1066: if (SEC_NAME == n->sec) {
1067: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 1068: hash_put(hash, buf, TYPE_Nm);
1.1 kristaps 1069: return;
1070: } else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
1071: return;
1072:
1073: if (NULL == n->child)
1074: buf_append(buf, m->name);
1075:
1076: buf_appendmdoc(buf, n->child, 0);
1.8 schwarze 1077: hash_put(hash, buf, TYPE_Nm);
1.1 kristaps 1078: }
1079:
1080: static void
1.9 kristaps 1081: hash_put(DB *db, const struct buf *buf, uint64_t mask)
1.1 kristaps 1082: {
1083: DBT key, val;
1084: int rc;
1085:
1086: if (buf->len < 2)
1087: return;
1088:
1089: key.data = buf->cp;
1090: key.size = buf->len;
1091:
1092: if ((rc = (*db->get)(db, &key, &val, 0)) < 0) {
1093: perror("hash");
1094: exit((int)MANDOCLEVEL_SYSERR);
1095: } else if (0 == rc)
1.9 kristaps 1096: mask |= *(uint64_t *)val.data;
1.1 kristaps 1097:
1098: val.data = &mask;
1.9 kristaps 1099: val.size = sizeof(uint64_t);
1.1 kristaps 1100:
1101: if ((rc = (*db->put)(db, &key, &val, 0)) < 0) {
1102: perror("hash");
1103: exit((int)MANDOCLEVEL_SYSERR);
1104: }
1105: }
1106:
1107: static void
1108: dbt_put(DB *db, const char *dbn, DBT *key, DBT *val)
1109: {
1110:
1111: assert(key->size);
1112: assert(val->size);
1113:
1114: if (0 == (*db->put)(db, key, val, 0))
1115: return;
1116:
1117: perror(dbn);
1118: exit((int)MANDOCLEVEL_SYSERR);
1119: /* NOTREACHED */
1120: }
1121:
1122: /*
1123: * Call out to per-macro handlers after clearing the persistent database
1124: * key. If the macro sets the database key, flush it to the database.
1125: */
1126: static void
1127: pmdoc_node(MDOC_ARGS)
1128: {
1129:
1130: if (NULL == n)
1131: return;
1132:
1133: switch (n->type) {
1134: case (MDOC_HEAD):
1135: /* FALLTHROUGH */
1136: case (MDOC_BODY):
1137: /* FALLTHROUGH */
1138: case (MDOC_TAIL):
1139: /* FALLTHROUGH */
1140: case (MDOC_BLOCK):
1141: /* FALLTHROUGH */
1142: case (MDOC_ELEM):
1143: if (NULL == mdocs[n->tok])
1144: break;
1145:
1146: buf->len = 0;
1147: (*mdocs[n->tok])(hash, buf, dbuf, n, m);
1148: break;
1149: default:
1150: break;
1151: }
1152:
1153: pmdoc_node(hash, buf, dbuf, n->child, m);
1154: pmdoc_node(hash, buf, dbuf, n->next, m);
1155: }
1156:
1157: static int
1158: pman_node(MAN_ARGS)
1159: {
1160: const struct man_node *head, *body;
1161: const char *start, *sv;
1162: size_t sz;
1163:
1164: if (NULL == n)
1165: return(0);
1166:
1167: /*
1168: * We're only searching for one thing: the first text child in
1169: * the BODY of a NAME section. Since we don't keep track of
1170: * sections in -man, run some hoops to find out whether we're in
1171: * the correct section or not.
1172: */
1173:
1174: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1175: body = n;
1176: assert(body->parent);
1177: if (NULL != (head = body->parent->head) &&
1178: 1 == head->nchild &&
1179: NULL != (head = (head->child)) &&
1180: MAN_TEXT == head->type &&
1181: 0 == strcmp(head->string, "NAME") &&
1182: NULL != (body = body->child) &&
1183: MAN_TEXT == body->type) {
1184:
1185: assert(body->string);
1186: start = sv = body->string;
1187:
1188: /*
1189: * Go through a special heuristic dance here.
1190: * This is why -man manuals are great!
1191: * (I'm being sarcastic: my eyes are bleeding.)
1192: * Conventionally, one or more manual names are
1193: * comma-specified prior to a whitespace, then a
1194: * dash, then a description. Try to puzzle out
1195: * the name parts here.
1196: */
1197:
1198: for ( ;; ) {
1199: sz = strcspn(start, " ,");
1200: if ('\0' == start[(int)sz])
1201: break;
1202:
1203: buf->len = 0;
1204: buf_appendb(buf, start, sz);
1205: buf_appendb(buf, "", 1);
1206:
1.8 schwarze 1207: hash_put(hash, buf, TYPE_Nm);
1.1 kristaps 1208:
1209: if (' ' == start[(int)sz]) {
1210: start += (int)sz + 1;
1211: break;
1212: }
1213:
1214: assert(',' == start[(int)sz]);
1215: start += (int)sz + 1;
1216: while (' ' == *start)
1217: start++;
1218: }
1219:
1220: buf->len = 0;
1221:
1222: if (sv == start) {
1223: buf_append(buf, start);
1224: return(1);
1225: }
1226:
1227: while (' ' == *start)
1228: start++;
1229:
1230: if (0 == strncmp(start, "-", 1))
1231: start += 1;
1232: else if (0 == strncmp(start, "\\-", 2))
1233: start += 2;
1234: else if (0 == strncmp(start, "\\(en", 4))
1235: start += 4;
1236: else if (0 == strncmp(start, "\\(em", 4))
1237: start += 4;
1238:
1239: while (' ' == *start)
1240: start++;
1241:
1242: sz = strlen(start) + 1;
1243: buf_appendb(dbuf, start, sz);
1244: buf_appendb(buf, start, sz);
1245:
1.8 schwarze 1246: hash_put(hash, buf, TYPE_Nd);
1.1 kristaps 1247: }
1248: }
1249:
1.7 schwarze 1250: for (n = n->child; n; n = n->next)
1251: if (pman_node(hash, buf, dbuf, n))
1252: return(1);
1.1 kristaps 1253:
1254: return(0);
1255: }
1256:
1.14 schwarze 1257: /*
1258: * Parse a formatted manual page.
1259: * By necessity, this involves rather crude guesswork.
1260: */
1261: static void
1262: pformatted(DB *hash, struct buf *buf, struct buf *dbuf,
1263: const struct of *of)
1264: {
1265: FILE *stream;
1266: char *line, *p;
1267: size_t len, plen;
1268:
1269: if (NULL == (stream = fopen(of->fname, "r"))) {
1270: perror(of->fname);
1271: return;
1272: }
1273:
1274: /*
1275: * Always use the title derived from the filename up front,
1276: * do not even try to find it in the file. This also makes
1277: * sure we don't end up with an orphan index record, even if
1278: * the file content turns out to be completely unintelligible.
1279: */
1280:
1281: buf->len = 0;
1282: buf_append(buf, of->title);
1283: hash_put(hash, buf, TYPE_Nm);
1284:
1285: while (NULL != (line = fgetln(stream, &len)) && '\n' != *line)
1286: /* Skip to first blank line. */ ;
1287:
1288: while (NULL != (line = fgetln(stream, &len)) &&
1289: ('\n' == *line || ' ' == *line))
1290: /* Skip to first section header. */ ;
1291:
1292: /*
1293: * If no page content can be found,
1294: * reuse the page title as the page description.
1295: */
1296:
1297: if (NULL == (line = fgetln(stream, &len))) {
1298: buf_appendb(dbuf, buf->cp, buf->size);
1299: hash_put(hash, buf, TYPE_Nd);
1300: fclose(stream);
1301: return;
1302: }
1303: fclose(stream);
1304:
1305: /*
1306: * If there is a dash, skip to the text following it.
1307: */
1308:
1309: for (p = line, plen = len; plen; p++, plen--)
1310: if ('-' == *p)
1311: break;
1312: for ( ; plen; p++, plen--)
1313: if ('-' != *p && ' ' != *p && 8 != *p)
1314: break;
1315: if (0 == plen) {
1316: p = line;
1317: plen = len;
1318: }
1319:
1320: /*
1321: * Copy the rest of the line, but no more than 70 bytes.
1322: */
1323:
1324: if (70 < plen)
1325: plen = 70;
1326: p[plen-1] = '\0';
1327: buf_appendb(dbuf, p, plen);
1328: buf->len = 0;
1329: buf_appendb(buf, p, plen);
1330: hash_put(hash, buf, TYPE_Nd);
1331: }
1332:
1.5 kristaps 1333: static void
1.16 ! schwarze 1334: ofile_argbuild(int argc, char *argv[], struct of **of)
1.5 kristaps 1335: {
1.12 schwarze 1336: char buf[MAXPATHLEN];
1337: char *sec, *arch, *title, *p;
1.14 schwarze 1338: int i, src_form;
1.5 kristaps 1339: struct of *nof;
1340:
1341: for (i = 0; i < argc; i++) {
1.12 schwarze 1342:
1343: /*
1344: * Try to infer the manual section, architecture and
1345: * page title from the path, assuming it looks like
1.14 schwarze 1346: * man*[/<arch>]/<title>.<section> or
1347: * cat<section>[/<arch>]/<title>.0
1.12 schwarze 1348: */
1349:
1350: if (strlcpy(buf, argv[i], sizeof(buf)) >= sizeof(buf)) {
1351: fprintf(stderr, "%s: Path too long\n", argv[i]);
1352: continue;
1353: }
1354: sec = arch = title = NULL;
1.14 schwarze 1355: src_form = 0;
1.12 schwarze 1356: p = strrchr(buf, '\0');
1357: while (p-- > buf) {
1358: if (NULL == sec && '.' == *p) {
1359: sec = p + 1;
1360: *p = '\0';
1.14 schwarze 1361: if ('0' == *sec)
1362: src_form |= MANDOC_FORM;
1363: else if ('1' <= *sec && '9' >= *sec)
1364: src_form |= MANDOC_SRC;
1.12 schwarze 1365: continue;
1366: }
1367: if ('/' != *p)
1368: continue;
1369: if (NULL == title) {
1370: title = p + 1;
1371: *p = '\0';
1372: continue;
1373: }
1.14 schwarze 1374: if (strncmp("man", p + 1, 3)) {
1375: src_form |= MANDOC_SRC;
1376: arch = p + 1;
1377: } else if (strncmp("cat", p + 1, 3)) {
1378: src_form |= MANDOC_FORM;
1.12 schwarze 1379: arch = p + 1;
1.14 schwarze 1380: }
1.12 schwarze 1381: break;
1382: }
1383: if (NULL == title)
1384: title = buf;
1385:
1386: /*
1387: * Build the file structure.
1388: */
1389:
1.5 kristaps 1390: nof = mandoc_calloc(1, sizeof(struct of));
1.12 schwarze 1391: nof->fname = mandoc_strdup(argv[i]);
1392: if (NULL != sec)
1393: nof->sec = mandoc_strdup(sec);
1394: if (NULL != arch)
1395: nof->arch = mandoc_strdup(arch);
1396: nof->title = mandoc_strdup(title);
1.14 schwarze 1397: nof->src_form = src_form;
1.12 schwarze 1398:
1399: /*
1400: * Add the structure to the list.
1401: */
1402:
1.5 kristaps 1403: if (verb > 2)
1404: printf("%s: Scheduling\n", argv[i]);
1405: if (NULL == *of) {
1406: *of = nof;
1407: (*of)->first = nof;
1408: } else {
1409: nof->first = (*of)->first;
1410: (*of)->next = nof;
1411: *of = nof;
1412: }
1413: }
1414: }
1415:
1.4 kristaps 1416: /*
1417: * Recursively build up a list of files to parse.
1418: * We use this instead of ftw() and so on because I don't want global
1419: * variables hanging around.
1420: * This ignores the mandoc.db and mandoc.index files, but assumes that
1421: * everything else is a manual.
1422: * Pass in a pointer to a NULL structure for the first invocation.
1423: */
1424: static int
1.12 schwarze 1425: ofile_dirbuild(const char *dir, const char* psec, const char *parch,
1.16 ! schwarze 1426: int p_src_form, struct of **of)
1.4 kristaps 1427: {
1.5 kristaps 1428: char buf[MAXPATHLEN];
1.14 schwarze 1429: struct stat sb;
1.5 kristaps 1430: size_t sz;
1.4 kristaps 1431: DIR *d;
1.12 schwarze 1432: const char *fn, *sec, *arch;
1.14 schwarze 1433: char *p, *q, *suffix;
1.4 kristaps 1434: struct of *nof;
1435: struct dirent *dp;
1.14 schwarze 1436: int src_form;
1.4 kristaps 1437:
1438: if (NULL == (d = opendir(dir))) {
1439: perror(dir);
1440: return(0);
1441: }
1442:
1443: while (NULL != (dp = readdir(d))) {
1444: fn = dp->d_name;
1.12 schwarze 1445:
1446: if ('.' == *fn)
1447: continue;
1448:
1.14 schwarze 1449: src_form = p_src_form;
1450:
1.4 kristaps 1451: if (DT_DIR == dp->d_type) {
1.12 schwarze 1452: sec = psec;
1453: arch = parch;
1454:
1455: /*
1456: * By default, only use directories called:
1.14 schwarze 1457: * man<section>/[<arch>/] or
1458: * cat<section>/[<arch>/]
1.12 schwarze 1459: */
1460:
1461: if (NULL == sec) {
1.14 schwarze 1462: if(0 == strncmp("man", fn, 3)) {
1463: src_form |= MANDOC_SRC;
1.12 schwarze 1464: sec = fn + 3;
1.14 schwarze 1465: } else if (0 == strncmp("cat", fn, 3)) {
1466: src_form |= MANDOC_FORM;
1467: sec = fn + 3;
1468: } else if (use_all)
1.12 schwarze 1469: sec = fn;
1470: else
1471: continue;
1472: } else if (NULL == arch && (use_all ||
1473: NULL == strchr(fn, '.')))
1474: arch = fn;
1475: else if (0 == use_all)
1.5 kristaps 1476: continue;
1477:
1478: buf[0] = '\0';
1479: strlcat(buf, dir, MAXPATHLEN);
1480: strlcat(buf, "/", MAXPATHLEN);
1481: sz = strlcat(buf, fn, MAXPATHLEN);
1482:
1.12 schwarze 1483: if (MAXPATHLEN <= sz) {
1484: fprintf(stderr, "%s: Path too long\n", dir);
1485: return(0);
1486: }
1487:
1488: if (verb > 2)
1489: printf("%s: Scanning\n", buf);
1490:
1491: if ( ! ofile_dirbuild(buf, sec, arch,
1.16 ! schwarze 1492: src_form, of))
1.12 schwarze 1493: return(0);
1494: }
1495: if (DT_REG != dp->d_type ||
1496: (NULL == psec && !use_all) ||
1497: !strcmp(MANDOC_DB, fn) ||
1498: !strcmp(MANDOC_IDX, fn))
1499: continue;
1500:
1501: /*
1502: * By default, skip files where the file name suffix
1503: * does not agree with the section directory
1504: * they are located in.
1505: */
1506:
1507: suffix = strrchr(fn, '.');
1508: if (0 == use_all) {
1509: if (NULL == suffix)
1.5 kristaps 1510: continue;
1.14 schwarze 1511: if ((MANDOC_SRC & src_form &&
1512: strcmp(suffix + 1, psec)) ||
1513: (MANDOC_FORM & src_form &&
1514: strcmp(suffix + 1, "0")))
1515: continue;
1516: }
1517: if (NULL != suffix) {
1518: if ('0' == suffix[1])
1519: src_form |= MANDOC_FORM;
1520: else if ('1' <= suffix[1] && '9' >= suffix[1])
1521: src_form |= MANDOC_SRC;
1522: }
1523:
1524:
1525: /*
1526: * Skip formatted manuals if a source version is
1527: * available. Ignore the age: it is very unlikely
1528: * that people install newer formatted base manuals
1529: * when they used to have source manuals before,
1530: * and in ports, old manuals get removed on update.
1531: */
1532: if (0 == use_all && MANDOC_FORM & src_form &&
1533: NULL != psec) {
1534: buf[0] = '\0';
1535: strlcat(buf, dir, MAXPATHLEN);
1536: p = strrchr(buf, '/');
1537: if (NULL == p)
1538: p = buf;
1539: else
1540: p++;
1541: if (0 == strncmp("cat", p, 3))
1542: memcpy(p, "man", 3);
1543: strlcat(buf, "/", MAXPATHLEN);
1544: sz = strlcat(buf, fn, MAXPATHLEN);
1545: if (sz >= MAXPATHLEN) {
1546: fprintf(stderr, "%s: Path too long\n", buf);
1.5 kristaps 1547: continue;
1.14 schwarze 1548: }
1549: q = strrchr(buf, '.');
1550: if (NULL != q && p < q++) {
1551: *q = '\0';
1552: sz = strlcat(buf, psec, MAXPATHLEN);
1553: if (sz >= MAXPATHLEN) {
1554: fprintf(stderr,
1555: "%s: Path too long\n", buf);
1556: continue;
1557: }
1558: if (0 == stat(buf, &sb))
1559: continue;
1560: }
1.5 kristaps 1561: }
1.4 kristaps 1562:
1.5 kristaps 1563: buf[0] = '\0';
1564: strlcat(buf, dir, MAXPATHLEN);
1565: strlcat(buf, "/", MAXPATHLEN);
1.6 schwarze 1566: sz = strlcat(buf, fn, MAXPATHLEN);
1.5 kristaps 1567: if (sz >= MAXPATHLEN) {
1568: fprintf(stderr, "%s: Path too long\n", dir);
1.14 schwarze 1569: continue;
1.5 kristaps 1570: }
1571:
1.4 kristaps 1572: nof = mandoc_calloc(1, sizeof(struct of));
1.5 kristaps 1573: nof->fname = mandoc_strdup(buf);
1.12 schwarze 1574: if (NULL != psec)
1575: nof->sec = mandoc_strdup(psec);
1576: if (NULL != parch)
1577: nof->arch = mandoc_strdup(parch);
1.14 schwarze 1578: nof->src_form = src_form;
1.12 schwarze 1579:
1580: /*
1581: * Remember the file name without the extension,
1582: * to be used as the page title in the database.
1583: */
1584:
1585: if (NULL != suffix)
1586: *suffix = '\0';
1587: nof->title = mandoc_strdup(fn);
1.5 kristaps 1588:
1.14 schwarze 1589: /*
1590: * Add the structure to the list.
1591: */
1592:
1.5 kristaps 1593: if (verb > 2)
1594: printf("%s: Scheduling\n", buf);
1.4 kristaps 1595: if (NULL == *of) {
1596: *of = nof;
1597: (*of)->first = nof;
1598: } else {
1.5 kristaps 1599: nof->first = (*of)->first;
1.4 kristaps 1600: (*of)->next = nof;
1601: *of = nof;
1602: }
1603: }
1604:
1.7 schwarze 1605: closedir(d);
1.4 kristaps 1606: return(1);
1607: }
1608:
1609: static void
1610: ofile_free(struct of *of)
1611: {
1612: struct of *nof;
1613:
1614: while (of) {
1615: nof = of->next;
1616: free(of->fname);
1.12 schwarze 1617: free(of->sec);
1618: free(of->arch);
1619: free(of->title);
1.4 kristaps 1620: free(of);
1621: of = nof;
1622: }
1623: }
1624:
1.1 kristaps 1625: static void
1626: usage(void)
1627: {
1628:
1.5 kristaps 1629: fprintf(stderr, "usage: %s [-v] "
1630: "[-d dir [files...] |"
1631: " -u dir [files...] |"
1632: " dir...]\n", progname);
1.1 kristaps 1633: }
CVSweb