Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.63
1.63 ! schwarze 1: /* $Id: mandocdb.c,v 1.62 2014/01/06 13:54:11 schwarze Exp $ */
1.1 schwarze 2: /*
1.47 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.52 schwarze 4: * Copyright (c) 2011, 2012, 2013, 2014 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: */
1.47 schwarze 18: #include <sys/stat.h>
1.1 schwarze 19:
20: #include <assert.h>
1.33 schwarze 21: #include <ctype.h>
1.34 schwarze 22: #include <errno.h>
1.1 schwarze 23: #include <fcntl.h>
1.47 schwarze 24: #include <fts.h>
1.1 schwarze 25: #include <getopt.h>
1.44 schwarze 26: #include <limits.h>
1.47 schwarze 27: #include <stddef.h>
1.1 schwarze 28: #include <stdio.h>
29: #include <stdint.h>
30: #include <stdlib.h>
31: #include <string.h>
1.14 schwarze 32: #include <unistd.h>
1.1 schwarze 33:
1.47 schwarze 34: #include <ohash.h>
35: #include <sqlite3.h>
36:
37: #include "mdoc.h"
1.1 schwarze 38: #include "man.h"
39: #include "mandoc.h"
1.10 schwarze 40: #include "manpath.h"
1.47 schwarze 41: #include "mansearch.h"
1.1 schwarze 42:
1.47 schwarze 43: #define SQL_EXEC(_v) \
44: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
45: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
46: #define SQL_BIND_TEXT(_s, _i, _v) \
47: if (SQLITE_OK != sqlite3_bind_text \
48: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
49: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
50: #define SQL_BIND_INT(_s, _i, _v) \
51: if (SQLITE_OK != sqlite3_bind_int \
52: ((_s), (_i)++, (_v))) \
53: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
54: #define SQL_BIND_INT64(_s, _i, _v) \
55: if (SQLITE_OK != sqlite3_bind_int64 \
56: ((_s), (_i)++, (_v))) \
57: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
58: #define SQL_STEP(_s) \
59: if (SQLITE_DONE != sqlite3_step((_s))) \
60: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
1.1 schwarze 61:
1.47 schwarze 62: enum op {
63: OP_DEFAULT = 0, /* new dbs from dir list or default config */
64: OP_CONFFILE, /* new databases from custom config file */
65: OP_UPDATE, /* delete/add entries in existing database */
66: OP_DELETE, /* delete entries from existing database */
67: OP_TEST /* change no databases, report potential problems */
68: };
1.11 schwarze 69:
1.47 schwarze 70: enum form {
71: FORM_NONE, /* format is unknown */
72: FORM_SRC, /* format is -man or -mdoc */
73: FORM_CAT /* format is cat */
74: };
1.28 schwarze 75:
1.47 schwarze 76: struct str {
1.53 schwarze 77: char *rendered; /* key in UTF-8 or ASCII form */
1.47 schwarze 78: const struct mpage *mpage; /* if set, the owning parse */
79: uint64_t mask; /* bitmask in sequence */
1.53 schwarze 80: char key[]; /* may contain escape sequences */
1.28 schwarze 81: };
82:
1.47 schwarze 83: struct inodev {
84: ino_t st_ino;
85: dev_t st_dev;
86: };
1.28 schwarze 87:
1.47 schwarze 88: struct mpage {
89: struct inodev inodev; /* used for hashing routine */
90: enum form form; /* format from file content */
91: char *sec; /* section from file content */
92: char *arch; /* architecture from file content */
93: char *title; /* title from file content */
94: char *desc; /* description from file content */
95: struct mlink *mlinks; /* singly linked list */
1.28 schwarze 96: };
97:
1.47 schwarze 98: struct mlink {
99: char file[PATH_MAX]; /* filename rel. to manpath */
100: enum form dform; /* format from directory */
101: enum form fform; /* format from file name suffix */
102: char *dsec; /* section from directory */
103: char *arch; /* architecture from directory */
104: char *name; /* name from file name (not empty) */
105: char *fsec; /* section from file name suffix */
106: struct mlink *next; /* singly linked list */
1.2 schwarze 107: };
108:
1.47 schwarze 109: enum stmt {
110: STMT_DELETE_PAGE = 0, /* delete mpage */
111: STMT_INSERT_PAGE, /* insert mpage */
112: STMT_INSERT_LINK, /* insert mlink */
113: STMT_INSERT_KEY, /* insert parsed key */
114: STMT__MAX
1.1 schwarze 115: };
116:
1.47 schwarze 117: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
1.1 schwarze 118:
1.19 schwarze 119: struct mdoc_handler {
1.47 schwarze 120: mdoc_fp fp; /* optional handler */
121: uint64_t mask; /* set unless handler returns 0 */
1.19 schwarze 122: };
123:
1.47 schwarze 124: static void dbclose(int);
1.62 schwarze 125: static void dbadd(const struct mpage *, struct mchars *);
1.47 schwarze 126: static int dbopen(int);
127: static void dbprune(void);
128: static void filescan(const char *);
129: static void *hash_alloc(size_t, void *);
130: static void hash_free(void *, size_t, void *);
131: static void *hash_halloc(size_t, void *);
132: static void mlink_add(struct mlink *, const struct stat *);
1.50 schwarze 133: static int mlink_check(struct mpage *, struct mlink *);
1.47 schwarze 134: static void mlink_free(struct mlink *);
135: static void mlinks_undupe(struct mpage *);
136: static void mpages_free(void);
1.58 schwarze 137: static void mpages_merge(struct mchars *, struct mparse *);
1.47 schwarze 138: static void parse_cat(struct mpage *);
139: static void parse_man(struct mpage *, const struct man_node *);
140: static void parse_mdoc(struct mpage *, const struct mdoc_node *);
141: static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
142: static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
143: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
144: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
145: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
146: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
147: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
148: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
149: static void putkey(const struct mpage *,
150: const char *, uint64_t);
151: static void putkeys(const struct mpage *,
152: const char *, size_t, uint64_t);
153: static void putmdockey(const struct mpage *,
154: const struct mdoc_node *, uint64_t);
1.53 schwarze 155: static void render_key(struct mchars *, struct str *);
1.47 schwarze 156: static void say(const char *, const char *, ...);
157: static int set_basedir(const char *);
158: static int treescan(void);
159: static size_t utf8(unsigned int, char [7]);
160:
161: static char *progname;
1.59 schwarze 162: static int nodb; /* no database changes */
163: static int quick; /* abort the parse early */
1.47 schwarze 164: static int use_all; /* use all found files */
165: static int verb; /* print what we're doing */
166: static int warnings; /* warn about crap */
1.52 schwarze 167: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 168: static int exitcode; /* to be returned by main */
169: static enum op op; /* operational mode */
170: static char basedir[PATH_MAX]; /* current base directory */
171: static struct ohash mpages; /* table of distinct manual pages */
172: static struct ohash mlinks; /* table of directory entries */
173: static struct ohash strings; /* table of all strings */
174: static sqlite3 *db = NULL; /* current database */
175: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
176:
1.19 schwarze 177: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 178: { NULL, 0 }, /* Ap */
179: { NULL, 0 }, /* Dd */
180: { NULL, 0 }, /* Dt */
181: { NULL, 0 }, /* Os */
182: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
183: { parse_mdoc_head, TYPE_Ss }, /* Ss */
184: { NULL, 0 }, /* Pp */
185: { NULL, 0 }, /* D1 */
186: { NULL, 0 }, /* Dl */
187: { NULL, 0 }, /* Bd */
188: { NULL, 0 }, /* Ed */
189: { NULL, 0 }, /* Bl */
190: { NULL, 0 }, /* El */
191: { NULL, 0 }, /* It */
192: { NULL, 0 }, /* Ad */
193: { NULL, TYPE_An }, /* An */
194: { NULL, TYPE_Ar }, /* Ar */
195: { NULL, TYPE_Cd }, /* Cd */
196: { NULL, TYPE_Cm }, /* Cm */
197: { NULL, TYPE_Dv }, /* Dv */
198: { NULL, TYPE_Er }, /* Er */
199: { NULL, TYPE_Ev }, /* Ev */
200: { NULL, 0 }, /* Ex */
201: { NULL, TYPE_Fa }, /* Fa */
202: { parse_mdoc_Fd, 0 }, /* Fd */
203: { NULL, TYPE_Fl }, /* Fl */
204: { parse_mdoc_Fn, 0 }, /* Fn */
205: { NULL, TYPE_Ft }, /* Ft */
206: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 207: { NULL, TYPE_In }, /* In */
1.47 schwarze 208: { NULL, TYPE_Li }, /* Li */
209: { parse_mdoc_Nd, TYPE_Nd }, /* Nd */
210: { parse_mdoc_Nm, TYPE_Nm }, /* Nm */
211: { NULL, 0 }, /* Op */
212: { NULL, 0 }, /* Ot */
213: { NULL, TYPE_Pa }, /* Pa */
214: { NULL, 0 }, /* Rv */
1.49 schwarze 215: { NULL, TYPE_St }, /* St */
1.47 schwarze 216: { NULL, TYPE_Va }, /* Va */
217: { parse_mdoc_body, TYPE_Va }, /* Vt */
218: { parse_mdoc_Xr, 0 }, /* Xr */
219: { NULL, 0 }, /* %A */
220: { NULL, 0 }, /* %B */
221: { NULL, 0 }, /* %D */
222: { NULL, 0 }, /* %I */
223: { NULL, 0 }, /* %J */
224: { NULL, 0 }, /* %N */
225: { NULL, 0 }, /* %O */
226: { NULL, 0 }, /* %P */
227: { NULL, 0 }, /* %R */
228: { NULL, 0 }, /* %T */
229: { NULL, 0 }, /* %V */
230: { NULL, 0 }, /* Ac */
231: { NULL, 0 }, /* Ao */
232: { NULL, 0 }, /* Aq */
233: { NULL, TYPE_At }, /* At */
234: { NULL, 0 }, /* Bc */
235: { NULL, 0 }, /* Bf */
236: { NULL, 0 }, /* Bo */
237: { NULL, 0 }, /* Bq */
238: { NULL, TYPE_Bsx }, /* Bsx */
239: { NULL, TYPE_Bx }, /* Bx */
240: { NULL, 0 }, /* Db */
241: { NULL, 0 }, /* Dc */
242: { NULL, 0 }, /* Do */
243: { NULL, 0 }, /* Dq */
244: { NULL, 0 }, /* Ec */
245: { NULL, 0 }, /* Ef */
246: { NULL, TYPE_Em }, /* Em */
247: { NULL, 0 }, /* Eo */
248: { NULL, TYPE_Fx }, /* Fx */
249: { NULL, TYPE_Ms }, /* Ms */
250: { NULL, 0 }, /* No */
251: { NULL, 0 }, /* Ns */
252: { NULL, TYPE_Nx }, /* Nx */
253: { NULL, TYPE_Ox }, /* Ox */
254: { NULL, 0 }, /* Pc */
255: { NULL, 0 }, /* Pf */
256: { NULL, 0 }, /* Po */
257: { NULL, 0 }, /* Pq */
258: { NULL, 0 }, /* Qc */
259: { NULL, 0 }, /* Ql */
260: { NULL, 0 }, /* Qo */
261: { NULL, 0 }, /* Qq */
262: { NULL, 0 }, /* Re */
263: { NULL, 0 }, /* Rs */
264: { NULL, 0 }, /* Sc */
265: { NULL, 0 }, /* So */
266: { NULL, 0 }, /* Sq */
267: { NULL, 0 }, /* Sm */
268: { NULL, 0 }, /* Sx */
269: { NULL, TYPE_Sy }, /* Sy */
270: { NULL, TYPE_Tn }, /* Tn */
271: { NULL, 0 }, /* Ux */
272: { NULL, 0 }, /* Xc */
273: { NULL, 0 }, /* Xo */
274: { parse_mdoc_head, 0 }, /* Fo */
275: { NULL, 0 }, /* Fc */
276: { NULL, 0 }, /* Oo */
277: { NULL, 0 }, /* Oc */
278: { NULL, 0 }, /* Bk */
279: { NULL, 0 }, /* Ek */
280: { NULL, 0 }, /* Bt */
281: { NULL, 0 }, /* Hf */
282: { NULL, 0 }, /* Fr */
283: { NULL, 0 }, /* Ud */
284: { NULL, TYPE_Lb }, /* Lb */
285: { NULL, 0 }, /* Lp */
286: { NULL, TYPE_Lk }, /* Lk */
287: { NULL, TYPE_Mt }, /* Mt */
288: { NULL, 0 }, /* Brq */
289: { NULL, 0 }, /* Bro */
290: { NULL, 0 }, /* Brc */
291: { NULL, 0 }, /* %C */
292: { NULL, 0 }, /* Es */
293: { NULL, 0 }, /* En */
294: { NULL, TYPE_Dx }, /* Dx */
295: { NULL, 0 }, /* %Q */
296: { NULL, 0 }, /* br */
297: { NULL, 0 }, /* sp */
298: { NULL, 0 }, /* %U */
299: { NULL, 0 }, /* Ta */
1.1 schwarze 300: };
301:
302: int
1.3 schwarze 303: mandocdb(int argc, char *argv[])
1.1 schwarze 304: {
1.47 schwarze 305: int ch, i;
306: size_t j, sz;
307: const char *path_arg;
308: struct mchars *mc;
309: struct manpaths dirs;
310: struct mparse *mp;
311: struct ohash_info mpages_info, mlinks_info;
312:
313: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
314: memset(&dirs, 0, sizeof(struct manpaths));
315:
316: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
317: mpages_info.halloc = mlinks_info.halloc = hash_halloc;
318: mpages_info.hfree = mlinks_info.hfree = hash_free;
319:
320: mpages_info.key_offset = offsetof(struct mpage, inodev);
321: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 322:
323: progname = strrchr(argv[0], '/');
324: if (progname == NULL)
325: progname = argv[0];
326: else
327: ++progname;
328:
1.47 schwarze 329: /*
330: * We accept a few different invocations.
331: * The CHECKOP macro makes sure that invocation styles don't
332: * clobber each other.
333: */
334: #define CHECKOP(_op, _ch) do \
335: if (OP_DEFAULT != (_op)) { \
336: fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
337: goto usage; \
338: } while (/*CONSTCOND*/0)
1.10 schwarze 339:
1.47 schwarze 340: path_arg = NULL;
1.28 schwarze 341: op = OP_DEFAULT;
1.1 schwarze 342:
1.59 schwarze 343: while (-1 != (ch = getopt(argc, argv, "aC:d:nQT:tu:vW")))
1.1 schwarze 344: switch (ch) {
1.6 schwarze 345: case ('a'):
346: use_all = 1;
347: break;
1.25 schwarze 348: case ('C'):
1.47 schwarze 349: CHECKOP(op, ch);
350: path_arg = optarg;
1.28 schwarze 351: op = OP_CONFFILE;
1.25 schwarze 352: break;
1.1 schwarze 353: case ('d'):
1.47 schwarze 354: CHECKOP(op, ch);
355: path_arg = optarg;
1.2 schwarze 356: op = OP_UPDATE;
1.1 schwarze 357: break;
1.47 schwarze 358: case ('n'):
359: nodb = 1;
360: break;
1.59 schwarze 361: case ('Q'):
362: quick = 1;
363: break;
1.52 schwarze 364: case ('T'):
365: if (strcmp(optarg, "utf8")) {
366: fprintf(stderr, "-T%s: Unsupported "
367: "output format\n", optarg);
368: goto usage;
369: }
370: write_utf8 = 1;
371: break;
1.28 schwarze 372: case ('t'):
1.47 schwarze 373: CHECKOP(op, ch);
1.28 schwarze 374: dup2(STDOUT_FILENO, STDERR_FILENO);
375: op = OP_TEST;
1.47 schwarze 376: nodb = warnings = 1;
1.28 schwarze 377: break;
1.2 schwarze 378: case ('u'):
1.47 schwarze 379: CHECKOP(op, ch);
380: path_arg = optarg;
1.1 schwarze 381: op = OP_DELETE;
382: break;
383: case ('v'):
384: verb++;
385: break;
1.28 schwarze 386: case ('W'):
387: warnings = 1;
388: break;
1.1 schwarze 389: default:
1.28 schwarze 390: goto usage;
1.1 schwarze 391: }
392:
393: argc -= optind;
394: argv += optind;
395:
1.28 schwarze 396: if (OP_CONFFILE == op && argc > 0) {
1.47 schwarze 397: fprintf(stderr, "-C: Too many arguments\n");
1.28 schwarze 398: goto usage;
399: }
400:
1.47 schwarze 401: exitcode = (int)MANDOCLEVEL_OK;
402: mp = mparse_alloc(MPARSE_AUTO,
1.59 schwarze 403: MANDOCLEVEL_FATAL, NULL, NULL, quick);
1.47 schwarze 404: mc = mchars_alloc();
1.2 schwarze 405:
1.47 schwarze 406: ohash_init(&mpages, 6, &mpages_info);
407: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 408:
1.47 schwarze 409: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
410: /*
411: * Force processing all files.
412: */
413: use_all = 1;
1.2 schwarze 414:
1.47 schwarze 415: /*
416: * All of these deal with a specific directory.
417: * Jump into that directory then collect files specified
418: * on the command-line.
419: */
420: if (0 == set_basedir(path_arg))
421: goto out;
422: for (i = 0; i < argc; i++)
423: filescan(argv[i]);
424: if (0 == dbopen(1))
425: goto out;
426: if (OP_TEST != op)
427: dbprune();
428: if (OP_DELETE != op)
1.58 schwarze 429: mpages_merge(mc, mp);
1.47 schwarze 430: dbclose(1);
431: } else {
432: /*
433: * If we have arguments, use them as our manpaths.
434: * If we don't, grok from manpath(1) or however else
435: * manpath_parse() wants to do it.
436: */
437: if (argc > 0) {
438: dirs.paths = mandoc_calloc
439: (argc, sizeof(char *));
440: dirs.sz = (size_t)argc;
441: for (i = 0; i < argc; i++)
442: dirs.paths[i] = mandoc_strdup(argv[i]);
443: } else
444: manpath_parse(&dirs, path_arg, NULL, NULL);
1.2 schwarze 445:
1.47 schwarze 446: /*
447: * First scan the tree rooted at a base directory, then
448: * build a new database and finally move it into place.
449: * Ignore zero-length directories and strip trailing
450: * slashes.
451: */
452: for (j = 0; j < dirs.sz; j++) {
453: sz = strlen(dirs.paths[j]);
454: if (sz && '/' == dirs.paths[j][sz - 1])
455: dirs.paths[j][--sz] = '\0';
456: if (0 == sz)
457: continue;
1.2 schwarze 458:
1.47 schwarze 459: if (j) {
460: ohash_init(&mpages, 6, &mpages_info);
461: ohash_init(&mlinks, 6, &mlinks_info);
462: }
1.1 schwarze 463:
1.47 schwarze 464: if (0 == set_basedir(dirs.paths[j]))
465: goto out;
466: if (0 == treescan())
467: goto out;
468: if (0 == set_basedir(dirs.paths[j]))
469: goto out;
470: if (0 == dbopen(0))
471: goto out;
1.41 deraadt 472:
1.58 schwarze 473: mpages_merge(mc, mp);
1.47 schwarze 474: dbclose(0);
1.28 schwarze 475:
1.47 schwarze 476: if (j + 1 < dirs.sz) {
477: mpages_free();
478: ohash_delete(&mpages);
479: ohash_delete(&mlinks);
480: }
1.2 schwarze 481: }
1.47 schwarze 482: }
483: out:
484: set_basedir(NULL);
485: manpath_free(&dirs);
486: mchars_free(mc);
487: mparse_free(mp);
488: mpages_free();
489: ohash_delete(&mpages);
490: ohash_delete(&mlinks);
491: return(exitcode);
492: usage:
1.59 schwarze 493: fprintf(stderr, "usage: %s [-anQvW] [-C file] [-Tutf8]\n"
494: " %s [-anQvW] [-Tutf8] dir ...\n"
495: " %s [-nQvW] [-Tutf8] -d dir [file ...]\n"
1.47 schwarze 496: " %s [-nvW] -u dir [file ...]\n"
1.59 schwarze 497: " %s [-Q] -t file ...\n",
1.47 schwarze 498: progname, progname, progname,
499: progname, progname);
1.1 schwarze 500:
1.47 schwarze 501: return((int)MANDOCLEVEL_BADARG);
502: }
1.1 schwarze 503:
1.47 schwarze 504: /*
505: * Scan a directory tree rooted at "basedir" for manpages.
506: * We use fts(), scanning directory parts along the way for clues to our
507: * section and architecture.
508: *
509: * If use_all has been specified, grok all files.
510: * If not, sanitise paths to the following:
511: *
512: * [./]man*[/<arch>]/<name>.<section>
513: * or
514: * [./]cat<section>[/<arch>]/<name>.0
515: *
516: * TODO: accomodate for multi-language directories.
517: */
518: static int
519: treescan(void)
520: {
521: FTS *f;
522: FTSENT *ff;
523: struct mlink *mlink;
524: int dform;
1.51 schwarze 525: char *dsec, *arch, *fsec, *cp;
526: const char *path;
1.47 schwarze 527: const char *argv[2];
1.1 schwarze 528:
1.47 schwarze 529: argv[0] = ".";
530: argv[1] = (char *)NULL;
1.1 schwarze 531:
1.47 schwarze 532: /*
533: * Walk through all components under the directory, using the
534: * logical descent of files.
535: */
536: f = fts_open((char * const *)argv, FTS_LOGICAL, NULL);
537: if (NULL == f) {
538: exitcode = (int)MANDOCLEVEL_SYSERR;
539: say("", NULL);
540: return(0);
541: }
1.2 schwarze 542:
1.47 schwarze 543: dsec = arch = NULL;
544: dform = FORM_NONE;
1.2 schwarze 545:
1.47 schwarze 546: while (NULL != (ff = fts_read(f))) {
547: path = ff->fts_path + 2;
1.15 schwarze 548: /*
1.47 schwarze 549: * If we're a regular file, add an mlink by using the
550: * stored directory data and handling the filename.
1.15 schwarze 551: */
1.47 schwarze 552: if (FTS_F == ff->fts_info) {
553: if (0 == strcmp(path, MANDOC_DB))
554: continue;
555: if ( ! use_all && ff->fts_level < 2) {
556: if (warnings)
557: say(path, "Extraneous file");
558: continue;
559: } else if (NULL == (fsec =
560: strrchr(ff->fts_name, '.'))) {
561: if ( ! use_all) {
562: if (warnings)
563: say(path,
564: "No filename suffix");
565: continue;
566: }
567: } else if (0 == strcmp(++fsec, "html")) {
568: if (warnings)
569: say(path, "Skip html");
570: continue;
571: } else if (0 == strcmp(fsec, "gz")) {
572: if (warnings)
573: say(path, "Skip gz");
574: continue;
575: } else if (0 == strcmp(fsec, "ps")) {
576: if (warnings)
577: say(path, "Skip ps");
578: continue;
579: } else if (0 == strcmp(fsec, "pdf")) {
580: if (warnings)
581: say(path, "Skip pdf");
582: continue;
583: } else if ( ! use_all &&
584: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
585: (FORM_CAT == dform && strcmp(fsec, "0")))) {
586: if (warnings)
587: say(path, "Wrong filename suffix");
588: continue;
589: } else
590: fsec[-1] = '\0';
1.51 schwarze 591:
1.47 schwarze 592: mlink = mandoc_calloc(1, sizeof(struct mlink));
593: strlcpy(mlink->file, path, sizeof(mlink->file));
594: mlink->dform = dform;
1.51 schwarze 595: mlink->dsec = dsec;
596: mlink->arch = arch;
597: mlink->name = ff->fts_name;
598: mlink->fsec = fsec;
1.47 schwarze 599: mlink_add(mlink, ff->fts_statp);
600: continue;
601: } else if (FTS_D != ff->fts_info &&
602: FTS_DP != ff->fts_info) {
603: if (warnings)
604: say(path, "Not a regular file");
605: continue;
606: }
607:
608: switch (ff->fts_level) {
609: case (0):
610: /* Ignore the root directory. */
611: break;
612: case (1):
613: /*
614: * This might contain manX/ or catX/.
615: * Try to infer this from the name.
616: * If we're not in use_all, enforce it.
617: */
618: cp = ff->fts_name;
619: if (FTS_DP == ff->fts_info)
620: break;
1.15 schwarze 621:
1.47 schwarze 622: if (0 == strncmp(cp, "man", 3)) {
623: dform = FORM_SRC;
624: dsec = cp + 3;
625: } else if (0 == strncmp(cp, "cat", 3)) {
626: dform = FORM_CAT;
627: dsec = cp + 3;
1.51 schwarze 628: } else {
629: dform = FORM_NONE;
630: dsec = NULL;
1.26 schwarze 631: }
1.47 schwarze 632:
633: if (NULL != dsec || use_all)
634: break;
635:
636: if (warnings)
637: say(path, "Unknown directory part");
638: fts_set(f, ff, FTS_SKIP);
639: break;
640: case (2):
641: /*
642: * Possibly our architecture.
643: * If we're descending, keep tabs on it.
644: */
645: if (FTS_DP != ff->fts_info && NULL != dsec)
646: arch = ff->fts_name;
1.51 schwarze 647: else
648: arch = NULL;
1.47 schwarze 649: break;
650: default:
651: if (FTS_DP == ff->fts_info || use_all)
652: break;
653: if (warnings)
654: say(path, "Extraneous directory part");
655: fts_set(f, ff, FTS_SKIP);
656: break;
1.14 schwarze 657: }
1.47 schwarze 658: }
659:
660: fts_close(f);
661: return(1);
662: }
1.1 schwarze 663:
1.47 schwarze 664: /*
665: * Add a file to the mlinks table.
666: * Do not verify that it's a "valid" looking manpage (we'll do that
667: * later).
668: *
669: * Try to infer the manual section, architecture, and page name from the
670: * path, assuming it looks like
671: *
672: * [./]man*[/<arch>]/<name>.<section>
673: * or
674: * [./]cat<section>[/<arch>]/<name>.0
675: *
676: * See treescan() for the fts(3) version of this.
677: */
678: static void
679: filescan(const char *file)
680: {
681: char buf[PATH_MAX];
682: struct stat st;
683: struct mlink *mlink;
684: char *p, *start;
685:
686: assert(use_all);
687:
688: if (0 == strncmp(file, "./", 2))
689: file += 2;
690:
691: if (NULL == realpath(file, buf)) {
692: exitcode = (int)MANDOCLEVEL_BADARG;
693: say(file, NULL);
694: return;
1.63 ! schwarze 695: }
! 696:
! 697: if (strstr(buf, basedir) == buf)
! 698: start = buf + strlen(basedir) + 1;
! 699: else if (OP_TEST == op)
! 700: start = buf;
! 701: else {
1.47 schwarze 702: exitcode = (int)MANDOCLEVEL_BADARG;
703: say("", "%s: outside base directory", buf);
704: return;
1.63 ! schwarze 705: }
! 706:
! 707: if (-1 == stat(buf, &st)) {
1.47 schwarze 708: exitcode = (int)MANDOCLEVEL_BADARG;
709: say(file, NULL);
710: return;
711: } else if ( ! (S_IFREG & st.st_mode)) {
712: exitcode = (int)MANDOCLEVEL_BADARG;
713: say(file, "Not a regular file");
714: return;
1.1 schwarze 715: }
1.63 ! schwarze 716:
1.47 schwarze 717: mlink = mandoc_calloc(1, sizeof(struct mlink));
718: strlcpy(mlink->file, start, sizeof(mlink->file));
1.1 schwarze 719:
1.10 schwarze 720: /*
1.47 schwarze 721: * First try to guess our directory structure.
722: * If we find a separator, try to look for man* or cat*.
723: * If we find one of these and what's underneath is a directory,
724: * assume it's an architecture.
1.10 schwarze 725: */
1.47 schwarze 726: if (NULL != (p = strchr(start, '/'))) {
727: *p++ = '\0';
728: if (0 == strncmp(start, "man", 3)) {
729: mlink->dform = FORM_SRC;
1.51 schwarze 730: mlink->dsec = start + 3;
1.47 schwarze 731: } else if (0 == strncmp(start, "cat", 3)) {
732: mlink->dform = FORM_CAT;
1.51 schwarze 733: mlink->dsec = start + 3;
1.47 schwarze 734: }
1.10 schwarze 735:
1.47 schwarze 736: start = p;
737: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
738: *p++ = '\0';
1.51 schwarze 739: mlink->arch = start;
1.47 schwarze 740: start = p;
1.41 deraadt 741: }
1.47 schwarze 742: }
1.7 schwarze 743:
1.47 schwarze 744: /*
745: * Now check the file suffix.
746: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
747: */
748: p = strrchr(start, '\0');
749: while (p-- > start && '/' != *p && '.' != *p)
750: /* Loop. */ ;
751:
752: if ('.' == *p) {
753: *p++ = '\0';
1.51 schwarze 754: mlink->fsec = p;
1.47 schwarze 755: }
1.41 deraadt 756:
1.47 schwarze 757: /*
758: * Now try to parse the name.
759: * Use the filename portion of the path.
760: */
761: mlink->name = start;
762: if (NULL != (p = strrchr(start, '/'))) {
763: mlink->name = p + 1;
764: *p = '\0';
765: }
766: mlink_add(mlink, &st);
767: }
1.2 schwarze 768:
1.47 schwarze 769: static void
770: mlink_add(struct mlink *mlink, const struct stat *st)
771: {
772: struct inodev inodev;
773: struct mpage *mpage;
774: unsigned int slot;
775:
776: assert(NULL != mlink->file);
777:
1.51 schwarze 778: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
779: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
780: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
781: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 782:
783: if ('0' == *mlink->fsec) {
784: free(mlink->fsec);
785: mlink->fsec = mandoc_strdup(mlink->dsec);
786: mlink->fform = FORM_CAT;
787: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
788: mlink->fform = FORM_SRC;
789: else
790: mlink->fform = FORM_NONE;
1.1 schwarze 791:
1.47 schwarze 792: slot = ohash_qlookup(&mlinks, mlink->file);
793: assert(NULL == ohash_find(&mlinks, slot));
794: ohash_insert(&mlinks, slot, mlink);
795:
796: inodev.st_ino = st->st_ino;
797: inodev.st_dev = st->st_dev;
798: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
799: sizeof(struct inodev), inodev.st_ino);
800: mpage = ohash_find(&mpages, slot);
801: if (NULL == mpage) {
802: mpage = mandoc_calloc(1, sizeof(struct mpage));
803: mpage->inodev.st_ino = inodev.st_ino;
804: mpage->inodev.st_dev = inodev.st_dev;
805: ohash_insert(&mpages, slot, mpage);
806: } else
807: mlink->next = mpage->mlinks;
808: mpage->mlinks = mlink;
809: }
1.1 schwarze 810:
1.47 schwarze 811: static void
812: mlink_free(struct mlink *mlink)
813: {
1.1 schwarze 814:
1.47 schwarze 815: free(mlink->dsec);
816: free(mlink->arch);
817: free(mlink->name);
818: free(mlink->fsec);
819: free(mlink);
820: }
1.1 schwarze 821:
1.47 schwarze 822: static void
823: mpages_free(void)
824: {
825: struct mpage *mpage;
826: struct mlink *mlink;
827: unsigned int slot;
828:
829: mpage = ohash_first(&mpages, &slot);
830: while (NULL != mpage) {
831: while (NULL != (mlink = mpage->mlinks)) {
832: mpage->mlinks = mlink->next;
833: mlink_free(mlink);
834: }
835: free(mpage->sec);
836: free(mpage->arch);
837: free(mpage->title);
838: free(mpage->desc);
839: free(mpage);
840: mpage = ohash_next(&mpages, &slot);
841: }
842: }
1.1 schwarze 843:
1.47 schwarze 844: /*
845: * For each mlink to the mpage, check whether the path looks like
846: * it is formatted, and if it does, check whether a source manual
847: * exists by the same name, ignoring the suffix.
848: * If both conditions hold, drop the mlink.
849: */
850: static void
851: mlinks_undupe(struct mpage *mpage)
852: {
853: char buf[PATH_MAX];
854: struct mlink **prev;
855: struct mlink *mlink;
856: char *bufp;
857:
858: mpage->form = FORM_CAT;
859: prev = &mpage->mlinks;
860: while (NULL != (mlink = *prev)) {
861: if (FORM_CAT != mlink->dform) {
862: mpage->form = FORM_NONE;
863: goto nextlink;
864: }
865: if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
866: if (warnings)
867: say(mlink->file, "Filename too long");
868: goto nextlink;
1.26 schwarze 869: }
1.47 schwarze 870: bufp = strstr(buf, "cat");
871: assert(NULL != bufp);
872: memcpy(bufp, "man", 3);
873: if (NULL != (bufp = strrchr(buf, '.')))
874: *++bufp = '\0';
875: strlcat(buf, mlink->dsec, PATH_MAX);
876: if (NULL == ohash_find(&mlinks,
877: ohash_qlookup(&mlinks, buf)))
878: goto nextlink;
879: if (warnings)
880: say(mlink->file, "Man source exists: %s", buf);
881: if (use_all)
882: goto nextlink;
883: *prev = mlink->next;
884: mlink_free(mlink);
885: continue;
886: nextlink:
887: prev = &(*prev)->next;
1.1 schwarze 888: }
1.47 schwarze 889: }
1.1 schwarze 890:
1.50 schwarze 891: static int
892: mlink_check(struct mpage *mpage, struct mlink *mlink)
893: {
894: int match;
895:
896: match = 1;
897:
898: /*
899: * Check whether the manual section given in a file
900: * agrees with the directory where the file is located.
901: * Some manuals have suffixes like (3p) on their
902: * section number either inside the file or in the
903: * directory name, some are linked into more than one
904: * section, like encrypt(1) = makekey(8).
905: */
906:
907: if (FORM_SRC == mpage->form &&
908: strcasecmp(mpage->sec, mlink->dsec)) {
909: match = 0;
910: say(mlink->file, "Section \"%s\" manual in %s directory",
911: mpage->sec, mlink->dsec);
912: }
913:
914: /*
915: * Manual page directories exist for each kernel
916: * architecture as returned by machine(1).
917: * However, many manuals only depend on the
918: * application architecture as returned by arch(1).
919: * For example, some (2/ARM) manuals are shared
920: * across the "armish" and "zaurus" kernel
921: * architectures.
922: * A few manuals are even shared across completely
923: * different architectures, for example fdformat(1)
924: * on amd64, i386, sparc, and sparc64.
925: */
926:
927: if (strcasecmp(mpage->arch, mlink->arch)) {
928: match = 0;
929: say(mlink->file, "Architecture \"%s\" manual in "
930: "\"%s\" directory", mpage->arch, mlink->arch);
931: }
932:
933: if (strcasecmp(mpage->title, mlink->name))
934: match = 0;
935:
936: return(match);
937: }
938:
1.47 schwarze 939: /*
940: * Run through the files in the global vector "mpages"
941: * and add them to the database specified in "basedir".
942: *
943: * This handles the parsing scheme itself, using the cues of directory
944: * and filename to determine whether the file is parsable or not.
945: */
946: static void
1.58 schwarze 947: mpages_merge(struct mchars *mc, struct mparse *mp)
1.47 schwarze 948: {
1.58 schwarze 949: struct ohash_info str_info;
1.47 schwarze 950: struct mpage *mpage;
1.50 schwarze 951: struct mlink *mlink;
1.47 schwarze 952: struct mdoc *mdoc;
953: struct man *man;
954: const char *cp;
955: int match;
1.58 schwarze 956: unsigned int pslot;
1.47 schwarze 957: enum mandoclevel lvl;
958:
959: str_info.alloc = hash_alloc;
960: str_info.halloc = hash_halloc;
961: str_info.hfree = hash_free;
962: str_info.key_offset = offsetof(struct str, key);
963:
964: mpage = ohash_first(&mpages, &pslot);
965: while (NULL != mpage) {
966: mlinks_undupe(mpage);
967: if (NULL == mpage->mlinks) {
968: mpage = ohash_next(&mpages, &pslot);
969: continue;
970: }
1.1 schwarze 971:
1.47 schwarze 972: ohash_init(&strings, 6, &str_info);
973: mparse_reset(mp);
974: mdoc = NULL;
975: man = NULL;
1.11 schwarze 976:
977: /*
1.24 schwarze 978: * Try interpreting the file as mdoc(7) or man(7)
979: * source code, unless it is already known to be
980: * formatted. Fall back to formatted mode.
1.11 schwarze 981: */
1.47 schwarze 982: if (FORM_CAT != mpage->mlinks->dform ||
983: FORM_CAT != mpage->mlinks->fform) {
984: lvl = mparse_readfd(mp, -1, mpage->mlinks->file);
985: if (lvl < MANDOCLEVEL_FATAL)
986: mparse_result(mp, &mdoc, &man);
987: }
1.11 schwarze 988:
989: if (NULL != mdoc) {
1.47 schwarze 990: mpage->form = FORM_SRC;
991: mpage->sec =
992: mandoc_strdup(mdoc_meta(mdoc)->msec);
993: mpage->arch = mdoc_meta(mdoc)->arch;
994: mpage->arch = mandoc_strdup(
995: NULL == mpage->arch ? "" : mpage->arch);
996: mpage->title =
997: mandoc_strdup(mdoc_meta(mdoc)->title);
1.11 schwarze 998: } else if (NULL != man) {
1.47 schwarze 999: mpage->form = FORM_SRC;
1000: mpage->sec =
1001: mandoc_strdup(man_meta(man)->msec);
1002: mpage->arch =
1003: mandoc_strdup(mpage->mlinks->arch);
1004: mpage->title =
1005: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1006: } else {
1.47 schwarze 1007: mpage->form = FORM_CAT;
1008: mpage->sec =
1009: mandoc_strdup(mpage->mlinks->dsec);
1010: mpage->arch =
1011: mandoc_strdup(mpage->mlinks->arch);
1012: mpage->title =
1013: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1014: }
1.54 schwarze 1015: putkey(mpage, mpage->sec, TYPE_sec);
1.55 schwarze 1016: putkey(mpage, '\0' == *mpage->arch ?
1017: "any" : mpage->arch, TYPE_arch);
1.1 schwarze 1018:
1.54 schwarze 1019: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1020: if ('\0' != *mlink->dsec)
1021: putkey(mpage, mlink->dsec, TYPE_sec);
1022: if ('\0' != *mlink->fsec)
1023: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1024: putkey(mpage, '\0' == *mlink->arch ?
1025: "any" : mlink->arch, TYPE_arch);
1.50 schwarze 1026: putkey(mpage, mlink->name, TYPE_Nm);
1.54 schwarze 1027: }
1.41 deraadt 1028:
1.50 schwarze 1029: if (warnings && !use_all) {
1.47 schwarze 1030: match = 0;
1.50 schwarze 1031: for (mlink = mpage->mlinks; mlink;
1032: mlink = mlink->next)
1033: if (mlink_check(mpage, mlink))
1034: match = 1;
1035: } else
1036: match = 1;
1.6 schwarze 1037:
1.47 schwarze 1038: if (NULL != mdoc) {
1039: if (NULL != (cp = mdoc_meta(mdoc)->name))
1040: putkey(mpage, cp, TYPE_Nm);
1041: assert(NULL == mpage->desc);
1042: parse_mdoc(mpage, mdoc_node(mdoc));
1043: putkey(mpage, NULL != mpage->desc ?
1044: mpage->desc : mpage->mlinks->name, TYPE_Nd);
1045: } else if (NULL != man)
1046: parse_man(mpage, man_node(man));
1047: else
1048: parse_cat(mpage);
1.6 schwarze 1049:
1.62 schwarze 1050: dbadd(mpage, mc);
1.47 schwarze 1051: ohash_delete(&strings);
1052: mpage = ohash_next(&mpages, &pslot);
1053: }
1054: }
1.6 schwarze 1055:
1.47 schwarze 1056: static void
1057: parse_cat(struct mpage *mpage)
1058: {
1059: FILE *stream;
1060: char *line, *p, *title;
1061: size_t len, plen, titlesz;
1.1 schwarze 1062:
1.47 schwarze 1063: if (NULL == (stream = fopen(mpage->mlinks->file, "r"))) {
1064: if (warnings)
1065: say(mpage->mlinks->file, NULL);
1066: return;
1067: }
1.1 schwarze 1068:
1.47 schwarze 1069: /* Skip to first blank line. */
1.1 schwarze 1070:
1.47 schwarze 1071: while (NULL != (line = fgetln(stream, &len)))
1072: if ('\n' == *line)
1073: break;
1.1 schwarze 1074:
1.47 schwarze 1075: /*
1076: * Assume the first line that is not indented
1077: * is the first section header. Skip to it.
1078: */
1.1 schwarze 1079:
1.47 schwarze 1080: while (NULL != (line = fgetln(stream, &len)))
1081: if ('\n' != *line && ' ' != *line)
1082: break;
1083:
1084: /*
1085: * Read up until the next section into a buffer.
1086: * Strip the leading and trailing newline from each read line,
1087: * appending a trailing space.
1088: * Ignore empty (whitespace-only) lines.
1089: */
1.28 schwarze 1090:
1.47 schwarze 1091: titlesz = 0;
1092: title = NULL;
1.38 schwarze 1093:
1.47 schwarze 1094: while (NULL != (line = fgetln(stream, &len))) {
1095: if (' ' != *line || '\n' != line[len - 1])
1096: break;
1097: while (len > 0 && isspace((unsigned char)*line)) {
1098: line++;
1099: len--;
1100: }
1101: if (1 == len)
1102: continue;
1103: title = mandoc_realloc(title, titlesz + len);
1104: memcpy(title + titlesz, line, len);
1105: titlesz += len;
1106: title[titlesz - 1] = ' ';
1107: }
1.28 schwarze 1108:
1.47 schwarze 1109: /*
1110: * If no page content can be found, or the input line
1111: * is already the next section header, or there is no
1112: * trailing newline, reuse the page title as the page
1113: * description.
1114: */
1.1 schwarze 1115:
1.47 schwarze 1116: if (NULL == title || '\0' == *title) {
1117: if (warnings)
1118: say(mpage->mlinks->file,
1119: "Cannot find NAME section");
1120: assert(NULL == mpage->desc);
1121: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1122: putkey(mpage, mpage->mlinks->name, TYPE_Nd);
1123: fclose(stream);
1124: free(title);
1125: return;
1126: }
1.24 schwarze 1127:
1.47 schwarze 1128: title = mandoc_realloc(title, titlesz + 1);
1129: title[titlesz] = '\0';
1.24 schwarze 1130:
1.47 schwarze 1131: /*
1132: * Skip to the first dash.
1133: * Use the remaining line as the description (no more than 70
1134: * bytes).
1135: */
1.28 schwarze 1136:
1.47 schwarze 1137: if (NULL != (p = strstr(title, "- "))) {
1138: for (p += 2; ' ' == *p || '\b' == *p; p++)
1139: /* Skip to next word. */ ;
1140: } else {
1141: if (warnings)
1142: say(mpage->mlinks->file,
1143: "No dash in title line");
1144: p = title;
1145: }
1.1 schwarze 1146:
1.47 schwarze 1147: plen = strlen(p);
1.1 schwarze 1148:
1.47 schwarze 1149: /* Strip backspace-encoding from line. */
1.1 schwarze 1150:
1.47 schwarze 1151: while (NULL != (line = memchr(p, '\b', plen))) {
1152: len = line - p;
1153: if (0 == len) {
1154: memmove(line, line + 1, plen--);
1155: continue;
1156: }
1157: memmove(line - 1, line + 1, plen - len);
1158: plen -= 2;
1159: }
1.1 schwarze 1160:
1.47 schwarze 1161: assert(NULL == mpage->desc);
1162: mpage->desc = mandoc_strdup(p);
1163: putkey(mpage, mpage->desc, TYPE_Nd);
1164: fclose(stream);
1165: free(title);
1166: }
1.16 schwarze 1167:
1.47 schwarze 1168: /*
1169: * Put a type/word pair into the word database for this particular file.
1170: */
1171: static void
1172: putkey(const struct mpage *mpage, const char *value, uint64_t type)
1173: {
1.37 schwarze 1174:
1.47 schwarze 1175: assert(NULL != value);
1176: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1177: }
1178:
1179: /*
1.47 schwarze 1180: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1181: */
1182: static void
1.47 schwarze 1183: putmdockey(const struct mpage *mpage,
1184: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1185: {
1.16 schwarze 1186:
1.47 schwarze 1187: for ( ; NULL != n; n = n->next) {
1188: if (NULL != n->child)
1189: putmdockey(mpage, n->child, m);
1190: if (MDOC_TEXT == n->type)
1191: putkey(mpage, n->string, m);
1192: }
1193: }
1.16 schwarze 1194:
1.47 schwarze 1195: static void
1196: parse_man(struct mpage *mpage, const struct man_node *n)
1197: {
1198: const struct man_node *head, *body;
1199: char *start, *sv, *title;
1200: char byte;
1201: size_t sz, titlesz;
1.16 schwarze 1202:
1.47 schwarze 1203: if (NULL == n)
1204: return;
1.16 schwarze 1205:
1.47 schwarze 1206: /*
1207: * We're only searching for one thing: the first text child in
1208: * the BODY of a NAME section. Since we don't keep track of
1209: * sections in -man, run some hoops to find out whether we're in
1210: * the correct section or not.
1211: */
1.16 schwarze 1212:
1.47 schwarze 1213: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1214: body = n;
1215: assert(body->parent);
1216: if (NULL != (head = body->parent->head) &&
1217: 1 == head->nchild &&
1218: NULL != (head = (head->child)) &&
1219: MAN_TEXT == head->type &&
1220: 0 == strcmp(head->string, "NAME") &&
1221: NULL != (body = body->child) &&
1222: MAN_TEXT == body->type) {
1.2 schwarze 1223:
1.47 schwarze 1224: title = NULL;
1225: titlesz = 0;
1.2 schwarze 1226:
1.47 schwarze 1227: /*
1228: * Suck the entire NAME section into memory.
1229: * Yes, we might run away.
1230: * But too many manuals have big, spread-out
1231: * NAME sections over many lines.
1232: */
1.2 schwarze 1233:
1.47 schwarze 1234: for ( ; NULL != body; body = body->next) {
1235: if (MAN_TEXT != body->type)
1236: break;
1237: if (0 == (sz = strlen(body->string)))
1238: continue;
1239: title = mandoc_realloc
1240: (title, titlesz + sz + 1);
1241: memcpy(title + titlesz, body->string, sz);
1242: titlesz += sz + 1;
1243: title[titlesz - 1] = ' ';
1244: }
1245: if (NULL == title)
1246: return;
1.16 schwarze 1247:
1.47 schwarze 1248: title = mandoc_realloc(title, titlesz + 1);
1249: title[titlesz] = '\0';
1.16 schwarze 1250:
1.47 schwarze 1251: /* Skip leading space. */
1.16 schwarze 1252:
1.47 schwarze 1253: sv = title;
1254: while (isspace((unsigned char)*sv))
1255: sv++;
1.16 schwarze 1256:
1.47 schwarze 1257: if (0 == (sz = strlen(sv))) {
1258: free(title);
1259: return;
1260: }
1.1 schwarze 1261:
1.47 schwarze 1262: /* Erase trailing space. */
1.1 schwarze 1263:
1.47 schwarze 1264: start = &sv[sz - 1];
1265: while (start > sv && isspace((unsigned char)*start))
1266: *start-- = '\0';
1.1 schwarze 1267:
1.47 schwarze 1268: if (start == sv) {
1269: free(title);
1270: return;
1271: }
1.1 schwarze 1272:
1.47 schwarze 1273: start = sv;
1.16 schwarze 1274:
1.47 schwarze 1275: /*
1276: * Go through a special heuristic dance here.
1277: * Conventionally, one or more manual names are
1278: * comma-specified prior to a whitespace, then a
1279: * dash, then a description. Try to puzzle out
1280: * the name parts here.
1281: */
1.16 schwarze 1282:
1.47 schwarze 1283: for ( ;; ) {
1284: sz = strcspn(start, " ,");
1285: if ('\0' == start[sz])
1286: break;
1.1 schwarze 1287:
1.47 schwarze 1288: byte = start[sz];
1289: start[sz] = '\0';
1.1 schwarze 1290:
1.47 schwarze 1291: putkey(mpage, start, TYPE_Nm);
1.1 schwarze 1292:
1.47 schwarze 1293: if (' ' == byte) {
1294: start += sz + 1;
1295: break;
1296: }
1.1 schwarze 1297:
1.47 schwarze 1298: assert(',' == byte);
1299: start += sz + 1;
1300: while (' ' == *start)
1301: start++;
1302: }
1.1 schwarze 1303:
1.47 schwarze 1304: if (sv == start) {
1305: putkey(mpage, start, TYPE_Nm);
1306: free(title);
1307: return;
1308: }
1.1 schwarze 1309:
1.47 schwarze 1310: while (isspace((unsigned char)*start))
1311: start++;
1.1 schwarze 1312:
1.47 schwarze 1313: if (0 == strncmp(start, "-", 1))
1314: start += 1;
1315: else if (0 == strncmp(start, "\\-\\-", 4))
1316: start += 4;
1317: else if (0 == strncmp(start, "\\-", 2))
1318: start += 2;
1319: else if (0 == strncmp(start, "\\(en", 4))
1320: start += 4;
1321: else if (0 == strncmp(start, "\\(em", 4))
1322: start += 4;
1.1 schwarze 1323:
1.47 schwarze 1324: while (' ' == *start)
1325: start++;
1.1 schwarze 1326:
1.47 schwarze 1327: assert(NULL == mpage->desc);
1328: mpage->desc = mandoc_strdup(start);
1329: putkey(mpage, mpage->desc, TYPE_Nd);
1330: free(title);
1331: return;
1332: }
1333: }
1.1 schwarze 1334:
1.47 schwarze 1335: for (n = n->child; n; n = n->next) {
1336: if (NULL != mpage->desc)
1337: break;
1338: parse_man(mpage, n);
1.1 schwarze 1339: }
1340: }
1341:
1342: static void
1.47 schwarze 1343: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1344: {
1345:
1.47 schwarze 1346: assert(NULL != n);
1347: for (n = n->child; NULL != n; n = n->next) {
1348: switch (n->type) {
1349: case (MDOC_ELEM):
1350: /* FALLTHROUGH */
1351: case (MDOC_BLOCK):
1352: /* FALLTHROUGH */
1353: case (MDOC_HEAD):
1354: /* FALLTHROUGH */
1355: case (MDOC_BODY):
1356: /* FALLTHROUGH */
1357: case (MDOC_TAIL):
1358: if (NULL != mdocs[n->tok].fp)
1359: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1360: break;
1361: if (mdocs[n->tok].mask)
1362: putmdockey(mpage, n->child,
1363: mdocs[n->tok].mask);
1364: break;
1365: default:
1366: assert(MDOC_ROOT != n->type);
1367: continue;
1368: }
1369: if (NULL != n->child)
1370: parse_mdoc(mpage, n);
1.1 schwarze 1371: }
1372: }
1373:
1.19 schwarze 1374: static int
1.47 schwarze 1375: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1376: {
1377: const char *start, *end;
1378: size_t sz;
1.19 schwarze 1379:
1.47 schwarze 1380: if (SEC_SYNOPSIS != n->sec ||
1381: NULL == (n = n->child) ||
1382: MDOC_TEXT != n->type)
1.19 schwarze 1383: return(0);
1.1 schwarze 1384:
1385: /*
1386: * Only consider those `Fd' macro fields that begin with an
1387: * "inclusion" token (versus, e.g., #define).
1388: */
1.47 schwarze 1389:
1.1 schwarze 1390: if (strcmp("#include", n->string))
1.19 schwarze 1391: return(0);
1.1 schwarze 1392:
1393: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1394: return(0);
1.1 schwarze 1395:
1396: /*
1397: * Strip away the enclosing angle brackets and make sure we're
1398: * not zero-length.
1399: */
1400:
1401: start = n->string;
1402: if ('<' == *start || '"' == *start)
1403: start++;
1404:
1405: if (0 == (sz = strlen(start)))
1.19 schwarze 1406: return(0);
1.1 schwarze 1407:
1408: end = &start[(int)sz - 1];
1409: if ('>' == *end || '"' == *end)
1410: end--;
1411:
1.47 schwarze 1412: if (end > start)
1413: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1414: return(0);
1.1 schwarze 1415: }
1416:
1.19 schwarze 1417: static int
1.47 schwarze 1418: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1419: {
1420: const char *cp;
1421:
1.47 schwarze 1422: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.19 schwarze 1423: return(0);
1424:
1.47 schwarze 1425: /*
1426: * Parse: .Fn "struct type *name" "char *arg".
1427: * First strip away pointer symbol.
1428: * Then store the function name, then type.
1429: * Finally, store the arguments.
1430: */
1.1 schwarze 1431:
1.47 schwarze 1432: if (NULL == (cp = strrchr(n->string, ' ')))
1433: cp = n->string;
1.1 schwarze 1434:
1435: while ('*' == *cp)
1436: cp++;
1437:
1.47 schwarze 1438: putkey(mpage, cp, TYPE_Fn);
1.19 schwarze 1439:
1.47 schwarze 1440: if (n->string < cp)
1441: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.19 schwarze 1442:
1.47 schwarze 1443: for (n = n->next; NULL != n; n = n->next)
1444: if (MDOC_TEXT == n->type)
1445: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1446:
1447: return(0);
1.1 schwarze 1448: }
1449:
1.19 schwarze 1450: static int
1.47 schwarze 1451: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1452: {
1.47 schwarze 1453: char *cp;
1.1 schwarze 1454:
1455: if (NULL == (n = n->child))
1.19 schwarze 1456: return(0);
1.1 schwarze 1457:
1.47 schwarze 1458: if (NULL == n->next) {
1459: putkey(mpage, n->string, TYPE_Xr);
1460: return(0);
1461: }
1.1 schwarze 1462:
1.47 schwarze 1463: if (-1 == asprintf(&cp, "%s(%s)", n->string, n->next->string)) {
1464: perror(NULL);
1465: exit((int)MANDOCLEVEL_SYSERR);
1466: }
1467: putkey(mpage, cp, TYPE_Xr);
1468: free(cp);
1469: return(0);
1.1 schwarze 1470: }
1471:
1.19 schwarze 1472: static int
1.47 schwarze 1473: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1474: {
1.47 schwarze 1475: size_t sz;
1.1 schwarze 1476:
1477: if (MDOC_BODY != n->type)
1.19 schwarze 1478: return(0);
1.1 schwarze 1479:
1.47 schwarze 1480: /*
1481: * Special-case the `Nd' because we need to put the description
1482: * into the document table.
1483: */
1484:
1485: for (n = n->child; NULL != n; n = n->next) {
1486: if (MDOC_TEXT == n->type) {
1487: if (NULL != mpage->desc) {
1488: sz = strlen(mpage->desc) +
1489: strlen(n->string) + 2;
1490: mpage->desc = mandoc_realloc(
1491: mpage->desc, sz);
1492: strlcat(mpage->desc, " ", sz);
1493: strlcat(mpage->desc, n->string, sz);
1494: } else
1495: mpage->desc = mandoc_strdup(n->string);
1496: }
1497: if (NULL != n->child)
1498: parse_mdoc_Nd(mpage, n);
1499: }
1.19 schwarze 1500: return(1);
1.1 schwarze 1501: }
1502:
1.19 schwarze 1503: static int
1.47 schwarze 1504: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1505: {
1506:
1.49 schwarze 1507: return(SEC_NAME == n->sec ||
1508: (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type));
1.1 schwarze 1509: }
1510:
1.19 schwarze 1511: static int
1.47 schwarze 1512: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1513: {
1514:
1.19 schwarze 1515: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1516: }
1517:
1.47 schwarze 1518: static int
1519: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1520: {
1521:
1.47 schwarze 1522: return(MDOC_HEAD == n->type);
1523: }
1.1 schwarze 1524:
1.47 schwarze 1525: static int
1526: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1527: {
1.1 schwarze 1528:
1.47 schwarze 1529: return(MDOC_BODY == n->type);
1.1 schwarze 1530: }
1531:
1.47 schwarze 1532: /*
1533: * Add a string to the hash table for the current manual.
1534: * Each string has a bitmask telling which macros it belongs to.
1535: * When we finish the manual, we'll dump the table.
1536: */
1.1 schwarze 1537: static void
1.47 schwarze 1538: putkeys(const struct mpage *mpage,
1539: const char *cp, size_t sz, uint64_t v)
1.1 schwarze 1540: {
1.47 schwarze 1541: struct str *s;
1542: unsigned int slot;
1543: const char *end;
1.1 schwarze 1544:
1.47 schwarze 1545: if (0 == sz)
1546: return;
1547:
1548: end = cp + sz;
1549: slot = ohash_qlookupi(&strings, cp, &end);
1550: s = ohash_find(&strings, slot);
1.1 schwarze 1551:
1.47 schwarze 1552: if (NULL != s && mpage == s->mpage) {
1553: s->mask |= v;
1.1 schwarze 1554: return;
1.47 schwarze 1555: } else if (NULL == s) {
1556: s = mandoc_calloc(sizeof(struct str) + sz + 1, 1);
1557: memcpy(s->key, cp, sz);
1558: ohash_insert(&strings, slot, s);
1559: }
1560: s->mpage = mpage;
1561: s->mask = v;
1.1 schwarze 1562: }
1563:
1564: /*
1.47 schwarze 1565: * Take a Unicode codepoint and produce its UTF-8 encoding.
1566: * This isn't the best way to do this, but it works.
1567: * The magic numbers are from the UTF-8 packaging.
1568: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1569: */
1.47 schwarze 1570: static size_t
1571: utf8(unsigned int cp, char out[7])
1.1 schwarze 1572: {
1.47 schwarze 1573: size_t rc;
1.1 schwarze 1574:
1.47 schwarze 1575: rc = 0;
1576: if (cp <= 0x0000007F) {
1577: rc = 1;
1578: out[0] = (char)cp;
1579: } else if (cp <= 0x000007FF) {
1580: rc = 2;
1581: out[0] = (cp >> 6 & 31) | 192;
1582: out[1] = (cp & 63) | 128;
1583: } else if (cp <= 0x0000FFFF) {
1584: rc = 3;
1585: out[0] = (cp >> 12 & 15) | 224;
1586: out[1] = (cp >> 6 & 63) | 128;
1587: out[2] = (cp & 63) | 128;
1588: } else if (cp <= 0x001FFFFF) {
1589: rc = 4;
1590: out[0] = (cp >> 18 & 7) | 240;
1591: out[1] = (cp >> 12 & 63) | 128;
1592: out[2] = (cp >> 6 & 63) | 128;
1593: out[3] = (cp & 63) | 128;
1594: } else if (cp <= 0x03FFFFFF) {
1595: rc = 5;
1596: out[0] = (cp >> 24 & 3) | 248;
1597: out[1] = (cp >> 18 & 63) | 128;
1598: out[2] = (cp >> 12 & 63) | 128;
1599: out[3] = (cp >> 6 & 63) | 128;
1600: out[4] = (cp & 63) | 128;
1601: } else if (cp <= 0x7FFFFFFF) {
1602: rc = 6;
1603: out[0] = (cp >> 30 & 1) | 252;
1604: out[1] = (cp >> 24 & 63) | 128;
1605: out[2] = (cp >> 18 & 63) | 128;
1606: out[3] = (cp >> 12 & 63) | 128;
1607: out[4] = (cp >> 6 & 63) | 128;
1608: out[5] = (cp & 63) | 128;
1609: } else
1610: return(0);
1.19 schwarze 1611:
1.47 schwarze 1612: out[rc] = '\0';
1613: return(rc);
1.1 schwarze 1614: }
1615:
1.47 schwarze 1616: /*
1.53 schwarze 1617: * Store the rendered version of a key, or alias the pointer
1618: * if the key contains no escape sequences.
1.47 schwarze 1619: */
1620: static void
1.53 schwarze 1621: render_key(struct mchars *mc, struct str *key)
1.1 schwarze 1622: {
1.47 schwarze 1623: size_t sz, bsz, pos;
1624: char utfbuf[7], res[5];
1625: char *buf;
1626: const char *seq, *cpp, *val;
1627: int len, u;
1628: enum mandoc_esc esc;
1629:
1.53 schwarze 1630: assert(NULL == key->rendered);
1.47 schwarze 1631:
1632: res[0] = '\\';
1633: res[1] = '\t';
1634: res[2] = ASCII_NBRSP;
1635: res[3] = ASCII_HYPH;
1636: res[4] = '\0';
1.1 schwarze 1637:
1.47 schwarze 1638: val = key->key;
1639: bsz = strlen(val);
1.1 schwarze 1640:
1641: /*
1.47 schwarze 1642: * Pre-check: if we have no stop-characters, then set the
1643: * pointer as ourselvse and get out of here.
1.1 schwarze 1644: */
1.47 schwarze 1645: if (strcspn(val, res) == bsz) {
1.53 schwarze 1646: key->rendered = key->key;
1.47 schwarze 1647: return;
1648: }
1.1 schwarze 1649:
1.47 schwarze 1650: /* Pre-allocate by the length of the input */
1.39 schwarze 1651:
1.47 schwarze 1652: buf = mandoc_malloc(++bsz);
1653: pos = 0;
1.39 schwarze 1654:
1.47 schwarze 1655: while ('\0' != *val) {
1656: /*
1657: * Halt on the first escape sequence.
1658: * This also halts on the end of string, in which case
1659: * we just copy, fallthrough, and exit the loop.
1660: */
1661: if ((sz = strcspn(val, res)) > 0) {
1662: memcpy(&buf[pos], val, sz);
1663: pos += sz;
1664: val += sz;
1665: }
1.39 schwarze 1666:
1.47 schwarze 1667: if (ASCII_HYPH == *val) {
1668: buf[pos++] = '-';
1669: val++;
1670: continue;
1671: } else if ('\t' == *val || ASCII_NBRSP == *val) {
1672: buf[pos++] = ' ';
1673: val++;
1674: continue;
1675: } else if ('\\' != *val)
1676: break;
1.39 schwarze 1677:
1.47 schwarze 1678: /* Read past the slash. */
1.39 schwarze 1679:
1.47 schwarze 1680: val++;
1.39 schwarze 1681:
1.47 schwarze 1682: /*
1683: * Parse the escape sequence and see if it's a
1684: * predefined character or special character.
1685: */
1.52 schwarze 1686:
1.47 schwarze 1687: esc = mandoc_escape
1688: ((const char **)&val, &seq, &len);
1689: if (ESCAPE_ERROR == esc)
1690: break;
1691: if (ESCAPE_SPECIAL != esc)
1692: continue;
1.39 schwarze 1693:
1.47 schwarze 1694: /*
1.52 schwarze 1695: * Render the special character
1696: * as either UTF-8 or ASCII.
1.47 schwarze 1697: */
1.52 schwarze 1698:
1699: if (write_utf8) {
1700: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1701: continue;
1702: cpp = utfbuf;
1703: if (0 == (sz = utf8(u, utfbuf)))
1704: continue;
1705: sz = strlen(cpp);
1706: } else {
1707: cpp = mchars_spec2str(mc, seq, len, &sz);
1708: if (NULL == cpp)
1709: continue;
1710: if (ASCII_NBRSP == *cpp) {
1711: cpp = " ";
1712: sz = 1;
1713: }
1714: }
1.1 schwarze 1715:
1.47 schwarze 1716: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1717:
1.47 schwarze 1718: bsz += sz;
1719: buf = mandoc_realloc(buf, bsz);
1720: memcpy(&buf[pos], cpp, sz);
1721: pos += sz;
1.1 schwarze 1722: }
1723:
1.47 schwarze 1724: buf[pos] = '\0';
1.53 schwarze 1725: key->rendered = buf;
1.1 schwarze 1726: }
1727:
1.11 schwarze 1728: /*
1.47 schwarze 1729: * Flush the current page's terms (and their bits) into the database.
1730: * Wrap the entire set of additions in a transaction to make sqlite be a
1731: * little faster.
1.53 schwarze 1732: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 1733: */
1734: static void
1.62 schwarze 1735: dbadd(const struct mpage *mpage, struct mchars *mc)
1.11 schwarze 1736: {
1.47 schwarze 1737: struct mlink *mlink;
1738: struct str *key;
1739: int64_t recno;
1740: size_t i;
1741: unsigned int slot;
1742:
1743: if (verb)
1.62 schwarze 1744: say(mpage->mlinks->file, "Adding to database");
1.11 schwarze 1745:
1.47 schwarze 1746: if (nodb)
1.11 schwarze 1747: return;
1.47 schwarze 1748:
1749: SQL_EXEC("BEGIN TRANSACTION");
1.22 schwarze 1750:
1.47 schwarze 1751: i = 1;
1752: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
1753: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1754: recno = sqlite3_last_insert_rowid(db);
1755: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1756:
1757: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1758: i = 1;
1759: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1760: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1761: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1762: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, recno);
1763: SQL_STEP(stmts[STMT_INSERT_LINK]);
1764: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1765: }
1766:
1767: for (key = ohash_first(&strings, &slot); NULL != key;
1768: key = ohash_next(&strings, &slot)) {
1769: assert(key->mpage == mpage);
1.53 schwarze 1770: if (NULL == key->rendered)
1771: render_key(mc, key);
1.47 schwarze 1772: i = 1;
1773: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.53 schwarze 1774: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.47 schwarze 1775: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno);
1776: SQL_STEP(stmts[STMT_INSERT_KEY]);
1777: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.53 schwarze 1778: if (key->rendered != key->key)
1779: free(key->rendered);
1.47 schwarze 1780: free(key);
1.33 schwarze 1781: }
1782:
1.47 schwarze 1783: SQL_EXEC("END TRANSACTION");
1784: }
1.41 deraadt 1785:
1.47 schwarze 1786: static void
1787: dbprune(void)
1788: {
1789: struct mpage *mpage;
1790: struct mlink *mlink;
1791: size_t i;
1792: unsigned int slot;
1.11 schwarze 1793:
1.63 ! schwarze 1794: if (0 == nodb)
! 1795: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 1796:
1.63 ! schwarze 1797: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
! 1798: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 1799: mlink = mpage->mlinks;
1800: if (verb)
1.63 ! schwarze 1801: say(mlink->file, "Deleting from database");
! 1802: if (nodb)
! 1803: continue;
! 1804: for ( ; NULL != mlink; mlink = mlink->next) {
! 1805: i = 1;
! 1806: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
! 1807: i, mlink->dsec);
! 1808: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
! 1809: i, mlink->arch);
! 1810: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
! 1811: i, mlink->name);
! 1812: SQL_STEP(stmts[STMT_DELETE_PAGE]);
! 1813: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
! 1814: }
1.11 schwarze 1815: }
1.63 ! schwarze 1816:
! 1817: if (0 == nodb)
! 1818: SQL_EXEC("END TRANSACTION");
1.47 schwarze 1819: }
1.22 schwarze 1820:
1.47 schwarze 1821: /*
1822: * Close an existing database and its prepared statements.
1823: * If "real" is not set, rename the temporary file into the real one.
1824: */
1825: static void
1826: dbclose(int real)
1827: {
1828: size_t i;
1.11 schwarze 1829:
1.47 schwarze 1830: if (nodb)
1831: return;
1.11 schwarze 1832:
1.47 schwarze 1833: for (i = 0; i < STMT__MAX; i++) {
1834: sqlite3_finalize(stmts[i]);
1835: stmts[i] = NULL;
1.28 schwarze 1836: }
1.22 schwarze 1837:
1.47 schwarze 1838: sqlite3_close(db);
1839: db = NULL;
1.11 schwarze 1840:
1.47 schwarze 1841: if (real)
1842: return;
1.22 schwarze 1843:
1.47 schwarze 1844: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
1845: exitcode = (int)MANDOCLEVEL_SYSERR;
1846: say(MANDOC_DB, NULL);
1.22 schwarze 1847: }
1.11 schwarze 1848: }
1849:
1.47 schwarze 1850: /*
1851: * This is straightforward stuff.
1852: * Open a database connection to a "temporary" database, then open a set
1853: * of prepared statements we'll use over and over again.
1854: * If "real" is set, we use the existing database; if not, we truncate a
1855: * temporary one.
1856: * Must be matched by dbclose().
1857: */
1858: static int
1859: dbopen(int real)
1.2 schwarze 1860: {
1.47 schwarze 1861: const char *file, *sql;
1862: int rc, ofl;
1.6 schwarze 1863:
1.47 schwarze 1864: if (nodb)
1865: return(1);
1.6 schwarze 1866:
1.47 schwarze 1867: ofl = SQLITE_OPEN_READWRITE;
1868: if (0 == real) {
1869: file = MANDOC_DB "~";
1870: if (-1 == remove(file) && ENOENT != errno) {
1871: exitcode = (int)MANDOCLEVEL_SYSERR;
1872: say(file, NULL);
1873: return(0);
1.28 schwarze 1874: }
1.47 schwarze 1875: ofl |= SQLITE_OPEN_EXCLUSIVE;
1876: } else
1877: file = MANDOC_DB;
1.6 schwarze 1878:
1.47 schwarze 1879: rc = sqlite3_open_v2(file, &db, ofl, NULL);
1880: if (SQLITE_OK == rc)
1881: goto prepare_statements;
1882: if (SQLITE_CANTOPEN != rc) {
1883: exitcode = (int)MANDOCLEVEL_SYSERR;
1884: say(file, NULL);
1885: return(0);
1886: }
1.6 schwarze 1887:
1.47 schwarze 1888: sqlite3_close(db);
1889: db = NULL;
1.6 schwarze 1890:
1.47 schwarze 1891: if (SQLITE_OK != (rc = sqlite3_open(file, &db))) {
1892: exitcode = (int)MANDOCLEVEL_SYSERR;
1893: say(file, NULL);
1894: return(0);
1.2 schwarze 1895: }
1896:
1.47 schwarze 1897: sql = "CREATE TABLE \"mpages\" (\n"
1898: " \"form\" INTEGER NOT NULL,\n"
1899: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1900: ");\n"
1901: "\n"
1902: "CREATE TABLE \"mlinks\" (\n"
1903: " \"sec\" TEXT NOT NULL,\n"
1904: " \"arch\" TEXT NOT NULL,\n"
1905: " \"name\" TEXT NOT NULL,\n"
1906: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1907: "ON DELETE CASCADE,\n"
1908: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1909: ");\n"
1910: "\n"
1911: "CREATE TABLE \"keys\" (\n"
1912: " \"bits\" INTEGER NOT NULL,\n"
1913: " \"key\" TEXT NOT NULL,\n"
1914: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1915: "ON DELETE CASCADE,\n"
1916: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1917: ");\n"
1918: "\n"
1919: "CREATE INDEX \"key_index\" ON keys (key);\n";
1920:
1921: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1922: exitcode = (int)MANDOCLEVEL_SYSERR;
1923: say(file, "%s", sqlite3_errmsg(db));
1924: return(0);
1.2 schwarze 1925: }
1926:
1.47 schwarze 1927: prepare_statements:
1928: SQL_EXEC("PRAGMA foreign_keys = ON");
1.63 ! schwarze 1929: sql = "DELETE FROM mpages WHERE id IN "
! 1930: "(SELECT pageid FROM mlinks WHERE "
! 1931: "sec=? AND arch=? AND name=?)";
1.47 schwarze 1932: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
1933: sql = "INSERT INTO mpages "
1.60 schwarze 1934: "(form) VALUES (?)";
1.47 schwarze 1935: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
1936: sql = "INSERT INTO mlinks "
1.61 schwarze 1937: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 1938: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1939: sql = "INSERT INTO keys "
1940: "(bits,key,pageid) VALUES (?,?,?)";
1941: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 1942:
1.47 schwarze 1943: /*
1944: * When opening a new database, we can turn off
1945: * synchronous mode for much better performance.
1946: */
1.6 schwarze 1947:
1.47 schwarze 1948: if (real)
1949: SQL_EXEC("PRAGMA synchronous = OFF");
1.11 schwarze 1950:
1.47 schwarze 1951: return(1);
1952: }
1.6 schwarze 1953:
1.47 schwarze 1954: static void *
1955: hash_halloc(size_t sz, void *arg)
1956: {
1.6 schwarze 1957:
1.47 schwarze 1958: return(mandoc_calloc(sz, 1));
1959: }
1.2 schwarze 1960:
1.47 schwarze 1961: static void *
1962: hash_alloc(size_t sz, void *arg)
1963: {
1.28 schwarze 1964:
1.47 schwarze 1965: return(mandoc_malloc(sz));
1966: }
1.6 schwarze 1967:
1.47 schwarze 1968: static void
1969: hash_free(void *p, size_t sz, void *arg)
1970: {
1.26 schwarze 1971:
1.47 schwarze 1972: free(p);
1973: }
1.6 schwarze 1974:
1.47 schwarze 1975: static int
1976: set_basedir(const char *targetdir)
1977: {
1978: static char startdir[PATH_MAX];
1979: static int fd;
1.6 schwarze 1980:
1.47 schwarze 1981: /*
1982: * Remember where we started by keeping a fd open to the origin
1983: * path component: throughout this utility, we chdir() a lot to
1984: * handle relative paths, and by doing this, we can return to
1985: * the starting point.
1986: */
1987: if ('\0' == *startdir) {
1988: if (NULL == getcwd(startdir, PATH_MAX)) {
1989: exitcode = (int)MANDOCLEVEL_SYSERR;
1990: if (NULL != targetdir)
1991: say(".", NULL);
1992: return(0);
1993: }
1994: if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
1995: exitcode = (int)MANDOCLEVEL_SYSERR;
1996: say(startdir, NULL);
1997: return(0);
1.11 schwarze 1998: }
1.47 schwarze 1999: if (NULL == targetdir)
2000: targetdir = startdir;
2001: } else {
2002: if (-1 == fd)
2003: return(0);
2004: if (-1 == fchdir(fd)) {
2005: close(fd);
2006: basedir[0] = '\0';
2007: exitcode = (int)MANDOCLEVEL_SYSERR;
2008: say(startdir, NULL);
2009: return(0);
1.2 schwarze 2010: }
1.47 schwarze 2011: if (NULL == targetdir) {
2012: close(fd);
2013: return(1);
1.2 schwarze 2014: }
2015: }
1.47 schwarze 2016: if (NULL == realpath(targetdir, basedir)) {
2017: basedir[0] = '\0';
2018: exitcode = (int)MANDOCLEVEL_BADARG;
2019: say(targetdir, NULL);
2020: return(0);
2021: } else if (-1 == chdir(basedir)) {
2022: exitcode = (int)MANDOCLEVEL_BADARG;
2023: say("", NULL);
2024: return(0);
2025: }
2026: return(1);
1.2 schwarze 2027: }
2028:
2029: static void
1.47 schwarze 2030: say(const char *file, const char *format, ...)
1.2 schwarze 2031: {
1.47 schwarze 2032: va_list ap;
1.2 schwarze 2033:
1.47 schwarze 2034: if ('\0' != *basedir)
2035: fprintf(stderr, "%s", basedir);
2036: if ('\0' != *basedir && '\0' != *file)
2037: fputs("//", stderr);
2038: if ('\0' != *file)
2039: fprintf(stderr, "%s", file);
2040: fputs(": ", stderr);
1.31 schwarze 2041:
1.47 schwarze 2042: if (NULL == format) {
2043: perror(NULL);
2044: return;
1.2 schwarze 2045: }
1.47 schwarze 2046:
2047: va_start(ap, format);
2048: vfprintf(stderr, format, ap);
2049: va_end(ap);
2050:
2051: fputc('\n', stderr);
1.1 schwarze 2052: }