Annotation of mandoc/mandocdb.c, Revision 1.106
1.106 ! schwarze 1: /* $Id: mandocdb.c,v 1.105 2014/01/06 13:54:17 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.95 schwarze 4: * Copyright (c) 2011, 2012, 2013, 2014 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 {
1.96 schwarze 85: char *rendered; /* key in UTF-8 or ASCII form */
1.78 schwarze 86: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 87: uint64_t mask; /* bitmask in sequence */
1.96 schwarze 88: char key[]; /* may contain escape sequences */
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:
1.50 kristaps 117: enum stmt {
1.81 schwarze 118: STMT_DELETE_PAGE = 0, /* delete mpage */
119: STMT_INSERT_PAGE, /* insert mpage */
120: STMT_INSERT_LINK, /* insert mlink */
121: STMT_INSERT_KEY, /* insert parsed key */
1.50 kristaps 122: STMT__MAX
1.1 kristaps 123: };
124:
1.78 schwarze 125: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
1.1 kristaps 126:
1.50 kristaps 127: struct mdoc_handler {
128: mdoc_fp fp; /* optional handler */
129: uint64_t mask; /* set unless handler returns 0 */
1.1 kristaps 130: };
131:
1.59 schwarze 132: static void dbclose(int);
1.105 schwarze 133: static void dbadd(const struct mpage *, struct mchars *);
1.59 schwarze 134: static int dbopen(int);
135: static void dbprune(void);
136: static void filescan(const char *);
1.50 kristaps 137: static void *hash_alloc(size_t, void *);
138: static void hash_free(void *, size_t, void *);
139: static void *hash_halloc(size_t, void *);
1.84 schwarze 140: static void mlink_add(struct mlink *, const struct stat *);
1.93 schwarze 141: static int mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 142: static void mlink_free(struct mlink *);
1.89 schwarze 143: static void mlinks_undupe(struct mpage *);
1.78 schwarze 144: static void mpages_free(void);
1.101 schwarze 145: static void mpages_merge(struct mchars *, struct mparse *);
1.78 schwarze 146: static void parse_cat(struct mpage *);
147: static void parse_man(struct mpage *, const struct man_node *);
148: static void parse_mdoc(struct mpage *, const struct mdoc_node *);
149: static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
150: static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
151: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
152: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
153: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
154: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
155: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
156: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
157: static void putkey(const struct mpage *,
1.50 kristaps 158: const char *, uint64_t);
1.78 schwarze 159: static void putkeys(const struct mpage *,
1.64 schwarze 160: const char *, size_t, uint64_t);
1.78 schwarze 161: static void putmdockey(const struct mpage *,
1.50 kristaps 162: const struct mdoc_node *, uint64_t);
1.96 schwarze 163: static void render_key(struct mchars *, struct str *);
1.59 schwarze 164: static void say(const char *, const char *, ...);
1.80 schwarze 165: static int set_basedir(const char *);
1.59 schwarze 166: static int treescan(void);
1.50 kristaps 167: static size_t utf8(unsigned int, char [7]);
168:
169: static char *progname;
1.102 schwarze 170: static int nodb; /* no database changes */
171: static int quick; /* abort the parse early */
1.50 kristaps 172: static int use_all; /* use all found files */
173: static int verb; /* print what we're doing */
174: static int warnings; /* warn about crap */
1.95 schwarze 175: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 176: static int exitcode; /* to be returned by main */
1.50 kristaps 177: static enum op op; /* operational mode */
1.59 schwarze 178: static char basedir[PATH_MAX]; /* current base directory */
1.78 schwarze 179: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 180: static struct ohash mlinks; /* table of directory entries */
1.50 kristaps 181: static struct ohash strings; /* table of all strings */
182: static sqlite3 *db = NULL; /* current database */
183: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.25 schwarze 184:
185: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 186: { NULL, 0 }, /* Ap */
187: { NULL, 0 }, /* Dd */
188: { NULL, 0 }, /* Dt */
189: { NULL, 0 }, /* Os */
190: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
191: { parse_mdoc_head, TYPE_Ss }, /* Ss */
192: { NULL, 0 }, /* Pp */
193: { NULL, 0 }, /* D1 */
194: { NULL, 0 }, /* Dl */
195: { NULL, 0 }, /* Bd */
196: { NULL, 0 }, /* Ed */
197: { NULL, 0 }, /* Bl */
198: { NULL, 0 }, /* El */
199: { NULL, 0 }, /* It */
200: { NULL, 0 }, /* Ad */
201: { NULL, TYPE_An }, /* An */
202: { NULL, TYPE_Ar }, /* Ar */
203: { NULL, TYPE_Cd }, /* Cd */
204: { NULL, TYPE_Cm }, /* Cm */
205: { NULL, TYPE_Dv }, /* Dv */
206: { NULL, TYPE_Er }, /* Er */
207: { NULL, TYPE_Ev }, /* Ev */
208: { NULL, 0 }, /* Ex */
209: { NULL, TYPE_Fa }, /* Fa */
210: { parse_mdoc_Fd, 0 }, /* Fd */
211: { NULL, TYPE_Fl }, /* Fl */
212: { parse_mdoc_Fn, 0 }, /* Fn */
213: { NULL, TYPE_Ft }, /* Ft */
214: { NULL, TYPE_Ic }, /* Ic */
1.92 schwarze 215: { NULL, TYPE_In }, /* In */
1.68 schwarze 216: { NULL, TYPE_Li }, /* Li */
217: { parse_mdoc_Nd, TYPE_Nd }, /* Nd */
218: { parse_mdoc_Nm, TYPE_Nm }, /* Nm */
219: { NULL, 0 }, /* Op */
220: { NULL, 0 }, /* Ot */
221: { NULL, TYPE_Pa }, /* Pa */
222: { NULL, 0 }, /* Rv */
1.92 schwarze 223: { NULL, TYPE_St }, /* St */
1.68 schwarze 224: { NULL, TYPE_Va }, /* Va */
225: { parse_mdoc_body, TYPE_Va }, /* Vt */
226: { parse_mdoc_Xr, 0 }, /* Xr */
227: { NULL, 0 }, /* %A */
228: { NULL, 0 }, /* %B */
229: { NULL, 0 }, /* %D */
230: { NULL, 0 }, /* %I */
231: { NULL, 0 }, /* %J */
232: { NULL, 0 }, /* %N */
233: { NULL, 0 }, /* %O */
234: { NULL, 0 }, /* %P */
235: { NULL, 0 }, /* %R */
236: { NULL, 0 }, /* %T */
237: { NULL, 0 }, /* %V */
238: { NULL, 0 }, /* Ac */
239: { NULL, 0 }, /* Ao */
240: { NULL, 0 }, /* Aq */
241: { NULL, TYPE_At }, /* At */
242: { NULL, 0 }, /* Bc */
243: { NULL, 0 }, /* Bf */
244: { NULL, 0 }, /* Bo */
245: { NULL, 0 }, /* Bq */
246: { NULL, TYPE_Bsx }, /* Bsx */
247: { NULL, TYPE_Bx }, /* Bx */
248: { NULL, 0 }, /* Db */
249: { NULL, 0 }, /* Dc */
250: { NULL, 0 }, /* Do */
251: { NULL, 0 }, /* Dq */
252: { NULL, 0 }, /* Ec */
253: { NULL, 0 }, /* Ef */
254: { NULL, TYPE_Em }, /* Em */
255: { NULL, 0 }, /* Eo */
256: { NULL, TYPE_Fx }, /* Fx */
257: { NULL, TYPE_Ms }, /* Ms */
258: { NULL, 0 }, /* No */
259: { NULL, 0 }, /* Ns */
260: { NULL, TYPE_Nx }, /* Nx */
261: { NULL, TYPE_Ox }, /* Ox */
262: { NULL, 0 }, /* Pc */
263: { NULL, 0 }, /* Pf */
264: { NULL, 0 }, /* Po */
265: { NULL, 0 }, /* Pq */
266: { NULL, 0 }, /* Qc */
267: { NULL, 0 }, /* Ql */
268: { NULL, 0 }, /* Qo */
269: { NULL, 0 }, /* Qq */
270: { NULL, 0 }, /* Re */
271: { NULL, 0 }, /* Rs */
272: { NULL, 0 }, /* Sc */
273: { NULL, 0 }, /* So */
274: { NULL, 0 }, /* Sq */
275: { NULL, 0 }, /* Sm */
276: { NULL, 0 }, /* Sx */
277: { NULL, TYPE_Sy }, /* Sy */
278: { NULL, TYPE_Tn }, /* Tn */
279: { NULL, 0 }, /* Ux */
280: { NULL, 0 }, /* Xc */
281: { NULL, 0 }, /* Xo */
282: { parse_mdoc_head, 0 }, /* Fo */
283: { NULL, 0 }, /* Fc */
284: { NULL, 0 }, /* Oo */
285: { NULL, 0 }, /* Oc */
286: { NULL, 0 }, /* Bk */
287: { NULL, 0 }, /* Ek */
288: { NULL, 0 }, /* Bt */
289: { NULL, 0 }, /* Hf */
290: { NULL, 0 }, /* Fr */
291: { NULL, 0 }, /* Ud */
292: { NULL, TYPE_Lb }, /* Lb */
293: { NULL, 0 }, /* Lp */
294: { NULL, TYPE_Lk }, /* Lk */
295: { NULL, TYPE_Mt }, /* Mt */
296: { NULL, 0 }, /* Brq */
297: { NULL, 0 }, /* Bro */
298: { NULL, 0 }, /* Brc */
299: { NULL, 0 }, /* %C */
300: { NULL, 0 }, /* Es */
301: { NULL, 0 }, /* En */
302: { NULL, TYPE_Dx }, /* Dx */
303: { NULL, 0 }, /* %Q */
304: { NULL, 0 }, /* br */
305: { NULL, 0 }, /* sp */
306: { NULL, 0 }, /* %U */
307: { NULL, 0 }, /* Ta */
1.1 kristaps 308: };
309:
310: int
311: main(int argc, char *argv[])
312: {
1.59 schwarze 313: int ch, i;
1.50 kristaps 314: size_t j, sz;
1.59 schwarze 315: const char *path_arg;
1.50 kristaps 316: struct mchars *mc;
317: struct manpaths dirs;
318: struct mparse *mp;
1.83 schwarze 319: struct ohash_info mpages_info, mlinks_info;
1.50 kristaps 320:
321: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
322: memset(&dirs, 0, sizeof(struct manpaths));
323:
1.83 schwarze 324: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
325: mpages_info.halloc = mlinks_info.halloc = hash_halloc;
326: mpages_info.hfree = mlinks_info.hfree = hash_free;
1.50 kristaps 327:
1.79 schwarze 328: mpages_info.key_offset = offsetof(struct mpage, inodev);
1.83 schwarze 329: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 kristaps 330:
331: progname = strrchr(argv[0], '/');
332: if (progname == NULL)
333: progname = argv[0];
334: else
335: ++progname;
336:
1.50 kristaps 337: /*
338: * We accept a few different invocations.
339: * The CHECKOP macro makes sure that invocation styles don't
340: * clobber each other.
341: */
342: #define CHECKOP(_op, _ch) do \
343: if (OP_DEFAULT != (_op)) { \
344: fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
345: goto usage; \
346: } while (/*CONSTCOND*/0)
1.10 kristaps 347:
1.59 schwarze 348: path_arg = NULL;
1.38 schwarze 349: op = OP_DEFAULT;
1.1 kristaps 350:
1.102 schwarze 351: while (-1 != (ch = getopt(argc, argv, "aC:d:nQT:tu:vW")))
1.1 kristaps 352: switch (ch) {
1.12 schwarze 353: case ('a'):
354: use_all = 1;
355: break;
1.34 schwarze 356: case ('C'):
1.50 kristaps 357: CHECKOP(op, ch);
1.59 schwarze 358: path_arg = optarg;
1.38 schwarze 359: op = OP_CONFFILE;
1.34 schwarze 360: break;
1.5 kristaps 361: case ('d'):
1.50 kristaps 362: CHECKOP(op, ch);
1.59 schwarze 363: path_arg = optarg;
1.5 kristaps 364: op = OP_UPDATE;
365: break;
1.50 kristaps 366: case ('n'):
367: nodb = 1;
368: break;
1.102 schwarze 369: case ('Q'):
370: quick = 1;
371: break;
1.95 schwarze 372: case ('T'):
373: if (strcmp(optarg, "utf8")) {
374: fprintf(stderr, "-T%s: Unsupported "
375: "output format\n", optarg);
376: goto usage;
377: }
378: write_utf8 = 1;
379: break;
1.38 schwarze 380: case ('t'):
1.50 kristaps 381: CHECKOP(op, ch);
1.38 schwarze 382: dup2(STDOUT_FILENO, STDERR_FILENO);
383: op = OP_TEST;
1.50 kristaps 384: nodb = warnings = 1;
1.38 schwarze 385: break;
1.5 kristaps 386: case ('u'):
1.50 kristaps 387: CHECKOP(op, ch);
1.59 schwarze 388: path_arg = optarg;
1.5 kristaps 389: op = OP_DELETE;
390: break;
391: case ('v'):
392: verb++;
393: break;
1.38 schwarze 394: case ('W'):
395: warnings = 1;
396: break;
1.1 kristaps 397: default:
1.38 schwarze 398: goto usage;
1.1 kristaps 399: }
400:
401: argc -= optind;
402: argv += optind;
403:
1.38 schwarze 404: if (OP_CONFFILE == op && argc > 0) {
1.50 kristaps 405: fprintf(stderr, "-C: Too many arguments\n");
1.38 schwarze 406: goto usage;
407: }
408:
1.59 schwarze 409: exitcode = (int)MANDOCLEVEL_OK;
1.50 kristaps 410: mp = mparse_alloc(MPARSE_AUTO,
1.102 schwarze 411: MANDOCLEVEL_FATAL, NULL, NULL, quick);
1.50 kristaps 412: mc = mchars_alloc();
413:
1.78 schwarze 414: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 415: ohash_init(&mlinks, 6, &mlinks_info);
1.50 kristaps 416:
417: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
418: /*
419: * Force processing all files.
420: */
421: use_all = 1;
1.59 schwarze 422:
1.50 kristaps 423: /*
424: * All of these deal with a specific directory.
425: * Jump into that directory then collect files specified
426: * on the command-line.
427: */
1.59 schwarze 428: if (0 == set_basedir(path_arg))
1.50 kristaps 429: goto out;
430: for (i = 0; i < argc; i++)
1.59 schwarze 431: filescan(argv[i]);
432: if (0 == dbopen(1))
1.50 kristaps 433: goto out;
434: if (OP_TEST != op)
1.59 schwarze 435: dbprune();
1.50 kristaps 436: if (OP_DELETE != op)
1.101 schwarze 437: mpages_merge(mc, mp);
1.59 schwarze 438: dbclose(1);
1.50 kristaps 439: } else {
440: /*
441: * If we have arguments, use them as our manpaths.
442: * If we don't, grok from manpath(1) or however else
443: * manpath_parse() wants to do it.
444: */
445: if (argc > 0) {
446: dirs.paths = mandoc_calloc
447: (argc, sizeof(char *));
448: dirs.sz = (size_t)argc;
449: for (i = 0; i < argc; i++)
450: dirs.paths[i] = mandoc_strdup(argv[i]);
451: } else
1.59 schwarze 452: manpath_parse(&dirs, path_arg, NULL, NULL);
1.50 kristaps 453:
454: /*
1.69 schwarze 455: * First scan the tree rooted at a base directory, then
456: * build a new database and finally move it into place.
1.50 kristaps 457: * Ignore zero-length directories and strip trailing
458: * slashes.
459: */
460: for (j = 0; j < dirs.sz; j++) {
461: sz = strlen(dirs.paths[j]);
462: if (sz && '/' == dirs.paths[j][sz - 1])
463: dirs.paths[j][--sz] = '\0';
464: if (0 == sz)
465: continue;
1.66 schwarze 466:
467: if (j) {
1.78 schwarze 468: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 469: ohash_init(&mlinks, 6, &mlinks_info);
1.66 schwarze 470: }
471:
1.59 schwarze 472: if (0 == set_basedir(dirs.paths[j]))
1.50 kristaps 473: goto out;
1.59 schwarze 474: if (0 == treescan())
1.50 kristaps 475: goto out;
1.59 schwarze 476: if (0 == set_basedir(dirs.paths[j]))
1.50 kristaps 477: goto out;
1.59 schwarze 478: if (0 == dbopen(0))
1.50 kristaps 479: goto out;
1.52 kristaps 480:
1.101 schwarze 481: mpages_merge(mc, mp);
1.59 schwarze 482: dbclose(0);
1.66 schwarze 483:
484: if (j + 1 < dirs.sz) {
1.80 schwarze 485: mpages_free();
1.78 schwarze 486: ohash_delete(&mpages);
1.83 schwarze 487: ohash_delete(&mlinks);
1.66 schwarze 488: }
1.50 kristaps 489: }
490: }
491: out:
1.59 schwarze 492: set_basedir(NULL);
1.50 kristaps 493: manpath_free(&dirs);
494: mchars_free(mc);
495: mparse_free(mp);
1.80 schwarze 496: mpages_free();
1.78 schwarze 497: ohash_delete(&mpages);
1.83 schwarze 498: ohash_delete(&mlinks);
1.59 schwarze 499: return(exitcode);
1.50 kristaps 500: usage:
1.102 schwarze 501: fprintf(stderr, "usage: %s [-anQvW] [-C file] [-Tutf8]\n"
502: " %s [-anQvW] [-Tutf8] dir ...\n"
503: " %s [-nQvW] [-Tutf8] -d dir [file ...]\n"
1.50 kristaps 504: " %s [-nvW] -u dir [file ...]\n"
1.102 schwarze 505: " %s [-Q] -t file ...\n",
1.50 kristaps 506: progname, progname, progname,
507: progname, progname);
508:
1.59 schwarze 509: return((int)MANDOCLEVEL_BADARG);
1.50 kristaps 510: }
511:
512: /*
1.59 schwarze 513: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 514: * We use fts(), scanning directory parts along the way for clues to our
515: * section and architecture.
516: *
517: * If use_all has been specified, grok all files.
518: * If not, sanitise paths to the following:
519: *
520: * [./]man*[/<arch>]/<name>.<section>
521: * or
522: * [./]cat<section>[/<arch>]/<name>.0
523: *
524: * TODO: accomodate for multi-language directories.
525: */
526: static int
1.59 schwarze 527: treescan(void)
1.50 kristaps 528: {
529: FTS *f;
530: FTSENT *ff;
1.84 schwarze 531: struct mlink *mlink;
1.50 kristaps 532: int dform;
1.94 schwarze 533: char *dsec, *arch, *fsec, *cp;
534: const char *path;
1.50 kristaps 535: const char *argv[2];
536:
537: argv[0] = ".";
538: argv[1] = (char *)NULL;
539:
540: /*
541: * Walk through all components under the directory, using the
542: * logical descent of files.
543: */
544: f = fts_open((char * const *)argv, FTS_LOGICAL, NULL);
545: if (NULL == f) {
1.59 schwarze 546: exitcode = (int)MANDOCLEVEL_SYSERR;
547: say("", NULL);
1.50 kristaps 548: return(0);
549: }
550:
551: dsec = arch = NULL;
552: dform = FORM_NONE;
553:
554: while (NULL != (ff = fts_read(f))) {
555: path = ff->fts_path + 2;
556: /*
1.83 schwarze 557: * If we're a regular file, add an mlink by using the
1.50 kristaps 558: * stored directory data and handling the filename.
559: */
560: if (FTS_F == ff->fts_info) {
1.60 schwarze 561: if (0 == strcmp(path, MANDOC_DB))
562: continue;
1.50 kristaps 563: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 564: if (warnings)
1.59 schwarze 565: say(path, "Extraneous file");
1.50 kristaps 566: continue;
1.84 schwarze 567: } else if (NULL == (fsec =
1.60 schwarze 568: strrchr(ff->fts_name, '.'))) {
569: if ( ! use_all) {
1.56 schwarze 570: if (warnings)
1.60 schwarze 571: say(path,
572: "No filename suffix");
1.50 kristaps 573: continue;
574: }
1.84 schwarze 575: } else if (0 == strcmp(++fsec, "html")) {
1.60 schwarze 576: if (warnings)
577: say(path, "Skip html");
578: continue;
1.84 schwarze 579: } else if (0 == strcmp(fsec, "gz")) {
1.60 schwarze 580: if (warnings)
581: say(path, "Skip gz");
582: continue;
1.84 schwarze 583: } else if (0 == strcmp(fsec, "ps")) {
1.60 schwarze 584: if (warnings)
585: say(path, "Skip ps");
586: continue;
1.84 schwarze 587: } else if (0 == strcmp(fsec, "pdf")) {
1.60 schwarze 588: if (warnings)
589: say(path, "Skip pdf");
590: continue;
591: } else if ( ! use_all &&
1.84 schwarze 592: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
593: (FORM_CAT == dform && strcmp(fsec, "0")))) {
1.60 schwarze 594: if (warnings)
595: say(path, "Wrong filename suffix");
596: continue;
1.66 schwarze 597: } else
1.84 schwarze 598: fsec[-1] = '\0';
1.94 schwarze 599:
1.84 schwarze 600: mlink = mandoc_calloc(1, sizeof(struct mlink));
601: strlcpy(mlink->file, path, sizeof(mlink->file));
602: mlink->dform = dform;
1.94 schwarze 603: mlink->dsec = dsec;
604: mlink->arch = arch;
605: mlink->name = ff->fts_name;
606: mlink->fsec = fsec;
1.84 schwarze 607: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 608: continue;
1.84 schwarze 609: } else if (FTS_D != ff->fts_info &&
1.60 schwarze 610: FTS_DP != ff->fts_info) {
611: if (warnings)
612: say(path, "Not a regular file");
1.50 kristaps 613: continue;
1.60 schwarze 614: }
1.1 kristaps 615:
1.50 kristaps 616: switch (ff->fts_level) {
617: case (0):
618: /* Ignore the root directory. */
619: break;
620: case (1):
621: /*
622: * This might contain manX/ or catX/.
623: * Try to infer this from the name.
624: * If we're not in use_all, enforce it.
625: */
626: cp = ff->fts_name;
627: if (FTS_DP == ff->fts_info)
628: break;
1.1 kristaps 629:
1.50 kristaps 630: if (0 == strncmp(cp, "man", 3)) {
631: dform = FORM_SRC;
1.66 schwarze 632: dsec = cp + 3;
1.50 kristaps 633: } else if (0 == strncmp(cp, "cat", 3)) {
634: dform = FORM_CAT;
1.66 schwarze 635: dsec = cp + 3;
1.94 schwarze 636: } else {
637: dform = FORM_NONE;
638: dsec = NULL;
1.50 kristaps 639: }
1.1 kristaps 640:
1.50 kristaps 641: if (NULL != dsec || use_all)
642: break;
1.1 kristaps 643:
1.56 schwarze 644: if (warnings)
1.59 schwarze 645: say(path, "Unknown directory part");
1.50 kristaps 646: fts_set(f, ff, FTS_SKIP);
647: break;
648: case (2):
649: /*
650: * Possibly our architecture.
651: * If we're descending, keep tabs on it.
652: */
653: if (FTS_DP != ff->fts_info && NULL != dsec)
1.66 schwarze 654: arch = ff->fts_name;
1.94 schwarze 655: else
656: arch = NULL;
1.50 kristaps 657: break;
658: default:
659: if (FTS_DP == ff->fts_info || use_all)
660: break;
1.56 schwarze 661: if (warnings)
1.59 schwarze 662: say(path, "Extraneous directory part");
1.50 kristaps 663: fts_set(f, ff, FTS_SKIP);
664: break;
1.5 kristaps 665: }
1.50 kristaps 666: }
667:
668: fts_close(f);
669: return(1);
670: }
1.5 kristaps 671:
1.50 kristaps 672: /*
1.89 schwarze 673: * Add a file to the mlinks table.
1.50 kristaps 674: * Do not verify that it's a "valid" looking manpage (we'll do that
675: * later).
676: *
677: * Try to infer the manual section, architecture, and page name from the
678: * path, assuming it looks like
679: *
680: * [./]man*[/<arch>]/<name>.<section>
681: * or
682: * [./]cat<section>[/<arch>]/<name>.0
683: *
684: * See treescan() for the fts(3) version of this.
685: */
686: static void
1.59 schwarze 687: filescan(const char *file)
1.50 kristaps 688: {
1.59 schwarze 689: char buf[PATH_MAX];
1.84 schwarze 690: struct stat st;
691: struct mlink *mlink;
1.59 schwarze 692: char *p, *start;
1.5 kristaps 693:
1.50 kristaps 694: assert(use_all);
1.5 kristaps 695:
1.50 kristaps 696: if (0 == strncmp(file, "./", 2))
697: file += 2;
1.5 kristaps 698:
1.59 schwarze 699: if (NULL == realpath(file, buf)) {
700: exitcode = (int)MANDOCLEVEL_BADARG;
701: say(file, NULL);
702: return;
1.106 ! schwarze 703: }
! 704:
! 705: if (strstr(buf, basedir) == buf)
! 706: start = buf + strlen(basedir) + 1;
! 707: else if (OP_TEST == op)
! 708: start = buf;
! 709: else {
1.59 schwarze 710: exitcode = (int)MANDOCLEVEL_BADARG;
711: say("", "%s: outside base directory", buf);
712: return;
1.106 ! schwarze 713: }
! 714:
! 715: if (-1 == stat(buf, &st)) {
1.59 schwarze 716: exitcode = (int)MANDOCLEVEL_BADARG;
717: say(file, NULL);
1.50 kristaps 718: return;
719: } else if ( ! (S_IFREG & st.st_mode)) {
1.59 schwarze 720: exitcode = (int)MANDOCLEVEL_BADARG;
721: say(file, "Not a regular file");
1.50 kristaps 722: return;
723: }
1.106 ! schwarze 724:
1.84 schwarze 725: mlink = mandoc_calloc(1, sizeof(struct mlink));
726: strlcpy(mlink->file, start, sizeof(mlink->file));
1.17 schwarze 727:
1.50 kristaps 728: /*
729: * First try to guess our directory structure.
730: * If we find a separator, try to look for man* or cat*.
731: * If we find one of these and what's underneath is a directory,
732: * assume it's an architecture.
733: */
734: if (NULL != (p = strchr(start, '/'))) {
735: *p++ = '\0';
736: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 737: mlink->dform = FORM_SRC;
1.94 schwarze 738: mlink->dsec = start + 3;
1.50 kristaps 739: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 740: mlink->dform = FORM_CAT;
1.94 schwarze 741: mlink->dsec = start + 3;
1.17 schwarze 742: }
1.5 kristaps 743:
1.50 kristaps 744: start = p;
1.84 schwarze 745: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 746: *p++ = '\0';
1.94 schwarze 747: mlink->arch = start;
1.50 kristaps 748: start = p;
1.84 schwarze 749: }
1.50 kristaps 750: }
751:
752: /*
753: * Now check the file suffix.
754: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
755: */
756: p = strrchr(start, '\0');
757: while (p-- > start && '/' != *p && '.' != *p)
758: /* Loop. */ ;
759:
760: if ('.' == *p) {
761: *p++ = '\0';
1.94 schwarze 762: mlink->fsec = p;
1.5 kristaps 763: }
764:
1.10 kristaps 765: /*
1.50 kristaps 766: * Now try to parse the name.
767: * Use the filename portion of the path.
1.10 kristaps 768: */
1.84 schwarze 769: mlink->name = start;
1.50 kristaps 770: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 771: mlink->name = p + 1;
1.50 kristaps 772: *p = '\0';
1.84 schwarze 773: }
774: mlink_add(mlink, &st);
1.50 kristaps 775: }
1.5 kristaps 776:
1.50 kristaps 777: static void
1.84 schwarze 778: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 779: {
1.83 schwarze 780: struct inodev inodev;
1.78 schwarze 781: struct mpage *mpage;
1.83 schwarze 782: unsigned int slot;
1.50 kristaps 783:
1.84 schwarze 784: assert(NULL != mlink->file);
1.50 kristaps 785:
1.94 schwarze 786: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
787: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
788: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
789: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 790:
791: if ('0' == *mlink->fsec) {
792: free(mlink->fsec);
793: mlink->fsec = mandoc_strdup(mlink->dsec);
794: mlink->fform = FORM_CAT;
795: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
796: mlink->fform = FORM_SRC;
1.74 schwarze 797: else
1.84 schwarze 798: mlink->fform = FORM_NONE;
1.82 schwarze 799:
1.83 schwarze 800: slot = ohash_qlookup(&mlinks, mlink->file);
801: assert(NULL == ohash_find(&mlinks, slot));
802: ohash_insert(&mlinks, slot, mlink);
803:
804: inodev.st_ino = st->st_ino;
805: inodev.st_dev = st->st_dev;
806: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
807: sizeof(struct inodev), inodev.st_ino);
808: mpage = ohash_find(&mpages, slot);
809: if (NULL == mpage) {
810: mpage = mandoc_calloc(1, sizeof(struct mpage));
811: mpage->inodev.st_ino = inodev.st_ino;
812: mpage->inodev.st_dev = inodev.st_dev;
813: ohash_insert(&mpages, slot, mpage);
814: } else
1.85 schwarze 815: mlink->next = mpage->mlinks;
1.82 schwarze 816: mpage->mlinks = mlink;
817: }
818:
819: static void
820: mlink_free(struct mlink *mlink)
821: {
822:
823: free(mlink->dsec);
824: free(mlink->arch);
825: free(mlink->name);
826: free(mlink->fsec);
827: free(mlink);
1.50 kristaps 828: }
1.3 kristaps 829:
1.50 kristaps 830: static void
1.78 schwarze 831: mpages_free(void)
1.50 kristaps 832: {
1.78 schwarze 833: struct mpage *mpage;
1.82 schwarze 834: struct mlink *mlink;
1.80 schwarze 835: unsigned int slot;
1.3 kristaps 836:
1.80 schwarze 837: mpage = ohash_first(&mpages, &slot);
838: while (NULL != mpage) {
1.82 schwarze 839: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 840: mpage->mlinks = mlink->next;
1.82 schwarze 841: mlink_free(mlink);
842: }
1.78 schwarze 843: free(mpage->sec);
844: free(mpage->arch);
1.82 schwarze 845: free(mpage->title);
846: free(mpage->desc);
1.78 schwarze 847: free(mpage);
1.80 schwarze 848: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 849: }
850: }
1.38 schwarze 851:
1.50 kristaps 852: /*
1.89 schwarze 853: * For each mlink to the mpage, check whether the path looks like
854: * it is formatted, and if it does, check whether a source manual
855: * exists by the same name, ignoring the suffix.
856: * If both conditions hold, drop the mlink.
857: */
858: static void
859: mlinks_undupe(struct mpage *mpage)
860: {
861: char buf[PATH_MAX];
862: struct mlink **prev;
863: struct mlink *mlink;
864: char *bufp;
865:
866: mpage->form = FORM_CAT;
1.90 schwarze 867: prev = &mpage->mlinks;
868: while (NULL != (mlink = *prev)) {
1.89 schwarze 869: if (FORM_CAT != mlink->dform) {
870: mpage->form = FORM_NONE;
1.90 schwarze 871: goto nextlink;
1.89 schwarze 872: }
873: if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
874: if (warnings)
875: say(mlink->file, "Filename too long");
1.90 schwarze 876: goto nextlink;
1.89 schwarze 877: }
878: bufp = strstr(buf, "cat");
879: assert(NULL != bufp);
880: memcpy(bufp, "man", 3);
881: if (NULL != (bufp = strrchr(buf, '.')))
882: *++bufp = '\0';
883: strlcat(buf, mlink->dsec, PATH_MAX);
884: if (NULL == ohash_find(&mlinks,
885: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 886: goto nextlink;
1.89 schwarze 887: if (warnings)
888: say(mlink->file, "Man source exists: %s", buf);
889: if (use_all)
1.90 schwarze 890: goto nextlink;
1.89 schwarze 891: *prev = mlink->next;
892: mlink_free(mlink);
1.90 schwarze 893: continue;
894: nextlink:
895: prev = &(*prev)->next;
1.89 schwarze 896: }
897: }
898:
1.93 schwarze 899: static int
900: mlink_check(struct mpage *mpage, struct mlink *mlink)
901: {
902: int match;
903:
904: match = 1;
905:
906: /*
907: * Check whether the manual section given in a file
908: * agrees with the directory where the file is located.
909: * Some manuals have suffixes like (3p) on their
910: * section number either inside the file or in the
911: * directory name, some are linked into more than one
912: * section, like encrypt(1) = makekey(8).
913: */
914:
915: if (FORM_SRC == mpage->form &&
916: strcasecmp(mpage->sec, mlink->dsec)) {
917: match = 0;
918: say(mlink->file, "Section \"%s\" manual in %s directory",
919: mpage->sec, mlink->dsec);
920: }
921:
922: /*
923: * Manual page directories exist for each kernel
924: * architecture as returned by machine(1).
925: * However, many manuals only depend on the
926: * application architecture as returned by arch(1).
927: * For example, some (2/ARM) manuals are shared
928: * across the "armish" and "zaurus" kernel
929: * architectures.
930: * A few manuals are even shared across completely
931: * different architectures, for example fdformat(1)
932: * on amd64, i386, sparc, and sparc64.
933: */
934:
935: if (strcasecmp(mpage->arch, mlink->arch)) {
936: match = 0;
937: say(mlink->file, "Architecture \"%s\" manual in "
938: "\"%s\" directory", mpage->arch, mlink->arch);
939: }
940:
941: if (strcasecmp(mpage->title, mlink->name))
942: match = 0;
943:
944: return(match);
945: }
946:
1.89 schwarze 947: /*
1.80 schwarze 948: * Run through the files in the global vector "mpages"
949: * and add them to the database specified in "basedir".
1.50 kristaps 950: *
951: * This handles the parsing scheme itself, using the cues of directory
952: * and filename to determine whether the file is parsable or not.
953: */
1.59 schwarze 954: static void
1.101 schwarze 955: mpages_merge(struct mchars *mc, struct mparse *mp)
1.50 kristaps 956: {
1.101 schwarze 957: struct ohash_info str_info;
1.78 schwarze 958: struct mpage *mpage;
1.93 schwarze 959: struct mlink *mlink;
1.69 schwarze 960: struct mdoc *mdoc;
961: struct man *man;
1.82 schwarze 962: const char *cp;
1.69 schwarze 963: int match;
1.101 schwarze 964: unsigned int pslot;
1.69 schwarze 965: enum mandoclevel lvl;
966:
1.71 schwarze 967: str_info.alloc = hash_alloc;
968: str_info.halloc = hash_halloc;
969: str_info.hfree = hash_free;
970: str_info.key_offset = offsetof(struct str, key);
971:
1.80 schwarze 972: mpage = ohash_first(&mpages, &pslot);
973: while (NULL != mpage) {
1.89 schwarze 974: mlinks_undupe(mpage);
975: if (NULL == mpage->mlinks) {
976: mpage = ohash_next(&mpages, &pslot);
977: continue;
1.50 kristaps 978: }
979:
1.71 schwarze 980: ohash_init(&strings, 6, &str_info);
1.50 kristaps 981: mparse_reset(mp);
982: mdoc = NULL;
983: man = NULL;
1.14 schwarze 984:
985: /*
1.33 schwarze 986: * Try interpreting the file as mdoc(7) or man(7)
987: * source code, unless it is already known to be
988: * formatted. Fall back to formatted mode.
1.14 schwarze 989: */
1.82 schwarze 990: if (FORM_CAT != mpage->mlinks->dform ||
1.84 schwarze 991: FORM_CAT != mpage->mlinks->fform) {
1.82 schwarze 992: lvl = mparse_readfd(mp, -1, mpage->mlinks->file);
1.50 kristaps 993: if (lvl < MANDOCLEVEL_FATAL)
994: mparse_result(mp, &mdoc, &man);
1.76 schwarze 995: }
1.14 schwarze 996:
997: if (NULL != mdoc) {
1.82 schwarze 998: mpage->form = FORM_SRC;
999: mpage->sec =
1000: mandoc_strdup(mdoc_meta(mdoc)->msec);
1001: mpage->arch = mdoc_meta(mdoc)->arch;
1.84 schwarze 1002: mpage->arch = mandoc_strdup(
1003: NULL == mpage->arch ? "" : mpage->arch);
1.82 schwarze 1004: mpage->title =
1005: mandoc_strdup(mdoc_meta(mdoc)->title);
1.14 schwarze 1006: } else if (NULL != man) {
1.82 schwarze 1007: mpage->form = FORM_SRC;
1008: mpage->sec =
1009: mandoc_strdup(man_meta(man)->msec);
1010: mpage->arch =
1011: mandoc_strdup(mpage->mlinks->arch);
1012: mpage->title =
1013: mandoc_strdup(man_meta(man)->title);
1.76 schwarze 1014: } else {
1.82 schwarze 1015: mpage->form = FORM_CAT;
1016: mpage->sec =
1017: mandoc_strdup(mpage->mlinks->dsec);
1018: mpage->arch =
1019: mandoc_strdup(mpage->mlinks->arch);
1020: mpage->title =
1021: mandoc_strdup(mpage->mlinks->name);
1.76 schwarze 1022: }
1.97 schwarze 1023: putkey(mpage, mpage->sec, TYPE_sec);
1.98 schwarze 1024: putkey(mpage, '\0' == *mpage->arch ?
1025: "any" : mpage->arch, TYPE_arch);
1.50 kristaps 1026:
1.97 schwarze 1027: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1028: if ('\0' != *mlink->dsec)
1029: putkey(mpage, mlink->dsec, TYPE_sec);
1030: if ('\0' != *mlink->fsec)
1031: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1032: putkey(mpage, '\0' == *mlink->arch ?
1033: "any" : mlink->arch, TYPE_arch);
1.93 schwarze 1034: putkey(mpage, mlink->name, TYPE_Nm);
1.97 schwarze 1035: }
1.12 schwarze 1036:
1.93 schwarze 1037: if (warnings && !use_all) {
1.69 schwarze 1038: match = 0;
1.93 schwarze 1039: for (mlink = mpage->mlinks; mlink;
1040: mlink = mlink->next)
1041: if (mlink_check(mpage, mlink))
1042: match = 1;
1043: } else
1044: match = 1;
1.12 schwarze 1045:
1.50 kristaps 1046: if (NULL != mdoc) {
1047: if (NULL != (cp = mdoc_meta(mdoc)->name))
1.78 schwarze 1048: putkey(mpage, cp, TYPE_Nm);
1049: assert(NULL == mpage->desc);
1050: parse_mdoc(mpage, mdoc_node(mdoc));
1051: putkey(mpage, NULL != mpage->desc ?
1.82 schwarze 1052: mpage->desc : mpage->mlinks->name, TYPE_Nd);
1.50 kristaps 1053: } else if (NULL != man)
1.78 schwarze 1054: parse_man(mpage, man_node(man));
1.50 kristaps 1055: else
1.78 schwarze 1056: parse_cat(mpage);
1.44 kristaps 1057:
1.105 schwarze 1058: dbadd(mpage, mc);
1.66 schwarze 1059: ohash_delete(&strings);
1.80 schwarze 1060: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1061: }
1062: }
1.12 schwarze 1063:
1.50 kristaps 1064: static void
1.78 schwarze 1065: parse_cat(struct mpage *mpage)
1.50 kristaps 1066: {
1067: FILE *stream;
1068: char *line, *p, *title;
1069: size_t len, plen, titlesz;
1.12 schwarze 1070:
1.82 schwarze 1071: if (NULL == (stream = fopen(mpage->mlinks->file, "r"))) {
1.56 schwarze 1072: if (warnings)
1.82 schwarze 1073: say(mpage->mlinks->file, NULL);
1.50 kristaps 1074: return;
1075: }
1.1 kristaps 1076:
1.50 kristaps 1077: /* Skip to first blank line. */
1.1 kristaps 1078:
1.50 kristaps 1079: while (NULL != (line = fgetln(stream, &len)))
1080: if ('\n' == *line)
1081: break;
1.1 kristaps 1082:
1.50 kristaps 1083: /*
1084: * Assume the first line that is not indented
1085: * is the first section header. Skip to it.
1086: */
1.1 kristaps 1087:
1.50 kristaps 1088: while (NULL != (line = fgetln(stream, &len)))
1089: if ('\n' != *line && ' ' != *line)
1090: break;
1091:
1092: /*
1093: * Read up until the next section into a buffer.
1094: * Strip the leading and trailing newline from each read line,
1095: * appending a trailing space.
1096: * Ignore empty (whitespace-only) lines.
1097: */
1.1 kristaps 1098:
1.50 kristaps 1099: titlesz = 0;
1100: title = NULL;
1.38 schwarze 1101:
1.50 kristaps 1102: while (NULL != (line = fgetln(stream, &len))) {
1103: if (' ' != *line || '\n' != line[len - 1])
1104: break;
1105: while (len > 0 && isspace((unsigned char)*line)) {
1106: line++;
1107: len--;
1108: }
1109: if (1 == len)
1.38 schwarze 1110: continue;
1.50 kristaps 1111: title = mandoc_realloc(title, titlesz + len);
1112: memcpy(title + titlesz, line, len);
1113: titlesz += len;
1114: title[titlesz - 1] = ' ';
1115: }
1.38 schwarze 1116:
1.50 kristaps 1117: /*
1118: * If no page content can be found, or the input line
1119: * is already the next section header, or there is no
1120: * trailing newline, reuse the page title as the page
1121: * description.
1122: */
1.44 kristaps 1123:
1.50 kristaps 1124: if (NULL == title || '\0' == *title) {
1.56 schwarze 1125: if (warnings)
1.82 schwarze 1126: say(mpage->mlinks->file,
1127: "Cannot find NAME section");
1.78 schwarze 1128: assert(NULL == mpage->desc);
1.82 schwarze 1129: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1130: putkey(mpage, mpage->mlinks->name, TYPE_Nd);
1.50 kristaps 1131: fclose(stream);
1132: free(title);
1133: return;
1134: }
1.1 kristaps 1135:
1.50 kristaps 1136: title = mandoc_realloc(title, titlesz + 1);
1137: title[titlesz] = '\0';
1.33 schwarze 1138:
1.50 kristaps 1139: /*
1140: * Skip to the first dash.
1141: * Use the remaining line as the description (no more than 70
1142: * bytes).
1143: */
1.33 schwarze 1144:
1.50 kristaps 1145: if (NULL != (p = strstr(title, "- "))) {
1146: for (p += 2; ' ' == *p || '\b' == *p; p++)
1147: /* Skip to next word. */ ;
1148: } else {
1.56 schwarze 1149: if (warnings)
1.82 schwarze 1150: say(mpage->mlinks->file,
1151: "No dash in title line");
1.50 kristaps 1152: p = title;
1153: }
1.38 schwarze 1154:
1.50 kristaps 1155: plen = strlen(p);
1.1 kristaps 1156:
1.50 kristaps 1157: /* Strip backspace-encoding from line. */
1.1 kristaps 1158:
1.50 kristaps 1159: while (NULL != (line = memchr(p, '\b', plen))) {
1160: len = line - p;
1161: if (0 == len) {
1162: memmove(line, line + 1, plen--);
1163: continue;
1164: }
1165: memmove(line - 1, line + 1, plen - len);
1166: plen -= 2;
1167: }
1.1 kristaps 1168:
1.78 schwarze 1169: assert(NULL == mpage->desc);
1170: mpage->desc = mandoc_strdup(p);
1171: putkey(mpage, mpage->desc, TYPE_Nd);
1.50 kristaps 1172: fclose(stream);
1173: free(title);
1174: }
1.1 kristaps 1175:
1.50 kristaps 1176: /*
1177: * Put a type/word pair into the word database for this particular file.
1178: */
1179: static void
1.78 schwarze 1180: putkey(const struct mpage *mpage, const char *value, uint64_t type)
1.50 kristaps 1181: {
1.18 kristaps 1182:
1.50 kristaps 1183: assert(NULL != value);
1.78 schwarze 1184: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1185: }
1186:
1187: /*
1.50 kristaps 1188: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1189: */
1190: static void
1.78 schwarze 1191: putmdockey(const struct mpage *mpage,
1192: const struct mdoc_node *n, uint64_t m)
1.3 kristaps 1193: {
1.18 kristaps 1194:
1.50 kristaps 1195: for ( ; NULL != n; n = n->next) {
1196: if (NULL != n->child)
1.78 schwarze 1197: putmdockey(mpage, n->child, m);
1.50 kristaps 1198: if (MDOC_TEXT == n->type)
1.78 schwarze 1199: putkey(mpage, n->string, m);
1.50 kristaps 1200: }
1201: }
1.18 kristaps 1202:
1.61 schwarze 1203: static void
1.78 schwarze 1204: parse_man(struct mpage *mpage, const struct man_node *n)
1.50 kristaps 1205: {
1206: const struct man_node *head, *body;
1207: char *start, *sv, *title;
1208: char byte;
1209: size_t sz, titlesz;
1.18 kristaps 1210:
1.50 kristaps 1211: if (NULL == n)
1.61 schwarze 1212: return;
1.18 kristaps 1213:
1.50 kristaps 1214: /*
1215: * We're only searching for one thing: the first text child in
1216: * the BODY of a NAME section. Since we don't keep track of
1217: * sections in -man, run some hoops to find out whether we're in
1218: * the correct section or not.
1219: */
1.18 kristaps 1220:
1.50 kristaps 1221: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1222: body = n;
1223: assert(body->parent);
1224: if (NULL != (head = body->parent->head) &&
1225: 1 == head->nchild &&
1226: NULL != (head = (head->child)) &&
1227: MAN_TEXT == head->type &&
1228: 0 == strcmp(head->string, "NAME") &&
1229: NULL != (body = body->child) &&
1230: MAN_TEXT == body->type) {
1.3 kristaps 1231:
1.50 kristaps 1232: title = NULL;
1233: titlesz = 0;
1.3 kristaps 1234:
1.50 kristaps 1235: /*
1236: * Suck the entire NAME section into memory.
1237: * Yes, we might run away.
1238: * But too many manuals have big, spread-out
1239: * NAME sections over many lines.
1240: */
1.3 kristaps 1241:
1.50 kristaps 1242: for ( ; NULL != body; body = body->next) {
1243: if (MAN_TEXT != body->type)
1244: break;
1245: if (0 == (sz = strlen(body->string)))
1246: continue;
1247: title = mandoc_realloc
1248: (title, titlesz + sz + 1);
1249: memcpy(title + titlesz, body->string, sz);
1250: titlesz += sz + 1;
1251: title[titlesz - 1] = ' ';
1252: }
1253: if (NULL == title)
1.61 schwarze 1254: return;
1.18 kristaps 1255:
1.50 kristaps 1256: title = mandoc_realloc(title, titlesz + 1);
1257: title[titlesz] = '\0';
1.18 kristaps 1258:
1.50 kristaps 1259: /* Skip leading space. */
1.18 kristaps 1260:
1.50 kristaps 1261: sv = title;
1262: while (isspace((unsigned char)*sv))
1263: sv++;
1.18 kristaps 1264:
1.50 kristaps 1265: if (0 == (sz = strlen(sv))) {
1266: free(title);
1.61 schwarze 1267: return;
1.50 kristaps 1268: }
1.1 kristaps 1269:
1.50 kristaps 1270: /* Erase trailing space. */
1.1 kristaps 1271:
1.50 kristaps 1272: start = &sv[sz - 1];
1273: while (start > sv && isspace((unsigned char)*start))
1274: *start-- = '\0';
1.1 kristaps 1275:
1.50 kristaps 1276: if (start == sv) {
1277: free(title);
1.61 schwarze 1278: return;
1.50 kristaps 1279: }
1.1 kristaps 1280:
1.50 kristaps 1281: start = sv;
1.18 kristaps 1282:
1.50 kristaps 1283: /*
1284: * Go through a special heuristic dance here.
1285: * Conventionally, one or more manual names are
1286: * comma-specified prior to a whitespace, then a
1287: * dash, then a description. Try to puzzle out
1288: * the name parts here.
1289: */
1.18 kristaps 1290:
1.50 kristaps 1291: for ( ;; ) {
1292: sz = strcspn(start, " ,");
1293: if ('\0' == start[sz])
1294: break;
1.1 kristaps 1295:
1.50 kristaps 1296: byte = start[sz];
1297: start[sz] = '\0';
1.1 kristaps 1298:
1.78 schwarze 1299: putkey(mpage, start, TYPE_Nm);
1.1 kristaps 1300:
1.50 kristaps 1301: if (' ' == byte) {
1302: start += sz + 1;
1303: break;
1304: }
1.1 kristaps 1305:
1.50 kristaps 1306: assert(',' == byte);
1307: start += sz + 1;
1308: while (' ' == *start)
1309: start++;
1310: }
1.1 kristaps 1311:
1.50 kristaps 1312: if (sv == start) {
1.78 schwarze 1313: putkey(mpage, start, TYPE_Nm);
1.50 kristaps 1314: free(title);
1.61 schwarze 1315: return;
1.50 kristaps 1316: }
1.1 kristaps 1317:
1.50 kristaps 1318: while (isspace((unsigned char)*start))
1319: start++;
1.1 kristaps 1320:
1.50 kristaps 1321: if (0 == strncmp(start, "-", 1))
1322: start += 1;
1323: else if (0 == strncmp(start, "\\-\\-", 4))
1324: start += 4;
1325: else if (0 == strncmp(start, "\\-", 2))
1326: start += 2;
1327: else if (0 == strncmp(start, "\\(en", 4))
1328: start += 4;
1329: else if (0 == strncmp(start, "\\(em", 4))
1330: start += 4;
1.1 kristaps 1331:
1.50 kristaps 1332: while (' ' == *start)
1333: start++;
1.1 kristaps 1334:
1.78 schwarze 1335: assert(NULL == mpage->desc);
1336: mpage->desc = mandoc_strdup(start);
1337: putkey(mpage, mpage->desc, TYPE_Nd);
1.50 kristaps 1338: free(title);
1.61 schwarze 1339: return;
1.50 kristaps 1340: }
1341: }
1.1 kristaps 1342:
1.77 schwarze 1343: for (n = n->child; n; n = n->next) {
1.78 schwarze 1344: if (NULL != mpage->desc)
1.77 schwarze 1345: break;
1.78 schwarze 1346: parse_man(mpage, n);
1.77 schwarze 1347: }
1.1 kristaps 1348: }
1349:
1350: static void
1.78 schwarze 1351: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1352: {
1353:
1.50 kristaps 1354: assert(NULL != n);
1355: for (n = n->child; NULL != n; n = n->next) {
1356: switch (n->type) {
1357: case (MDOC_ELEM):
1358: /* FALLTHROUGH */
1359: case (MDOC_BLOCK):
1360: /* FALLTHROUGH */
1361: case (MDOC_HEAD):
1362: /* FALLTHROUGH */
1363: case (MDOC_BODY):
1364: /* FALLTHROUGH */
1365: case (MDOC_TAIL):
1366: if (NULL != mdocs[n->tok].fp)
1.78 schwarze 1367: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1.50 kristaps 1368: break;
1.68 schwarze 1369: if (mdocs[n->tok].mask)
1.78 schwarze 1370: putmdockey(mpage, n->child,
1371: mdocs[n->tok].mask);
1.50 kristaps 1372: break;
1373: default:
1374: assert(MDOC_ROOT != n->type);
1375: continue;
1376: }
1377: if (NULL != n->child)
1.78 schwarze 1378: parse_mdoc(mpage, n);
1.1 kristaps 1379: }
1380: }
1381:
1.25 schwarze 1382: static int
1.78 schwarze 1383: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1384: {
1385: const char *start, *end;
1386: size_t sz;
1.25 schwarze 1387:
1.50 kristaps 1388: if (SEC_SYNOPSIS != n->sec ||
1389: NULL == (n = n->child) ||
1390: MDOC_TEXT != n->type)
1.25 schwarze 1391: return(0);
1.1 kristaps 1392:
1393: /*
1394: * Only consider those `Fd' macro fields that begin with an
1395: * "inclusion" token (versus, e.g., #define).
1396: */
1.50 kristaps 1397:
1.1 kristaps 1398: if (strcmp("#include", n->string))
1.25 schwarze 1399: return(0);
1.1 kristaps 1400:
1401: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.25 schwarze 1402: return(0);
1.1 kristaps 1403:
1404: /*
1405: * Strip away the enclosing angle brackets and make sure we're
1406: * not zero-length.
1407: */
1408:
1409: start = n->string;
1410: if ('<' == *start || '"' == *start)
1411: start++;
1412:
1413: if (0 == (sz = strlen(start)))
1.25 schwarze 1414: return(0);
1.1 kristaps 1415:
1416: end = &start[(int)sz - 1];
1417: if ('>' == *end || '"' == *end)
1418: end--;
1419:
1.50 kristaps 1420: if (end > start)
1.78 schwarze 1421: putkeys(mpage, start, end - start + 1, TYPE_In);
1.92 schwarze 1422: return(0);
1.1 kristaps 1423: }
1424:
1.25 schwarze 1425: static int
1.78 schwarze 1426: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1427: {
1428: const char *cp;
1429:
1.50 kristaps 1430: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.25 schwarze 1431: return(0);
1432:
1.50 kristaps 1433: /*
1434: * Parse: .Fn "struct type *name" "char *arg".
1435: * First strip away pointer symbol.
1436: * Then store the function name, then type.
1437: * Finally, store the arguments.
1438: */
1.1 kristaps 1439:
1.50 kristaps 1440: if (NULL == (cp = strrchr(n->string, ' ')))
1441: cp = n->string;
1.1 kristaps 1442:
1443: while ('*' == *cp)
1444: cp++;
1445:
1.78 schwarze 1446: putkey(mpage, cp, TYPE_Fn);
1.25 schwarze 1447:
1.50 kristaps 1448: if (n->string < cp)
1.78 schwarze 1449: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.25 schwarze 1450:
1.50 kristaps 1451: for (n = n->next; NULL != n; n = n->next)
1452: if (MDOC_TEXT == n->type)
1.78 schwarze 1453: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1454:
1455: return(0);
1.1 kristaps 1456: }
1457:
1.25 schwarze 1458: static int
1.78 schwarze 1459: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1460: {
1.67 schwarze 1461: char *cp;
1.1 kristaps 1462:
1463: if (NULL == (n = n->child))
1.25 schwarze 1464: return(0);
1.1 kristaps 1465:
1.67 schwarze 1466: if (NULL == n->next) {
1.78 schwarze 1467: putkey(mpage, n->string, TYPE_Xr);
1.67 schwarze 1468: return(0);
1469: }
1470:
1471: if (-1 == asprintf(&cp, "%s(%s)", n->string, n->next->string)) {
1472: perror(NULL);
1473: exit((int)MANDOCLEVEL_SYSERR);
1474: }
1.78 schwarze 1475: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1476: free(cp);
1477: return(0);
1.1 kristaps 1478: }
1479:
1.25 schwarze 1480: static int
1.78 schwarze 1481: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1482: {
1.50 kristaps 1483: size_t sz;
1.1 kristaps 1484:
1485: if (MDOC_BODY != n->type)
1.25 schwarze 1486: return(0);
1.1 kristaps 1487:
1.50 kristaps 1488: /*
1489: * Special-case the `Nd' because we need to put the description
1490: * into the document table.
1491: */
1492:
1493: for (n = n->child; NULL != n; n = n->next) {
1494: if (MDOC_TEXT == n->type) {
1.78 schwarze 1495: if (NULL != mpage->desc) {
1496: sz = strlen(mpage->desc) +
1.66 schwarze 1497: strlen(n->string) + 2;
1.78 schwarze 1498: mpage->desc = mandoc_realloc(
1499: mpage->desc, sz);
1500: strlcat(mpage->desc, " ", sz);
1501: strlcat(mpage->desc, n->string, sz);
1.66 schwarze 1502: } else
1.78 schwarze 1503: mpage->desc = mandoc_strdup(n->string);
1.50 kristaps 1504: }
1505: if (NULL != n->child)
1.78 schwarze 1506: parse_mdoc_Nd(mpage, n);
1.50 kristaps 1507: }
1.25 schwarze 1508: return(1);
1.1 kristaps 1509: }
1510:
1.25 schwarze 1511: static int
1.78 schwarze 1512: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1513: {
1514:
1.92 schwarze 1515: return(SEC_NAME == n->sec ||
1516: (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type));
1.1 kristaps 1517: }
1518:
1.25 schwarze 1519: static int
1.78 schwarze 1520: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1521: {
1522:
1.25 schwarze 1523: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 kristaps 1524: }
1525:
1.50 kristaps 1526: static int
1.78 schwarze 1527: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1528: {
1529:
1.50 kristaps 1530: return(MDOC_HEAD == n->type);
1531: }
1.1 kristaps 1532:
1.50 kristaps 1533: static int
1.78 schwarze 1534: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1.50 kristaps 1535: {
1.1 kristaps 1536:
1.50 kristaps 1537: return(MDOC_BODY == n->type);
1.1 kristaps 1538: }
1539:
1.50 kristaps 1540: /*
1.66 schwarze 1541: * Add a string to the hash table for the current manual.
1542: * Each string has a bitmask telling which macros it belongs to.
1543: * When we finish the manual, we'll dump the table.
1.50 kristaps 1544: */
1545: static void
1.78 schwarze 1546: putkeys(const struct mpage *mpage,
1547: const char *cp, size_t sz, uint64_t v)
1.50 kristaps 1548: {
1549: struct str *s;
1.65 schwarze 1550: unsigned int slot;
1.50 kristaps 1551: const char *end;
1.25 schwarze 1552:
1.50 kristaps 1553: if (0 == sz)
1554: return;
1.25 schwarze 1555:
1.65 schwarze 1556: end = cp + sz;
1557: slot = ohash_qlookupi(&strings, cp, &end);
1558: s = ohash_find(&strings, slot);
1.25 schwarze 1559:
1.78 schwarze 1560: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1561: s->mask |= v;
1562: return;
1563: } else if (NULL == s) {
1.51 kristaps 1564: s = mandoc_calloc(sizeof(struct str) + sz + 1, 1);
1.50 kristaps 1565: memcpy(s->key, cp, sz);
1.65 schwarze 1566: ohash_insert(&strings, slot, s);
1.1 kristaps 1567: }
1.78 schwarze 1568: s->mpage = mpage;
1.50 kristaps 1569: s->mask = v;
1.1 kristaps 1570: }
1571:
1.50 kristaps 1572: /*
1573: * Take a Unicode codepoint and produce its UTF-8 encoding.
1574: * This isn't the best way to do this, but it works.
1575: * The magic numbers are from the UTF-8 packaging.
1576: * They're not as scary as they seem: read the UTF-8 spec for details.
1577: */
1578: static size_t
1579: utf8(unsigned int cp, char out[7])
1.1 kristaps 1580: {
1.50 kristaps 1581: size_t rc;
1.1 kristaps 1582:
1.50 kristaps 1583: rc = 0;
1584: if (cp <= 0x0000007F) {
1585: rc = 1;
1586: out[0] = (char)cp;
1587: } else if (cp <= 0x000007FF) {
1588: rc = 2;
1589: out[0] = (cp >> 6 & 31) | 192;
1590: out[1] = (cp & 63) | 128;
1591: } else if (cp <= 0x0000FFFF) {
1592: rc = 3;
1593: out[0] = (cp >> 12 & 15) | 224;
1594: out[1] = (cp >> 6 & 63) | 128;
1595: out[2] = (cp & 63) | 128;
1596: } else if (cp <= 0x001FFFFF) {
1597: rc = 4;
1598: out[0] = (cp >> 18 & 7) | 240;
1599: out[1] = (cp >> 12 & 63) | 128;
1600: out[2] = (cp >> 6 & 63) | 128;
1601: out[3] = (cp & 63) | 128;
1602: } else if (cp <= 0x03FFFFFF) {
1603: rc = 5;
1604: out[0] = (cp >> 24 & 3) | 248;
1605: out[1] = (cp >> 18 & 63) | 128;
1606: out[2] = (cp >> 12 & 63) | 128;
1607: out[3] = (cp >> 6 & 63) | 128;
1608: out[4] = (cp & 63) | 128;
1609: } else if (cp <= 0x7FFFFFFF) {
1610: rc = 6;
1611: out[0] = (cp >> 30 & 1) | 252;
1612: out[1] = (cp >> 24 & 63) | 128;
1613: out[2] = (cp >> 18 & 63) | 128;
1614: out[3] = (cp >> 12 & 63) | 128;
1615: out[4] = (cp >> 6 & 63) | 128;
1616: out[5] = (cp & 63) | 128;
1617: } else
1.1 kristaps 1618: return(0);
1619:
1.50 kristaps 1620: out[rc] = '\0';
1621: return(rc);
1622: }
1.1 kristaps 1623:
1.50 kristaps 1624: /*
1.96 schwarze 1625: * Store the rendered version of a key, or alias the pointer
1626: * if the key contains no escape sequences.
1.50 kristaps 1627: */
1628: static void
1.96 schwarze 1629: render_key(struct mchars *mc, struct str *key)
1.50 kristaps 1630: {
1631: size_t sz, bsz, pos;
1632: char utfbuf[7], res[5];
1633: char *buf;
1634: const char *seq, *cpp, *val;
1635: int len, u;
1636: enum mandoc_esc esc;
1637:
1.96 schwarze 1638: assert(NULL == key->rendered);
1.50 kristaps 1639:
1640: res[0] = '\\';
1641: res[1] = '\t';
1642: res[2] = ASCII_NBRSP;
1643: res[3] = ASCII_HYPH;
1644: res[4] = '\0';
1.1 kristaps 1645:
1.50 kristaps 1646: val = key->key;
1647: bsz = strlen(val);
1.46 kristaps 1648:
1.50 kristaps 1649: /*
1650: * Pre-check: if we have no stop-characters, then set the
1651: * pointer as ourselvse and get out of here.
1652: */
1653: if (strcspn(val, res) == bsz) {
1.96 schwarze 1654: key->rendered = key->key;
1.50 kristaps 1655: return;
1656: }
1.46 kristaps 1657:
1.50 kristaps 1658: /* Pre-allocate by the length of the input */
1.46 kristaps 1659:
1.50 kristaps 1660: buf = mandoc_malloc(++bsz);
1661: pos = 0;
1.46 kristaps 1662:
1.50 kristaps 1663: while ('\0' != *val) {
1664: /*
1665: * Halt on the first escape sequence.
1666: * This also halts on the end of string, in which case
1667: * we just copy, fallthrough, and exit the loop.
1668: */
1669: if ((sz = strcspn(val, res)) > 0) {
1670: memcpy(&buf[pos], val, sz);
1671: pos += sz;
1672: val += sz;
1673: }
1.46 kristaps 1674:
1.50 kristaps 1675: if (ASCII_HYPH == *val) {
1676: buf[pos++] = '-';
1677: val++;
1678: continue;
1679: } else if ('\t' == *val || ASCII_NBRSP == *val) {
1680: buf[pos++] = ' ';
1681: val++;
1682: continue;
1683: } else if ('\\' != *val)
1684: break;
1.46 kristaps 1685:
1.50 kristaps 1686: /* Read past the slash. */
1.46 kristaps 1687:
1.50 kristaps 1688: val++;
1.46 kristaps 1689:
1.50 kristaps 1690: /*
1691: * Parse the escape sequence and see if it's a
1692: * predefined character or special character.
1693: */
1.95 schwarze 1694:
1.50 kristaps 1695: esc = mandoc_escape
1696: ((const char **)&val, &seq, &len);
1697: if (ESCAPE_ERROR == esc)
1698: break;
1699: if (ESCAPE_SPECIAL != esc)
1700: continue;
1.1 kristaps 1701:
1.50 kristaps 1702: /*
1.95 schwarze 1703: * Render the special character
1704: * as either UTF-8 or ASCII.
1.50 kristaps 1705: */
1.95 schwarze 1706:
1707: if (write_utf8) {
1708: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1709: continue;
1710: cpp = utfbuf;
1711: if (0 == (sz = utf8(u, utfbuf)))
1712: continue;
1713: sz = strlen(cpp);
1714: } else {
1715: cpp = mchars_spec2str(mc, seq, len, &sz);
1716: if (NULL == cpp)
1717: continue;
1718: if (ASCII_NBRSP == *cpp) {
1719: cpp = " ";
1720: sz = 1;
1721: }
1722: }
1.1 kristaps 1723:
1.50 kristaps 1724: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1725:
1.50 kristaps 1726: bsz += sz;
1727: buf = mandoc_realloc(buf, bsz);
1728: memcpy(&buf[pos], cpp, sz);
1729: pos += sz;
1.1 kristaps 1730: }
1731:
1.50 kristaps 1732: buf[pos] = '\0';
1.96 schwarze 1733: key->rendered = buf;
1.1 kristaps 1734: }
1735:
1.14 schwarze 1736: /*
1.50 kristaps 1737: * Flush the current page's terms (and their bits) into the database.
1738: * Wrap the entire set of additions in a transaction to make sqlite be a
1739: * little faster.
1.96 schwarze 1740: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 1741: */
1742: static void
1.105 schwarze 1743: dbadd(const struct mpage *mpage, struct mchars *mc)
1.14 schwarze 1744: {
1.87 schwarze 1745: struct mlink *mlink;
1.50 kristaps 1746: struct str *key;
1747: int64_t recno;
1.52 kristaps 1748: size_t i;
1.66 schwarze 1749: unsigned int slot;
1.14 schwarze 1750:
1.56 schwarze 1751: if (verb)
1.105 schwarze 1752: say(mpage->mlinks->file, "Adding to database");
1.43 kristaps 1753:
1.50 kristaps 1754: if (nodb)
1.14 schwarze 1755: return;
1.28 kristaps 1756:
1.52 kristaps 1757: SQL_EXEC("BEGIN TRANSACTION");
1.50 kristaps 1758:
1.52 kristaps 1759: i = 1;
1.82 schwarze 1760: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
1.81 schwarze 1761: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.50 kristaps 1762: recno = sqlite3_last_insert_rowid(db);
1.81 schwarze 1763: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1764:
1.87 schwarze 1765: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1766: i = 1;
1767: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1768: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1.88 schwarze 1769: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.87 schwarze 1770: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, recno);
1771: SQL_STEP(stmts[STMT_INSERT_LINK]);
1772: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1773: }
1.50 kristaps 1774:
1.66 schwarze 1775: for (key = ohash_first(&strings, &slot); NULL != key;
1776: key = ohash_next(&strings, &slot)) {
1.78 schwarze 1777: assert(key->mpage == mpage);
1.96 schwarze 1778: if (NULL == key->rendered)
1779: render_key(mc, key);
1.52 kristaps 1780: i = 1;
1781: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.96 schwarze 1782: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.52 kristaps 1783: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno);
1784: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 1785: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.96 schwarze 1786: if (key->rendered != key->key)
1787: free(key->rendered);
1.66 schwarze 1788: free(key);
1.38 schwarze 1789: }
1.28 kristaps 1790:
1.52 kristaps 1791: SQL_EXEC("END TRANSACTION");
1.14 schwarze 1792: }
1793:
1.5 kristaps 1794: static void
1.59 schwarze 1795: dbprune(void)
1.5 kristaps 1796: {
1.78 schwarze 1797: struct mpage *mpage;
1.82 schwarze 1798: struct mlink *mlink;
1.52 kristaps 1799: size_t i;
1.80 schwarze 1800: unsigned int slot;
1.5 kristaps 1801:
1.106 ! schwarze 1802: if (0 == nodb)
! 1803: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 1804:
1.106 ! schwarze 1805: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
! 1806: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 1807: mlink = mpage->mlinks;
1.56 schwarze 1808: if (verb)
1.106 ! schwarze 1809: say(mlink->file, "Deleting from database");
! 1810: if (nodb)
! 1811: continue;
! 1812: for ( ; NULL != mlink; mlink = mlink->next) {
! 1813: i = 1;
! 1814: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
! 1815: i, mlink->dsec);
! 1816: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
! 1817: i, mlink->arch);
! 1818: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
! 1819: i, mlink->name);
! 1820: SQL_STEP(stmts[STMT_DELETE_PAGE]);
! 1821: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
! 1822: }
1.5 kristaps 1823: }
1.106 ! schwarze 1824:
! 1825: if (0 == nodb)
! 1826: SQL_EXEC("END TRANSACTION");
1.5 kristaps 1827: }
1828:
1.4 kristaps 1829: /*
1.50 kristaps 1830: * Close an existing database and its prepared statements.
1831: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 1832: */
1.35 kristaps 1833: static void
1.59 schwarze 1834: dbclose(int real)
1.4 kristaps 1835: {
1.50 kristaps 1836: size_t i;
1.4 kristaps 1837:
1.50 kristaps 1838: if (nodb)
1.38 schwarze 1839: return;
1.50 kristaps 1840:
1841: for (i = 0; i < STMT__MAX; i++) {
1842: sqlite3_finalize(stmts[i]);
1843: stmts[i] = NULL;
1.4 kristaps 1844: }
1845:
1.50 kristaps 1846: sqlite3_close(db);
1847: db = NULL;
1.12 schwarze 1848:
1.50 kristaps 1849: if (real)
1850: return;
1.12 schwarze 1851:
1.63 schwarze 1852: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
1.59 schwarze 1853: exitcode = (int)MANDOCLEVEL_SYSERR;
1854: say(MANDOC_DB, NULL);
1855: }
1.50 kristaps 1856: }
1.14 schwarze 1857:
1.50 kristaps 1858: /*
1859: * This is straightforward stuff.
1860: * Open a database connection to a "temporary" database, then open a set
1861: * of prepared statements we'll use over and over again.
1862: * If "real" is set, we use the existing database; if not, we truncate a
1863: * temporary one.
1864: * Must be matched by dbclose().
1865: */
1866: static int
1.59 schwarze 1867: dbopen(int real)
1.50 kristaps 1868: {
1.63 schwarze 1869: const char *file, *sql;
1.50 kristaps 1870: int rc, ofl;
1.12 schwarze 1871:
1.50 kristaps 1872: if (nodb)
1873: return(1);
1.12 schwarze 1874:
1.63 schwarze 1875: ofl = SQLITE_OPEN_READWRITE;
1876: if (0 == real) {
1877: file = MANDOC_DB "~";
1878: if (-1 == remove(file) && ENOENT != errno) {
1879: exitcode = (int)MANDOCLEVEL_SYSERR;
1880: say(file, NULL);
1881: return(0);
1882: }
1883: ofl |= SQLITE_OPEN_EXCLUSIVE;
1884: } else
1885: file = MANDOC_DB;
1.45 kristaps 1886:
1.50 kristaps 1887: rc = sqlite3_open_v2(file, &db, ofl, NULL);
1888: if (SQLITE_OK == rc)
1.57 schwarze 1889: goto prepare_statements;
1.50 kristaps 1890: if (SQLITE_CANTOPEN != rc) {
1.59 schwarze 1891: exitcode = (int)MANDOCLEVEL_SYSERR;
1892: say(file, NULL);
1.50 kristaps 1893: return(0);
1894: }
1.12 schwarze 1895:
1.50 kristaps 1896: sqlite3_close(db);
1897: db = NULL;
1.12 schwarze 1898:
1.50 kristaps 1899: if (SQLITE_OK != (rc = sqlite3_open(file, &db))) {
1.59 schwarze 1900: exitcode = (int)MANDOCLEVEL_SYSERR;
1901: say(file, NULL);
1.50 kristaps 1902: return(0);
1903: }
1.12 schwarze 1904:
1.81 schwarze 1905: sql = "CREATE TABLE \"mpages\" (\n"
1.50 kristaps 1906: " \"form\" INTEGER NOT NULL,\n"
1907: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1908: ");\n"
1909: "\n"
1.81 schwarze 1910: "CREATE TABLE \"mlinks\" (\n"
1911: " \"sec\" TEXT NOT NULL,\n"
1912: " \"arch\" TEXT NOT NULL,\n"
1913: " \"name\" 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"
1.50 kristaps 1919: "CREATE TABLE \"keys\" (\n"
1920: " \"bits\" INTEGER NOT NULL,\n"
1921: " \"key\" TEXT NOT NULL,\n"
1.81 schwarze 1922: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1923: "ON DELETE CASCADE,\n"
1.50 kristaps 1924: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1925: ");\n"
1926: "\n"
1927: "CREATE INDEX \"key_index\" ON keys (key);\n";
1.14 schwarze 1928:
1.50 kristaps 1929: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 1930: exitcode = (int)MANDOCLEVEL_SYSERR;
1931: say(file, "%s", sqlite3_errmsg(db));
1.50 kristaps 1932: return(0);
1933: }
1.4 kristaps 1934:
1.57 schwarze 1935: prepare_statements:
1936: SQL_EXEC("PRAGMA foreign_keys = ON");
1.106 ! schwarze 1937: sql = "DELETE FROM mpages WHERE id IN "
! 1938: "(SELECT pageid FROM mlinks WHERE "
! 1939: "sec=? AND arch=? AND name=?)";
1.81 schwarze 1940: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
1941: sql = "INSERT INTO mpages "
1.103 schwarze 1942: "(form) VALUES (?)";
1.81 schwarze 1943: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
1944: sql = "INSERT INTO mlinks "
1.104 schwarze 1945: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 1946: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.50 kristaps 1947: sql = "INSERT INTO keys "
1.81 schwarze 1948: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 1949: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 1950:
1951: #ifndef __APPLE__
1952: /*
1953: * When opening a new database, we can turn off
1954: * synchronous mode for much better performance.
1955: */
1956:
1957: if (real)
1958: SQL_EXEC("PRAGMA synchronous = OFF");
1959: #endif
1960:
1.50 kristaps 1961: return(1);
1962: }
1.5 kristaps 1963:
1.50 kristaps 1964: static void *
1965: hash_halloc(size_t sz, void *arg)
1966: {
1.12 schwarze 1967:
1.50 kristaps 1968: return(mandoc_calloc(sz, 1));
1969: }
1.12 schwarze 1970:
1.50 kristaps 1971: static void *
1972: hash_alloc(size_t sz, void *arg)
1973: {
1.5 kristaps 1974:
1.50 kristaps 1975: return(mandoc_malloc(sz));
1976: }
1.41 kristaps 1977:
1.50 kristaps 1978: static void
1979: hash_free(void *p, size_t sz, void *arg)
1980: {
1.4 kristaps 1981:
1.50 kristaps 1982: free(p);
1.4 kristaps 1983: }
1984:
1.50 kristaps 1985: static int
1.59 schwarze 1986: set_basedir(const char *targetdir)
1.4 kristaps 1987: {
1.59 schwarze 1988: static char startdir[PATH_MAX];
1989: static int fd;
1.4 kristaps 1990:
1.59 schwarze 1991: /*
1992: * Remember where we started by keeping a fd open to the origin
1993: * path component: throughout this utility, we chdir() a lot to
1994: * handle relative paths, and by doing this, we can return to
1995: * the starting point.
1996: */
1997: if ('\0' == *startdir) {
1998: if (NULL == getcwd(startdir, PATH_MAX)) {
1999: exitcode = (int)MANDOCLEVEL_SYSERR;
2000: if (NULL != targetdir)
2001: say(".", NULL);
2002: return(0);
2003: }
2004: if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
2005: exitcode = (int)MANDOCLEVEL_SYSERR;
2006: say(startdir, NULL);
2007: return(0);
2008: }
2009: if (NULL == targetdir)
2010: targetdir = startdir;
2011: } else {
2012: if (-1 == fd)
2013: return(0);
2014: if (-1 == fchdir(fd)) {
2015: close(fd);
2016: basedir[0] = '\0';
2017: exitcode = (int)MANDOCLEVEL_SYSERR;
2018: say(startdir, NULL);
2019: return(0);
2020: }
2021: if (NULL == targetdir) {
2022: close(fd);
2023: return(1);
2024: }
2025: }
2026: if (NULL == realpath(targetdir, basedir)) {
2027: basedir[0] = '\0';
2028: exitcode = (int)MANDOCLEVEL_BADARG;
2029: say(targetdir, NULL);
1.50 kristaps 2030: return(0);
1.59 schwarze 2031: } else if (-1 == chdir(basedir)) {
2032: exitcode = (int)MANDOCLEVEL_BADARG;
2033: say("", NULL);
1.50 kristaps 2034: return(0);
1.4 kristaps 2035: }
1.50 kristaps 2036: return(1);
1.56 schwarze 2037: }
2038:
2039: static void
1.59 schwarze 2040: say(const char *file, const char *format, ...)
1.56 schwarze 2041: {
2042: va_list ap;
2043:
1.59 schwarze 2044: if ('\0' != *basedir)
2045: fprintf(stderr, "%s", basedir);
2046: if ('\0' != *basedir && '\0' != *file)
2047: fputs("//", stderr);
1.56 schwarze 2048: if ('\0' != *file)
1.59 schwarze 2049: fprintf(stderr, "%s", file);
1.56 schwarze 2050: fputs(": ", stderr);
1.59 schwarze 2051:
2052: if (NULL == format) {
2053: perror(NULL);
2054: return;
2055: }
1.56 schwarze 2056:
2057: va_start(ap, format);
2058: vfprintf(stderr, format, ap);
2059: va_end(ap);
2060:
2061: fputc('\n', stderr);
1.1 kristaps 2062: }
CVSweb