Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.15
1.15 ! schwarze 1: /* $Id: mandocdb.c,v 1.14 2011/11/28 00:57:28 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.10 schwarze 4: * Copyright (c) 2011 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17: */
18: #include <sys/param.h>
1.11 schwarze 19: #include <sys/types.h>
20: #include <sys/stat.h>
1.1 schwarze 21:
22: #include <assert.h>
1.2 schwarze 23: #include <dirent.h>
1.1 schwarze 24: #include <fcntl.h>
25: #include <getopt.h>
26: #include <stdio.h>
27: #include <stdint.h>
28: #include <stdlib.h>
29: #include <string.h>
1.14 schwarze 30: #include <unistd.h>
1.1 schwarze 31: #include <db.h>
32:
33: #include "man.h"
34: #include "mdoc.h"
35: #include "mandoc.h"
1.5 schwarze 36: #include "mandocdb.h"
1.10 schwarze 37: #include "manpath.h"
1.1 schwarze 38:
39: #define MANDOC_BUFSZ BUFSIZ
40: #define MANDOC_SLOP 1024
41:
1.11 schwarze 42: #define MANDOC_SRC 0x1
43: #define MANDOC_FORM 0x2
44:
1.2 schwarze 45: /* Tiny list for files. No need to bring in QUEUE. */
46:
47: struct of {
48: char *fname; /* heap-allocated */
1.6 schwarze 49: char *sec;
50: char *arch;
51: char *title;
1.11 schwarze 52: int src_form;
1.2 schwarze 53: struct of *next; /* NULL for last one */
54: struct of *first; /* first in list */
55: };
56:
1.1 schwarze 57: /* Buffer for storing growable data. */
58:
59: struct buf {
60: char *cp;
1.2 schwarze 61: size_t len; /* current length */
62: size_t size; /* total buffer size */
1.1 schwarze 63: };
64:
65: /* Operation we're going to perform. */
66:
67: enum op {
68: OP_NEW = 0, /* new database */
1.2 schwarze 69: OP_UPDATE, /* delete/add entries in existing database */
1.1 schwarze 70: OP_DELETE /* delete entries from existing database */
71: };
72:
73: #define MAN_ARGS DB *hash, \
74: struct buf *buf, \
75: struct buf *dbuf, \
76: const struct man_node *n
77: #define MDOC_ARGS DB *hash, \
78: struct buf *buf, \
79: struct buf *dbuf, \
80: const struct mdoc_node *n, \
81: const struct mdoc_meta *m
82:
83: static void buf_appendmdoc(struct buf *,
84: const struct mdoc_node *, int);
85: static void buf_append(struct buf *, const char *);
86: static void buf_appendb(struct buf *,
87: const void *, size_t);
88: static void dbt_put(DB *, const char *, DBT *, DBT *);
1.9 schwarze 89: static void hash_put(DB *, const struct buf *, uint64_t);
1.1 schwarze 90: static void hash_reset(DB **);
1.2 schwarze 91: static void index_merge(const struct of *, struct mparse *,
1.13 schwarze 92: struct buf *, struct buf *, DB *,
93: DB *, const char *, DB *, const char *,
1.2 schwarze 94: recno_t, const recno_t *, size_t);
95: static void index_prune(const struct of *, DB *,
96: const char *, DB *, const char *,
1.13 schwarze 97: recno_t *, recno_t **, size_t *);
98: static void ofile_argbuild(int, char *[], struct of **);
1.6 schwarze 99: static int ofile_dirbuild(const char *, const char *,
1.13 schwarze 100: const char *, int, struct of **);
1.2 schwarze 101: static void ofile_free(struct of *);
1.11 schwarze 102: static void pformatted(DB *, struct buf *, struct buf *,
103: const struct of *);
1.1 schwarze 104: static int pman_node(MAN_ARGS);
105: static void pmdoc_node(MDOC_ARGS);
106: static void pmdoc_An(MDOC_ARGS);
107: static void pmdoc_Cd(MDOC_ARGS);
108: static void pmdoc_Er(MDOC_ARGS);
109: static void pmdoc_Ev(MDOC_ARGS);
110: static void pmdoc_Fd(MDOC_ARGS);
111: static void pmdoc_In(MDOC_ARGS);
112: static void pmdoc_Fn(MDOC_ARGS);
113: static void pmdoc_Fo(MDOC_ARGS);
114: static void pmdoc_Nd(MDOC_ARGS);
115: static void pmdoc_Nm(MDOC_ARGS);
116: static void pmdoc_Pa(MDOC_ARGS);
117: static void pmdoc_St(MDOC_ARGS);
118: static void pmdoc_Vt(MDOC_ARGS);
119: static void pmdoc_Xr(MDOC_ARGS);
120: static void usage(void);
121:
122: typedef void (*pmdoc_nf)(MDOC_ARGS);
123:
124: static const pmdoc_nf mdocs[MDOC_MAX] = {
125: NULL, /* Ap */
126: NULL, /* Dd */
127: NULL, /* Dt */
128: NULL, /* Os */
129: NULL, /* Sh */
130: NULL, /* Ss */
131: NULL, /* Pp */
132: NULL, /* D1 */
133: NULL, /* Dl */
134: NULL, /* Bd */
135: NULL, /* Ed */
136: NULL, /* Bl */
137: NULL, /* El */
138: NULL, /* It */
139: NULL, /* Ad */
140: pmdoc_An, /* An */
141: NULL, /* Ar */
142: pmdoc_Cd, /* Cd */
143: NULL, /* Cm */
144: NULL, /* Dv */
145: pmdoc_Er, /* Er */
146: pmdoc_Ev, /* Ev */
147: NULL, /* Ex */
148: NULL, /* Fa */
149: pmdoc_Fd, /* Fd */
150: NULL, /* Fl */
151: pmdoc_Fn, /* Fn */
152: NULL, /* Ft */
153: NULL, /* Ic */
154: pmdoc_In, /* In */
155: NULL, /* Li */
156: pmdoc_Nd, /* Nd */
157: pmdoc_Nm, /* Nm */
158: NULL, /* Op */
159: NULL, /* Ot */
160: pmdoc_Pa, /* Pa */
161: NULL, /* Rv */
162: pmdoc_St, /* St */
163: pmdoc_Vt, /* Va */
164: pmdoc_Vt, /* Vt */
165: pmdoc_Xr, /* Xr */
166: NULL, /* %A */
167: NULL, /* %B */
168: NULL, /* %D */
169: NULL, /* %I */
170: NULL, /* %J */
171: NULL, /* %N */
172: NULL, /* %O */
173: NULL, /* %P */
174: NULL, /* %R */
175: NULL, /* %T */
176: NULL, /* %V */
177: NULL, /* Ac */
178: NULL, /* Ao */
179: NULL, /* Aq */
180: NULL, /* At */
181: NULL, /* Bc */
182: NULL, /* Bf */
183: NULL, /* Bo */
184: NULL, /* Bq */
185: NULL, /* Bsx */
186: NULL, /* Bx */
187: NULL, /* Db */
188: NULL, /* Dc */
189: NULL, /* Do */
190: NULL, /* Dq */
191: NULL, /* Ec */
192: NULL, /* Ef */
193: NULL, /* Em */
194: NULL, /* Eo */
195: NULL, /* Fx */
196: NULL, /* Ms */
197: NULL, /* No */
198: NULL, /* Ns */
199: NULL, /* Nx */
200: NULL, /* Ox */
201: NULL, /* Pc */
202: NULL, /* Pf */
203: NULL, /* Po */
204: NULL, /* Pq */
205: NULL, /* Qc */
206: NULL, /* Ql */
207: NULL, /* Qo */
208: NULL, /* Qq */
209: NULL, /* Re */
210: NULL, /* Rs */
211: NULL, /* Sc */
212: NULL, /* So */
213: NULL, /* Sq */
214: NULL, /* Sm */
215: NULL, /* Sx */
216: NULL, /* Sy */
217: NULL, /* Tn */
218: NULL, /* Ux */
219: NULL, /* Xc */
220: NULL, /* Xo */
221: pmdoc_Fo, /* Fo */
222: NULL, /* Fc */
223: NULL, /* Oo */
224: NULL, /* Oc */
225: NULL, /* Bk */
226: NULL, /* Ek */
227: NULL, /* Bt */
228: NULL, /* Hf */
229: NULL, /* Fr */
230: NULL, /* Ud */
231: NULL, /* Lb */
232: NULL, /* Lp */
233: NULL, /* Lk */
234: NULL, /* Mt */
235: NULL, /* Brq */
236: NULL, /* Bro */
237: NULL, /* Brc */
238: NULL, /* %C */
239: NULL, /* Es */
240: NULL, /* En */
241: NULL, /* Dx */
242: NULL, /* %Q */
243: NULL, /* br */
244: NULL, /* sp */
245: NULL, /* %U */
246: NULL, /* Ta */
247: };
248:
249: static const char *progname;
1.13 schwarze 250: static int use_all; /* Use all directories and files. */
251: static int verb; /* Output verbosity level. */
1.1 schwarze 252:
253: int
1.3 schwarze 254: mandocdb(int argc, char *argv[])
1.1 schwarze 255: {
256: struct mparse *mp; /* parse sequence */
1.10 schwarze 257: struct manpaths dirs;
1.1 schwarze 258: enum op op; /* current operation */
1.2 schwarze 259: const char *dir;
1.1 schwarze 260: char ibuf[MAXPATHLEN], /* index fname */
1.2 schwarze 261: fbuf[MAXPATHLEN]; /* btree fname */
1.13 schwarze 262: int ch, i, flags;
1.1 schwarze 263: DB *idx, /* index database */
264: *db, /* keyword database */
265: *hash; /* temporary keyword hashtable */
266: BTREEINFO info; /* btree configuration */
1.8 schwarze 267: recno_t maxrec; /* last record number in the index */
268: recno_t *recs; /* the numbers of all empty records */
1.2 schwarze 269: size_t sz1, sz2,
1.8 schwarze 270: recsz, /* number of allocated slots in recs */
271: reccur; /* current number of empty records */
1.1 schwarze 272: struct buf buf, /* keyword buffer */
273: dbuf; /* description buffer */
1.2 schwarze 274: struct of *of; /* list of files for processing */
1.1 schwarze 275: extern int optind;
276: extern char *optarg;
277:
278: progname = strrchr(argv[0], '/');
279: if (progname == NULL)
280: progname = argv[0];
281: else
282: ++progname;
283:
1.10 schwarze 284: memset(&dirs, 0, sizeof(struct manpaths));
285:
1.1 schwarze 286: verb = 0;
1.6 schwarze 287: use_all = 0;
1.2 schwarze 288: of = NULL;
1.1 schwarze 289: db = idx = NULL;
290: mp = NULL;
291: hash = NULL;
292: recs = NULL;
293: recsz = reccur = 0;
294: maxrec = 0;
295: op = OP_NEW;
1.2 schwarze 296: dir = NULL;
1.1 schwarze 297:
1.6 schwarze 298: while (-1 != (ch = getopt(argc, argv, "ad:u:v")))
1.1 schwarze 299: switch (ch) {
1.6 schwarze 300: case ('a'):
301: use_all = 1;
302: break;
1.1 schwarze 303: case ('d'):
304: dir = optarg;
1.2 schwarze 305: op = OP_UPDATE;
1.1 schwarze 306: break;
1.2 schwarze 307: case ('u'):
308: dir = optarg;
1.1 schwarze 309: op = OP_DELETE;
310: break;
311: case ('v'):
312: verb++;
313: break;
314: default:
315: usage();
316: return((int)MANDOCLEVEL_BADARG);
317: }
318:
319: argc -= optind;
320: argv += optind;
321:
1.2 schwarze 322: memset(&info, 0, sizeof(BTREEINFO));
323: info.flags = R_DUP;
324:
325: mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL);
326:
327: memset(&buf, 0, sizeof(struct buf));
328: memset(&dbuf, 0, sizeof(struct buf));
329:
330: buf.size = dbuf.size = MANDOC_BUFSZ;
331:
332: buf.cp = mandoc_malloc(buf.size);
333: dbuf.cp = mandoc_malloc(dbuf.size);
334:
335: flags = OP_NEW == op ? O_CREAT|O_TRUNC|O_RDWR : O_CREAT|O_RDWR;
1.1 schwarze 336:
1.2 schwarze 337: if (OP_UPDATE == op || OP_DELETE == op) {
338: ibuf[0] = fbuf[0] = '\0';
1.1 schwarze 339:
1.2 schwarze 340: strlcat(fbuf, dir, MAXPATHLEN);
341: strlcat(fbuf, "/", MAXPATHLEN);
342: sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
343:
344: strlcat(ibuf, dir, MAXPATHLEN);
345: strlcat(ibuf, "/", MAXPATHLEN);
346: sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
347:
348: if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) {
349: fprintf(stderr, "%s: Path too long\n", dir);
350: exit((int)MANDOCLEVEL_BADARG);
351: }
1.1 schwarze 352:
1.2 schwarze 353: db = dbopen(fbuf, flags, 0644, DB_BTREE, &info);
354: idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL);
1.1 schwarze 355:
1.2 schwarze 356: if (NULL == db) {
357: perror(fbuf);
358: exit((int)MANDOCLEVEL_SYSERR);
1.8 schwarze 359: } else if (NULL == idx) {
1.2 schwarze 360: perror(ibuf);
361: exit((int)MANDOCLEVEL_SYSERR);
362: }
1.1 schwarze 363:
1.2 schwarze 364: if (verb > 2) {
365: printf("%s: Opened\n", fbuf);
366: printf("%s: Opened\n", ibuf);
367: }
1.1 schwarze 368:
1.13 schwarze 369: ofile_argbuild(argc, argv, &of);
1.2 schwarze 370: if (NULL == of)
371: goto out;
372:
373: of = of->first;
374:
1.13 schwarze 375: index_prune(of, db, fbuf, idx, ibuf,
1.2 schwarze 376: &maxrec, &recs, &recsz);
377:
1.15 ! schwarze 378: /*
! 379: * Go to the root of the respective manual tree
! 380: * such that .so links work. In case of failure,
! 381: * just prod on, even though .so links won't work.
! 382: */
! 383:
1.14 schwarze 384: if (OP_UPDATE == op) {
385: chdir(dir);
1.10 schwarze 386: index_merge(of, mp, &dbuf, &buf, hash,
1.13 schwarze 387: db, fbuf, idx, ibuf,
388: maxrec, recs, reccur);
1.14 schwarze 389: }
1.1 schwarze 390:
391: goto out;
392: }
393:
1.10 schwarze 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
406: manpath_parse(&dirs, NULL, NULL);
1.7 schwarze 407:
1.10 schwarze 408: for (i = 0; i < dirs.sz; i++) {
1.2 schwarze 409: ibuf[0] = fbuf[0] = '\0';
410:
1.10 schwarze 411: strlcat(fbuf, dirs.paths[i], MAXPATHLEN);
1.2 schwarze 412: strlcat(fbuf, "/", MAXPATHLEN);
413: sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
414:
1.10 schwarze 415: strlcat(ibuf, dirs.paths[i], MAXPATHLEN);
1.2 schwarze 416: strlcat(ibuf, "/", MAXPATHLEN);
417: sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
418:
419: if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) {
1.10 schwarze 420: fprintf(stderr, "%s: Path too long\n",
421: dirs.paths[i]);
1.2 schwarze 422: exit((int)MANDOCLEVEL_BADARG);
423: }
424:
1.7 schwarze 425: if (db)
426: (*db->close)(db);
427: if (idx)
428: (*idx->close)(idx);
429:
1.2 schwarze 430: db = dbopen(fbuf, flags, 0644, DB_BTREE, &info);
431: idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL);
1.1 schwarze 432:
1.2 schwarze 433: if (NULL == db) {
434: perror(fbuf);
435: exit((int)MANDOCLEVEL_SYSERR);
1.8 schwarze 436: } else if (NULL == idx) {
1.2 schwarze 437: perror(ibuf);
438: exit((int)MANDOCLEVEL_SYSERR);
439: }
1.1 schwarze 440:
1.2 schwarze 441: if (verb > 2) {
442: printf("%s: Truncated\n", fbuf);
443: printf("%s: Truncated\n", ibuf);
444: }
1.1 schwarze 445:
1.2 schwarze 446: ofile_free(of);
447: of = NULL;
1.1 schwarze 448:
1.10 schwarze 449: if ( ! ofile_dirbuild(dirs.paths[i], NULL, NULL,
1.13 schwarze 450: 0, &of))
1.2 schwarze 451: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 452:
1.2 schwarze 453: if (NULL == of)
454: continue;
1.1 schwarze 455:
1.2 schwarze 456: of = of->first;
1.15 ! schwarze 457:
! 458: /*
! 459: * Go to the root of the respective manual tree
! 460: * such that .so links work. In case of failure,
! 461: * just prod on, even though .so links won't work.
! 462: */
1.1 schwarze 463:
1.14 schwarze 464: chdir(dirs.paths[i]);
1.10 schwarze 465: index_merge(of, mp, &dbuf, &buf, hash, db, fbuf,
1.13 schwarze 466: idx, ibuf, maxrec, recs, reccur);
1.1 schwarze 467: }
468:
1.2 schwarze 469: out:
470: if (db)
471: (*db->close)(db);
472: if (idx)
473: (*idx->close)(idx);
474: if (hash)
475: (*hash->close)(hash);
476: if (mp)
477: mparse_free(mp);
1.1 schwarze 478:
1.10 schwarze 479: manpath_free(&dirs);
1.2 schwarze 480: ofile_free(of);
481: free(buf.cp);
482: free(dbuf.cp);
483: free(recs);
1.1 schwarze 484:
1.2 schwarze 485: return(MANDOCLEVEL_OK);
486: }
1.1 schwarze 487:
1.2 schwarze 488: void
489: index_merge(const struct of *of, struct mparse *mp,
1.13 schwarze 490: struct buf *dbuf, struct buf *buf, DB *hash,
491: DB *db, const char *dbf, DB *idx, const char *idxf,
1.2 schwarze 492: recno_t maxrec, const recno_t *recs, size_t reccur)
493: {
494: recno_t rec;
495: int ch;
496: DBT key, val;
497: struct mdoc *mdoc;
498: struct man *man;
499: const char *fn, *msec, *mtitle, *arch;
500: size_t sv;
501: unsigned seq;
1.9 schwarze 502: struct db_val vbuf;
1.1 schwarze 503:
1.2 schwarze 504: for (rec = 0; of; of = of->next) {
505: fn = of->fname;
1.11 schwarze 506:
507: /*
508: * Reclaim an empty index record, if available.
509: */
510:
1.2 schwarze 511: if (reccur > 0) {
512: --reccur;
513: rec = recs[(int)reccur];
514: } else if (maxrec > 0) {
515: rec = maxrec;
516: maxrec = 0;
1.1 schwarze 517: } else
518: rec++;
519:
520: mparse_reset(mp);
521: hash_reset(&hash);
1.11 schwarze 522: mdoc = NULL;
523: man = NULL;
1.1 schwarze 524:
1.11 schwarze 525: /*
526: * Try interpreting the file as mdoc(7) or man(7)
527: * source code, unless it is already known to be
528: * formatted. Fall back to formatted mode.
529: */
530:
531: if ((MANDOC_SRC & of->src_form ||
532: ! (MANDOC_FORM & of->src_form)) &&
533: MANDOCLEVEL_FATAL > mparse_readfd(mp, -1, fn))
534: mparse_result(mp, &mdoc, &man);
535:
536: if (NULL != mdoc) {
537: msec = mdoc_meta(mdoc)->msec;
538: arch = mdoc_meta(mdoc)->arch;
539: mtitle = mdoc_meta(mdoc)->title;
540: } else if (NULL != man) {
541: msec = man_meta(man)->msec;
542: arch = NULL;
543: mtitle = man_meta(man)->title;
544: } else {
545: msec = of->sec;
546: arch = of->arch;
547: mtitle = of->title;
1.1 schwarze 548: }
549:
1.6 schwarze 550: /*
1.8 schwarze 551: * By default, skip a file if the manual section
552: * and architecture given in the file disagree
553: * with the directory where the file is located.
1.6 schwarze 554: */
555:
556: if (0 == use_all) {
557: assert(of->sec);
558: assert(msec);
559: if (strcmp(msec, of->sec))
560: continue;
561:
562: if (NULL == arch) {
563: if (NULL != of->arch)
564: continue;
565: } else if (NULL == of->arch ||
566: strcmp(arch, of->arch))
567: continue;
568: }
569:
1.1 schwarze 570: if (NULL == arch)
571: arch = "";
572:
573: /*
1.8 schwarze 574: * By default, skip a file if the title given
575: * in the file disagrees with the file name.
576: * If both agree, use the file name as the title,
577: * because the one in the file usually is all caps.
1.6 schwarze 578: */
579:
580: assert(of->title);
581: assert(mtitle);
582:
583: if (0 == strcasecmp(mtitle, of->title))
584: mtitle = of->title;
585: else if (0 == use_all)
586: continue;
587:
588: /*
1.1 schwarze 589: * The index record value consists of a nil-terminated
590: * filename, a nil-terminated manual section, and a
591: * nil-terminated description. Since the description
592: * may not be set, we set a sentinel to see if we're
593: * going to write a nil byte in its place.
594: */
595:
1.2 schwarze 596: dbuf->len = 0;
1.12 schwarze 597: buf_append(dbuf, mdoc ? "mdoc" : (man ? "man" : "cat"));
1.2 schwarze 598: buf_appendb(dbuf, fn, strlen(fn) + 1);
599: buf_appendb(dbuf, msec, strlen(msec) + 1);
600: buf_appendb(dbuf, mtitle, strlen(mtitle) + 1);
601: buf_appendb(dbuf, arch, strlen(arch) + 1);
1.1 schwarze 602:
1.2 schwarze 603: sv = dbuf->len;
1.1 schwarze 604:
605: /* Fix the record number in the btree value. */
606:
607: if (mdoc)
1.2 schwarze 608: pmdoc_node(hash, buf, dbuf,
1.1 schwarze 609: mdoc_node(mdoc), mdoc_meta(mdoc));
1.11 schwarze 610: else if (man)
1.2 schwarze 611: pman_node(hash, buf, dbuf, man_node(man));
1.11 schwarze 612: else
613: pformatted(hash, buf, dbuf, of);
1.1 schwarze 614:
615: /*
616: * Copy from the in-memory hashtable of pending keywords
617: * into the database.
618: */
619:
1.9 schwarze 620: vbuf.rec = rec;
1.1 schwarze 621: seq = R_FIRST;
622: while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) {
623: seq = R_NEXT;
624:
1.9 schwarze 625: vbuf.mask = *(uint64_t *)val.data;
626: val.size = sizeof(struct db_val);
627: val.data = &vbuf;
1.1 schwarze 628:
629: if (verb > 1)
1.2 schwarze 630: printf("%s: Added keyword: %s\n",
631: fn, (char *)key.data);
632: dbt_put(db, dbf, &key, &val);
1.1 schwarze 633: }
634: if (ch < 0) {
635: perror("hash");
636: exit((int)MANDOCLEVEL_SYSERR);
637: }
638:
639: /*
640: * Apply to the index. If we haven't had a description
641: * set, put an empty one in now.
642: */
643:
1.2 schwarze 644: if (dbuf->len == sv)
645: buf_appendb(dbuf, "", 1);
1.1 schwarze 646:
647: key.data = &rec;
648: key.size = sizeof(recno_t);
649:
1.2 schwarze 650: val.data = dbuf->cp;
651: val.size = dbuf->len;
1.1 schwarze 652:
1.2 schwarze 653: if (verb)
1.1 schwarze 654: printf("%s: Added index\n", fn);
1.2 schwarze 655: dbt_put(idx, idxf, &key, &val);
656: }
657: }
658:
659: /*
660: * Scan through all entries in the index file `idx' and prune those
661: * entries in `ofile'.
662: * Pruning consists of removing from `db', then invalidating the entry
663: * in `idx' (zeroing its value size).
664: */
665: static void
666: index_prune(const struct of *ofile, DB *db, const char *dbf,
1.13 schwarze 667: DB *idx, const char *idxf,
1.2 schwarze 668: recno_t *maxrec, recno_t **recs, size_t *recsz)
669: {
670: const struct of *of;
671: const char *fn;
1.9 schwarze 672: struct db_val *vbuf;
1.2 schwarze 673: unsigned seq, sseq;
674: DBT key, val;
675: size_t reccur;
676: int ch;
677:
678: reccur = 0;
679: seq = R_FIRST;
680: while (0 == (ch = (*idx->seq)(idx, &key, &val, seq))) {
681: seq = R_NEXT;
682: *maxrec = *(recno_t *)key.data;
683: if (0 == val.size) {
684: if (reccur >= *recsz) {
685: *recsz += MANDOC_SLOP;
686: *recs = mandoc_realloc(*recs,
687: *recsz * sizeof(recno_t));
688: }
689: (*recs)[(int)reccur] = *maxrec;
690: reccur++;
691: continue;
692: }
693:
694: fn = (char *)val.data;
695: for (of = ofile; of; of = of->next)
696: if (0 == strcmp(fn, of->fname))
697: break;
698:
699: if (NULL == of)
700: continue;
701:
702: sseq = R_FIRST;
703: while (0 == (ch = (*db->seq)(db, &key, &val, sseq))) {
704: sseq = R_NEXT;
1.9 schwarze 705: assert(sizeof(struct db_val) == val.size);
706: vbuf = val.data;
707: if (*maxrec != vbuf->rec)
1.2 schwarze 708: continue;
709: if (verb)
710: printf("%s: Deleted keyword: %s\n",
711: fn, (char *)key.data);
712: ch = (*db->del)(db, &key, R_CURSOR);
713: if (ch < 0)
714: break;
715: }
716: if (ch < 0) {
717: perror(dbf);
718: exit((int)MANDOCLEVEL_SYSERR);
719: }
1.1 schwarze 720:
1.2 schwarze 721: if (verb)
722: printf("%s: Deleted index\n", fn);
1.1 schwarze 723:
1.2 schwarze 724: val.size = 0;
725: ch = (*idx->put)(idx, &key, &val, R_CURSOR);
726: if (ch < 0) {
727: perror(idxf);
728: exit((int)MANDOCLEVEL_SYSERR);
729: }
1.1 schwarze 730:
1.2 schwarze 731: if (reccur >= *recsz) {
732: *recsz += MANDOC_SLOP;
733: *recs = mandoc_realloc
734: (*recs, *recsz * sizeof(recno_t));
735: }
1.1 schwarze 736:
1.2 schwarze 737: (*recs)[(int)reccur] = *maxrec;
738: reccur++;
739: }
740: (*maxrec)++;
1.1 schwarze 741: }
742:
743: /*
744: * Grow the buffer (if necessary) and copy in a binary string.
745: */
746: static void
747: buf_appendb(struct buf *buf, const void *cp, size_t sz)
748: {
749:
750: /* Overshoot by MANDOC_BUFSZ. */
751:
752: while (buf->len + sz >= buf->size) {
753: buf->size = buf->len + sz + MANDOC_BUFSZ;
754: buf->cp = mandoc_realloc(buf->cp, buf->size);
755: }
756:
757: memcpy(buf->cp + (int)buf->len, cp, sz);
758: buf->len += sz;
759: }
760:
761: /*
762: * Append a nil-terminated string to the buffer.
763: * This can be invoked multiple times.
764: * The buffer string will be nil-terminated.
765: * If invoked multiple times, a space is put between strings.
766: */
767: static void
768: buf_append(struct buf *buf, const char *cp)
769: {
770: size_t sz;
771:
772: if (0 == (sz = strlen(cp)))
773: return;
774:
775: if (buf->len)
776: buf->cp[(int)buf->len - 1] = ' ';
777:
778: buf_appendb(buf, cp, sz + 1);
779: }
780:
781: /*
782: * Recursively add all text from a given node.
783: * This is optimised for general mdoc nodes in this context, which do
784: * not consist of subexpressions and having a recursive call for n->next
785: * would be wasteful.
786: * The "f" variable should be 0 unless called from pmdoc_Nd for the
787: * description buffer, which does not start at the beginning of the
788: * buffer.
789: */
790: static void
791: buf_appendmdoc(struct buf *buf, const struct mdoc_node *n, int f)
792: {
793:
794: for ( ; n; n = n->next) {
795: if (n->child)
796: buf_appendmdoc(buf, n->child, f);
797:
798: if (MDOC_TEXT == n->type && f) {
799: f = 0;
800: buf_appendb(buf, n->string,
801: strlen(n->string) + 1);
802: } else if (MDOC_TEXT == n->type)
803: buf_append(buf, n->string);
804:
805: }
806: }
807:
808: /* ARGSUSED */
809: static void
810: pmdoc_An(MDOC_ARGS)
811: {
812:
813: if (SEC_AUTHORS != n->sec)
814: return;
815:
816: buf_appendmdoc(buf, n->child, 0);
1.5 schwarze 817: hash_put(hash, buf, TYPE_An);
1.1 schwarze 818: }
819:
820: static void
821: hash_reset(DB **db)
822: {
823: DB *hash;
824:
825: if (NULL != (hash = *db))
826: (*hash->close)(hash);
827:
1.2 schwarze 828: *db = dbopen(NULL, O_CREAT|O_RDWR, 0644, DB_HASH, NULL);
1.1 schwarze 829: if (NULL == *db) {
830: perror("hash");
831: exit((int)MANDOCLEVEL_SYSERR);
832: }
833: }
834:
835: /* ARGSUSED */
836: static void
837: pmdoc_Fd(MDOC_ARGS)
838: {
839: const char *start, *end;
840: size_t sz;
841:
842: if (SEC_SYNOPSIS != n->sec)
843: return;
844: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
845: return;
846:
847: /*
848: * Only consider those `Fd' macro fields that begin with an
849: * "inclusion" token (versus, e.g., #define).
850: */
851: if (strcmp("#include", n->string))
852: return;
853:
854: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
855: return;
856:
857: /*
858: * Strip away the enclosing angle brackets and make sure we're
859: * not zero-length.
860: */
861:
862: start = n->string;
863: if ('<' == *start || '"' == *start)
864: start++;
865:
866: if (0 == (sz = strlen(start)))
867: return;
868:
869: end = &start[(int)sz - 1];
870: if ('>' == *end || '"' == *end)
871: end--;
872:
873: assert(end >= start);
874:
875: buf_appendb(buf, start, (size_t)(end - start + 1));
876: buf_appendb(buf, "", 1);
877:
1.5 schwarze 878: hash_put(hash, buf, TYPE_In);
1.1 schwarze 879: }
880:
881: /* ARGSUSED */
882: static void
883: pmdoc_Cd(MDOC_ARGS)
884: {
885:
886: if (SEC_SYNOPSIS != n->sec)
887: return;
888:
889: buf_appendmdoc(buf, n->child, 0);
1.5 schwarze 890: hash_put(hash, buf, TYPE_Cd);
1.1 schwarze 891: }
892:
893: /* ARGSUSED */
894: static void
895: pmdoc_In(MDOC_ARGS)
896: {
897:
898: if (SEC_SYNOPSIS != n->sec)
899: return;
900: if (NULL == n->child || MDOC_TEXT != n->child->type)
901: return;
902:
903: buf_append(buf, n->child->string);
1.5 schwarze 904: hash_put(hash, buf, TYPE_In);
1.1 schwarze 905: }
906:
907: /* ARGSUSED */
908: static void
909: pmdoc_Fn(MDOC_ARGS)
910: {
911: const char *cp;
912:
913: if (SEC_SYNOPSIS != n->sec)
914: return;
915: if (NULL == n->child || MDOC_TEXT != n->child->type)
916: return;
917:
918: /* .Fn "struct type *arg" "foo" */
919:
920: cp = strrchr(n->child->string, ' ');
921: if (NULL == cp)
922: cp = n->child->string;
923:
924: /* Strip away pointer symbol. */
925:
926: while ('*' == *cp)
927: cp++;
928:
929: buf_append(buf, cp);
1.5 schwarze 930: hash_put(hash, buf, TYPE_Fn);
1.1 schwarze 931: }
932:
933: /* ARGSUSED */
934: static void
935: pmdoc_St(MDOC_ARGS)
936: {
937:
938: if (SEC_STANDARDS != n->sec)
939: return;
940: if (NULL == n->child || MDOC_TEXT != n->child->type)
941: return;
942:
943: buf_append(buf, n->child->string);
1.5 schwarze 944: hash_put(hash, buf, TYPE_St);
1.1 schwarze 945: }
946:
947: /* ARGSUSED */
948: static void
949: pmdoc_Xr(MDOC_ARGS)
950: {
951:
952: if (NULL == (n = n->child))
953: return;
954:
955: buf_appendb(buf, n->string, strlen(n->string));
956:
957: if (NULL != (n = n->next)) {
958: buf_appendb(buf, ".", 1);
959: buf_appendb(buf, n->string, strlen(n->string) + 1);
960: } else
961: buf_appendb(buf, ".", 2);
962:
1.5 schwarze 963: hash_put(hash, buf, TYPE_Xr);
1.1 schwarze 964: }
965:
966: /* ARGSUSED */
967: static void
968: pmdoc_Vt(MDOC_ARGS)
969: {
970: const char *start;
971: size_t sz;
972:
973: if (SEC_SYNOPSIS != n->sec)
974: return;
975: if (MDOC_Vt == n->tok && MDOC_BODY != n->type)
976: return;
977: if (NULL == n->last || MDOC_TEXT != n->last->type)
978: return;
979:
980: /*
981: * Strip away leading pointer symbol '*' and trailing ';'.
982: */
983:
984: start = n->last->string;
985:
986: while ('*' == *start)
987: start++;
988:
989: if (0 == (sz = strlen(start)))
990: return;
991:
992: if (';' == start[(int)sz - 1])
993: sz--;
994:
995: if (0 == sz)
996: return;
997:
998: buf_appendb(buf, start, sz);
999: buf_appendb(buf, "", 1);
1.5 schwarze 1000: hash_put(hash, buf, TYPE_Va);
1.1 schwarze 1001: }
1002:
1003: /* ARGSUSED */
1004: static void
1005: pmdoc_Fo(MDOC_ARGS)
1006: {
1007:
1008: if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
1009: return;
1010: if (NULL == n->child || MDOC_TEXT != n->child->type)
1011: return;
1012:
1013: buf_append(buf, n->child->string);
1.5 schwarze 1014: hash_put(hash, buf, TYPE_Fn);
1.1 schwarze 1015: }
1016:
1017:
1018: /* ARGSUSED */
1019: static void
1020: pmdoc_Nd(MDOC_ARGS)
1021: {
1022:
1023: if (MDOC_BODY != n->type)
1024: return;
1025:
1026: buf_appendmdoc(dbuf, n->child, 1);
1027: buf_appendmdoc(buf, n->child, 0);
1028:
1.5 schwarze 1029: hash_put(hash, buf, TYPE_Nd);
1.1 schwarze 1030: }
1031:
1032: /* ARGSUSED */
1033: static void
1034: pmdoc_Er(MDOC_ARGS)
1035: {
1036:
1037: if (SEC_ERRORS != n->sec)
1038: return;
1039:
1040: buf_appendmdoc(buf, n->child, 0);
1.5 schwarze 1041: hash_put(hash, buf, TYPE_Er);
1.1 schwarze 1042: }
1043:
1044: /* ARGSUSED */
1045: static void
1046: pmdoc_Ev(MDOC_ARGS)
1047: {
1048:
1049: if (SEC_ENVIRONMENT != n->sec)
1050: return;
1051:
1052: buf_appendmdoc(buf, n->child, 0);
1.5 schwarze 1053: hash_put(hash, buf, TYPE_Ev);
1.1 schwarze 1054: }
1055:
1056: /* ARGSUSED */
1057: static void
1058: pmdoc_Pa(MDOC_ARGS)
1059: {
1060:
1061: if (SEC_FILES != n->sec)
1062: return;
1063:
1064: buf_appendmdoc(buf, n->child, 0);
1.5 schwarze 1065: hash_put(hash, buf, TYPE_Pa);
1.1 schwarze 1066: }
1067:
1068: /* ARGSUSED */
1069: static void
1070: pmdoc_Nm(MDOC_ARGS)
1071: {
1072:
1073: if (SEC_NAME == n->sec) {
1074: buf_appendmdoc(buf, n->child, 0);
1.5 schwarze 1075: hash_put(hash, buf, TYPE_Nm);
1.1 schwarze 1076: return;
1077: } else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
1078: return;
1079:
1080: if (NULL == n->child)
1081: buf_append(buf, m->name);
1082:
1083: buf_appendmdoc(buf, n->child, 0);
1.5 schwarze 1084: hash_put(hash, buf, TYPE_Nm);
1.1 schwarze 1085: }
1086:
1087: static void
1.9 schwarze 1088: hash_put(DB *db, const struct buf *buf, uint64_t mask)
1.1 schwarze 1089: {
1090: DBT key, val;
1091: int rc;
1092:
1093: if (buf->len < 2)
1094: return;
1095:
1096: key.data = buf->cp;
1097: key.size = buf->len;
1098:
1099: if ((rc = (*db->get)(db, &key, &val, 0)) < 0) {
1100: perror("hash");
1101: exit((int)MANDOCLEVEL_SYSERR);
1102: } else if (0 == rc)
1.9 schwarze 1103: mask |= *(uint64_t *)val.data;
1.1 schwarze 1104:
1105: val.data = &mask;
1.9 schwarze 1106: val.size = sizeof(uint64_t);
1.1 schwarze 1107:
1108: if ((rc = (*db->put)(db, &key, &val, 0)) < 0) {
1109: perror("hash");
1110: exit((int)MANDOCLEVEL_SYSERR);
1111: }
1112: }
1113:
1114: static void
1115: dbt_put(DB *db, const char *dbn, DBT *key, DBT *val)
1116: {
1117:
1118: assert(key->size);
1119: assert(val->size);
1120:
1121: if (0 == (*db->put)(db, key, val, 0))
1122: return;
1123:
1124: perror(dbn);
1125: exit((int)MANDOCLEVEL_SYSERR);
1126: /* NOTREACHED */
1127: }
1128:
1129: /*
1130: * Call out to per-macro handlers after clearing the persistent database
1131: * key. If the macro sets the database key, flush it to the database.
1132: */
1133: static void
1134: pmdoc_node(MDOC_ARGS)
1135: {
1136:
1137: if (NULL == n)
1138: return;
1139:
1140: switch (n->type) {
1141: case (MDOC_HEAD):
1142: /* FALLTHROUGH */
1143: case (MDOC_BODY):
1144: /* FALLTHROUGH */
1145: case (MDOC_TAIL):
1146: /* FALLTHROUGH */
1147: case (MDOC_BLOCK):
1148: /* FALLTHROUGH */
1149: case (MDOC_ELEM):
1150: if (NULL == mdocs[n->tok])
1151: break;
1152:
1153: buf->len = 0;
1154: (*mdocs[n->tok])(hash, buf, dbuf, n, m);
1155: break;
1156: default:
1157: break;
1158: }
1159:
1160: pmdoc_node(hash, buf, dbuf, n->child, m);
1161: pmdoc_node(hash, buf, dbuf, n->next, m);
1162: }
1163:
1164: static int
1165: pman_node(MAN_ARGS)
1166: {
1167: const struct man_node *head, *body;
1168: const char *start, *sv;
1169: size_t sz;
1170:
1171: if (NULL == n)
1172: return(0);
1173:
1174: /*
1175: * We're only searching for one thing: the first text child in
1176: * the BODY of a NAME section. Since we don't keep track of
1177: * sections in -man, run some hoops to find out whether we're in
1178: * the correct section or not.
1179: */
1180:
1181: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1182: body = n;
1183: assert(body->parent);
1184: if (NULL != (head = body->parent->head) &&
1185: 1 == head->nchild &&
1186: NULL != (head = (head->child)) &&
1187: MAN_TEXT == head->type &&
1188: 0 == strcmp(head->string, "NAME") &&
1189: NULL != (body = body->child) &&
1190: MAN_TEXT == body->type) {
1191:
1192: assert(body->string);
1193: start = sv = body->string;
1194:
1195: /*
1196: * Go through a special heuristic dance here.
1197: * This is why -man manuals are great!
1198: * (I'm being sarcastic: my eyes are bleeding.)
1199: * Conventionally, one or more manual names are
1200: * comma-specified prior to a whitespace, then a
1201: * dash, then a description. Try to puzzle out
1202: * the name parts here.
1203: */
1204:
1205: for ( ;; ) {
1206: sz = strcspn(start, " ,");
1207: if ('\0' == start[(int)sz])
1208: break;
1209:
1210: buf->len = 0;
1211: buf_appendb(buf, start, sz);
1212: buf_appendb(buf, "", 1);
1213:
1.5 schwarze 1214: hash_put(hash, buf, TYPE_Nm);
1.1 schwarze 1215:
1216: if (' ' == start[(int)sz]) {
1217: start += (int)sz + 1;
1218: break;
1219: }
1220:
1221: assert(',' == start[(int)sz]);
1222: start += (int)sz + 1;
1223: while (' ' == *start)
1224: start++;
1225: }
1226:
1227: buf->len = 0;
1228:
1229: if (sv == start) {
1230: buf_append(buf, start);
1231: return(1);
1232: }
1233:
1234: while (' ' == *start)
1235: start++;
1236:
1237: if (0 == strncmp(start, "-", 1))
1238: start += 1;
1239: else if (0 == strncmp(start, "\\-", 2))
1240: start += 2;
1241: else if (0 == strncmp(start, "\\(en", 4))
1242: start += 4;
1243: else if (0 == strncmp(start, "\\(em", 4))
1244: start += 4;
1245:
1246: while (' ' == *start)
1247: start++;
1248:
1249: sz = strlen(start) + 1;
1250: buf_appendb(dbuf, start, sz);
1251: buf_appendb(buf, start, sz);
1252:
1.5 schwarze 1253: hash_put(hash, buf, TYPE_Nd);
1.1 schwarze 1254: }
1255: }
1256:
1.4 schwarze 1257: for (n = n->child; n; n = n->next)
1258: if (pman_node(hash, buf, dbuf, n))
1259: return(1);
1.1 schwarze 1260:
1261: return(0);
1262: }
1263:
1.11 schwarze 1264: /*
1265: * Parse a formatted manual page.
1266: * By necessity, this involves rather crude guesswork.
1267: */
1268: static void
1269: pformatted(DB *hash, struct buf *buf, struct buf *dbuf,
1270: const struct of *of)
1271: {
1272: FILE *stream;
1273: char *line, *p;
1274: size_t len, plen;
1275:
1276: if (NULL == (stream = fopen(of->fname, "r"))) {
1277: perror(of->fname);
1278: return;
1279: }
1280:
1281: /*
1282: * Always use the title derived from the filename up front,
1283: * do not even try to find it in the file. This also makes
1284: * sure we don't end up with an orphan index record, even if
1285: * the file content turns out to be completely unintelligible.
1286: */
1287:
1288: buf->len = 0;
1289: buf_append(buf, of->title);
1290: hash_put(hash, buf, TYPE_Nm);
1291:
1292: while (NULL != (line = fgetln(stream, &len)) && '\n' != *line)
1293: /* Skip to first blank line. */ ;
1294:
1295: while (NULL != (line = fgetln(stream, &len)) &&
1296: ('\n' == *line || ' ' == *line))
1297: /* Skip to first section header. */ ;
1298:
1299: /*
1300: * If no page content can be found,
1301: * reuse the page title as the page description.
1302: */
1303:
1304: if (NULL == (line = fgetln(stream, &len))) {
1305: buf_appendb(dbuf, buf->cp, buf->size);
1306: hash_put(hash, buf, TYPE_Nd);
1307: fclose(stream);
1308: return;
1309: }
1310: fclose(stream);
1311:
1312: /*
1313: * If there is a dash, skip to the text following it.
1314: */
1315:
1316: for (p = line, plen = len; plen; p++, plen--)
1317: if ('-' == *p)
1318: break;
1319: for ( ; plen; p++, plen--)
1320: if ('-' != *p && ' ' != *p && 8 != *p)
1321: break;
1322: if (0 == plen) {
1323: p = line;
1324: plen = len;
1325: }
1326:
1327: /*
1328: * Copy the rest of the line, but no more than 70 bytes.
1329: */
1330:
1331: if (70 < plen)
1332: plen = 70;
1333: p[plen-1] = '\0';
1334: buf_appendb(dbuf, p, plen);
1335: buf->len = 0;
1336: buf_appendb(buf, p, plen);
1337: hash_put(hash, buf, TYPE_Nd);
1338: }
1339:
1.1 schwarze 1340: static void
1.13 schwarze 1341: ofile_argbuild(int argc, char *argv[], struct of **of)
1.2 schwarze 1342: {
1.6 schwarze 1343: char buf[MAXPATHLEN];
1344: char *sec, *arch, *title, *p;
1.11 schwarze 1345: int i, src_form;
1.2 schwarze 1346: struct of *nof;
1347:
1348: for (i = 0; i < argc; i++) {
1.6 schwarze 1349:
1350: /*
1.8 schwarze 1351: * Try to infer the manual section, architecture and
1352: * page title from the path, assuming it looks like
1.11 schwarze 1353: * man*[/<arch>]/<title>.<section> or
1354: * cat<section>[/<arch>]/<title>.0
1.6 schwarze 1355: */
1356:
1357: if (strlcpy(buf, argv[i], sizeof(buf)) >= sizeof(buf)) {
1358: fprintf(stderr, "%s: Path too long\n", argv[i]);
1359: continue;
1360: }
1361: sec = arch = title = NULL;
1.11 schwarze 1362: src_form = 0;
1.6 schwarze 1363: p = strrchr(buf, '\0');
1364: while (p-- > buf) {
1365: if (NULL == sec && '.' == *p) {
1366: sec = p + 1;
1367: *p = '\0';
1.11 schwarze 1368: if ('0' == *sec)
1369: src_form |= MANDOC_FORM;
1370: else if ('1' <= *sec && '9' >= *sec)
1371: src_form |= MANDOC_SRC;
1.6 schwarze 1372: continue;
1373: }
1374: if ('/' != *p)
1375: continue;
1376: if (NULL == title) {
1377: title = p + 1;
1378: *p = '\0';
1379: continue;
1380: }
1.11 schwarze 1381: if (strncmp("man", p + 1, 3)) {
1382: src_form |= MANDOC_SRC;
1383: arch = p + 1;
1384: } else if (strncmp("cat", p + 1, 3)) {
1385: src_form |= MANDOC_FORM;
1.6 schwarze 1386: arch = p + 1;
1.11 schwarze 1387: }
1.6 schwarze 1388: break;
1389: }
1390: if (NULL == title)
1391: title = buf;
1392:
1393: /*
1394: * Build the file structure.
1395: */
1396:
1.2 schwarze 1397: nof = mandoc_calloc(1, sizeof(struct of));
1.6 schwarze 1398: nof->fname = mandoc_strdup(argv[i]);
1399: if (NULL != sec)
1400: nof->sec = mandoc_strdup(sec);
1401: if (NULL != arch)
1402: nof->arch = mandoc_strdup(arch);
1403: nof->title = mandoc_strdup(title);
1.11 schwarze 1404: nof->src_form = src_form;
1.6 schwarze 1405:
1406: /*
1407: * Add the structure to the list.
1408: */
1409:
1.2 schwarze 1410: if (verb > 2)
1411: printf("%s: Scheduling\n", argv[i]);
1412: if (NULL == *of) {
1413: *of = nof;
1414: (*of)->first = nof;
1415: } else {
1416: nof->first = (*of)->first;
1417: (*of)->next = nof;
1418: *of = nof;
1419: }
1420: }
1421: }
1422:
1423: /*
1424: * Recursively build up a list of files to parse.
1425: * We use this instead of ftw() and so on because I don't want global
1426: * variables hanging around.
1427: * This ignores the mandoc.db and mandoc.index files, but assumes that
1428: * everything else is a manual.
1429: * Pass in a pointer to a NULL structure for the first invocation.
1430: */
1431: static int
1.6 schwarze 1432: ofile_dirbuild(const char *dir, const char* psec, const char *parch,
1.13 schwarze 1433: int p_src_form, struct of **of)
1.2 schwarze 1434: {
1435: char buf[MAXPATHLEN];
1.11 schwarze 1436: struct stat sb;
1.2 schwarze 1437: size_t sz;
1438: DIR *d;
1.6 schwarze 1439: const char *fn, *sec, *arch;
1.11 schwarze 1440: char *p, *q, *suffix;
1.2 schwarze 1441: struct of *nof;
1442: struct dirent *dp;
1.11 schwarze 1443: int src_form;
1.2 schwarze 1444:
1445: if (NULL == (d = opendir(dir))) {
1446: perror(dir);
1447: return(0);
1448: }
1449:
1450: while (NULL != (dp = readdir(d))) {
1451: fn = dp->d_name;
1.6 schwarze 1452:
1453: if ('.' == *fn)
1454: continue;
1455:
1.11 schwarze 1456: src_form = p_src_form;
1457:
1.2 schwarze 1458: if (DT_DIR == dp->d_type) {
1.6 schwarze 1459: sec = psec;
1460: arch = parch;
1461:
1462: /*
1.8 schwarze 1463: * By default, only use directories called:
1.11 schwarze 1464: * man<section>/[<arch>/] or
1465: * cat<section>/[<arch>/]
1.6 schwarze 1466: */
1467:
1468: if (NULL == sec) {
1.11 schwarze 1469: if(0 == strncmp("man", fn, 3)) {
1470: src_form |= MANDOC_SRC;
1.6 schwarze 1471: sec = fn + 3;
1.11 schwarze 1472: } else if (0 == strncmp("cat", fn, 3)) {
1473: src_form |= MANDOC_FORM;
1474: sec = fn + 3;
1475: } else if (use_all)
1.6 schwarze 1476: sec = fn;
1477: else
1478: continue;
1479: } else if (NULL == arch && (use_all ||
1480: NULL == strchr(fn, '.')))
1481: arch = fn;
1482: else if (0 == use_all)
1.2 schwarze 1483: continue;
1484:
1485: buf[0] = '\0';
1486: strlcat(buf, dir, MAXPATHLEN);
1487: strlcat(buf, "/", MAXPATHLEN);
1488: sz = strlcat(buf, fn, MAXPATHLEN);
1489:
1.6 schwarze 1490: if (MAXPATHLEN <= sz) {
1491: fprintf(stderr, "%s: Path too long\n", dir);
1492: return(0);
1493: }
1494:
1495: if (verb > 2)
1496: printf("%s: Scanning\n", buf);
1497:
1498: if ( ! ofile_dirbuild(buf, sec, arch,
1.13 schwarze 1499: src_form, of))
1.6 schwarze 1500: return(0);
1501: }
1502: if (DT_REG != dp->d_type ||
1503: (NULL == psec && !use_all) ||
1504: !strcmp(MANDOC_DB, fn) ||
1505: !strcmp(MANDOC_IDX, fn))
1506: continue;
1507:
1508: /*
1.8 schwarze 1509: * By default, skip files where the file name suffix
1510: * does not agree with the section directory
1511: * they are located in.
1.6 schwarze 1512: */
1513:
1514: suffix = strrchr(fn, '.');
1515: if (0 == use_all) {
1516: if (NULL == suffix)
1.2 schwarze 1517: continue;
1.11 schwarze 1518: if ((MANDOC_SRC & src_form &&
1519: strcmp(suffix + 1, psec)) ||
1520: (MANDOC_FORM & src_form &&
1521: strcmp(suffix + 1, "0")))
1522: continue;
1523: }
1524: if (NULL != suffix) {
1525: if ('0' == suffix[1])
1526: src_form |= MANDOC_FORM;
1527: else if ('1' <= suffix[1] && '9' >= suffix[1])
1528: src_form |= MANDOC_SRC;
1529: }
1530:
1531:
1532: /*
1533: * Skip formatted manuals if a source version is
1534: * available. Ignore the age: it is very unlikely
1535: * that people install newer formatted base manuals
1536: * when they used to have source manuals before,
1537: * and in ports, old manuals get removed on update.
1538: */
1539: if (0 == use_all && MANDOC_FORM & src_form &&
1540: NULL != psec) {
1541: buf[0] = '\0';
1542: strlcat(buf, dir, MAXPATHLEN);
1543: p = strrchr(buf, '/');
1544: if (NULL == p)
1545: p = buf;
1546: else
1547: p++;
1548: if (0 == strncmp("cat", p, 3))
1549: memcpy(p, "man", 3);
1550: strlcat(buf, "/", MAXPATHLEN);
1551: sz = strlcat(buf, fn, MAXPATHLEN);
1552: if (sz >= MAXPATHLEN) {
1553: fprintf(stderr, "%s: Path too long\n", buf);
1.2 schwarze 1554: continue;
1.11 schwarze 1555: }
1556: q = strrchr(buf, '.');
1557: if (NULL != q && p < q++) {
1558: *q = '\0';
1559: sz = strlcat(buf, psec, MAXPATHLEN);
1560: if (sz >= MAXPATHLEN) {
1561: fprintf(stderr,
1562: "%s: Path too long\n", buf);
1563: continue;
1564: }
1565: if (0 == stat(buf, &sb))
1566: continue;
1567: }
1.2 schwarze 1568: }
1569:
1570: buf[0] = '\0';
1571: strlcat(buf, dir, MAXPATHLEN);
1572: strlcat(buf, "/", MAXPATHLEN);
1573: sz = strlcat(buf, fn, MAXPATHLEN);
1574: if (sz >= MAXPATHLEN) {
1575: fprintf(stderr, "%s: Path too long\n", dir);
1.11 schwarze 1576: continue;
1.2 schwarze 1577: }
1578:
1579: nof = mandoc_calloc(1, sizeof(struct of));
1580: nof->fname = mandoc_strdup(buf);
1.6 schwarze 1581: if (NULL != psec)
1582: nof->sec = mandoc_strdup(psec);
1583: if (NULL != parch)
1584: nof->arch = mandoc_strdup(parch);
1.11 schwarze 1585: nof->src_form = src_form;
1.8 schwarze 1586:
1587: /*
1588: * Remember the file name without the extension,
1589: * to be used as the page title in the database.
1590: */
1591:
1.6 schwarze 1592: if (NULL != suffix)
1593: *suffix = '\0';
1594: nof->title = mandoc_strdup(fn);
1.2 schwarze 1595:
1.11 schwarze 1596: /*
1597: * Add the structure to the list.
1598: */
1599:
1.2 schwarze 1600: if (verb > 2)
1601: printf("%s: Scheduling\n", buf);
1602: if (NULL == *of) {
1603: *of = nof;
1604: (*of)->first = nof;
1605: } else {
1606: nof->first = (*of)->first;
1607: (*of)->next = nof;
1608: *of = nof;
1609: }
1610: }
1611:
1.4 schwarze 1612: closedir(d);
1.2 schwarze 1613: return(1);
1614: }
1615:
1616: static void
1617: ofile_free(struct of *of)
1618: {
1619: struct of *nof;
1620:
1621: while (of) {
1622: nof = of->next;
1623: free(of->fname);
1.6 schwarze 1624: free(of->sec);
1625: free(of->arch);
1626: free(of->title);
1.2 schwarze 1627: free(of);
1628: of = nof;
1629: }
1630: }
1631:
1632: static void
1.1 schwarze 1633: usage(void)
1634: {
1635:
1.2 schwarze 1636: fprintf(stderr, "usage: %s [-v] "
1637: "[-d dir [files...] |"
1638: " -u dir [files...] |"
1639: " dir...]\n", progname);
1.1 schwarze 1640: }