Annotation of mandoc/mandocdb.c, Revision 1.86
1.86 ! schwarze 1: /* $Id: mandocdb.c,v 1.85 2013/12/27 15:48:09 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.56 schwarze 4: * Copyright (c) 2011, 2012, 2013 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:
1.50 kristaps 22: #include <sys/stat.h>
1.1 kristaps 23:
24: #include <assert.h>
1.43 kristaps 25: #include <ctype.h>
1.63 schwarze 26: #include <errno.h>
1.1 kristaps 27: #include <fcntl.h>
1.50 kristaps 28: #include <fts.h>
1.1 kristaps 29: #include <getopt.h>
1.58 schwarze 30: #include <limits.h>
1.50 kristaps 31: #include <stddef.h>
1.59 schwarze 32: #include <stdio.h>
1.1 kristaps 33: #include <stdint.h>
34: #include <stdlib.h>
35: #include <string.h>
1.17 schwarze 36: #include <unistd.h>
1.1 kristaps 37:
1.53 kristaps 38: #ifdef HAVE_OHASH
1.50 kristaps 39: #include <ohash.h>
1.53 kristaps 40: #else
41: #include "compat_ohash.h"
42: #endif
1.50 kristaps 43: #include <sqlite3.h>
1.1 kristaps 44:
1.50 kristaps 45: #include "mdoc.h"
1.1 kristaps 46: #include "man.h"
47: #include "mandoc.h"
1.10 kristaps 48: #include "manpath.h"
1.55 kristaps 49: #include "mansearch.h"
1.1 kristaps 50:
1.52 kristaps 51: #define SQL_EXEC(_v) \
52: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
53: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
54: #define SQL_BIND_TEXT(_s, _i, _v) \
55: if (SQLITE_OK != sqlite3_bind_text \
56: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
57: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
58: #define SQL_BIND_INT(_s, _i, _v) \
59: if (SQLITE_OK != sqlite3_bind_int \
60: ((_s), (_i)++, (_v))) \
61: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
62: #define SQL_BIND_INT64(_s, _i, _v) \
63: if (SQLITE_OK != sqlite3_bind_int64 \
64: ((_s), (_i)++, (_v))) \
65: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
66: #define SQL_STEP(_s) \
67: if (SQLITE_DONE != sqlite3_step((_s))) \
68: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
69:
1.50 kristaps 70: enum op {
71: OP_DEFAULT = 0, /* new dbs from dir list or default config */
72: OP_CONFFILE, /* new databases from custom config file */
73: OP_UPDATE, /* delete/add entries in existing database */
74: OP_DELETE, /* delete entries from existing database */
75: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 76: };
77:
1.50 kristaps 78: enum form {
1.84 schwarze 79: FORM_NONE, /* format is unknown */
80: FORM_SRC, /* format is -man or -mdoc */
81: FORM_CAT /* format is cat */
1.50 kristaps 82: };
1.38 schwarze 83:
1.50 kristaps 84: struct str {
85: char *utf8; /* key in UTF-8 form */
1.78 schwarze 86: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 87: uint64_t mask; /* bitmask in sequence */
1.51 kristaps 88: char key[]; /* the string itself */
1.38 schwarze 89: };
90:
1.79 schwarze 91: struct inodev {
92: ino_t st_ino;
93: dev_t st_dev;
1.50 kristaps 94: };
1.5 kristaps 95:
1.78 schwarze 96: struct mpage {
1.79 schwarze 97: struct inodev inodev; /* used for hashing routine */
1.82 schwarze 98: enum form form; /* format from file content */
99: char *sec; /* section from file content */
100: char *arch; /* architecture from file content */
101: char *title; /* title from file content */
102: char *desc; /* description from file content */
103: struct mlink *mlinks; /* singly linked list */
104: };
105:
106: struct mlink {
1.58 schwarze 107: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 108: enum form dform; /* format from directory */
1.84 schwarze 109: enum form fform; /* format from file name suffix */
1.82 schwarze 110: char *dsec; /* section from directory */
111: char *arch; /* architecture from directory */
112: char *name; /* name from file name (not empty) */
113: char *fsec; /* section from file name suffix */
1.85 schwarze 114: struct mlink *next; /* singly linked list */
1.3 kristaps 115: };
116:
1.69 schwarze 117: struct title {
118: char *title; /* name(sec/arch) given inside the file */
119: char *file; /* file name in case of mismatch */
120: };
121:
1.50 kristaps 122: enum stmt {
1.81 schwarze 123: STMT_DELETE_PAGE = 0, /* delete mpage */
124: STMT_INSERT_PAGE, /* insert mpage */
125: STMT_INSERT_LINK, /* insert mlink */
126: STMT_INSERT_KEY, /* insert parsed key */
1.50 kristaps 127: STMT__MAX
1.1 kristaps 128: };
129:
1.78 schwarze 130: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
1.1 kristaps 131:
1.50 kristaps 132: struct mdoc_handler {
133: mdoc_fp fp; /* optional handler */
134: uint64_t mask; /* set unless handler returns 0 */
1.1 kristaps 135: };
136:
1.59 schwarze 137: static void dbclose(int);
1.82 schwarze 138: static void dbindex(const struct mpage *, struct mchars *);
1.59 schwarze 139: static int dbopen(int);
140: static void dbprune(void);
141: static void filescan(const char *);
1.50 kristaps 142: static void *hash_alloc(size_t, void *);
143: static void hash_free(void *, size_t, void *);
144: static void *hash_halloc(size_t, void *);
1.84 schwarze 145: static void mlink_add(struct mlink *, const struct stat *);
1.82 schwarze 146: static void mlink_free(struct mlink *);
1.78 schwarze 147: static void mpages_free(void);
148: static void mpages_merge(struct mchars *, struct mparse *, int);
149: static void parse_cat(struct mpage *);
150: static void parse_man(struct mpage *, const struct man_node *);
151: static void parse_mdoc(struct mpage *, const struct mdoc_node *);
152: static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
153: static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
154: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
155: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
156: static int parse_mdoc_In(struct mpage *, const struct mdoc_node *);
157: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
158: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
159: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
160: static int parse_mdoc_St(struct mpage *, const struct mdoc_node *);
161: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
162: static void putkey(const struct mpage *,
1.50 kristaps 163: const char *, uint64_t);
1.78 schwarze 164: static void putkeys(const struct mpage *,
1.64 schwarze 165: const char *, size_t, uint64_t);
1.78 schwarze 166: static void putmdockey(const struct mpage *,
1.50 kristaps 167: const struct mdoc_node *, uint64_t);
1.59 schwarze 168: static void say(const char *, const char *, ...);
1.80 schwarze 169: static int set_basedir(const char *);
1.59 schwarze 170: static int treescan(void);
1.50 kristaps 171: static size_t utf8(unsigned int, char [7]);
172: static void utf8key(struct mchars *, struct str *);
173:
174: static char *progname;
175: static int use_all; /* use all found files */
176: static int nodb; /* no database changes */
177: static int verb; /* print what we're doing */
178: static int warnings; /* warn about crap */
1.59 schwarze 179: static int exitcode; /* to be returned by main */
1.50 kristaps 180: static enum op op; /* operational mode */
1.59 schwarze 181: static char basedir[PATH_MAX]; /* current base directory */
1.78 schwarze 182: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 183: static struct ohash mlinks; /* table of directory entries */
1.50 kristaps 184: static struct ohash strings; /* table of all strings */
185: static sqlite3 *db = NULL; /* current database */
186: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.25 schwarze 187:
188: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 189: { NULL, 0 }, /* Ap */
190: { NULL, 0 }, /* Dd */
191: { NULL, 0 }, /* Dt */
192: { NULL, 0 }, /* Os */
193: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
194: { parse_mdoc_head, TYPE_Ss }, /* Ss */
195: { NULL, 0 }, /* Pp */
196: { NULL, 0 }, /* D1 */
197: { NULL, 0 }, /* Dl */
198: { NULL, 0 }, /* Bd */
199: { NULL, 0 }, /* Ed */
200: { NULL, 0 }, /* Bl */
201: { NULL, 0 }, /* El */
202: { NULL, 0 }, /* It */
203: { NULL, 0 }, /* Ad */
204: { NULL, TYPE_An }, /* An */
205: { NULL, TYPE_Ar }, /* Ar */
206: { NULL, TYPE_Cd }, /* Cd */
207: { NULL, TYPE_Cm }, /* Cm */
208: { NULL, TYPE_Dv }, /* Dv */
209: { NULL, TYPE_Er }, /* Er */
210: { NULL, TYPE_Ev }, /* Ev */
211: { NULL, 0 }, /* Ex */
212: { NULL, TYPE_Fa }, /* Fa */
213: { parse_mdoc_Fd, 0 }, /* Fd */
214: { NULL, TYPE_Fl }, /* Fl */
215: { parse_mdoc_Fn, 0 }, /* Fn */
216: { NULL, TYPE_Ft }, /* Ft */
217: { NULL, TYPE_Ic }, /* Ic */
218: { parse_mdoc_In, TYPE_In }, /* In */
219: { NULL, TYPE_Li }, /* Li */
220: { parse_mdoc_Nd, TYPE_Nd }, /* Nd */
221: { parse_mdoc_Nm, TYPE_Nm }, /* Nm */
222: { NULL, 0 }, /* Op */
223: { NULL, 0 }, /* Ot */
224: { NULL, TYPE_Pa }, /* Pa */
225: { NULL, 0 }, /* Rv */
226: { parse_mdoc_St, 0 }, /* St */
227: { NULL, TYPE_Va }, /* Va */
228: { parse_mdoc_body, TYPE_Va }, /* Vt */
229: { parse_mdoc_Xr, 0 }, /* Xr */
230: { NULL, 0 }, /* %A */
231: { NULL, 0 }, /* %B */
232: { NULL, 0 }, /* %D */
233: { NULL, 0 }, /* %I */
234: { NULL, 0 }, /* %J */
235: { NULL, 0 }, /* %N */
236: { NULL, 0 }, /* %O */
237: { NULL, 0 }, /* %P */
238: { NULL, 0 }, /* %R */
239: { NULL, 0 }, /* %T */
240: { NULL, 0 }, /* %V */
241: { NULL, 0 }, /* Ac */
242: { NULL, 0 }, /* Ao */
243: { NULL, 0 }, /* Aq */
244: { NULL, TYPE_At }, /* At */
245: { NULL, 0 }, /* Bc */
246: { NULL, 0 }, /* Bf */
247: { NULL, 0 }, /* Bo */
248: { NULL, 0 }, /* Bq */
249: { NULL, TYPE_Bsx }, /* Bsx */
250: { NULL, TYPE_Bx }, /* Bx */
251: { NULL, 0 }, /* Db */
252: { NULL, 0 }, /* Dc */
253: { NULL, 0 }, /* Do */
254: { NULL, 0 }, /* Dq */
255: { NULL, 0 }, /* Ec */
256: { NULL, 0 }, /* Ef */
257: { NULL, TYPE_Em }, /* Em */
258: { NULL, 0 }, /* Eo */
259: { NULL, TYPE_Fx }, /* Fx */
260: { NULL, TYPE_Ms }, /* Ms */
261: { NULL, 0 }, /* No */
262: { NULL, 0 }, /* Ns */
263: { NULL, TYPE_Nx }, /* Nx */
264: { NULL, TYPE_Ox }, /* Ox */
265: { NULL, 0 }, /* Pc */
266: { NULL, 0 }, /* Pf */
267: { NULL, 0 }, /* Po */
268: { NULL, 0 }, /* Pq */
269: { NULL, 0 }, /* Qc */
270: { NULL, 0 }, /* Ql */
271: { NULL, 0 }, /* Qo */
272: { NULL, 0 }, /* Qq */
273: { NULL, 0 }, /* Re */
274: { NULL, 0 }, /* Rs */
275: { NULL, 0 }, /* Sc */
276: { NULL, 0 }, /* So */
277: { NULL, 0 }, /* Sq */
278: { NULL, 0 }, /* Sm */
279: { NULL, 0 }, /* Sx */
280: { NULL, TYPE_Sy }, /* Sy */
281: { NULL, TYPE_Tn }, /* Tn */
282: { NULL, 0 }, /* Ux */
283: { NULL, 0 }, /* Xc */
284: { NULL, 0 }, /* Xo */
285: { parse_mdoc_head, 0 }, /* Fo */
286: { NULL, 0 }, /* Fc */
287: { NULL, 0 }, /* Oo */
288: { NULL, 0 }, /* Oc */
289: { NULL, 0 }, /* Bk */
290: { NULL, 0 }, /* Ek */
291: { NULL, 0 }, /* Bt */
292: { NULL, 0 }, /* Hf */
293: { NULL, 0 }, /* Fr */
294: { NULL, 0 }, /* Ud */
295: { NULL, TYPE_Lb }, /* Lb */
296: { NULL, 0 }, /* Lp */
297: { NULL, TYPE_Lk }, /* Lk */
298: { NULL, TYPE_Mt }, /* Mt */
299: { NULL, 0 }, /* Brq */
300: { NULL, 0 }, /* Bro */
301: { NULL, 0 }, /* Brc */
302: { NULL, 0 }, /* %C */
303: { NULL, 0 }, /* Es */
304: { NULL, 0 }, /* En */
305: { NULL, TYPE_Dx }, /* Dx */
306: { NULL, 0 }, /* %Q */
307: { NULL, 0 }, /* br */
308: { NULL, 0 }, /* sp */
309: { NULL, 0 }, /* %U */
310: { NULL, 0 }, /* Ta */
1.1 kristaps 311: };
312:
313: int
314: main(int argc, char *argv[])
315: {
1.59 schwarze 316: int ch, i;
1.50 kristaps 317: size_t j, sz;
1.59 schwarze 318: const char *path_arg;
1.50 kristaps 319: struct mchars *mc;
320: struct manpaths dirs;
321: struct mparse *mp;
1.83 schwarze 322: struct ohash_info mpages_info, mlinks_info;
1.50 kristaps 323:
324: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
325: memset(&dirs, 0, sizeof(struct manpaths));
326:
1.83 schwarze 327: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
328: mpages_info.halloc = mlinks_info.halloc = hash_halloc;
329: mpages_info.hfree = mlinks_info.hfree = hash_free;
1.50 kristaps 330:
1.79 schwarze 331: mpages_info.key_offset = offsetof(struct mpage, inodev);
1.83 schwarze 332: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 kristaps 333:
334: progname = strrchr(argv[0], '/');
335: if (progname == NULL)
336: progname = argv[0];
337: else
338: ++progname;
339:
1.50 kristaps 340: /*
341: * We accept a few different invocations.
342: * The CHECKOP macro makes sure that invocation styles don't
343: * clobber each other.
344: */
345: #define CHECKOP(_op, _ch) do \
346: if (OP_DEFAULT != (_op)) { \
347: fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
348: goto usage; \
349: } while (/*CONSTCOND*/0)
1.10 kristaps 350:
1.59 schwarze 351: path_arg = NULL;
1.38 schwarze 352: op = OP_DEFAULT;
1.1 kristaps 353:
1.50 kristaps 354: while (-1 != (ch = getopt(argc, argv, "aC:d:ntu:vW")))
1.1 kristaps 355: switch (ch) {
1.12 schwarze 356: case ('a'):
357: use_all = 1;
358: break;
1.34 schwarze 359: case ('C'):
1.50 kristaps 360: CHECKOP(op, ch);
1.59 schwarze 361: path_arg = optarg;
1.38 schwarze 362: op = OP_CONFFILE;
1.34 schwarze 363: break;
1.5 kristaps 364: case ('d'):
1.50 kristaps 365: CHECKOP(op, ch);
1.59 schwarze 366: path_arg = optarg;
1.5 kristaps 367: op = OP_UPDATE;
368: break;
1.50 kristaps 369: case ('n'):
370: nodb = 1;
371: break;
1.38 schwarze 372: case ('t'):
1.50 kristaps 373: CHECKOP(op, ch);
1.38 schwarze 374: dup2(STDOUT_FILENO, STDERR_FILENO);
375: op = OP_TEST;
1.50 kristaps 376: nodb = warnings = 1;
1.38 schwarze 377: break;
1.5 kristaps 378: case ('u'):
1.50 kristaps 379: CHECKOP(op, ch);
1.59 schwarze 380: path_arg = optarg;
1.5 kristaps 381: op = OP_DELETE;
382: break;
383: case ('v'):
384: verb++;
385: break;
1.38 schwarze 386: case ('W'):
387: warnings = 1;
388: break;
1.1 kristaps 389: default:
1.38 schwarze 390: goto usage;
1.1 kristaps 391: }
392:
393: argc -= optind;
394: argv += optind;
395:
1.38 schwarze 396: if (OP_CONFFILE == op && argc > 0) {
1.50 kristaps 397: fprintf(stderr, "-C: Too many arguments\n");
1.38 schwarze 398: goto usage;
399: }
400:
1.59 schwarze 401: exitcode = (int)MANDOCLEVEL_OK;
1.50 kristaps 402: mp = mparse_alloc(MPARSE_AUTO,
403: MANDOCLEVEL_FATAL, NULL, NULL, NULL);
404: mc = mchars_alloc();
405:
1.78 schwarze 406: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 407: ohash_init(&mlinks, 6, &mlinks_info);
1.50 kristaps 408:
409: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
410: /*
411: * Force processing all files.
412: */
413: use_all = 1;
1.59 schwarze 414:
1.50 kristaps 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: */
1.59 schwarze 420: if (0 == set_basedir(path_arg))
1.50 kristaps 421: goto out;
422: for (i = 0; i < argc; i++)
1.59 schwarze 423: filescan(argv[i]);
424: if (0 == dbopen(1))
1.50 kristaps 425: goto out;
426: if (OP_TEST != op)
1.59 schwarze 427: dbprune();
1.50 kristaps 428: if (OP_DELETE != op)
1.78 schwarze 429: mpages_merge(mc, mp, 0);
1.59 schwarze 430: dbclose(1);
1.50 kristaps 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
1.59 schwarze 444: manpath_parse(&dirs, path_arg, NULL, NULL);
1.50 kristaps 445:
446: /*
1.69 schwarze 447: * First scan the tree rooted at a base directory, then
448: * build a new database and finally move it into place.
1.50 kristaps 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;
1.66 schwarze 458:
459: if (j) {
1.78 schwarze 460: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 461: ohash_init(&mlinks, 6, &mlinks_info);
1.66 schwarze 462: }
463:
1.59 schwarze 464: if (0 == set_basedir(dirs.paths[j]))
1.50 kristaps 465: goto out;
1.59 schwarze 466: if (0 == treescan())
1.50 kristaps 467: goto out;
1.59 schwarze 468: if (0 == set_basedir(dirs.paths[j]))
1.50 kristaps 469: goto out;
1.59 schwarze 470: if (0 == dbopen(0))
1.50 kristaps 471: goto out;
1.52 kristaps 472:
1.78 schwarze 473: mpages_merge(mc, mp, warnings && !use_all);
1.59 schwarze 474: dbclose(0);
1.66 schwarze 475:
476: if (j + 1 < dirs.sz) {
1.80 schwarze 477: mpages_free();
1.78 schwarze 478: ohash_delete(&mpages);
1.83 schwarze 479: ohash_delete(&mlinks);
1.66 schwarze 480: }
1.50 kristaps 481: }
482: }
483: out:
1.59 schwarze 484: set_basedir(NULL);
1.50 kristaps 485: manpath_free(&dirs);
486: mchars_free(mc);
487: mparse_free(mp);
1.80 schwarze 488: mpages_free();
1.78 schwarze 489: ohash_delete(&mpages);
1.83 schwarze 490: ohash_delete(&mlinks);
1.59 schwarze 491: return(exitcode);
1.50 kristaps 492: usage:
493: fprintf(stderr, "usage: %s [-anvW] [-C file]\n"
494: " %s [-anvW] dir ...\n"
495: " %s [-nvW] -d dir [file ...]\n"
496: " %s [-nvW] -u dir [file ...]\n"
497: " %s -t file ...\n",
498: progname, progname, progname,
499: progname, progname);
500:
1.59 schwarze 501: return((int)MANDOCLEVEL_BADARG);
1.50 kristaps 502: }
503:
504: /*
1.59 schwarze 505: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 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
1.59 schwarze 519: treescan(void)
1.50 kristaps 520: {
521: FTS *f;
522: FTSENT *ff;
1.84 schwarze 523: struct mlink *mlink;
1.50 kristaps 524: int dform;
1.84 schwarze 525: char *fsec;
1.66 schwarze 526: const char *dsec, *arch, *cp, *path;
1.50 kristaps 527: const char *argv[2];
528:
529: argv[0] = ".";
530: argv[1] = (char *)NULL;
531:
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) {
1.59 schwarze 538: exitcode = (int)MANDOCLEVEL_SYSERR;
539: say("", NULL);
1.50 kristaps 540: return(0);
541: }
542:
543: dsec = arch = NULL;
544: dform = FORM_NONE;
545:
546: while (NULL != (ff = fts_read(f))) {
547: path = ff->fts_path + 2;
548: /*
1.83 schwarze 549: * If we're a regular file, add an mlink by using the
1.50 kristaps 550: * stored directory data and handling the filename.
551: * Disallow duplicate (hard-linked) files.
552: */
553: if (FTS_F == ff->fts_info) {
1.60 schwarze 554: if (0 == strcmp(path, MANDOC_DB))
555: continue;
1.50 kristaps 556: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 557: if (warnings)
1.59 schwarze 558: say(path, "Extraneous file");
1.50 kristaps 559: continue;
1.84 schwarze 560: } else if (NULL == (fsec =
1.60 schwarze 561: strrchr(ff->fts_name, '.'))) {
562: if ( ! use_all) {
1.56 schwarze 563: if (warnings)
1.60 schwarze 564: say(path,
565: "No filename suffix");
1.50 kristaps 566: continue;
567: }
1.84 schwarze 568: } else if (0 == strcmp(++fsec, "html")) {
1.60 schwarze 569: if (warnings)
570: say(path, "Skip html");
571: continue;
1.84 schwarze 572: } else if (0 == strcmp(fsec, "gz")) {
1.60 schwarze 573: if (warnings)
574: say(path, "Skip gz");
575: continue;
1.84 schwarze 576: } else if (0 == strcmp(fsec, "ps")) {
1.60 schwarze 577: if (warnings)
578: say(path, "Skip ps");
579: continue;
1.84 schwarze 580: } else if (0 == strcmp(fsec, "pdf")) {
1.60 schwarze 581: if (warnings)
582: say(path, "Skip pdf");
583: continue;
584: } else if ( ! use_all &&
1.84 schwarze 585: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
586: (FORM_CAT == dform && strcmp(fsec, "0")))) {
1.60 schwarze 587: if (warnings)
588: say(path, "Wrong filename suffix");
589: continue;
1.66 schwarze 590: } else
1.84 schwarze 591: fsec[-1] = '\0';
592: mlink = mandoc_calloc(1, sizeof(struct mlink));
593: strlcpy(mlink->file, path, sizeof(mlink->file));
594: mlink->dform = dform;
595: if (NULL != dsec)
596: mlink->dsec = mandoc_strdup(dsec);
597: if (NULL != arch)
598: mlink->arch = mandoc_strdup(arch);
599: mlink->name = mandoc_strdup(ff->fts_name);
600: if (NULL != fsec)
601: mlink->fsec = mandoc_strdup(fsec);
602: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 603: continue;
1.84 schwarze 604: } else if (FTS_D != ff->fts_info &&
1.60 schwarze 605: FTS_DP != ff->fts_info) {
606: if (warnings)
607: say(path, "Not a regular file");
1.50 kristaps 608: continue;
1.60 schwarze 609: }
1.1 kristaps 610:
1.50 kristaps 611: switch (ff->fts_level) {
612: case (0):
613: /* Ignore the root directory. */
614: break;
615: case (1):
616: /*
617: * This might contain manX/ or catX/.
618: * Try to infer this from the name.
619: * If we're not in use_all, enforce it.
620: */
621: dsec = NULL;
622: dform = FORM_NONE;
623: cp = ff->fts_name;
624: if (FTS_DP == ff->fts_info)
625: break;
1.1 kristaps 626:
1.50 kristaps 627: if (0 == strncmp(cp, "man", 3)) {
628: dform = FORM_SRC;
1.66 schwarze 629: dsec = cp + 3;
1.50 kristaps 630: } else if (0 == strncmp(cp, "cat", 3)) {
631: dform = FORM_CAT;
1.66 schwarze 632: dsec = cp + 3;
1.50 kristaps 633: }
1.1 kristaps 634:
1.50 kristaps 635: if (NULL != dsec || use_all)
636: break;
1.1 kristaps 637:
1.56 schwarze 638: if (warnings)
1.59 schwarze 639: say(path, "Unknown directory part");
1.50 kristaps 640: fts_set(f, ff, FTS_SKIP);
641: break;
642: case (2):
643: /*
644: * Possibly our architecture.
645: * If we're descending, keep tabs on it.
646: */
647: arch = NULL;
648: if (FTS_DP != ff->fts_info && NULL != dsec)
1.66 schwarze 649: arch = ff->fts_name;
1.50 kristaps 650: break;
651: default:
652: if (FTS_DP == ff->fts_info || use_all)
653: break;
1.56 schwarze 654: if (warnings)
1.59 schwarze 655: say(path, "Extraneous directory part");
1.50 kristaps 656: fts_set(f, ff, FTS_SKIP);
657: break;
1.5 kristaps 658: }
1.50 kristaps 659: }
660:
661: fts_close(f);
662: return(1);
663: }
1.5 kristaps 664:
1.50 kristaps 665: /*
666: * Add a file to the file vector.
667: * Do not verify that it's a "valid" looking manpage (we'll do that
668: * later).
669: *
670: * Try to infer the manual section, architecture, and page name from the
671: * path, assuming it looks like
672: *
673: * [./]man*[/<arch>]/<name>.<section>
674: * or
675: * [./]cat<section>[/<arch>]/<name>.0
676: *
1.83 schwarze 677: * Stuff this information directly into the mlink vector.
1.50 kristaps 678: * See treescan() for the fts(3) version of this.
679: */
680: static void
1.59 schwarze 681: filescan(const char *file)
1.50 kristaps 682: {
1.59 schwarze 683: char buf[PATH_MAX];
1.84 schwarze 684: struct stat st;
685: struct mlink *mlink;
1.59 schwarze 686: char *p, *start;
1.5 kristaps 687:
1.50 kristaps 688: assert(use_all);
1.5 kristaps 689:
1.50 kristaps 690: if (0 == strncmp(file, "./", 2))
691: file += 2;
1.5 kristaps 692:
1.59 schwarze 693: if (NULL == realpath(file, buf)) {
694: exitcode = (int)MANDOCLEVEL_BADARG;
695: say(file, NULL);
696: return;
1.72 schwarze 697: } else if (OP_TEST != op && strstr(buf, basedir) != buf) {
1.59 schwarze 698: exitcode = (int)MANDOCLEVEL_BADARG;
699: say("", "%s: outside base directory", buf);
700: return;
701: } else if (-1 == stat(buf, &st)) {
702: exitcode = (int)MANDOCLEVEL_BADARG;
703: say(file, NULL);
1.50 kristaps 704: return;
705: } else if ( ! (S_IFREG & st.st_mode)) {
1.59 schwarze 706: exitcode = (int)MANDOCLEVEL_BADARG;
707: say(file, "Not a regular file");
1.50 kristaps 708: return;
709: }
1.59 schwarze 710: start = buf + strlen(basedir);
1.84 schwarze 711: mlink = mandoc_calloc(1, sizeof(struct mlink));
712: strlcpy(mlink->file, start, sizeof(mlink->file));
1.17 schwarze 713:
1.50 kristaps 714: /*
715: * First try to guess our directory structure.
716: * If we find a separator, try to look for man* or cat*.
717: * If we find one of these and what's underneath is a directory,
718: * assume it's an architecture.
719: */
720: if (NULL != (p = strchr(start, '/'))) {
721: *p++ = '\0';
722: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 723: mlink->dform = FORM_SRC;
724: mlink->dsec = mandoc_strdup(start + 3);
1.50 kristaps 725: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 726: mlink->dform = FORM_CAT;
727: mlink->dsec = mandoc_strdup(start + 3);
1.17 schwarze 728: }
1.5 kristaps 729:
1.50 kristaps 730: start = p;
1.84 schwarze 731: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 732: *p++ = '\0';
1.84 schwarze 733: mlink->arch = mandoc_strdup(start);
1.50 kristaps 734: start = p;
1.84 schwarze 735: }
1.50 kristaps 736: }
737:
738: /*
739: * Now check the file suffix.
740: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
741: */
742: p = strrchr(start, '\0');
743: while (p-- > start && '/' != *p && '.' != *p)
744: /* Loop. */ ;
745:
746: if ('.' == *p) {
747: *p++ = '\0';
1.84 schwarze 748: mlink->fsec = mandoc_strdup(p);
1.5 kristaps 749: }
750:
1.10 kristaps 751: /*
1.50 kristaps 752: * Now try to parse the name.
753: * Use the filename portion of the path.
1.10 kristaps 754: */
1.84 schwarze 755: mlink->name = start;
1.50 kristaps 756: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 757: mlink->name = p + 1;
1.50 kristaps 758: *p = '\0';
1.84 schwarze 759: }
760: mlink->name = mandoc_strdup(mlink->name);
1.50 kristaps 761:
1.84 schwarze 762: mlink_add(mlink, &st);
1.50 kristaps 763: }
1.5 kristaps 764:
1.50 kristaps 765: static void
1.84 schwarze 766: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 767: {
1.83 schwarze 768: struct inodev inodev;
1.78 schwarze 769: struct mpage *mpage;
1.83 schwarze 770: unsigned int slot;
1.50 kristaps 771:
1.84 schwarze 772: assert(NULL != mlink->file);
1.50 kristaps 773:
1.84 schwarze 774: if (NULL == mlink->dsec)
775: mlink->dsec = mandoc_strdup("");
776: if (NULL == mlink->arch)
777: mlink->arch = mandoc_strdup("");
778: if (NULL == mlink->name)
779: mlink->name = mandoc_strdup("");
780: if (NULL == mlink->fsec)
781: mlink->fsec = mandoc_strdup("");
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;
1.74 schwarze 789: else
1.84 schwarze 790: mlink->fform = FORM_NONE;
1.82 schwarze 791:
1.83 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
1.85 schwarze 807: mlink->next = mpage->mlinks;
1.82 schwarze 808: mpage->mlinks = mlink;
809: }
810:
811: static void
812: mlink_free(struct mlink *mlink)
813: {
814:
815: free(mlink->dsec);
816: free(mlink->arch);
817: free(mlink->name);
818: free(mlink->fsec);
819: free(mlink);
1.50 kristaps 820: }
1.3 kristaps 821:
1.50 kristaps 822: static void
1.78 schwarze 823: mpages_free(void)
1.50 kristaps 824: {
1.78 schwarze 825: struct mpage *mpage;
1.82 schwarze 826: struct mlink *mlink;
1.80 schwarze 827: unsigned int slot;
1.3 kristaps 828:
1.80 schwarze 829: mpage = ohash_first(&mpages, &slot);
830: while (NULL != mpage) {
1.82 schwarze 831: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 832: mpage->mlinks = mlink->next;
1.82 schwarze 833: mlink_free(mlink);
834: }
1.78 schwarze 835: free(mpage->sec);
836: free(mpage->arch);
1.82 schwarze 837: free(mpage->title);
838: free(mpage->desc);
1.78 schwarze 839: free(mpage);
1.80 schwarze 840: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 841: }
842: }
1.38 schwarze 843:
1.50 kristaps 844: /*
1.80 schwarze 845: * Run through the files in the global vector "mpages"
846: * and add them to the database specified in "basedir".
1.50 kristaps 847: *
848: * This handles the parsing scheme itself, using the cues of directory
849: * and filename to determine whether the file is parsable or not.
850: */
1.59 schwarze 851: static void
1.78 schwarze 852: mpages_merge(struct mchars *mc, struct mparse *mp, int check_reachable)
1.50 kristaps 853: {
1.69 schwarze 854: struct ohash title_table;
1.71 schwarze 855: struct ohash_info title_info, str_info;
1.69 schwarze 856: char buf[PATH_MAX];
1.78 schwarze 857: struct mpage *mpage;
1.69 schwarze 858: struct mdoc *mdoc;
859: struct man *man;
860: struct title *title_entry;
861: char *bufp, *title_str;
1.82 schwarze 862: const char *cp;
1.69 schwarze 863: size_t sz;
864: int match;
1.80 schwarze 865: unsigned int pslot, tslot;
1.69 schwarze 866: enum mandoclevel lvl;
867:
1.71 schwarze 868: str_info.alloc = hash_alloc;
869: str_info.halloc = hash_halloc;
870: str_info.hfree = hash_free;
871: str_info.key_offset = offsetof(struct str, key);
872:
1.69 schwarze 873: if (check_reachable) {
874: title_info.alloc = hash_alloc;
875: title_info.halloc = hash_halloc;
876: title_info.hfree = hash_free;
877: title_info.key_offset = offsetof(struct title, title);
878: ohash_init(&title_table, 6, &title_info);
879: }
1.50 kristaps 880:
1.80 schwarze 881: mpage = ohash_first(&mpages, &pslot);
882: while (NULL != mpage) {
1.50 kristaps 883: /*
884: * If we're a catpage (as defined by our path), then see
885: * if a manpage exists by the same name (ignoring the
886: * suffix).
887: * If it does, then we want to use it instead of our
888: * own.
889: */
1.82 schwarze 890: if ( ! use_all && FORM_CAT == mpage->mlinks->dform) {
891: sz = strlcpy(buf, mpage->mlinks->file, PATH_MAX);
1.58 schwarze 892: if (sz >= PATH_MAX) {
1.56 schwarze 893: if (warnings)
1.82 schwarze 894: say(mpage->mlinks->file,
1.78 schwarze 895: "Filename too long");
1.80 schwarze 896: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 897: continue;
898: }
899: bufp = strstr(buf, "cat");
900: assert(NULL != bufp);
901: memcpy(bufp, "man", 3);
902: if (NULL != (bufp = strrchr(buf, '.')))
903: *++bufp = '\0';
1.82 schwarze 904: strlcat(buf, mpage->mlinks->dsec, PATH_MAX);
1.83 schwarze 905: if (NULL != ohash_find(&mlinks,
906: ohash_qlookup(&mlinks, buf))) {
1.56 schwarze 907: if (warnings)
1.82 schwarze 908: say(mpage->mlinks->file, "Man "
1.56 schwarze 909: "source exists: %s", buf);
1.80 schwarze 910: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 911: continue;
912: }
913: }
914:
1.71 schwarze 915: ohash_init(&strings, 6, &str_info);
1.50 kristaps 916: mparse_reset(mp);
917: mdoc = NULL;
918: man = NULL;
1.69 schwarze 919: match = 1;
1.14 schwarze 920:
921: /*
1.33 schwarze 922: * Try interpreting the file as mdoc(7) or man(7)
923: * source code, unless it is already known to be
924: * formatted. Fall back to formatted mode.
1.14 schwarze 925: */
1.82 schwarze 926: if (FORM_CAT != mpage->mlinks->dform ||
1.84 schwarze 927: FORM_CAT != mpage->mlinks->fform) {
1.82 schwarze 928: lvl = mparse_readfd(mp, -1, mpage->mlinks->file);
1.50 kristaps 929: if (lvl < MANDOCLEVEL_FATAL)
930: mparse_result(mp, &mdoc, &man);
1.76 schwarze 931: }
1.14 schwarze 932:
933: if (NULL != mdoc) {
1.82 schwarze 934: mpage->form = FORM_SRC;
935: mpage->sec =
936: mandoc_strdup(mdoc_meta(mdoc)->msec);
937: mpage->arch = mdoc_meta(mdoc)->arch;
1.84 schwarze 938: mpage->arch = mandoc_strdup(
939: NULL == mpage->arch ? "" : mpage->arch);
1.82 schwarze 940: mpage->title =
941: mandoc_strdup(mdoc_meta(mdoc)->title);
1.14 schwarze 942: } else if (NULL != man) {
1.82 schwarze 943: mpage->form = FORM_SRC;
944: mpage->sec =
945: mandoc_strdup(man_meta(man)->msec);
946: mpage->arch =
947: mandoc_strdup(mpage->mlinks->arch);
948: mpage->title =
949: mandoc_strdup(man_meta(man)->title);
1.76 schwarze 950: } else {
1.82 schwarze 951: mpage->form = FORM_CAT;
952: mpage->sec =
953: mandoc_strdup(mpage->mlinks->dsec);
954: mpage->arch =
955: mandoc_strdup(mpage->mlinks->arch);
956: mpage->title =
957: mandoc_strdup(mpage->mlinks->name);
1.76 schwarze 958: }
1.50 kristaps 959:
1.12 schwarze 960: /*
1.44 kristaps 961: * Check whether the manual section given in a file
962: * agrees with the directory where the file is located.
963: * Some manuals have suffixes like (3p) on their
964: * section number either inside the file or in the
965: * directory name, some are linked into more than one
966: * section, like encrypt(1) = makekey(8). Do not skip
967: * manuals for such reasons.
1.12 schwarze 968: */
1.82 schwarze 969: if (warnings && !use_all && FORM_SRC == mpage->form &&
970: strcasecmp(mpage->sec, mpage->mlinks->dsec)) {
1.69 schwarze 971: match = 0;
1.82 schwarze 972: say(mpage->mlinks->file, "Section \"%s\" "
1.78 schwarze 973: "manual in %s directory",
1.82 schwarze 974: mpage->sec, mpage->mlinks->dsec);
1.69 schwarze 975: }
1.12 schwarze 976:
1.42 schwarze 977: /*
978: * Manual page directories exist for each kernel
979: * architecture as returned by machine(1).
980: * However, many manuals only depend on the
981: * application architecture as returned by arch(1).
982: * For example, some (2/ARM) manuals are shared
983: * across the "armish" and "zaurus" kernel
984: * architectures.
985: * A few manuals are even shared across completely
986: * different architectures, for example fdformat(1)
987: * on amd64, i386, sparc, and sparc64.
988: * Thus, warn about architecture mismatches,
989: * but don't skip manuals for this reason.
990: */
1.82 schwarze 991: if (warnings && !use_all &&
992: strcasecmp(mpage->arch, mpage->mlinks->arch)) {
1.69 schwarze 993: match = 0;
1.82 schwarze 994: say(mpage->mlinks->file, "Architecture \"%s\" "
1.47 schwarze 995: "manual in \"%s\" directory",
1.82 schwarze 996: mpage->arch, mpage->mlinks->arch);
1.69 schwarze 997: }
1.82 schwarze 998: if (warnings && !use_all &&
999: strcasecmp(mpage->title, mpage->mlinks->name))
1.69 schwarze 1000: match = 0;
1.12 schwarze 1001:
1.82 schwarze 1002: putkey(mpage, mpage->mlinks->name, TYPE_Nm);
1.12 schwarze 1003:
1.50 kristaps 1004: if (NULL != mdoc) {
1005: if (NULL != (cp = mdoc_meta(mdoc)->name))
1.78 schwarze 1006: putkey(mpage, cp, TYPE_Nm);
1007: assert(NULL == mpage->desc);
1008: parse_mdoc(mpage, mdoc_node(mdoc));
1009: putkey(mpage, NULL != mpage->desc ?
1.82 schwarze 1010: mpage->desc : mpage->mlinks->name, TYPE_Nd);
1.50 kristaps 1011: } else if (NULL != man)
1.78 schwarze 1012: parse_man(mpage, man_node(man));
1.50 kristaps 1013: else
1.78 schwarze 1014: parse_cat(mpage);
1.44 kristaps 1015:
1.69 schwarze 1016: /*
1017: * Build a title string for the file. If it matches
1018: * the location of the file, remember the title as
1019: * found; else, remember it as missing.
1020: */
1021:
1022: if (check_reachable) {
1.82 schwarze 1023: if (-1 == asprintf(&title_str, "%s(%s%s%s)",
1024: mpage->title, mpage->sec,
1025: '\0' == *mpage->arch ? "" : "/",
1026: mpage->arch)) {
1.69 schwarze 1027: perror(NULL);
1028: exit((int)MANDOCLEVEL_SYSERR);
1029: }
1.80 schwarze 1030: tslot = ohash_qlookup(&title_table, title_str);
1031: title_entry = ohash_find(&title_table, tslot);
1.69 schwarze 1032: if (NULL == title_entry) {
1033: title_entry = mandoc_malloc(
1034: sizeof(struct title));
1035: title_entry->title = title_str;
1036: title_entry->file = mandoc_strdup(
1.82 schwarze 1037: match ? "" : mpage->mlinks->file);
1.80 schwarze 1038: ohash_insert(&title_table, tslot,
1.69 schwarze 1039: title_entry);
1040: } else {
1041: if (match)
1042: *title_entry->file = '\0';
1043: free(title_str);
1044: }
1045: }
1046:
1.82 schwarze 1047: dbindex(mpage, mc);
1.66 schwarze 1048: ohash_delete(&strings);
1.80 schwarze 1049: mpage = ohash_next(&mpages, &pslot);
1.69 schwarze 1050: }
1051:
1052: if (check_reachable) {
1.80 schwarze 1053: title_entry = ohash_first(&title_table, &tslot);
1.69 schwarze 1054: while (NULL != title_entry) {
1055: if ('\0' != *title_entry->file)
1056: say(title_entry->file,
1057: "Probably unreachable, title is %s",
1058: title_entry->title);
1059: free(title_entry->title);
1060: free(title_entry->file);
1061: free(title_entry);
1.80 schwarze 1062: title_entry = ohash_next(&title_table, &tslot);
1.69 schwarze 1063: }
1064: ohash_delete(&title_table);
1.50 kristaps 1065: }
1066: }
1.12 schwarze 1067:
1.50 kristaps 1068: static void
1.78 schwarze 1069: parse_cat(struct mpage *mpage)
1.50 kristaps 1070: {
1071: FILE *stream;
1072: char *line, *p, *title;
1073: size_t len, plen, titlesz;
1.12 schwarze 1074:
1.82 schwarze 1075: if (NULL == (stream = fopen(mpage->mlinks->file, "r"))) {
1.56 schwarze 1076: if (warnings)
1.82 schwarze 1077: say(mpage->mlinks->file, NULL);
1.50 kristaps 1078: return;
1079: }
1.1 kristaps 1080:
1.50 kristaps 1081: /* Skip to first blank line. */
1.1 kristaps 1082:
1.50 kristaps 1083: while (NULL != (line = fgetln(stream, &len)))
1084: if ('\n' == *line)
1085: break;
1.1 kristaps 1086:
1.50 kristaps 1087: /*
1088: * Assume the first line that is not indented
1089: * is the first section header. Skip to it.
1090: */
1.1 kristaps 1091:
1.50 kristaps 1092: while (NULL != (line = fgetln(stream, &len)))
1093: if ('\n' != *line && ' ' != *line)
1094: break;
1095:
1096: /*
1097: * Read up until the next section into a buffer.
1098: * Strip the leading and trailing newline from each read line,
1099: * appending a trailing space.
1100: * Ignore empty (whitespace-only) lines.
1101: */
1.1 kristaps 1102:
1.50 kristaps 1103: titlesz = 0;
1104: title = NULL;
1.38 schwarze 1105:
1.50 kristaps 1106: while (NULL != (line = fgetln(stream, &len))) {
1107: if (' ' != *line || '\n' != line[len - 1])
1108: break;
1109: while (len > 0 && isspace((unsigned char)*line)) {
1110: line++;
1111: len--;
1112: }
1113: if (1 == len)
1.38 schwarze 1114: continue;
1.50 kristaps 1115: title = mandoc_realloc(title, titlesz + len);
1116: memcpy(title + titlesz, line, len);
1117: titlesz += len;
1118: title[titlesz - 1] = ' ';
1119: }
1.38 schwarze 1120:
1.50 kristaps 1121: /*
1122: * If no page content can be found, or the input line
1123: * is already the next section header, or there is no
1124: * trailing newline, reuse the page title as the page
1125: * description.
1126: */
1.44 kristaps 1127:
1.50 kristaps 1128: if (NULL == title || '\0' == *title) {
1.56 schwarze 1129: if (warnings)
1.82 schwarze 1130: say(mpage->mlinks->file,
1131: "Cannot find NAME section");
1.78 schwarze 1132: assert(NULL == mpage->desc);
1.82 schwarze 1133: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1134: putkey(mpage, mpage->mlinks->name, TYPE_Nd);
1.50 kristaps 1135: fclose(stream);
1136: free(title);
1137: return;
1138: }
1.1 kristaps 1139:
1.50 kristaps 1140: title = mandoc_realloc(title, titlesz + 1);
1141: title[titlesz] = '\0';
1.33 schwarze 1142:
1.50 kristaps 1143: /*
1144: * Skip to the first dash.
1145: * Use the remaining line as the description (no more than 70
1146: * bytes).
1147: */
1.33 schwarze 1148:
1.50 kristaps 1149: if (NULL != (p = strstr(title, "- "))) {
1150: for (p += 2; ' ' == *p || '\b' == *p; p++)
1151: /* Skip to next word. */ ;
1152: } else {
1.56 schwarze 1153: if (warnings)
1.82 schwarze 1154: say(mpage->mlinks->file,
1155: "No dash in title line");
1.50 kristaps 1156: p = title;
1157: }
1.38 schwarze 1158:
1.50 kristaps 1159: plen = strlen(p);
1.1 kristaps 1160:
1.50 kristaps 1161: /* Strip backspace-encoding from line. */
1.1 kristaps 1162:
1.50 kristaps 1163: while (NULL != (line = memchr(p, '\b', plen))) {
1164: len = line - p;
1165: if (0 == len) {
1166: memmove(line, line + 1, plen--);
1167: continue;
1168: }
1169: memmove(line - 1, line + 1, plen - len);
1170: plen -= 2;
1171: }
1.1 kristaps 1172:
1.78 schwarze 1173: assert(NULL == mpage->desc);
1174: mpage->desc = mandoc_strdup(p);
1175: putkey(mpage, mpage->desc, TYPE_Nd);
1.50 kristaps 1176: fclose(stream);
1177: free(title);
1178: }
1.1 kristaps 1179:
1.50 kristaps 1180: /*
1181: * Put a type/word pair into the word database for this particular file.
1182: */
1183: static void
1.78 schwarze 1184: putkey(const struct mpage *mpage, const char *value, uint64_t type)
1.50 kristaps 1185: {
1.18 kristaps 1186:
1.50 kristaps 1187: assert(NULL != value);
1.78 schwarze 1188: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1189: }
1190:
1191: /*
1.50 kristaps 1192: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1193: */
1194: static void
1.78 schwarze 1195: putmdockey(const struct mpage *mpage,
1196: const struct mdoc_node *n, uint64_t m)
1.3 kristaps 1197: {
1.18 kristaps 1198:
1.50 kristaps 1199: for ( ; NULL != n; n = n->next) {
1200: if (NULL != n->child)
1.78 schwarze 1201: putmdockey(mpage, n->child, m);
1.50 kristaps 1202: if (MDOC_TEXT == n->type)
1.78 schwarze 1203: putkey(mpage, n->string, m);
1.50 kristaps 1204: }
1205: }
1.18 kristaps 1206:
1.61 schwarze 1207: static void
1.78 schwarze 1208: parse_man(struct mpage *mpage, const struct man_node *n)
1.50 kristaps 1209: {
1210: const struct man_node *head, *body;
1211: char *start, *sv, *title;
1212: char byte;
1213: size_t sz, titlesz;
1.18 kristaps 1214:
1.50 kristaps 1215: if (NULL == n)
1.61 schwarze 1216: return;
1.18 kristaps 1217:
1.50 kristaps 1218: /*
1219: * We're only searching for one thing: the first text child in
1220: * the BODY of a NAME section. Since we don't keep track of
1221: * sections in -man, run some hoops to find out whether we're in
1222: * the correct section or not.
1223: */
1.18 kristaps 1224:
1.50 kristaps 1225: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1226: body = n;
1227: assert(body->parent);
1228: if (NULL != (head = body->parent->head) &&
1229: 1 == head->nchild &&
1230: NULL != (head = (head->child)) &&
1231: MAN_TEXT == head->type &&
1232: 0 == strcmp(head->string, "NAME") &&
1233: NULL != (body = body->child) &&
1234: MAN_TEXT == body->type) {
1.3 kristaps 1235:
1.50 kristaps 1236: title = NULL;
1237: titlesz = 0;
1.3 kristaps 1238:
1.50 kristaps 1239: /*
1240: * Suck the entire NAME section into memory.
1241: * Yes, we might run away.
1242: * But too many manuals have big, spread-out
1243: * NAME sections over many lines.
1244: */
1.3 kristaps 1245:
1.50 kristaps 1246: for ( ; NULL != body; body = body->next) {
1247: if (MAN_TEXT != body->type)
1248: break;
1249: if (0 == (sz = strlen(body->string)))
1250: continue;
1251: title = mandoc_realloc
1252: (title, titlesz + sz + 1);
1253: memcpy(title + titlesz, body->string, sz);
1254: titlesz += sz + 1;
1255: title[titlesz - 1] = ' ';
1256: }
1257: if (NULL == title)
1.61 schwarze 1258: return;
1.18 kristaps 1259:
1.50 kristaps 1260: title = mandoc_realloc(title, titlesz + 1);
1261: title[titlesz] = '\0';
1.18 kristaps 1262:
1.50 kristaps 1263: /* Skip leading space. */
1.18 kristaps 1264:
1.50 kristaps 1265: sv = title;
1266: while (isspace((unsigned char)*sv))
1267: sv++;
1.18 kristaps 1268:
1.50 kristaps 1269: if (0 == (sz = strlen(sv))) {
1270: free(title);
1.61 schwarze 1271: return;
1.50 kristaps 1272: }
1.1 kristaps 1273:
1.50 kristaps 1274: /* Erase trailing space. */
1.1 kristaps 1275:
1.50 kristaps 1276: start = &sv[sz - 1];
1277: while (start > sv && isspace((unsigned char)*start))
1278: *start-- = '\0';
1.1 kristaps 1279:
1.50 kristaps 1280: if (start == sv) {
1281: free(title);
1.61 schwarze 1282: return;
1.50 kristaps 1283: }
1.1 kristaps 1284:
1.50 kristaps 1285: start = sv;
1.18 kristaps 1286:
1.50 kristaps 1287: /*
1288: * Go through a special heuristic dance here.
1289: * Conventionally, one or more manual names are
1290: * comma-specified prior to a whitespace, then a
1291: * dash, then a description. Try to puzzle out
1292: * the name parts here.
1293: */
1.18 kristaps 1294:
1.50 kristaps 1295: for ( ;; ) {
1296: sz = strcspn(start, " ,");
1297: if ('\0' == start[sz])
1298: break;
1.1 kristaps 1299:
1.50 kristaps 1300: byte = start[sz];
1301: start[sz] = '\0';
1.1 kristaps 1302:
1.78 schwarze 1303: putkey(mpage, start, TYPE_Nm);
1.1 kristaps 1304:
1.50 kristaps 1305: if (' ' == byte) {
1306: start += sz + 1;
1307: break;
1308: }
1.1 kristaps 1309:
1.50 kristaps 1310: assert(',' == byte);
1311: start += sz + 1;
1312: while (' ' == *start)
1313: start++;
1314: }
1.1 kristaps 1315:
1.50 kristaps 1316: if (sv == start) {
1.78 schwarze 1317: putkey(mpage, start, TYPE_Nm);
1.50 kristaps 1318: free(title);
1.61 schwarze 1319: return;
1.50 kristaps 1320: }
1.1 kristaps 1321:
1.50 kristaps 1322: while (isspace((unsigned char)*start))
1323: start++;
1.1 kristaps 1324:
1.50 kristaps 1325: if (0 == strncmp(start, "-", 1))
1326: start += 1;
1327: else if (0 == strncmp(start, "\\-\\-", 4))
1328: start += 4;
1329: else if (0 == strncmp(start, "\\-", 2))
1330: start += 2;
1331: else if (0 == strncmp(start, "\\(en", 4))
1332: start += 4;
1333: else if (0 == strncmp(start, "\\(em", 4))
1334: start += 4;
1.1 kristaps 1335:
1.50 kristaps 1336: while (' ' == *start)
1337: start++;
1.1 kristaps 1338:
1.78 schwarze 1339: assert(NULL == mpage->desc);
1340: mpage->desc = mandoc_strdup(start);
1341: putkey(mpage, mpage->desc, TYPE_Nd);
1.50 kristaps 1342: free(title);
1.61 schwarze 1343: return;
1.50 kristaps 1344: }
1345: }
1.1 kristaps 1346:
1.77 schwarze 1347: for (n = n->child; n; n = n->next) {
1.78 schwarze 1348: if (NULL != mpage->desc)
1.77 schwarze 1349: break;
1.78 schwarze 1350: parse_man(mpage, n);
1.77 schwarze 1351: }
1.1 kristaps 1352: }
1353:
1354: static void
1.78 schwarze 1355: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1356: {
1357:
1.50 kristaps 1358: assert(NULL != n);
1359: for (n = n->child; NULL != n; n = n->next) {
1360: switch (n->type) {
1361: case (MDOC_ELEM):
1362: /* FALLTHROUGH */
1363: case (MDOC_BLOCK):
1364: /* FALLTHROUGH */
1365: case (MDOC_HEAD):
1366: /* FALLTHROUGH */
1367: case (MDOC_BODY):
1368: /* FALLTHROUGH */
1369: case (MDOC_TAIL):
1370: if (NULL != mdocs[n->tok].fp)
1.78 schwarze 1371: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1.50 kristaps 1372: break;
1.68 schwarze 1373: if (mdocs[n->tok].mask)
1.78 schwarze 1374: putmdockey(mpage, n->child,
1375: mdocs[n->tok].mask);
1.50 kristaps 1376: break;
1377: default:
1378: assert(MDOC_ROOT != n->type);
1379: continue;
1380: }
1381: if (NULL != n->child)
1.78 schwarze 1382: parse_mdoc(mpage, n);
1.1 kristaps 1383: }
1384: }
1385:
1.25 schwarze 1386: static int
1.78 schwarze 1387: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1388: {
1389: const char *start, *end;
1390: size_t sz;
1.25 schwarze 1391:
1.50 kristaps 1392: if (SEC_SYNOPSIS != n->sec ||
1393: NULL == (n = n->child) ||
1394: MDOC_TEXT != n->type)
1.25 schwarze 1395: return(0);
1.1 kristaps 1396:
1397: /*
1398: * Only consider those `Fd' macro fields that begin with an
1399: * "inclusion" token (versus, e.g., #define).
1400: */
1.50 kristaps 1401:
1.1 kristaps 1402: if (strcmp("#include", n->string))
1.25 schwarze 1403: return(0);
1.1 kristaps 1404:
1405: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.25 schwarze 1406: return(0);
1.1 kristaps 1407:
1408: /*
1409: * Strip away the enclosing angle brackets and make sure we're
1410: * not zero-length.
1411: */
1412:
1413: start = n->string;
1414: if ('<' == *start || '"' == *start)
1415: start++;
1416:
1417: if (0 == (sz = strlen(start)))
1.25 schwarze 1418: return(0);
1.1 kristaps 1419:
1420: end = &start[(int)sz - 1];
1421: if ('>' == *end || '"' == *end)
1422: end--;
1423:
1.50 kristaps 1424: if (end > start)
1.78 schwarze 1425: putkeys(mpage, start, end - start + 1, TYPE_In);
1.25 schwarze 1426: return(1);
1.1 kristaps 1427: }
1428:
1.25 schwarze 1429: static int
1.78 schwarze 1430: parse_mdoc_In(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1431: {
1432:
1.50 kristaps 1433: if (NULL != n->child && MDOC_TEXT == n->child->type)
1.25 schwarze 1434: return(0);
1.1 kristaps 1435:
1.78 schwarze 1436: putkey(mpage, n->child->string, TYPE_In);
1.25 schwarze 1437: return(1);
1.1 kristaps 1438: }
1439:
1.25 schwarze 1440: static int
1.78 schwarze 1441: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1442: {
1443: const char *cp;
1444:
1.50 kristaps 1445: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.25 schwarze 1446: return(0);
1447:
1.50 kristaps 1448: /*
1449: * Parse: .Fn "struct type *name" "char *arg".
1450: * First strip away pointer symbol.
1451: * Then store the function name, then type.
1452: * Finally, store the arguments.
1453: */
1.1 kristaps 1454:
1.50 kristaps 1455: if (NULL == (cp = strrchr(n->string, ' ')))
1456: cp = n->string;
1.1 kristaps 1457:
1458: while ('*' == *cp)
1459: cp++;
1460:
1.78 schwarze 1461: putkey(mpage, cp, TYPE_Fn);
1.25 schwarze 1462:
1.50 kristaps 1463: if (n->string < cp)
1.78 schwarze 1464: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.25 schwarze 1465:
1.50 kristaps 1466: for (n = n->next; NULL != n; n = n->next)
1467: if (MDOC_TEXT == n->type)
1.78 schwarze 1468: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1469:
1470: return(0);
1.1 kristaps 1471: }
1472:
1.25 schwarze 1473: static int
1.78 schwarze 1474: parse_mdoc_St(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1475: {
1.25 schwarze 1476:
1.1 kristaps 1477: if (NULL == n->child || MDOC_TEXT != n->child->type)
1.25 schwarze 1478: return(0);
1.1 kristaps 1479:
1.78 schwarze 1480: putkey(mpage, n->child->string, TYPE_St);
1.25 schwarze 1481: return(1);
1.1 kristaps 1482: }
1483:
1.25 schwarze 1484: static int
1.78 schwarze 1485: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1486: {
1.67 schwarze 1487: char *cp;
1.1 kristaps 1488:
1489: if (NULL == (n = n->child))
1.25 schwarze 1490: return(0);
1.1 kristaps 1491:
1.67 schwarze 1492: if (NULL == n->next) {
1.78 schwarze 1493: putkey(mpage, n->string, TYPE_Xr);
1.67 schwarze 1494: return(0);
1495: }
1496:
1497: if (-1 == asprintf(&cp, "%s(%s)", n->string, n->next->string)) {
1498: perror(NULL);
1499: exit((int)MANDOCLEVEL_SYSERR);
1500: }
1.78 schwarze 1501: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1502: free(cp);
1503: return(0);
1.1 kristaps 1504: }
1505:
1.25 schwarze 1506: static int
1.78 schwarze 1507: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1508: {
1.50 kristaps 1509: size_t sz;
1.1 kristaps 1510:
1511: if (MDOC_BODY != n->type)
1.25 schwarze 1512: return(0);
1.1 kristaps 1513:
1.50 kristaps 1514: /*
1515: * Special-case the `Nd' because we need to put the description
1516: * into the document table.
1517: */
1518:
1519: for (n = n->child; NULL != n; n = n->next) {
1520: if (MDOC_TEXT == n->type) {
1.78 schwarze 1521: if (NULL != mpage->desc) {
1522: sz = strlen(mpage->desc) +
1.66 schwarze 1523: strlen(n->string) + 2;
1.78 schwarze 1524: mpage->desc = mandoc_realloc(
1525: mpage->desc, sz);
1526: strlcat(mpage->desc, " ", sz);
1527: strlcat(mpage->desc, n->string, sz);
1.66 schwarze 1528: } else
1.78 schwarze 1529: mpage->desc = mandoc_strdup(n->string);
1.50 kristaps 1530: }
1531: if (NULL != n->child)
1.78 schwarze 1532: parse_mdoc_Nd(mpage, n);
1.50 kristaps 1533: }
1.25 schwarze 1534: return(1);
1.1 kristaps 1535: }
1536:
1.25 schwarze 1537: static int
1.78 schwarze 1538: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1539: {
1540:
1.25 schwarze 1541: if (SEC_NAME == n->sec)
1542: return(1);
1543: else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
1544: return(0);
1.1 kristaps 1545:
1.25 schwarze 1546: return(1);
1.1 kristaps 1547: }
1548:
1.25 schwarze 1549: static int
1.78 schwarze 1550: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1551: {
1552:
1.25 schwarze 1553: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 kristaps 1554: }
1555:
1.50 kristaps 1556: static int
1.78 schwarze 1557: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1558: {
1559:
1.50 kristaps 1560: return(MDOC_HEAD == n->type);
1561: }
1.1 kristaps 1562:
1.50 kristaps 1563: static int
1.78 schwarze 1564: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1.50 kristaps 1565: {
1.1 kristaps 1566:
1.50 kristaps 1567: return(MDOC_BODY == n->type);
1.1 kristaps 1568: }
1569:
1.50 kristaps 1570: /*
1.66 schwarze 1571: * Add a string to the hash table for the current manual.
1572: * Each string has a bitmask telling which macros it belongs to.
1573: * When we finish the manual, we'll dump the table.
1.50 kristaps 1574: */
1575: static void
1.78 schwarze 1576: putkeys(const struct mpage *mpage,
1577: const char *cp, size_t sz, uint64_t v)
1.50 kristaps 1578: {
1579: struct str *s;
1.65 schwarze 1580: unsigned int slot;
1.50 kristaps 1581: const char *end;
1.25 schwarze 1582:
1.50 kristaps 1583: if (0 == sz)
1584: return;
1.25 schwarze 1585:
1.65 schwarze 1586: end = cp + sz;
1587: slot = ohash_qlookupi(&strings, cp, &end);
1588: s = ohash_find(&strings, slot);
1.25 schwarze 1589:
1.78 schwarze 1590: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1591: s->mask |= v;
1592: return;
1593: } else if (NULL == s) {
1.51 kristaps 1594: s = mandoc_calloc(sizeof(struct str) + sz + 1, 1);
1.50 kristaps 1595: memcpy(s->key, cp, sz);
1.65 schwarze 1596: ohash_insert(&strings, slot, s);
1.1 kristaps 1597: }
1.78 schwarze 1598: s->mpage = mpage;
1.50 kristaps 1599: s->mask = v;
1.1 kristaps 1600: }
1601:
1.50 kristaps 1602: /*
1603: * Take a Unicode codepoint and produce its UTF-8 encoding.
1604: * This isn't the best way to do this, but it works.
1605: * The magic numbers are from the UTF-8 packaging.
1606: * They're not as scary as they seem: read the UTF-8 spec for details.
1607: */
1608: static size_t
1609: utf8(unsigned int cp, char out[7])
1.1 kristaps 1610: {
1.50 kristaps 1611: size_t rc;
1.1 kristaps 1612:
1.50 kristaps 1613: rc = 0;
1614: if (cp <= 0x0000007F) {
1615: rc = 1;
1616: out[0] = (char)cp;
1617: } else if (cp <= 0x000007FF) {
1618: rc = 2;
1619: out[0] = (cp >> 6 & 31) | 192;
1620: out[1] = (cp & 63) | 128;
1621: } else if (cp <= 0x0000FFFF) {
1622: rc = 3;
1623: out[0] = (cp >> 12 & 15) | 224;
1624: out[1] = (cp >> 6 & 63) | 128;
1625: out[2] = (cp & 63) | 128;
1626: } else if (cp <= 0x001FFFFF) {
1627: rc = 4;
1628: out[0] = (cp >> 18 & 7) | 240;
1629: out[1] = (cp >> 12 & 63) | 128;
1630: out[2] = (cp >> 6 & 63) | 128;
1631: out[3] = (cp & 63) | 128;
1632: } else if (cp <= 0x03FFFFFF) {
1633: rc = 5;
1634: out[0] = (cp >> 24 & 3) | 248;
1635: out[1] = (cp >> 18 & 63) | 128;
1636: out[2] = (cp >> 12 & 63) | 128;
1637: out[3] = (cp >> 6 & 63) | 128;
1638: out[4] = (cp & 63) | 128;
1639: } else if (cp <= 0x7FFFFFFF) {
1640: rc = 6;
1641: out[0] = (cp >> 30 & 1) | 252;
1642: out[1] = (cp >> 24 & 63) | 128;
1643: out[2] = (cp >> 18 & 63) | 128;
1644: out[3] = (cp >> 12 & 63) | 128;
1645: out[4] = (cp >> 6 & 63) | 128;
1646: out[5] = (cp & 63) | 128;
1647: } else
1.1 kristaps 1648: return(0);
1649:
1.50 kristaps 1650: out[rc] = '\0';
1651: return(rc);
1652: }
1.1 kristaps 1653:
1.50 kristaps 1654: /*
1655: * Store the UTF-8 version of a key, or alias the pointer if the key has
1656: * no UTF-8 transcription marks in it.
1657: */
1658: static void
1659: utf8key(struct mchars *mc, struct str *key)
1660: {
1661: size_t sz, bsz, pos;
1662: char utfbuf[7], res[5];
1663: char *buf;
1664: const char *seq, *cpp, *val;
1665: int len, u;
1666: enum mandoc_esc esc;
1667:
1668: assert(NULL == key->utf8);
1669:
1670: res[0] = '\\';
1671: res[1] = '\t';
1672: res[2] = ASCII_NBRSP;
1673: res[3] = ASCII_HYPH;
1674: res[4] = '\0';
1.1 kristaps 1675:
1.50 kristaps 1676: val = key->key;
1677: bsz = strlen(val);
1.46 kristaps 1678:
1.50 kristaps 1679: /*
1680: * Pre-check: if we have no stop-characters, then set the
1681: * pointer as ourselvse and get out of here.
1682: */
1683: if (strcspn(val, res) == bsz) {
1684: key->utf8 = key->key;
1685: return;
1686: }
1.46 kristaps 1687:
1.50 kristaps 1688: /* Pre-allocate by the length of the input */
1.46 kristaps 1689:
1.50 kristaps 1690: buf = mandoc_malloc(++bsz);
1691: pos = 0;
1.46 kristaps 1692:
1.50 kristaps 1693: while ('\0' != *val) {
1694: /*
1695: * Halt on the first escape sequence.
1696: * This also halts on the end of string, in which case
1697: * we just copy, fallthrough, and exit the loop.
1698: */
1699: if ((sz = strcspn(val, res)) > 0) {
1700: memcpy(&buf[pos], val, sz);
1701: pos += sz;
1702: val += sz;
1703: }
1.46 kristaps 1704:
1.50 kristaps 1705: if (ASCII_HYPH == *val) {
1706: buf[pos++] = '-';
1707: val++;
1708: continue;
1709: } else if ('\t' == *val || ASCII_NBRSP == *val) {
1710: buf[pos++] = ' ';
1711: val++;
1712: continue;
1713: } else if ('\\' != *val)
1714: break;
1.46 kristaps 1715:
1.50 kristaps 1716: /* Read past the slash. */
1.46 kristaps 1717:
1.50 kristaps 1718: val++;
1719: u = 0;
1.46 kristaps 1720:
1.50 kristaps 1721: /*
1722: * Parse the escape sequence and see if it's a
1723: * predefined character or special character.
1724: */
1725: esc = mandoc_escape
1726: ((const char **)&val, &seq, &len);
1727: if (ESCAPE_ERROR == esc)
1728: break;
1.1 kristaps 1729:
1.50 kristaps 1730: if (ESCAPE_SPECIAL != esc)
1731: continue;
1732: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1733: continue;
1.1 kristaps 1734:
1.50 kristaps 1735: /*
1736: * If we have a Unicode codepoint, try to convert that
1737: * to a UTF-8 byte string.
1738: */
1739: cpp = utfbuf;
1740: if (0 == (sz = utf8(u, utfbuf)))
1741: continue;
1.1 kristaps 1742:
1.50 kristaps 1743: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1744:
1.50 kristaps 1745: sz = strlen(cpp);
1746: bsz += sz;
1.1 kristaps 1747:
1.50 kristaps 1748: buf = mandoc_realloc(buf, bsz);
1.1 kristaps 1749:
1.50 kristaps 1750: memcpy(&buf[pos], cpp, sz);
1751: pos += sz;
1.1 kristaps 1752: }
1753:
1.50 kristaps 1754: buf[pos] = '\0';
1755: key->utf8 = buf;
1.1 kristaps 1756: }
1757:
1.14 schwarze 1758: /*
1.50 kristaps 1759: * Flush the current page's terms (and their bits) into the database.
1760: * Wrap the entire set of additions in a transaction to make sqlite be a
1761: * little faster.
1762: * Also, UTF-8-encode the description at the last possible moment.
1.14 schwarze 1763: */
1764: static void
1.82 schwarze 1765: dbindex(const struct mpage *mpage, struct mchars *mc)
1.14 schwarze 1766: {
1.50 kristaps 1767: struct str *key;
1768: const char *desc;
1769: int64_t recno;
1.52 kristaps 1770: size_t i;
1.66 schwarze 1771: unsigned int slot;
1.14 schwarze 1772:
1.56 schwarze 1773: if (verb)
1.82 schwarze 1774: say(mpage->mlinks->file, "Adding to index");
1.43 kristaps 1775:
1.50 kristaps 1776: if (nodb)
1.14 schwarze 1777: return;
1.28 kristaps 1778:
1.50 kristaps 1779: desc = "";
1.78 schwarze 1780: if (NULL != mpage->desc && '\0' != *mpage->desc) {
1.65 schwarze 1781: key = ohash_find(&strings,
1.78 schwarze 1782: ohash_qlookup(&strings, mpage->desc));
1.50 kristaps 1783: assert(NULL != key);
1784: if (NULL == key->utf8)
1785: utf8key(mc, key);
1786: desc = key->utf8;
1787: }
1788:
1.52 kristaps 1789: SQL_EXEC("BEGIN TRANSACTION");
1.50 kristaps 1790:
1.52 kristaps 1791: i = 1;
1.81 schwarze 1792: /*
1793: * XXX The following three lines are obsolete
1794: * and only kept for backward compatibility
1795: * until apropos(1) and friends have caught up.
1796: */
1.82 schwarze 1797: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->mlinks->file);
1798: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->mlinks->dsec);
1799: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->mlinks->arch);
1.81 schwarze 1800: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, desc);
1.82 schwarze 1801: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
1.81 schwarze 1802: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.50 kristaps 1803: recno = sqlite3_last_insert_rowid(db);
1.81 schwarze 1804: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1805:
1806: i = 1;
1.82 schwarze 1807: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mpage->mlinks->dsec);
1808: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mpage->mlinks->arch);
1809: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mpage->mlinks->file);
1.81 schwarze 1810: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, recno);
1811: SQL_STEP(stmts[STMT_INSERT_LINK]);
1812: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.50 kristaps 1813:
1.66 schwarze 1814: for (key = ohash_first(&strings, &slot); NULL != key;
1815: key = ohash_next(&strings, &slot)) {
1.78 schwarze 1816: assert(key->mpage == mpage);
1.50 kristaps 1817: if (NULL == key->utf8)
1818: utf8key(mc, key);
1.52 kristaps 1819: i = 1;
1820: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1821: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->utf8);
1822: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno);
1823: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 1824: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 1825: if (key->utf8 != key->key)
1826: free(key->utf8);
1827: free(key);
1.38 schwarze 1828: }
1.28 kristaps 1829:
1.52 kristaps 1830: SQL_EXEC("END TRANSACTION");
1.14 schwarze 1831: }
1832:
1.5 kristaps 1833: static void
1.59 schwarze 1834: dbprune(void)
1.5 kristaps 1835: {
1.78 schwarze 1836: struct mpage *mpage;
1.82 schwarze 1837: struct mlink *mlink;
1.52 kristaps 1838: size_t i;
1.80 schwarze 1839: unsigned int slot;
1.5 kristaps 1840:
1.50 kristaps 1841: if (nodb)
1842: return;
1.12 schwarze 1843:
1.80 schwarze 1844: mpage = ohash_first(&mpages, &slot);
1845: while (NULL != mpage) {
1.82 schwarze 1846: mlink = mpage->mlinks;
1.52 kristaps 1847: i = 1;
1.82 schwarze 1848: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], i, mlink->file);
1.81 schwarze 1849: SQL_STEP(stmts[STMT_DELETE_PAGE]);
1850: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
1.56 schwarze 1851: if (verb)
1.82 schwarze 1852: say(mlink->file, "Deleted from index");
1.80 schwarze 1853: mpage = ohash_next(&mpages, &slot);
1.5 kristaps 1854: }
1855: }
1856:
1.4 kristaps 1857: /*
1.50 kristaps 1858: * Close an existing database and its prepared statements.
1859: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 1860: */
1.35 kristaps 1861: static void
1.59 schwarze 1862: dbclose(int real)
1.4 kristaps 1863: {
1.50 kristaps 1864: size_t i;
1.4 kristaps 1865:
1.50 kristaps 1866: if (nodb)
1.38 schwarze 1867: return;
1.50 kristaps 1868:
1869: for (i = 0; i < STMT__MAX; i++) {
1870: sqlite3_finalize(stmts[i]);
1871: stmts[i] = NULL;
1.4 kristaps 1872: }
1873:
1.50 kristaps 1874: sqlite3_close(db);
1875: db = NULL;
1.12 schwarze 1876:
1.50 kristaps 1877: if (real)
1878: return;
1.12 schwarze 1879:
1.63 schwarze 1880: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
1.59 schwarze 1881: exitcode = (int)MANDOCLEVEL_SYSERR;
1882: say(MANDOC_DB, NULL);
1883: }
1.50 kristaps 1884: }
1.14 schwarze 1885:
1.50 kristaps 1886: /*
1887: * This is straightforward stuff.
1888: * Open a database connection to a "temporary" database, then open a set
1889: * of prepared statements we'll use over and over again.
1890: * If "real" is set, we use the existing database; if not, we truncate a
1891: * temporary one.
1892: * Must be matched by dbclose().
1893: */
1894: static int
1.59 schwarze 1895: dbopen(int real)
1.50 kristaps 1896: {
1.63 schwarze 1897: const char *file, *sql;
1.50 kristaps 1898: int rc, ofl;
1.12 schwarze 1899:
1.50 kristaps 1900: if (nodb)
1901: return(1);
1.12 schwarze 1902:
1.63 schwarze 1903: ofl = SQLITE_OPEN_READWRITE;
1904: if (0 == real) {
1905: file = MANDOC_DB "~";
1906: if (-1 == remove(file) && ENOENT != errno) {
1907: exitcode = (int)MANDOCLEVEL_SYSERR;
1908: say(file, NULL);
1909: return(0);
1910: }
1911: ofl |= SQLITE_OPEN_EXCLUSIVE;
1912: } else
1913: file = MANDOC_DB;
1.45 kristaps 1914:
1.50 kristaps 1915: rc = sqlite3_open_v2(file, &db, ofl, NULL);
1916: if (SQLITE_OK == rc)
1.57 schwarze 1917: goto prepare_statements;
1.50 kristaps 1918: if (SQLITE_CANTOPEN != rc) {
1.59 schwarze 1919: exitcode = (int)MANDOCLEVEL_SYSERR;
1920: say(file, NULL);
1.50 kristaps 1921: return(0);
1922: }
1.12 schwarze 1923:
1.50 kristaps 1924: sqlite3_close(db);
1925: db = NULL;
1.12 schwarze 1926:
1.50 kristaps 1927: if (SQLITE_OK != (rc = sqlite3_open(file, &db))) {
1.59 schwarze 1928: exitcode = (int)MANDOCLEVEL_SYSERR;
1929: say(file, NULL);
1.50 kristaps 1930: return(0);
1931: }
1.12 schwarze 1932:
1.81 schwarze 1933: /*
1934: * XXX The first three columns in table mpages are obsolete
1935: * and only kept for backward compatibility
1936: * until apropos(1) and friends have caught up.
1937: */
1938: sql = "CREATE TABLE \"mpages\" (\n"
1.50 kristaps 1939: " \"file\" TEXT NOT NULL,\n"
1940: " \"sec\" TEXT NOT NULL,\n"
1941: " \"arch\" TEXT NOT NULL,\n"
1942: " \"desc\" TEXT NOT NULL,\n"
1943: " \"form\" INTEGER NOT NULL,\n"
1944: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1945: ");\n"
1946: "\n"
1.81 schwarze 1947: "CREATE TABLE \"mlinks\" (\n"
1948: " \"sec\" TEXT NOT NULL,\n"
1949: " \"arch\" TEXT NOT NULL,\n"
1950: " \"name\" TEXT NOT NULL,\n"
1951: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1952: "ON DELETE CASCADE,\n"
1953: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1954: ");\n"
1955: "\n"
1.50 kristaps 1956: "CREATE TABLE \"keys\" (\n"
1957: " \"bits\" INTEGER NOT NULL,\n"
1958: " \"key\" TEXT NOT NULL,\n"
1.81 schwarze 1959: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1960: "ON DELETE CASCADE,\n"
1.50 kristaps 1961: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1962: ");\n"
1963: "\n"
1964: "CREATE INDEX \"key_index\" ON keys (key);\n";
1.14 schwarze 1965:
1.50 kristaps 1966: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 1967: exitcode = (int)MANDOCLEVEL_SYSERR;
1968: say(file, "%s", sqlite3_errmsg(db));
1.50 kristaps 1969: return(0);
1970: }
1.4 kristaps 1971:
1.57 schwarze 1972: prepare_statements:
1973: SQL_EXEC("PRAGMA foreign_keys = ON");
1.81 schwarze 1974: sql = "DELETE FROM mpages where file=?";
1975: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
1976: sql = "INSERT INTO mpages "
1.50 kristaps 1977: "(file,sec,arch,desc,form) VALUES (?,?,?,?,?)";
1.81 schwarze 1978: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
1979: sql = "INSERT INTO mlinks "
1980: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1981: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.50 kristaps 1982: sql = "INSERT INTO keys "
1.81 schwarze 1983: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 1984: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 1985:
1986: #ifndef __APPLE__
1987: /*
1988: * When opening a new database, we can turn off
1989: * synchronous mode for much better performance.
1990: */
1991:
1992: if (real)
1993: SQL_EXEC("PRAGMA synchronous = OFF");
1994: #endif
1995:
1.50 kristaps 1996: return(1);
1997: }
1.5 kristaps 1998:
1.50 kristaps 1999: static void *
2000: hash_halloc(size_t sz, void *arg)
2001: {
1.12 schwarze 2002:
1.50 kristaps 2003: return(mandoc_calloc(sz, 1));
2004: }
1.12 schwarze 2005:
1.50 kristaps 2006: static void *
2007: hash_alloc(size_t sz, void *arg)
2008: {
1.5 kristaps 2009:
1.50 kristaps 2010: return(mandoc_malloc(sz));
2011: }
1.41 kristaps 2012:
1.50 kristaps 2013: static void
2014: hash_free(void *p, size_t sz, void *arg)
2015: {
1.4 kristaps 2016:
1.50 kristaps 2017: free(p);
1.4 kristaps 2018: }
2019:
1.50 kristaps 2020: static int
1.59 schwarze 2021: set_basedir(const char *targetdir)
1.4 kristaps 2022: {
1.59 schwarze 2023: static char startdir[PATH_MAX];
2024: static int fd;
1.4 kristaps 2025:
1.59 schwarze 2026: /*
2027: * Remember where we started by keeping a fd open to the origin
2028: * path component: throughout this utility, we chdir() a lot to
2029: * handle relative paths, and by doing this, we can return to
2030: * the starting point.
2031: */
2032: if ('\0' == *startdir) {
2033: if (NULL == getcwd(startdir, PATH_MAX)) {
2034: exitcode = (int)MANDOCLEVEL_SYSERR;
2035: if (NULL != targetdir)
2036: say(".", NULL);
2037: return(0);
2038: }
2039: if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
2040: exitcode = (int)MANDOCLEVEL_SYSERR;
2041: say(startdir, NULL);
2042: return(0);
2043: }
2044: if (NULL == targetdir)
2045: targetdir = startdir;
2046: } else {
2047: if (-1 == fd)
2048: return(0);
2049: if (-1 == fchdir(fd)) {
2050: close(fd);
2051: basedir[0] = '\0';
2052: exitcode = (int)MANDOCLEVEL_SYSERR;
2053: say(startdir, NULL);
2054: return(0);
2055: }
2056: if (NULL == targetdir) {
2057: close(fd);
2058: return(1);
2059: }
2060: }
2061: if (NULL == realpath(targetdir, basedir)) {
2062: basedir[0] = '\0';
2063: exitcode = (int)MANDOCLEVEL_BADARG;
2064: say(targetdir, NULL);
1.50 kristaps 2065: return(0);
1.59 schwarze 2066: } else if (-1 == chdir(basedir)) {
2067: exitcode = (int)MANDOCLEVEL_BADARG;
2068: say("", NULL);
1.50 kristaps 2069: return(0);
1.4 kristaps 2070: }
1.50 kristaps 2071: return(1);
1.56 schwarze 2072: }
2073:
2074: static void
1.59 schwarze 2075: say(const char *file, const char *format, ...)
1.56 schwarze 2076: {
2077: va_list ap;
2078:
1.59 schwarze 2079: if ('\0' != *basedir)
2080: fprintf(stderr, "%s", basedir);
2081: if ('\0' != *basedir && '\0' != *file)
2082: fputs("//", stderr);
1.56 schwarze 2083: if ('\0' != *file)
1.59 schwarze 2084: fprintf(stderr, "%s", file);
1.56 schwarze 2085: fputs(": ", stderr);
1.59 schwarze 2086:
2087: if (NULL == format) {
2088: perror(NULL);
2089: return;
2090: }
1.56 schwarze 2091:
2092: va_start(ap, format);
2093: vfprintf(stderr, format, ap);
2094: va_end(ap);
2095:
2096: fputc('\n', stderr);
1.1 kristaps 2097: }
CVSweb