Annotation of mandoc/mandocdb.c, Revision 1.37
1.37 ! schwarze 1: /* $Id: mandocdb.c,v 1.36 2011/12/16 12:06:35 kristaps 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>
1.1 kristaps 24:
25: #include <assert.h>
1.4 kristaps 26: #include <dirent.h>
1.1 kristaps 27: #include <fcntl.h>
28: #include <getopt.h>
29: #include <stdio.h>
30: #include <stdint.h>
31: #include <stdlib.h>
32: #include <string.h>
1.17 schwarze 33: #include <unistd.h>
1.1 kristaps 34:
1.21 kristaps 35: #if defined(__linux__)
36: # include <endian.h>
1.1 kristaps 37: # include <db_185.h>
1.21 kristaps 38: #elif defined(__APPLE__)
39: # include <libkern/OSByteOrder.h>
40: # include <db.h>
1.1 kristaps 41: #else
42: # include <db.h>
43: #endif
44:
45: #include "man.h"
46: #include "mdoc.h"
47: #include "mandoc.h"
1.8 schwarze 48: #include "mandocdb.h"
1.10 kristaps 49: #include "manpath.h"
1.1 kristaps 50:
51: #define MANDOC_BUFSZ BUFSIZ
52: #define MANDOC_SLOP 1024
53:
1.14 schwarze 54: #define MANDOC_SRC 0x1
55: #define MANDOC_FORM 0x2
56:
1.5 kristaps 57: /* Tiny list for files. No need to bring in QUEUE. */
58:
1.3 kristaps 59: struct of {
1.5 kristaps 60: char *fname; /* heap-allocated */
1.12 schwarze 61: char *sec;
62: char *arch;
63: char *title;
1.14 schwarze 64: int src_form;
1.5 kristaps 65: struct of *next; /* NULL for last one */
66: struct of *first; /* first in list */
1.3 kristaps 67: };
68:
1.1 kristaps 69: /* Buffer for storing growable data. */
70:
71: struct buf {
72: char *cp;
1.5 kristaps 73: size_t len; /* current length */
74: size_t size; /* total buffer size */
1.1 kristaps 75: };
76:
77: /* Operation we're going to perform. */
78:
79: enum op {
80: OP_NEW = 0, /* new database */
1.5 kristaps 81: OP_UPDATE, /* delete/add entries in existing database */
1.1 kristaps 82: OP_DELETE /* delete entries from existing database */
83: };
84:
85: #define MAN_ARGS DB *hash, \
86: struct buf *buf, \
87: struct buf *dbuf, \
88: const struct man_node *n
89: #define MDOC_ARGS DB *hash, \
90: struct buf *buf, \
91: struct buf *dbuf, \
92: const struct mdoc_node *n, \
93: const struct mdoc_meta *m
94:
95: static void buf_appendmdoc(struct buf *,
96: const struct mdoc_node *, int);
97: static void buf_append(struct buf *, const char *);
98: static void buf_appendb(struct buf *,
99: const void *, size_t);
100: static void dbt_put(DB *, const char *, DBT *, DBT *);
1.9 kristaps 101: static void hash_put(DB *, const struct buf *, uint64_t);
1.1 kristaps 102: static void hash_reset(DB **);
1.3 kristaps 103: static void index_merge(const struct of *, struct mparse *,
1.16 schwarze 104: struct buf *, struct buf *, DB *,
105: DB *, const char *, DB *, const char *,
1.3 kristaps 106: recno_t, const recno_t *, size_t);
107: static void index_prune(const struct of *, DB *,
108: const char *, DB *, const char *,
1.27 schwarze 109: recno_t *, recno_t **, size_t *,
110: size_t *);
1.16 schwarze 111: static void ofile_argbuild(int, char *[], struct of **);
1.35 kristaps 112: static void ofile_dirbuild(const char *, const char *,
1.16 schwarze 113: const char *, int, struct of **);
1.4 kristaps 114: static void ofile_free(struct of *);
1.14 schwarze 115: static void pformatted(DB *, struct buf *, struct buf *,
116: const struct of *);
1.1 kristaps 117: static int pman_node(MAN_ARGS);
118: static void pmdoc_node(MDOC_ARGS);
1.25 schwarze 119: static int pmdoc_head(MDOC_ARGS);
120: static int pmdoc_body(MDOC_ARGS);
121: static int pmdoc_Fd(MDOC_ARGS);
122: static int pmdoc_In(MDOC_ARGS);
123: static int pmdoc_Fn(MDOC_ARGS);
124: static int pmdoc_Nd(MDOC_ARGS);
125: static int pmdoc_Nm(MDOC_ARGS);
126: static int pmdoc_Sh(MDOC_ARGS);
127: static int pmdoc_St(MDOC_ARGS);
128: static int pmdoc_Xr(MDOC_ARGS);
1.1 kristaps 129: static void usage(void);
130:
1.25 schwarze 131: #define MDOCF_CHILD 0x01 /* Automatically index child nodes. */
1.1 kristaps 132:
1.25 schwarze 133: struct mdoc_handler {
134: int (*fp)(MDOC_ARGS); /* Optional handler. */
135: uint64_t mask; /* Set unless handler returns 0. */
136: int flags; /* For use by pmdoc_node. */
137: };
138:
139: static const struct mdoc_handler mdocs[MDOC_MAX] = {
140: { NULL, 0, 0 }, /* Ap */
141: { NULL, 0, 0 }, /* Dd */
142: { NULL, 0, 0 }, /* Dt */
143: { NULL, 0, 0 }, /* Os */
144: { pmdoc_Sh, TYPE_Sh, MDOCF_CHILD }, /* Sh */
145: { pmdoc_head, TYPE_Ss, MDOCF_CHILD }, /* Ss */
146: { NULL, 0, 0 }, /* Pp */
147: { NULL, 0, 0 }, /* D1 */
148: { NULL, 0, 0 }, /* Dl */
149: { NULL, 0, 0 }, /* Bd */
150: { NULL, 0, 0 }, /* Ed */
151: { NULL, 0, 0 }, /* Bl */
152: { NULL, 0, 0 }, /* El */
153: { NULL, 0, 0 }, /* It */
154: { NULL, 0, 0 }, /* Ad */
155: { NULL, TYPE_An, MDOCF_CHILD }, /* An */
156: { NULL, TYPE_Ar, MDOCF_CHILD }, /* Ar */
157: { NULL, TYPE_Cd, MDOCF_CHILD }, /* Cd */
158: { NULL, TYPE_Cm, MDOCF_CHILD }, /* Cm */
159: { NULL, TYPE_Dv, MDOCF_CHILD }, /* Dv */
160: { NULL, TYPE_Er, MDOCF_CHILD }, /* Er */
161: { NULL, TYPE_Ev, MDOCF_CHILD }, /* Ev */
162: { NULL, 0, 0 }, /* Ex */
163: { NULL, TYPE_Fa, MDOCF_CHILD }, /* Fa */
164: { pmdoc_Fd, TYPE_In, 0 }, /* Fd */
165: { NULL, TYPE_Fl, MDOCF_CHILD }, /* Fl */
166: { pmdoc_Fn, 0, 0 }, /* Fn */
167: { NULL, TYPE_Ft, MDOCF_CHILD }, /* Ft */
168: { NULL, TYPE_Ic, MDOCF_CHILD }, /* Ic */
169: { pmdoc_In, TYPE_In, 0 }, /* In */
170: { NULL, TYPE_Li, MDOCF_CHILD }, /* Li */
171: { pmdoc_Nd, TYPE_Nd, MDOCF_CHILD }, /* Nd */
172: { pmdoc_Nm, TYPE_Nm, MDOCF_CHILD }, /* Nm */
173: { NULL, 0, 0 }, /* Op */
174: { NULL, 0, 0 }, /* Ot */
175: { NULL, TYPE_Pa, MDOCF_CHILD }, /* Pa */
176: { NULL, 0, 0 }, /* Rv */
177: { pmdoc_St, TYPE_St, 0 }, /* St */
178: { NULL, TYPE_Va, MDOCF_CHILD }, /* Va */
179: { pmdoc_body, TYPE_Va, MDOCF_CHILD }, /* Vt */
180: { pmdoc_Xr, TYPE_Xr, 0 }, /* Xr */
181: { NULL, 0, 0 }, /* %A */
182: { NULL, 0, 0 }, /* %B */
183: { NULL, 0, 0 }, /* %D */
184: { NULL, 0, 0 }, /* %I */
185: { NULL, 0, 0 }, /* %J */
186: { NULL, 0, 0 }, /* %N */
187: { NULL, 0, 0 }, /* %O */
188: { NULL, 0, 0 }, /* %P */
189: { NULL, 0, 0 }, /* %R */
190: { NULL, 0, 0 }, /* %T */
191: { NULL, 0, 0 }, /* %V */
192: { NULL, 0, 0 }, /* Ac */
193: { NULL, 0, 0 }, /* Ao */
194: { NULL, 0, 0 }, /* Aq */
195: { NULL, TYPE_At, MDOCF_CHILD }, /* At */
196: { NULL, 0, 0 }, /* Bc */
197: { NULL, 0, 0 }, /* Bf */
198: { NULL, 0, 0 }, /* Bo */
199: { NULL, 0, 0 }, /* Bq */
200: { NULL, TYPE_Bsx, MDOCF_CHILD }, /* Bsx */
201: { NULL, TYPE_Bx, MDOCF_CHILD }, /* Bx */
202: { NULL, 0, 0 }, /* Db */
203: { NULL, 0, 0 }, /* Dc */
204: { NULL, 0, 0 }, /* Do */
205: { NULL, 0, 0 }, /* Dq */
206: { NULL, 0, 0 }, /* Ec */
207: { NULL, 0, 0 }, /* Ef */
208: { NULL, TYPE_Em, MDOCF_CHILD }, /* Em */
209: { NULL, 0, 0 }, /* Eo */
210: { NULL, TYPE_Fx, MDOCF_CHILD }, /* Fx */
211: { NULL, TYPE_Ms, MDOCF_CHILD }, /* Ms */
212: { NULL, 0, 0 }, /* No */
213: { NULL, 0, 0 }, /* Ns */
214: { NULL, TYPE_Nx, MDOCF_CHILD }, /* Nx */
215: { NULL, TYPE_Ox, MDOCF_CHILD }, /* Ox */
216: { NULL, 0, 0 }, /* Pc */
217: { NULL, 0, 0 }, /* Pf */
218: { NULL, 0, 0 }, /* Po */
219: { NULL, 0, 0 }, /* Pq */
220: { NULL, 0, 0 }, /* Qc */
221: { NULL, 0, 0 }, /* Ql */
222: { NULL, 0, 0 }, /* Qo */
223: { NULL, 0, 0 }, /* Qq */
224: { NULL, 0, 0 }, /* Re */
225: { NULL, 0, 0 }, /* Rs */
226: { NULL, 0, 0 }, /* Sc */
227: { NULL, 0, 0 }, /* So */
228: { NULL, 0, 0 }, /* Sq */
229: { NULL, 0, 0 }, /* Sm */
230: { NULL, 0, 0 }, /* Sx */
231: { NULL, TYPE_Sy, MDOCF_CHILD }, /* Sy */
232: { NULL, TYPE_Tn, MDOCF_CHILD }, /* Tn */
233: { NULL, 0, 0 }, /* Ux */
234: { NULL, 0, 0 }, /* Xc */
235: { NULL, 0, 0 }, /* Xo */
236: { pmdoc_head, TYPE_Fn, 0 }, /* Fo */
237: { NULL, 0, 0 }, /* Fc */
238: { NULL, 0, 0 }, /* Oo */
239: { NULL, 0, 0 }, /* Oc */
240: { NULL, 0, 0 }, /* Bk */
241: { NULL, 0, 0 }, /* Ek */
242: { NULL, 0, 0 }, /* Bt */
243: { NULL, 0, 0 }, /* Hf */
244: { NULL, 0, 0 }, /* Fr */
245: { NULL, 0, 0 }, /* Ud */
246: { NULL, TYPE_Lb, MDOCF_CHILD }, /* Lb */
247: { NULL, 0, 0 }, /* Lp */
248: { NULL, TYPE_Lk, MDOCF_CHILD }, /* Lk */
249: { NULL, TYPE_Mt, MDOCF_CHILD }, /* Mt */
250: { NULL, 0, 0 }, /* Brq */
251: { NULL, 0, 0 }, /* Bro */
252: { NULL, 0, 0 }, /* Brc */
253: { NULL, 0, 0 }, /* %C */
254: { NULL, 0, 0 }, /* Es */
255: { NULL, 0, 0 }, /* En */
256: { NULL, TYPE_Dx, MDOCF_CHILD }, /* Dx */
257: { NULL, 0, 0 }, /* %Q */
258: { NULL, 0, 0 }, /* br */
259: { NULL, 0, 0 }, /* sp */
260: { NULL, 0, 0 }, /* %U */
261: { NULL, 0, 0 }, /* Ta */
1.1 kristaps 262: };
263:
264: static const char *progname;
1.16 schwarze 265: static int use_all; /* Use all directories and files. */
266: static int verb; /* Output verbosity level. */
1.1 kristaps 267:
268: int
269: main(int argc, char *argv[])
270: {
271: struct mparse *mp; /* parse sequence */
1.10 kristaps 272: struct manpaths dirs;
1.1 kristaps 273: enum op op; /* current operation */
1.5 kristaps 274: const char *dir;
1.34 schwarze 275: char *conf_file;
1.26 kristaps 276: char *cp;
277: char pbuf[PATH_MAX],
278: ibuf[MAXPATHLEN], /* index fname */
1.3 kristaps 279: fbuf[MAXPATHLEN]; /* btree fname */
1.16 schwarze 280: int ch, i, flags;
1.1 kristaps 281: DB *idx, /* index database */
282: *db, /* keyword database */
283: *hash; /* temporary keyword hashtable */
284: BTREEINFO info; /* btree configuration */
1.12 schwarze 285: recno_t maxrec; /* last record number in the index */
286: recno_t *recs; /* the numbers of all empty records */
1.5 kristaps 287: size_t sz1, sz2,
1.12 schwarze 288: recsz, /* number of allocated slots in recs */
289: reccur; /* current number of empty records */
1.1 kristaps 290: struct buf buf, /* keyword buffer */
291: dbuf; /* description buffer */
1.5 kristaps 292: struct of *of; /* list of files for processing */
1.1 kristaps 293: extern int optind;
294: extern char *optarg;
295:
296: progname = strrchr(argv[0], '/');
297: if (progname == NULL)
298: progname = argv[0];
299: else
300: ++progname;
301:
1.10 kristaps 302: memset(&dirs, 0, sizeof(struct manpaths));
303:
1.5 kristaps 304: verb = 0;
1.12 schwarze 305: use_all = 0;
1.4 kristaps 306: of = NULL;
1.1 kristaps 307: db = idx = NULL;
308: mp = NULL;
309: hash = NULL;
310: recs = NULL;
311: recsz = reccur = 0;
312: maxrec = 0;
313: op = OP_NEW;
1.5 kristaps 314: dir = NULL;
1.34 schwarze 315: conf_file = NULL;
1.1 kristaps 316:
1.34 schwarze 317: while (-1 != (ch = getopt(argc, argv, "aC:d:u:v")))
1.1 kristaps 318: switch (ch) {
1.12 schwarze 319: case ('a'):
320: use_all = 1;
321: break;
1.34 schwarze 322: case ('C'):
323: conf_file = optarg;
324: break;
1.5 kristaps 325: case ('d'):
326: dir = optarg;
327: op = OP_UPDATE;
328: break;
329: case ('u'):
330: dir = optarg;
331: op = OP_DELETE;
332: break;
333: case ('v'):
334: verb++;
335: break;
1.1 kristaps 336: default:
337: usage();
338: return((int)MANDOCLEVEL_BADARG);
339: }
340:
341: argc -= optind;
342: argv += optind;
343:
1.4 kristaps 344: memset(&info, 0, sizeof(BTREEINFO));
345: info.flags = R_DUP;
1.1 kristaps 346:
1.4 kristaps 347: mp = mparse_alloc(MPARSE_AUTO, MANDOCLEVEL_FATAL, NULL, NULL);
1.1 kristaps 348:
1.5 kristaps 349: memset(&buf, 0, sizeof(struct buf));
350: memset(&dbuf, 0, sizeof(struct buf));
1.1 kristaps 351:
1.4 kristaps 352: buf.size = dbuf.size = MANDOC_BUFSZ;
1.1 kristaps 353:
1.4 kristaps 354: buf.cp = mandoc_malloc(buf.size);
355: dbuf.cp = mandoc_malloc(dbuf.size);
1.1 kristaps 356:
1.5 kristaps 357: flags = OP_NEW == op ? O_CREAT|O_TRUNC|O_RDWR : O_CREAT|O_RDWR;
358:
359: if (OP_UPDATE == op || OP_DELETE == op) {
360: ibuf[0] = fbuf[0] = '\0';
361:
362: strlcat(fbuf, dir, MAXPATHLEN);
363: strlcat(fbuf, "/", MAXPATHLEN);
364: sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
365:
366: strlcat(ibuf, dir, MAXPATHLEN);
367: strlcat(ibuf, "/", MAXPATHLEN);
368: sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
369:
370: if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) {
371: fprintf(stderr, "%s: Path too long\n", dir);
372: exit((int)MANDOCLEVEL_BADARG);
373: }
374:
375: db = dbopen(fbuf, flags, 0644, DB_BTREE, &info);
376: idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL);
377:
378: if (NULL == db) {
379: perror(fbuf);
380: exit((int)MANDOCLEVEL_SYSERR);
1.12 schwarze 381: } else if (NULL == idx) {
1.5 kristaps 382: perror(ibuf);
383: exit((int)MANDOCLEVEL_SYSERR);
384: }
385:
1.35 kristaps 386: ofile_argbuild(argc, argv, &of);
1.5 kristaps 387:
388: if (NULL == of)
389: goto out;
390:
391: of = of->first;
392:
1.16 schwarze 393: index_prune(of, db, fbuf, idx, ibuf,
1.27 schwarze 394: &maxrec, &recs, &recsz, &reccur);
1.5 kristaps 395:
1.17 schwarze 396: /*
1.35 kristaps 397: * Go to the root of the respective manual tree.
398: * This must work or no manuals may be found (they're
399: * indexed relative to the root).
1.17 schwarze 400: */
401:
402: if (OP_UPDATE == op) {
1.35 kristaps 403: if (-1 == chdir(dir)) {
404: perror(dir);
405: exit((int)MANDOCLEVEL_SYSERR);
406: }
1.13 schwarze 407: index_merge(of, mp, &dbuf, &buf, hash,
1.16 schwarze 408: db, fbuf, idx, ibuf,
409: maxrec, recs, reccur);
1.17 schwarze 410: }
1.5 kristaps 411:
412: goto out;
413: }
414:
1.10 kristaps 415: /*
416: * Configure the directories we're going to scan.
417: * If we have command-line arguments, use them.
418: * If not, we use man(1)'s method (see mandocdb.8).
419: */
420:
421: if (argc > 0) {
1.26 kristaps 422: dirs.paths = mandoc_calloc(argc, sizeof(char *));
1.10 kristaps 423: dirs.sz = argc;
1.26 kristaps 424: for (i = 0; i < argc; i++) {
425: if (NULL == (cp = realpath(argv[i], pbuf))) {
426: perror(argv[i]);
427: goto out;
428: }
429: dirs.paths[i] = mandoc_strdup(cp);
430: }
1.10 kristaps 431: } else
1.34 schwarze 432: manpath_parse(&dirs, conf_file, NULL, NULL);
1.10 kristaps 433:
434: for (i = 0; i < dirs.sz; i++) {
1.5 kristaps 435: ibuf[0] = fbuf[0] = '\0';
1.1 kristaps 436:
1.10 kristaps 437: strlcat(fbuf, dirs.paths[i], MAXPATHLEN);
1.5 kristaps 438: strlcat(fbuf, "/", MAXPATHLEN);
439: sz1 = strlcat(fbuf, MANDOC_DB, MAXPATHLEN);
1.1 kristaps 440:
1.10 kristaps 441: strlcat(ibuf, dirs.paths[i], MAXPATHLEN);
1.5 kristaps 442: strlcat(ibuf, "/", MAXPATHLEN);
443: sz2 = strlcat(ibuf, MANDOC_IDX, MAXPATHLEN);
1.1 kristaps 444:
1.5 kristaps 445: if (sz1 >= MAXPATHLEN || sz2 >= MAXPATHLEN) {
1.13 schwarze 446: fprintf(stderr, "%s: Path too long\n",
1.10 kristaps 447: dirs.paths[i]);
1.5 kristaps 448: exit((int)MANDOCLEVEL_BADARG);
1.4 kristaps 449: }
1.3 kristaps 450:
1.13 schwarze 451: if (db)
452: (*db->close)(db);
453: if (idx)
454: (*idx->close)(idx);
455:
1.4 kristaps 456: db = dbopen(fbuf, flags, 0644, DB_BTREE, &info);
457: idx = dbopen(ibuf, flags, 0644, DB_RECNO, NULL);
1.3 kristaps 458:
1.4 kristaps 459: if (NULL == db) {
460: perror(fbuf);
1.5 kristaps 461: exit((int)MANDOCLEVEL_SYSERR);
1.12 schwarze 462: } else if (NULL == idx) {
1.4 kristaps 463: perror(ibuf);
1.5 kristaps 464: exit((int)MANDOCLEVEL_SYSERR);
465: }
466:
1.4 kristaps 467: ofile_free(of);
468: of = NULL;
1.1 kristaps 469:
1.35 kristaps 470: if (-1 == chdir(dirs.paths[i])) {
471: perror(dirs.paths[i]);
1.5 kristaps 472: exit((int)MANDOCLEVEL_SYSERR);
1.35 kristaps 473: }
474:
475: ofile_dirbuild(".", NULL, NULL, 0, &of);
1.1 kristaps 476:
1.5 kristaps 477: if (NULL == of)
478: continue;
1.1 kristaps 479:
1.5 kristaps 480: of = of->first;
1.1 kristaps 481:
1.17 schwarze 482: /*
1.35 kristaps 483: * Go to the root of the respective manual tree.
484: * This must work or no manuals may be found (they're
485: * indexed relative to the root).
1.17 schwarze 486: */
487:
1.35 kristaps 488: if (-1 == chdir(dirs.paths[i])) {
489: perror(dirs.paths[i]);
490: exit((int)MANDOCLEVEL_SYSERR);
491: }
492:
1.13 schwarze 493: index_merge(of, mp, &dbuf, &buf, hash, db, fbuf,
1.16 schwarze 494: idx, ibuf, maxrec, recs, reccur);
1.4 kristaps 495: }
1.3 kristaps 496:
1.5 kristaps 497: out:
1.3 kristaps 498: if (db)
499: (*db->close)(db);
500: if (idx)
501: (*idx->close)(idx);
502: if (hash)
503: (*hash->close)(hash);
504: if (mp)
505: mparse_free(mp);
506:
1.10 kristaps 507: manpath_free(&dirs);
1.4 kristaps 508: ofile_free(of);
1.3 kristaps 509: free(buf.cp);
510: free(dbuf.cp);
511: free(recs);
512:
1.5 kristaps 513: return(MANDOCLEVEL_OK);
1.3 kristaps 514: }
515:
516: void
517: index_merge(const struct of *of, struct mparse *mp,
1.16 schwarze 518: struct buf *dbuf, struct buf *buf, DB *hash,
519: DB *db, const char *dbf, DB *idx, const char *idxf,
1.3 kristaps 520: recno_t maxrec, const recno_t *recs, size_t reccur)
521: {
522: recno_t rec;
523: int ch;
524: DBT key, val;
525: struct mdoc *mdoc;
526: struct man *man;
527: const char *fn, *msec, *mtitle, *arch;
1.37 ! schwarze 528: uint64_t mask;
1.3 kristaps 529: size_t sv;
530: unsigned seq;
1.9 kristaps 531: struct db_val vbuf;
1.36 kristaps 532: char type;
1.3 kristaps 533:
534: for (rec = 0; of; of = of->next) {
535: fn = of->fname;
1.14 schwarze 536:
537: /*
1.33 schwarze 538: * Try interpreting the file as mdoc(7) or man(7)
539: * source code, unless it is already known to be
540: * formatted. Fall back to formatted mode.
1.14 schwarze 541: */
542:
1.1 kristaps 543: mparse_reset(mp);
1.14 schwarze 544: mdoc = NULL;
545: man = NULL;
1.1 kristaps 546:
1.14 schwarze 547: if ((MANDOC_SRC & of->src_form ||
548: ! (MANDOC_FORM & of->src_form)) &&
549: MANDOCLEVEL_FATAL > mparse_readfd(mp, -1, fn))
550: mparse_result(mp, &mdoc, &man);
551:
552: if (NULL != mdoc) {
553: msec = mdoc_meta(mdoc)->msec;
554: arch = mdoc_meta(mdoc)->arch;
555: mtitle = mdoc_meta(mdoc)->title;
556: } else if (NULL != man) {
557: msec = man_meta(man)->msec;
558: arch = NULL;
559: mtitle = man_meta(man)->title;
560: } else {
561: msec = of->sec;
562: arch = of->arch;
563: mtitle = of->title;
1.1 kristaps 564: }
565:
1.12 schwarze 566: /*
567: * By default, skip a file if the manual section
568: * and architecture given in the file disagree
569: * with the directory where the file is located.
570: */
571:
572: if (0 == use_all) {
573: assert(of->sec);
574: assert(msec);
1.32 schwarze 575: if (strcasecmp(msec, of->sec))
1.12 schwarze 576: continue;
577:
578: if (NULL == arch) {
579: if (NULL != of->arch)
580: continue;
581: } else if (NULL == of->arch ||
1.32 schwarze 582: strcasecmp(arch, of->arch))
1.12 schwarze 583: continue;
584: }
585:
1.1 kristaps 586: if (NULL == arch)
587: arch = "";
588:
589: /*
1.12 schwarze 590: * By default, skip a file if the title given
591: * in the file disagrees with the file name.
592: * If both agree, use the file name as the title,
593: * because the one in the file usually is all caps.
594: */
595:
596: assert(of->title);
597: assert(mtitle);
598:
599: if (0 == strcasecmp(mtitle, of->title))
600: mtitle = of->title;
601: else if (0 == use_all)
602: continue;
603:
604: /*
1.1 kristaps 605: * The index record value consists of a nil-terminated
606: * filename, a nil-terminated manual section, and a
607: * nil-terminated description. Since the description
608: * may not be set, we set a sentinel to see if we're
609: * going to write a nil byte in its place.
610: */
611:
1.3 kristaps 612: dbuf->len = 0;
1.36 kristaps 613: type = mdoc ? 'd' : (man ? 'a' : 'c');
614: buf_appendb(dbuf, &type, 1);
1.3 kristaps 615: buf_appendb(dbuf, fn, strlen(fn) + 1);
616: buf_appendb(dbuf, msec, strlen(msec) + 1);
617: buf_appendb(dbuf, mtitle, strlen(mtitle) + 1);
618: buf_appendb(dbuf, arch, strlen(arch) + 1);
1.1 kristaps 619:
1.3 kristaps 620: sv = dbuf->len;
1.1 kristaps 621:
1.33 schwarze 622: /*
623: * Collect keyword/mask pairs.
624: * Each pair will become a new btree node.
625: */
1.1 kristaps 626:
1.33 schwarze 627: hash_reset(&hash);
1.1 kristaps 628: if (mdoc)
1.3 kristaps 629: pmdoc_node(hash, buf, dbuf,
1.1 kristaps 630: mdoc_node(mdoc), mdoc_meta(mdoc));
1.14 schwarze 631: else if (man)
1.3 kristaps 632: pman_node(hash, buf, dbuf, man_node(man));
1.14 schwarze 633: else
634: pformatted(hash, buf, dbuf, of);
1.1 kristaps 635:
636: /*
1.33 schwarze 637: * Reclaim an empty index record, if available.
638: * Use its record number for all new btree nodes.
1.1 kristaps 639: */
640:
1.33 schwarze 641: if (reccur > 0) {
642: --reccur;
643: rec = recs[(int)reccur];
644: } else if (maxrec > 0) {
645: rec = maxrec;
646: maxrec = 0;
647: } else
648: rec++;
1.20 kristaps 649: vbuf.rec = htobe32(rec);
1.33 schwarze 650:
651: /*
652: * Copy from the in-memory hashtable of pending
653: * keyword/mask pairs into the database.
654: */
655:
1.1 kristaps 656: seq = R_FIRST;
657: while (0 == (ch = (*hash->seq)(hash, &key, &val, seq))) {
658: seq = R_NEXT;
1.37 ! schwarze 659: assert(sizeof(uint64_t) == val.size);
! 660: memcpy(&mask, val.data, val.size);
! 661: vbuf.mask = htobe64(mask);
1.9 kristaps 662: val.size = sizeof(struct db_val);
663: val.data = &vbuf;
1.3 kristaps 664: dbt_put(db, dbf, &key, &val);
1.1 kristaps 665: }
666: if (ch < 0) {
667: perror("hash");
668: exit((int)MANDOCLEVEL_SYSERR);
669: }
670:
671: /*
672: * Apply to the index. If we haven't had a description
673: * set, put an empty one in now.
674: */
675:
1.3 kristaps 676: if (dbuf->len == sv)
677: buf_appendb(dbuf, "", 1);
1.1 kristaps 678:
679: key.data = &rec;
680: key.size = sizeof(recno_t);
681:
1.3 kristaps 682: val.data = dbuf->cp;
683: val.size = dbuf->len;
1.1 kristaps 684:
1.5 kristaps 685: if (verb)
686: printf("%s: Added index\n", fn);
1.18 kristaps 687:
1.3 kristaps 688: dbt_put(idx, idxf, &key, &val);
689: }
690: }
691:
692: /*
693: * Scan through all entries in the index file `idx' and prune those
694: * entries in `ofile'.
695: * Pruning consists of removing from `db', then invalidating the entry
696: * in `idx' (zeroing its value size).
697: */
698: static void
699: index_prune(const struct of *ofile, DB *db, const char *dbf,
1.27 schwarze 700: DB *idx, const char *idxf, recno_t *maxrec,
701: recno_t **recs, size_t *recsz, size_t *reccur)
1.3 kristaps 702: {
703: const struct of *of;
1.36 kristaps 704: const char *fn;
1.9 kristaps 705: struct db_val *vbuf;
1.3 kristaps 706: unsigned seq, sseq;
707: DBT key, val;
708: int ch;
709:
1.27 schwarze 710: *reccur = 0;
1.3 kristaps 711: seq = R_FIRST;
712: while (0 == (ch = (*idx->seq)(idx, &key, &val, seq))) {
713: seq = R_NEXT;
1.37 ! schwarze 714: assert(sizeof(recno_t) == key.size);
! 715: memcpy(maxrec, key.data, key.size);
1.18 kristaps 716:
717: /* Deleted records are zero-sized. Skip them. */
718:
719: if (0 == val.size)
720: goto cont;
721:
722: /*
723: * Make sure we're sane.
724: * Read past our mdoc/man/cat type to the next string,
725: * then make sure it's bounded by a NUL.
726: * Failing any of these, we go into our error handler.
727: */
728:
1.36 kristaps 729: fn = (char *)val.data + 1;
730: if (NULL == memchr(fn, '\0', val.size - 1))
1.18 kristaps 731: break;
732:
733: /*
734: * Search for the file in those we care about.
735: * XXX: build this into a tree. Too slow.
736: */
1.3 kristaps 737:
738: for (of = ofile; of; of = of->next)
739: if (0 == strcmp(fn, of->fname))
740: break;
741:
742: if (NULL == of)
743: continue;
744:
1.18 kristaps 745: /*
746: * Search through the keyword database, throwing out all
747: * references to our file.
748: */
749:
1.3 kristaps 750: sseq = R_FIRST;
751: while (0 == (ch = (*db->seq)(db, &key, &val, sseq))) {
752: sseq = R_NEXT;
1.18 kristaps 753: if (sizeof(struct db_val) != val.size)
754: break;
755:
1.9 kristaps 756: vbuf = val.data;
1.20 kristaps 757: if (*maxrec != betoh32(vbuf->rec))
1.3 kristaps 758: continue;
1.18 kristaps 759:
760: if ((ch = (*db->del)(db, &key, R_CURSOR)) < 0)
1.3 kristaps 761: break;
762: }
1.18 kristaps 763:
1.3 kristaps 764: if (ch < 0) {
765: perror(dbf);
766: exit((int)MANDOCLEVEL_SYSERR);
1.18 kristaps 767: } else if (1 != ch) {
768: fprintf(stderr, "%s: Corrupt database\n", dbf);
769: exit((int)MANDOCLEVEL_SYSERR);
1.3 kristaps 770: }
1.1 kristaps 771:
1.5 kristaps 772: if (verb)
773: printf("%s: Deleted index\n", fn);
1.1 kristaps 774:
1.3 kristaps 775: val.size = 0;
776: ch = (*idx->put)(idx, &key, &val, R_CURSOR);
1.1 kristaps 777:
1.18 kristaps 778: if (ch < 0)
779: break;
780: cont:
1.27 schwarze 781: if (*reccur >= *recsz) {
1.3 kristaps 782: *recsz += MANDOC_SLOP;
783: *recs = mandoc_realloc
784: (*recs, *recsz * sizeof(recno_t));
785: }
1.1 kristaps 786:
1.27 schwarze 787: (*recs)[(int)*reccur] = *maxrec;
788: (*reccur)++;
1.3 kristaps 789: }
1.18 kristaps 790:
791: if (ch < 0) {
792: perror(idxf);
793: exit((int)MANDOCLEVEL_SYSERR);
794: } else if (1 != ch) {
795: fprintf(stderr, "%s: Corrupt index\n", idxf);
796: exit((int)MANDOCLEVEL_SYSERR);
797: }
798:
1.3 kristaps 799: (*maxrec)++;
1.1 kristaps 800: }
801:
802: /*
803: * Grow the buffer (if necessary) and copy in a binary string.
804: */
805: static void
806: buf_appendb(struct buf *buf, const void *cp, size_t sz)
807: {
808:
809: /* Overshoot by MANDOC_BUFSZ. */
810:
811: while (buf->len + sz >= buf->size) {
812: buf->size = buf->len + sz + MANDOC_BUFSZ;
813: buf->cp = mandoc_realloc(buf->cp, buf->size);
814: }
815:
816: memcpy(buf->cp + (int)buf->len, cp, sz);
817: buf->len += sz;
818: }
819:
820: /*
821: * Append a nil-terminated string to the buffer.
822: * This can be invoked multiple times.
823: * The buffer string will be nil-terminated.
824: * If invoked multiple times, a space is put between strings.
825: */
826: static void
827: buf_append(struct buf *buf, const char *cp)
828: {
829: size_t sz;
830:
831: if (0 == (sz = strlen(cp)))
832: return;
833:
834: if (buf->len)
835: buf->cp[(int)buf->len - 1] = ' ';
836:
837: buf_appendb(buf, cp, sz + 1);
838: }
839:
840: /*
841: * Recursively add all text from a given node.
842: * This is optimised for general mdoc nodes in this context, which do
843: * not consist of subexpressions and having a recursive call for n->next
844: * would be wasteful.
845: * The "f" variable should be 0 unless called from pmdoc_Nd for the
846: * description buffer, which does not start at the beginning of the
847: * buffer.
848: */
849: static void
850: buf_appendmdoc(struct buf *buf, const struct mdoc_node *n, int f)
851: {
852:
853: for ( ; n; n = n->next) {
854: if (n->child)
855: buf_appendmdoc(buf, n->child, f);
856:
857: if (MDOC_TEXT == n->type && f) {
858: f = 0;
859: buf_appendb(buf, n->string,
860: strlen(n->string) + 1);
861: } else if (MDOC_TEXT == n->type)
862: buf_append(buf, n->string);
863:
864: }
865: }
866:
867: static void
868: hash_reset(DB **db)
869: {
870: DB *hash;
871:
872: if (NULL != (hash = *db))
873: (*hash->close)(hash);
874:
1.5 kristaps 875: *db = dbopen(NULL, O_CREAT|O_RDWR, 0644, DB_HASH, NULL);
1.1 kristaps 876: if (NULL == *db) {
877: perror("hash");
878: exit((int)MANDOCLEVEL_SYSERR);
879: }
880: }
881:
882: /* ARGSUSED */
1.25 schwarze 883: static int
884: pmdoc_head(MDOC_ARGS)
885: {
886:
887: return(MDOC_HEAD == n->type);
888: }
889:
890: /* ARGSUSED */
891: static int
892: pmdoc_body(MDOC_ARGS)
893: {
894:
895: return(MDOC_BODY == n->type);
896: }
897:
898: /* ARGSUSED */
899: static int
1.1 kristaps 900: pmdoc_Fd(MDOC_ARGS)
901: {
902: const char *start, *end;
903: size_t sz;
1.25 schwarze 904:
1.1 kristaps 905: if (SEC_SYNOPSIS != n->sec)
1.25 schwarze 906: return(0);
1.1 kristaps 907: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.25 schwarze 908: return(0);
1.1 kristaps 909:
910: /*
911: * Only consider those `Fd' macro fields that begin with an
912: * "inclusion" token (versus, e.g., #define).
913: */
914: if (strcmp("#include", n->string))
1.25 schwarze 915: return(0);
1.1 kristaps 916:
917: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.25 schwarze 918: return(0);
1.1 kristaps 919:
920: /*
921: * Strip away the enclosing angle brackets and make sure we're
922: * not zero-length.
923: */
924:
925: start = n->string;
926: if ('<' == *start || '"' == *start)
927: start++;
928:
929: if (0 == (sz = strlen(start)))
1.25 schwarze 930: return(0);
1.1 kristaps 931:
932: end = &start[(int)sz - 1];
933: if ('>' == *end || '"' == *end)
934: end--;
935:
936: assert(end >= start);
937:
938: buf_appendb(buf, start, (size_t)(end - start + 1));
939: buf_appendb(buf, "", 1);
1.25 schwarze 940: return(1);
1.1 kristaps 941: }
942:
943: /* ARGSUSED */
1.25 schwarze 944: static int
945: pmdoc_In(MDOC_ARGS)
1.1 kristaps 946: {
947:
948: if (NULL == n->child || MDOC_TEXT != n->child->type)
1.25 schwarze 949: return(0);
1.1 kristaps 950:
951: buf_append(buf, n->child->string);
1.25 schwarze 952: return(1);
1.1 kristaps 953: }
954:
955: /* ARGSUSED */
1.25 schwarze 956: static int
1.1 kristaps 957: pmdoc_Fn(MDOC_ARGS)
958: {
1.25 schwarze 959: struct mdoc_node *nn;
1.1 kristaps 960: const char *cp;
961:
1.25 schwarze 962: nn = n->child;
963:
964: if (NULL == nn || MDOC_TEXT != nn->type)
965: return(0);
966:
967: /* .Fn "struct type *name" "char *arg" */
1.1 kristaps 968:
1.25 schwarze 969: cp = strrchr(nn->string, ' ');
1.1 kristaps 970: if (NULL == cp)
1.25 schwarze 971: cp = nn->string;
1.1 kristaps 972:
973: /* Strip away pointer symbol. */
974:
975: while ('*' == *cp)
976: cp++;
977:
1.25 schwarze 978: /* Store the function name. */
979:
1.1 kristaps 980: buf_append(buf, cp);
1.8 schwarze 981: hash_put(hash, buf, TYPE_Fn);
1.25 schwarze 982:
983: /* Store the function type. */
984:
985: if (nn->string < cp) {
986: buf->len = 0;
987: buf_appendb(buf, nn->string, cp - nn->string);
988: buf_appendb(buf, "", 1);
989: hash_put(hash, buf, TYPE_Ft);
990: }
991:
992: /* Store the arguments. */
993:
994: for (nn = nn->next; nn; nn = nn->next) {
995: if (MDOC_TEXT != nn->type)
996: continue;
997: buf->len = 0;
998: buf_append(buf, nn->string);
999: hash_put(hash, buf, TYPE_Fa);
1000: }
1001:
1002: return(0);
1.1 kristaps 1003: }
1004:
1005: /* ARGSUSED */
1.25 schwarze 1006: static int
1.1 kristaps 1007: pmdoc_St(MDOC_ARGS)
1008: {
1.25 schwarze 1009:
1.1 kristaps 1010: if (NULL == n->child || MDOC_TEXT != n->child->type)
1.25 schwarze 1011: return(0);
1.1 kristaps 1012:
1013: buf_append(buf, n->child->string);
1.25 schwarze 1014: return(1);
1.1 kristaps 1015: }
1016:
1017: /* ARGSUSED */
1.25 schwarze 1018: static int
1.1 kristaps 1019: pmdoc_Xr(MDOC_ARGS)
1020: {
1021:
1022: if (NULL == (n = n->child))
1.25 schwarze 1023: return(0);
1.1 kristaps 1024:
1025: buf_appendb(buf, n->string, strlen(n->string));
1026:
1027: if (NULL != (n = n->next)) {
1028: buf_appendb(buf, ".", 1);
1029: buf_appendb(buf, n->string, strlen(n->string) + 1);
1030: } else
1031: buf_appendb(buf, ".", 2);
1032:
1.25 schwarze 1033: return(1);
1.1 kristaps 1034: }
1035:
1036: /* ARGSUSED */
1.25 schwarze 1037: static int
1.1 kristaps 1038: pmdoc_Nd(MDOC_ARGS)
1039: {
1040:
1041: if (MDOC_BODY != n->type)
1.25 schwarze 1042: return(0);
1.1 kristaps 1043:
1044: buf_appendmdoc(dbuf, n->child, 1);
1.25 schwarze 1045: return(1);
1.1 kristaps 1046: }
1047:
1048: /* ARGSUSED */
1.25 schwarze 1049: static int
1050: pmdoc_Nm(MDOC_ARGS)
1.1 kristaps 1051: {
1052:
1.25 schwarze 1053: if (SEC_NAME == n->sec)
1054: return(1);
1055: else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
1056: return(0);
1.1 kristaps 1057:
1.25 schwarze 1058: if (NULL == n->child)
1059: buf_append(buf, m->name);
1.1 kristaps 1060:
1.25 schwarze 1061: return(1);
1.1 kristaps 1062: }
1063:
1064: /* ARGSUSED */
1.25 schwarze 1065: static int
1066: pmdoc_Sh(MDOC_ARGS)
1.1 kristaps 1067: {
1068:
1.25 schwarze 1069: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 kristaps 1070: }
1071:
1072: static void
1.9 kristaps 1073: hash_put(DB *db, const struct buf *buf, uint64_t mask)
1.1 kristaps 1074: {
1.37 ! schwarze 1075: uint64_t oldmask;
1.1 kristaps 1076: DBT key, val;
1077: int rc;
1078:
1079: if (buf->len < 2)
1080: return;
1081:
1082: key.data = buf->cp;
1083: key.size = buf->len;
1084:
1085: if ((rc = (*db->get)(db, &key, &val, 0)) < 0) {
1086: perror("hash");
1087: exit((int)MANDOCLEVEL_SYSERR);
1.37 ! schwarze 1088: } else if (0 == rc) {
! 1089: assert(sizeof(uint64_t) == val.size);
! 1090: memcpy(&oldmask, val.data, val.size);
! 1091: mask |= oldmask;
! 1092: }
1.1 kristaps 1093:
1094: val.data = &mask;
1.9 kristaps 1095: val.size = sizeof(uint64_t);
1.1 kristaps 1096:
1097: if ((rc = (*db->put)(db, &key, &val, 0)) < 0) {
1098: perror("hash");
1099: exit((int)MANDOCLEVEL_SYSERR);
1100: }
1101: }
1102:
1103: static void
1104: dbt_put(DB *db, const char *dbn, DBT *key, DBT *val)
1105: {
1106:
1107: assert(key->size);
1108: assert(val->size);
1109:
1110: if (0 == (*db->put)(db, key, val, 0))
1111: return;
1112:
1113: perror(dbn);
1114: exit((int)MANDOCLEVEL_SYSERR);
1115: /* NOTREACHED */
1116: }
1117:
1118: /*
1119: * Call out to per-macro handlers after clearing the persistent database
1120: * key. If the macro sets the database key, flush it to the database.
1121: */
1122: static void
1123: pmdoc_node(MDOC_ARGS)
1124: {
1125:
1126: if (NULL == n)
1127: return;
1128:
1129: switch (n->type) {
1130: case (MDOC_HEAD):
1131: /* FALLTHROUGH */
1132: case (MDOC_BODY):
1133: /* FALLTHROUGH */
1134: case (MDOC_TAIL):
1135: /* FALLTHROUGH */
1136: case (MDOC_BLOCK):
1137: /* FALLTHROUGH */
1138: case (MDOC_ELEM):
1.25 schwarze 1139: buf->len = 0;
1140:
1141: /*
1142: * Both NULL handlers and handlers returning true
1143: * request using the data. Only skip the element
1144: * when the handler returns false.
1145: */
1146:
1147: if (NULL != mdocs[n->tok].fp &&
1148: 0 == (*mdocs[n->tok].fp)(hash, buf, dbuf, n, m))
1.1 kristaps 1149: break;
1150:
1.25 schwarze 1151: /*
1152: * For many macros, use the text from all children.
1153: * Set zero flags for macros not needing this.
1154: * In that case, the handler must fill the buffer.
1155: */
1156:
1157: if (MDOCF_CHILD & mdocs[n->tok].flags)
1158: buf_appendmdoc(buf, n->child, 0);
1159:
1160: /*
1161: * Cover the most common case:
1162: * Automatically stage one string per element.
1163: * Set a zero mask for macros not needing this.
1164: * Additional staging can be done in the handler.
1165: */
1166:
1167: if (mdocs[n->tok].mask)
1168: hash_put(hash, buf, mdocs[n->tok].mask);
1.1 kristaps 1169: break;
1170: default:
1171: break;
1172: }
1173:
1174: pmdoc_node(hash, buf, dbuf, n->child, m);
1175: pmdoc_node(hash, buf, dbuf, n->next, m);
1176: }
1177:
1178: static int
1179: pman_node(MAN_ARGS)
1180: {
1181: const struct man_node *head, *body;
1182: const char *start, *sv;
1183: size_t sz;
1184:
1185: if (NULL == n)
1186: return(0);
1187:
1188: /*
1189: * We're only searching for one thing: the first text child in
1190: * the BODY of a NAME section. Since we don't keep track of
1191: * sections in -man, run some hoops to find out whether we're in
1192: * the correct section or not.
1193: */
1194:
1195: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1196: body = n;
1197: assert(body->parent);
1198: if (NULL != (head = body->parent->head) &&
1199: 1 == head->nchild &&
1200: NULL != (head = (head->child)) &&
1201: MAN_TEXT == head->type &&
1202: 0 == strcmp(head->string, "NAME") &&
1203: NULL != (body = body->child) &&
1204: MAN_TEXT == body->type) {
1205:
1206: assert(body->string);
1207: start = sv = body->string;
1208:
1209: /*
1210: * Go through a special heuristic dance here.
1211: * This is why -man manuals are great!
1212: * (I'm being sarcastic: my eyes are bleeding.)
1213: * Conventionally, one or more manual names are
1214: * comma-specified prior to a whitespace, then a
1215: * dash, then a description. Try to puzzle out
1216: * the name parts here.
1217: */
1218:
1219: for ( ;; ) {
1220: sz = strcspn(start, " ,");
1221: if ('\0' == start[(int)sz])
1222: break;
1223:
1224: buf->len = 0;
1225: buf_appendb(buf, start, sz);
1226: buf_appendb(buf, "", 1);
1227:
1.8 schwarze 1228: hash_put(hash, buf, TYPE_Nm);
1.1 kristaps 1229:
1230: if (' ' == start[(int)sz]) {
1231: start += (int)sz + 1;
1232: break;
1233: }
1234:
1235: assert(',' == start[(int)sz]);
1236: start += (int)sz + 1;
1237: while (' ' == *start)
1238: start++;
1239: }
1240:
1241: buf->len = 0;
1242:
1243: if (sv == start) {
1244: buf_append(buf, start);
1245: return(1);
1246: }
1247:
1248: while (' ' == *start)
1249: start++;
1250:
1251: if (0 == strncmp(start, "-", 1))
1252: start += 1;
1253: else if (0 == strncmp(start, "\\-", 2))
1254: start += 2;
1255: else if (0 == strncmp(start, "\\(en", 4))
1256: start += 4;
1257: else if (0 == strncmp(start, "\\(em", 4))
1258: start += 4;
1259:
1260: while (' ' == *start)
1261: start++;
1262:
1263: sz = strlen(start) + 1;
1264: buf_appendb(dbuf, start, sz);
1265: buf_appendb(buf, start, sz);
1266:
1.8 schwarze 1267: hash_put(hash, buf, TYPE_Nd);
1.1 kristaps 1268: }
1269: }
1270:
1.7 schwarze 1271: for (n = n->child; n; n = n->next)
1272: if (pman_node(hash, buf, dbuf, n))
1273: return(1);
1.1 kristaps 1274:
1275: return(0);
1276: }
1277:
1.14 schwarze 1278: /*
1279: * Parse a formatted manual page.
1280: * By necessity, this involves rather crude guesswork.
1281: */
1282: static void
1283: pformatted(DB *hash, struct buf *buf, struct buf *dbuf,
1284: const struct of *of)
1285: {
1286: FILE *stream;
1287: char *line, *p;
1288: size_t len, plen;
1289:
1290: if (NULL == (stream = fopen(of->fname, "r"))) {
1291: perror(of->fname);
1292: return;
1293: }
1294:
1295: /*
1296: * Always use the title derived from the filename up front,
1297: * do not even try to find it in the file. This also makes
1298: * sure we don't end up with an orphan index record, even if
1299: * the file content turns out to be completely unintelligible.
1300: */
1301:
1302: buf->len = 0;
1303: buf_append(buf, of->title);
1304: hash_put(hash, buf, TYPE_Nm);
1305:
1.31 schwarze 1306: /* Skip to first blank line. */
1.14 schwarze 1307:
1.28 kristaps 1308: while (NULL != (line = fgetln(stream, &len)))
1.31 schwarze 1309: if ('\n' == *line)
1.28 kristaps 1310: break;
1311:
1.31 schwarze 1312: /*
1313: * Assume the first line that is not indented
1314: * is the first section header. Skip to it.
1.28 kristaps 1315: */
1316:
1317: while (NULL != (line = fgetln(stream, &len)))
1.31 schwarze 1318: if ('\n' != *line && ' ' != *line)
1.28 kristaps 1319: break;
1.14 schwarze 1320:
1321: /*
1.31 schwarze 1322: * If no page content can be found, or the input line
1323: * is already the next section header, or there is no
1324: * trailing newline, reuse the page title as the page
1325: * description.
1.14 schwarze 1326: */
1327:
1.28 kristaps 1328: line = fgetln(stream, &len);
1.31 schwarze 1329: if (NULL == line || ' ' != *line || '\n' != line[(int)len - 1]) {
1.14 schwarze 1330: buf_appendb(dbuf, buf->cp, buf->size);
1331: hash_put(hash, buf, TYPE_Nd);
1332: fclose(stream);
1333: return;
1334: }
1335:
1.28 kristaps 1336: line[(int)--len] = '\0';
1337:
1.31 schwarze 1338: /*
1339: * Skip to the first dash.
1.28 kristaps 1340: * Use the remaining line as the description (no more than 70
1341: * bytes).
1.14 schwarze 1342: */
1343:
1.30 kristaps 1344: if (NULL != (p = strstr(line, "- "))) {
1345: for (p += 2; ' ' == *p || '\b' == *p; p++)
1.28 kristaps 1346: /* Skip to next word. */ ;
1347: } else
1.14 schwarze 1348: p = line;
1.28 kristaps 1349:
1350: if ((plen = strlen(p)) > 70) {
1351: plen = 70;
1352: p[plen] = '\0';
1.29 kristaps 1353: }
1354:
1355: /* Strip backspace-encoding from line. */
1356:
1357: while (NULL != (line = memchr(p, '\b', plen))) {
1358: len = line - p;
1359: if (0 == len) {
1360: memmove(line, line + 1, plen--);
1361: continue;
1362: }
1363: memmove(line - 1, line + 1, plen - len);
1364: plen -= 2;
1.14 schwarze 1365: }
1366:
1.28 kristaps 1367: buf_appendb(dbuf, p, plen + 1);
1.14 schwarze 1368: buf->len = 0;
1.28 kristaps 1369: buf_appendb(buf, p, plen + 1);
1.14 schwarze 1370: hash_put(hash, buf, TYPE_Nd);
1.28 kristaps 1371: fclose(stream);
1.14 schwarze 1372: }
1373:
1.5 kristaps 1374: static void
1.16 schwarze 1375: ofile_argbuild(int argc, char *argv[], struct of **of)
1.5 kristaps 1376: {
1.12 schwarze 1377: char buf[MAXPATHLEN];
1378: char *sec, *arch, *title, *p;
1.14 schwarze 1379: int i, src_form;
1.5 kristaps 1380: struct of *nof;
1381:
1382: for (i = 0; i < argc; i++) {
1.12 schwarze 1383:
1384: /*
1385: * Try to infer the manual section, architecture and
1386: * page title from the path, assuming it looks like
1.14 schwarze 1387: * man*[/<arch>]/<title>.<section> or
1388: * cat<section>[/<arch>]/<title>.0
1.12 schwarze 1389: */
1390:
1391: if (strlcpy(buf, argv[i], sizeof(buf)) >= sizeof(buf)) {
1392: fprintf(stderr, "%s: Path too long\n", argv[i]);
1393: continue;
1394: }
1395: sec = arch = title = NULL;
1.14 schwarze 1396: src_form = 0;
1.12 schwarze 1397: p = strrchr(buf, '\0');
1398: while (p-- > buf) {
1399: if (NULL == sec && '.' == *p) {
1400: sec = p + 1;
1401: *p = '\0';
1.14 schwarze 1402: if ('0' == *sec)
1403: src_form |= MANDOC_FORM;
1404: else if ('1' <= *sec && '9' >= *sec)
1405: src_form |= MANDOC_SRC;
1.12 schwarze 1406: continue;
1407: }
1408: if ('/' != *p)
1409: continue;
1410: if (NULL == title) {
1411: title = p + 1;
1412: *p = '\0';
1413: continue;
1414: }
1.24 schwarze 1415: if (0 == strncmp("man", p + 1, 3))
1.14 schwarze 1416: src_form |= MANDOC_SRC;
1.24 schwarze 1417: else if (0 == strncmp("cat", p + 1, 3))
1.14 schwarze 1418: src_form |= MANDOC_FORM;
1.24 schwarze 1419: else
1.12 schwarze 1420: arch = p + 1;
1421: break;
1422: }
1423: if (NULL == title)
1424: title = buf;
1425:
1426: /*
1427: * Build the file structure.
1428: */
1429:
1.5 kristaps 1430: nof = mandoc_calloc(1, sizeof(struct of));
1.12 schwarze 1431: nof->fname = mandoc_strdup(argv[i]);
1432: if (NULL != sec)
1433: nof->sec = mandoc_strdup(sec);
1434: if (NULL != arch)
1435: nof->arch = mandoc_strdup(arch);
1436: nof->title = mandoc_strdup(title);
1.14 schwarze 1437: nof->src_form = src_form;
1.12 schwarze 1438:
1439: /*
1440: * Add the structure to the list.
1441: */
1442:
1.5 kristaps 1443: if (NULL == *of) {
1444: *of = nof;
1445: (*of)->first = nof;
1446: } else {
1447: nof->first = (*of)->first;
1448: (*of)->next = nof;
1449: *of = nof;
1450: }
1451: }
1452: }
1453:
1.4 kristaps 1454: /*
1455: * Recursively build up a list of files to parse.
1456: * We use this instead of ftw() and so on because I don't want global
1457: * variables hanging around.
1458: * This ignores the mandoc.db and mandoc.index files, but assumes that
1459: * everything else is a manual.
1460: * Pass in a pointer to a NULL structure for the first invocation.
1461: */
1.35 kristaps 1462: static void
1.12 schwarze 1463: ofile_dirbuild(const char *dir, const char* psec, const char *parch,
1.16 schwarze 1464: int p_src_form, struct of **of)
1.4 kristaps 1465: {
1.5 kristaps 1466: char buf[MAXPATHLEN];
1467: size_t sz;
1.4 kristaps 1468: DIR *d;
1.12 schwarze 1469: const char *fn, *sec, *arch;
1.14 schwarze 1470: char *p, *q, *suffix;
1.4 kristaps 1471: struct of *nof;
1472: struct dirent *dp;
1.14 schwarze 1473: int src_form;
1.4 kristaps 1474:
1475: if (NULL == (d = opendir(dir))) {
1476: perror(dir);
1.35 kristaps 1477: exit((int)MANDOCLEVEL_SYSERR);
1.4 kristaps 1478: }
1479:
1480: while (NULL != (dp = readdir(d))) {
1481: fn = dp->d_name;
1.12 schwarze 1482:
1483: if ('.' == *fn)
1484: continue;
1485:
1.14 schwarze 1486: src_form = p_src_form;
1487:
1.4 kristaps 1488: if (DT_DIR == dp->d_type) {
1.12 schwarze 1489: sec = psec;
1490: arch = parch;
1491:
1492: /*
1493: * By default, only use directories called:
1.14 schwarze 1494: * man<section>/[<arch>/] or
1495: * cat<section>/[<arch>/]
1.12 schwarze 1496: */
1497:
1498: if (NULL == sec) {
1.14 schwarze 1499: if(0 == strncmp("man", fn, 3)) {
1500: src_form |= MANDOC_SRC;
1.12 schwarze 1501: sec = fn + 3;
1.14 schwarze 1502: } else if (0 == strncmp("cat", fn, 3)) {
1503: src_form |= MANDOC_FORM;
1504: sec = fn + 3;
1505: } else if (use_all)
1.12 schwarze 1506: sec = fn;
1507: else
1508: continue;
1509: } else if (NULL == arch && (use_all ||
1510: NULL == strchr(fn, '.')))
1511: arch = fn;
1512: else if (0 == use_all)
1.5 kristaps 1513: continue;
1514:
1515: buf[0] = '\0';
1516: strlcat(buf, dir, MAXPATHLEN);
1517: strlcat(buf, "/", MAXPATHLEN);
1518: sz = strlcat(buf, fn, MAXPATHLEN);
1519:
1.12 schwarze 1520: if (MAXPATHLEN <= sz) {
1521: fprintf(stderr, "%s: Path too long\n", dir);
1.35 kristaps 1522: exit((int)MANDOCLEVEL_SYSERR);
1.12 schwarze 1523: }
1524:
1.35 kristaps 1525: ofile_dirbuild(buf, sec, arch, src_form, of);
1526: }
1.12 schwarze 1527:
1528: if (DT_REG != dp->d_type ||
1.35 kristaps 1529: (NULL == psec && !use_all) ||
1530: ! strcmp(MANDOC_DB, fn) ||
1531: ! strcmp(MANDOC_IDX, fn))
1.12 schwarze 1532: continue;
1533:
1534: /*
1535: * By default, skip files where the file name suffix
1536: * does not agree with the section directory
1537: * they are located in.
1538: */
1539:
1540: suffix = strrchr(fn, '.');
1541: if (0 == use_all) {
1542: if (NULL == suffix)
1.5 kristaps 1543: continue;
1.14 schwarze 1544: if ((MANDOC_SRC & src_form &&
1545: strcmp(suffix + 1, psec)) ||
1546: (MANDOC_FORM & src_form &&
1547: strcmp(suffix + 1, "0")))
1548: continue;
1549: }
1550: if (NULL != suffix) {
1551: if ('0' == suffix[1])
1552: src_form |= MANDOC_FORM;
1553: else if ('1' <= suffix[1] && '9' >= suffix[1])
1554: src_form |= MANDOC_SRC;
1555: }
1556:
1557:
1558: /*
1559: * Skip formatted manuals if a source version is
1560: * available. Ignore the age: it is very unlikely
1561: * that people install newer formatted base manuals
1562: * when they used to have source manuals before,
1563: * and in ports, old manuals get removed on update.
1564: */
1565: if (0 == use_all && MANDOC_FORM & src_form &&
1566: NULL != psec) {
1567: buf[0] = '\0';
1568: strlcat(buf, dir, MAXPATHLEN);
1569: p = strrchr(buf, '/');
1.32 schwarze 1570: if (NULL != parch && NULL != p)
1571: for (p--; p > buf; p--)
1572: if ('/' == *p)
1573: break;
1.14 schwarze 1574: if (NULL == p)
1575: p = buf;
1576: else
1577: p++;
1578: if (0 == strncmp("cat", p, 3))
1579: memcpy(p, "man", 3);
1580: strlcat(buf, "/", MAXPATHLEN);
1581: sz = strlcat(buf, fn, MAXPATHLEN);
1582: if (sz >= MAXPATHLEN) {
1583: fprintf(stderr, "%s: Path too long\n", buf);
1.5 kristaps 1584: continue;
1.14 schwarze 1585: }
1586: q = strrchr(buf, '.');
1587: if (NULL != q && p < q++) {
1588: *q = '\0';
1589: sz = strlcat(buf, psec, MAXPATHLEN);
1590: if (sz >= MAXPATHLEN) {
1591: fprintf(stderr,
1592: "%s: Path too long\n", buf);
1593: continue;
1594: }
1.35 kristaps 1595: if (0 == access(buf, R_OK))
1.14 schwarze 1596: continue;
1597: }
1.5 kristaps 1598: }
1.4 kristaps 1599:
1.35 kristaps 1600: assert('.' == dir[0]);
1601: assert('/' == dir[1]);
1.5 kristaps 1602: buf[0] = '\0';
1.35 kristaps 1603: strlcat(buf, dir + 2, MAXPATHLEN);
1.5 kristaps 1604: strlcat(buf, "/", MAXPATHLEN);
1.6 schwarze 1605: sz = strlcat(buf, fn, MAXPATHLEN);
1.5 kristaps 1606: if (sz >= MAXPATHLEN) {
1607: fprintf(stderr, "%s: Path too long\n", dir);
1.14 schwarze 1608: continue;
1.5 kristaps 1609: }
1610:
1.4 kristaps 1611: nof = mandoc_calloc(1, sizeof(struct of));
1.5 kristaps 1612: nof->fname = mandoc_strdup(buf);
1.12 schwarze 1613: if (NULL != psec)
1614: nof->sec = mandoc_strdup(psec);
1615: if (NULL != parch)
1616: nof->arch = mandoc_strdup(parch);
1.14 schwarze 1617: nof->src_form = src_form;
1.12 schwarze 1618:
1619: /*
1620: * Remember the file name without the extension,
1621: * to be used as the page title in the database.
1622: */
1623:
1624: if (NULL != suffix)
1625: *suffix = '\0';
1626: nof->title = mandoc_strdup(fn);
1.5 kristaps 1627:
1.14 schwarze 1628: /*
1629: * Add the structure to the list.
1630: */
1631:
1.4 kristaps 1632: if (NULL == *of) {
1633: *of = nof;
1634: (*of)->first = nof;
1635: } else {
1.5 kristaps 1636: nof->first = (*of)->first;
1.4 kristaps 1637: (*of)->next = nof;
1638: *of = nof;
1639: }
1640: }
1641:
1.7 schwarze 1642: closedir(d);
1.4 kristaps 1643: }
1644:
1645: static void
1646: ofile_free(struct of *of)
1647: {
1648: struct of *nof;
1649:
1650: while (of) {
1651: nof = of->next;
1652: free(of->fname);
1.12 schwarze 1653: free(of->sec);
1654: free(of->arch);
1655: free(of->title);
1.4 kristaps 1656: free(of);
1657: of = nof;
1658: }
1659: }
1660:
1.1 kristaps 1661: static void
1662: usage(void)
1663: {
1664:
1.5 kristaps 1665: fprintf(stderr, "usage: %s [-v] "
1.34 schwarze 1666: "[-C file] |"
1667: " dir ... |"
1668: " -d dir [file ...] |"
1669: " -u dir [file ...]\n", progname);
1.1 kristaps 1670: }
CVSweb