Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.128
1.128 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.127 2014/12/04 17:35:42 schwarze Exp $ */
1.1 schwarze 2: /*
1.47 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.52 schwarze 4: * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17: */
1.117 schwarze 18: #include <sys/types.h>
1.47 schwarze 19: #include <sys/stat.h>
1.72 schwarze 20: #include <sys/wait.h>
1.1 schwarze 21:
22: #include <assert.h>
1.33 schwarze 23: #include <ctype.h>
1.34 schwarze 24: #include <errno.h>
1.1 schwarze 25: #include <fcntl.h>
1.47 schwarze 26: #include <fts.h>
1.1 schwarze 27: #include <getopt.h>
1.44 schwarze 28: #include <limits.h>
1.47 schwarze 29: #include <stddef.h>
1.1 schwarze 30: #include <stdio.h>
31: #include <stdint.h>
32: #include <stdlib.h>
33: #include <string.h>
1.14 schwarze 34: #include <unistd.h>
1.1 schwarze 35:
1.47 schwarze 36: #include <ohash.h>
37: #include <sqlite3.h>
38:
39: #include "mdoc.h"
1.1 schwarze 40: #include "man.h"
41: #include "mandoc.h"
1.76 schwarze 42: #include "mandoc_aux.h"
1.10 schwarze 43: #include "manpath.h"
1.47 schwarze 44: #include "mansearch.h"
1.1 schwarze 45:
1.68 schwarze 46: extern int mansearch_keymax;
47: extern const char *const mansearch_keynames[];
48:
1.47 schwarze 49: #define SQL_EXEC(_v) \
50: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.91 schwarze 51: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.47 schwarze 52: #define SQL_BIND_TEXT(_s, _i, _v) \
53: if (SQLITE_OK != sqlite3_bind_text \
54: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.91 schwarze 55: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 56: #define SQL_BIND_INT(_s, _i, _v) \
57: if (SQLITE_OK != sqlite3_bind_int \
58: ((_s), (_i)++, (_v))) \
1.91 schwarze 59: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 60: #define SQL_BIND_INT64(_s, _i, _v) \
61: if (SQLITE_OK != sqlite3_bind_int64 \
62: ((_s), (_i)++, (_v))) \
1.91 schwarze 63: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 64: #define SQL_STEP(_s) \
65: if (SQLITE_DONE != sqlite3_step((_s))) \
1.91 schwarze 66: say(mlink->file, "%s", sqlite3_errmsg(db))
1.1 schwarze 67:
1.47 schwarze 68: enum op {
69: OP_DEFAULT = 0, /* new dbs from dir list or default config */
70: OP_CONFFILE, /* new databases from custom config file */
71: OP_UPDATE, /* delete/add entries in existing database */
72: OP_DELETE, /* delete entries from existing database */
73: OP_TEST /* change no databases, report potential problems */
74: };
1.11 schwarze 75:
1.47 schwarze 76: struct str {
1.53 schwarze 77: char *rendered; /* key in UTF-8 or ASCII form */
1.47 schwarze 78: const struct mpage *mpage; /* if set, the owning parse */
79: uint64_t mask; /* bitmask in sequence */
1.53 schwarze 80: char key[]; /* may contain escape sequences */
1.28 schwarze 81: };
82:
1.47 schwarze 83: struct inodev {
84: ino_t st_ino;
85: dev_t st_dev;
86: };
1.28 schwarze 87:
1.47 schwarze 88: struct mpage {
89: struct inodev inodev; /* used for hashing routine */
1.94 schwarze 90: int64_t pageid; /* pageid in mpages SQL table */
1.47 schwarze 91: char *sec; /* section from file content */
92: char *arch; /* architecture from file content */
93: char *title; /* title from file content */
94: char *desc; /* description from file content */
95: struct mlink *mlinks; /* singly linked list */
1.116 schwarze 96: int form; /* format from file content */
1.28 schwarze 97: };
98:
1.47 schwarze 99: struct mlink {
100: char file[PATH_MAX]; /* filename rel. to manpath */
101: char *dsec; /* section from directory */
102: char *arch; /* architecture from directory */
103: char *name; /* name from file name (not empty) */
104: char *fsec; /* section from file name suffix */
105: struct mlink *next; /* singly linked list */
1.75 schwarze 106: struct mpage *mpage; /* parent */
1.116 schwarze 107: int dform; /* format from directory */
108: int fform; /* format from file name suffix */
1.81 schwarze 109: int gzip; /* filename has a .gz suffix */
1.2 schwarze 110: };
111:
1.47 schwarze 112: enum stmt {
113: STMT_DELETE_PAGE = 0, /* delete mpage */
114: STMT_INSERT_PAGE, /* insert mpage */
115: STMT_INSERT_LINK, /* insert mlink */
1.90 schwarze 116: STMT_INSERT_NAME, /* insert name */
1.47 schwarze 117: STMT_INSERT_KEY, /* insert parsed key */
118: STMT__MAX
1.1 schwarze 119: };
120:
1.127 schwarze 121: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_meta *,
122: const struct mdoc_node *);
1.1 schwarze 123:
1.19 schwarze 124: struct mdoc_handler {
1.47 schwarze 125: mdoc_fp fp; /* optional handler */
126: uint64_t mask; /* set unless handler returns 0 */
1.19 schwarze 127: };
128:
1.47 schwarze 129: static void dbclose(int);
1.75 schwarze 130: static void dbadd(struct mpage *, struct mchars *);
131: static void dbadd_mlink(const struct mlink *mlink);
1.124 schwarze 132: static void dbadd_mlink_name(const struct mlink *mlink);
1.47 schwarze 133: static int dbopen(int);
134: static void dbprune(void);
135: static void filescan(const char *);
136: static void *hash_alloc(size_t, void *);
1.106 espie 137: static void hash_free(void *, void *);
138: static void *hash_calloc(size_t, size_t, void *);
1.47 schwarze 139: static void mlink_add(struct mlink *, const struct stat *);
1.88 schwarze 140: static void mlink_check(struct mpage *, struct mlink *);
1.47 schwarze 141: static void mlink_free(struct mlink *);
142: static void mlinks_undupe(struct mpage *);
143: static void mpages_free(void);
1.58 schwarze 144: static void mpages_merge(struct mchars *, struct mparse *);
1.87 schwarze 145: static void names_check(void);
1.81 schwarze 146: static void parse_cat(struct mpage *, int);
1.47 schwarze 147: static void parse_man(struct mpage *, const struct man_node *);
1.127 schwarze 148: static void parse_mdoc(struct mpage *, const struct mdoc_meta *,
149: const struct mdoc_node *);
150: static int parse_mdoc_body(struct mpage *, const struct mdoc_meta *,
151: const struct mdoc_node *);
152: static int parse_mdoc_head(struct mpage *, const struct mdoc_meta *,
153: const struct mdoc_node *);
154: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_meta *,
155: const struct mdoc_node *);
156: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_meta *,
157: const struct mdoc_node *);
1.128 ! schwarze 158: static int parse_mdoc_Fo(struct mpage *, const struct mdoc_meta *,
! 159: const struct mdoc_node *);
1.127 schwarze 160: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_meta *,
161: const struct mdoc_node *);
162: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_meta *,
163: const struct mdoc_node *);
164: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_meta *,
165: const struct mdoc_node *);
166: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_meta *,
167: const struct mdoc_node *);
1.69 schwarze 168: static void putkey(const struct mpage *, char *, uint64_t);
1.47 schwarze 169: static void putkeys(const struct mpage *,
170: const char *, size_t, uint64_t);
171: static void putmdockey(const struct mpage *,
172: const struct mdoc_node *, uint64_t);
1.53 schwarze 173: static void render_key(struct mchars *, struct str *);
1.47 schwarze 174: static void say(const char *, const char *, ...);
1.120 schwarze 175: static int set_basedir(const char *, int);
1.47 schwarze 176: static int treescan(void);
177: static size_t utf8(unsigned int, char [7]);
178:
1.72 schwarze 179: static char tempfilename[32];
1.47 schwarze 180: static char *progname;
1.59 schwarze 181: static int nodb; /* no database changes */
1.73 schwarze 182: static int mparse_options; /* abort the parse early */
1.98 schwarze 183: static int use_all; /* use all found files */
184: static int debug; /* print what we're doing */
185: static int warnings; /* warn about crap */
1.52 schwarze 186: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 187: static int exitcode; /* to be returned by main */
1.98 schwarze 188: static enum op op; /* operational mode */
1.47 schwarze 189: static char basedir[PATH_MAX]; /* current base directory */
190: static struct ohash mpages; /* table of distinct manual pages */
191: static struct ohash mlinks; /* table of directory entries */
1.90 schwarze 192: static struct ohash names; /* table of all names */
1.47 schwarze 193: static struct ohash strings; /* table of all strings */
194: static sqlite3 *db = NULL; /* current database */
195: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.90 schwarze 196: static uint64_t name_mask;
1.47 schwarze 197:
1.19 schwarze 198: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 199: { NULL, 0 }, /* Ap */
200: { NULL, 0 }, /* Dd */
201: { NULL, 0 }, /* Dt */
202: { NULL, 0 }, /* Os */
203: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
204: { parse_mdoc_head, TYPE_Ss }, /* Ss */
205: { NULL, 0 }, /* Pp */
206: { NULL, 0 }, /* D1 */
207: { NULL, 0 }, /* Dl */
208: { NULL, 0 }, /* Bd */
209: { NULL, 0 }, /* Ed */
210: { NULL, 0 }, /* Bl */
211: { NULL, 0 }, /* El */
212: { NULL, 0 }, /* It */
213: { NULL, 0 }, /* Ad */
214: { NULL, TYPE_An }, /* An */
215: { NULL, TYPE_Ar }, /* Ar */
216: { NULL, TYPE_Cd }, /* Cd */
217: { NULL, TYPE_Cm }, /* Cm */
218: { NULL, TYPE_Dv }, /* Dv */
219: { NULL, TYPE_Er }, /* Er */
220: { NULL, TYPE_Ev }, /* Ev */
221: { NULL, 0 }, /* Ex */
222: { NULL, TYPE_Fa }, /* Fa */
223: { parse_mdoc_Fd, 0 }, /* Fd */
224: { NULL, TYPE_Fl }, /* Fl */
225: { parse_mdoc_Fn, 0 }, /* Fn */
226: { NULL, TYPE_Ft }, /* Ft */
227: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 228: { NULL, TYPE_In }, /* In */
1.47 schwarze 229: { NULL, TYPE_Li }, /* Li */
1.89 schwarze 230: { parse_mdoc_Nd, 0 }, /* Nd */
1.86 schwarze 231: { parse_mdoc_Nm, 0 }, /* Nm */
1.47 schwarze 232: { NULL, 0 }, /* Op */
233: { NULL, 0 }, /* Ot */
234: { NULL, TYPE_Pa }, /* Pa */
235: { NULL, 0 }, /* Rv */
1.49 schwarze 236: { NULL, TYPE_St }, /* St */
1.47 schwarze 237: { NULL, TYPE_Va }, /* Va */
238: { parse_mdoc_body, TYPE_Va }, /* Vt */
239: { parse_mdoc_Xr, 0 }, /* Xr */
240: { NULL, 0 }, /* %A */
241: { NULL, 0 }, /* %B */
242: { NULL, 0 }, /* %D */
243: { NULL, 0 }, /* %I */
244: { NULL, 0 }, /* %J */
245: { NULL, 0 }, /* %N */
246: { NULL, 0 }, /* %O */
247: { NULL, 0 }, /* %P */
248: { NULL, 0 }, /* %R */
249: { NULL, 0 }, /* %T */
250: { NULL, 0 }, /* %V */
251: { NULL, 0 }, /* Ac */
252: { NULL, 0 }, /* Ao */
253: { NULL, 0 }, /* Aq */
254: { NULL, TYPE_At }, /* At */
255: { NULL, 0 }, /* Bc */
256: { NULL, 0 }, /* Bf */
257: { NULL, 0 }, /* Bo */
258: { NULL, 0 }, /* Bq */
259: { NULL, TYPE_Bsx }, /* Bsx */
260: { NULL, TYPE_Bx }, /* Bx */
261: { NULL, 0 }, /* Db */
262: { NULL, 0 }, /* Dc */
263: { NULL, 0 }, /* Do */
264: { NULL, 0 }, /* Dq */
265: { NULL, 0 }, /* Ec */
266: { NULL, 0 }, /* Ef */
267: { NULL, TYPE_Em }, /* Em */
268: { NULL, 0 }, /* Eo */
269: { NULL, TYPE_Fx }, /* Fx */
270: { NULL, TYPE_Ms }, /* Ms */
271: { NULL, 0 }, /* No */
272: { NULL, 0 }, /* Ns */
273: { NULL, TYPE_Nx }, /* Nx */
274: { NULL, TYPE_Ox }, /* Ox */
275: { NULL, 0 }, /* Pc */
276: { NULL, 0 }, /* Pf */
277: { NULL, 0 }, /* Po */
278: { NULL, 0 }, /* Pq */
279: { NULL, 0 }, /* Qc */
280: { NULL, 0 }, /* Ql */
281: { NULL, 0 }, /* Qo */
282: { NULL, 0 }, /* Qq */
283: { NULL, 0 }, /* Re */
284: { NULL, 0 }, /* Rs */
285: { NULL, 0 }, /* Sc */
286: { NULL, 0 }, /* So */
287: { NULL, 0 }, /* Sq */
288: { NULL, 0 }, /* Sm */
289: { NULL, 0 }, /* Sx */
290: { NULL, TYPE_Sy }, /* Sy */
291: { NULL, TYPE_Tn }, /* Tn */
292: { NULL, 0 }, /* Ux */
293: { NULL, 0 }, /* Xc */
294: { NULL, 0 }, /* Xo */
1.128 ! schwarze 295: { parse_mdoc_Fo, 0 }, /* Fo */
1.47 schwarze 296: { NULL, 0 }, /* Fc */
297: { NULL, 0 }, /* Oo */
298: { NULL, 0 }, /* Oc */
299: { NULL, 0 }, /* Bk */
300: { NULL, 0 }, /* Ek */
301: { NULL, 0 }, /* Bt */
302: { NULL, 0 }, /* Hf */
303: { NULL, 0 }, /* Fr */
304: { NULL, 0 }, /* Ud */
305: { NULL, TYPE_Lb }, /* Lb */
306: { NULL, 0 }, /* Lp */
307: { NULL, TYPE_Lk }, /* Lk */
308: { NULL, TYPE_Mt }, /* Mt */
309: { NULL, 0 }, /* Brq */
310: { NULL, 0 }, /* Bro */
311: { NULL, 0 }, /* Brc */
312: { NULL, 0 }, /* %C */
313: { NULL, 0 }, /* Es */
314: { NULL, 0 }, /* En */
315: { NULL, TYPE_Dx }, /* Dx */
316: { NULL, 0 }, /* %Q */
317: { NULL, 0 }, /* br */
318: { NULL, 0 }, /* sp */
319: { NULL, 0 }, /* %U */
320: { NULL, 0 }, /* Ta */
1.121 schwarze 321: { NULL, 0 }, /* ll */
1.1 schwarze 322: };
323:
1.98 schwarze 324:
1.1 schwarze 325: int
1.3 schwarze 326: mandocdb(int argc, char *argv[])
1.1 schwarze 327: {
1.47 schwarze 328: int ch, i;
329: size_t j, sz;
330: const char *path_arg;
331: struct mchars *mc;
332: struct manpaths dirs;
333: struct mparse *mp;
334: struct ohash_info mpages_info, mlinks_info;
335:
336: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
337: memset(&dirs, 0, sizeof(struct manpaths));
338:
339: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
1.106 espie 340: mpages_info.calloc = mlinks_info.calloc = hash_calloc;
341: mpages_info.free = mlinks_info.free = hash_free;
1.47 schwarze 342:
343: mpages_info.key_offset = offsetof(struct mpage, inodev);
344: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 345:
346: progname = strrchr(argv[0], '/');
347: if (progname == NULL)
348: progname = argv[0];
349: else
350: ++progname;
351:
1.47 schwarze 352: /*
1.98 schwarze 353: * We accept a few different invocations.
1.47 schwarze 354: * The CHECKOP macro makes sure that invocation styles don't
355: * clobber each other.
356: */
357: #define CHECKOP(_op, _ch) do \
358: if (OP_DEFAULT != (_op)) { \
1.111 schwarze 359: fprintf(stderr, "%s: -%c: Conflicting option\n", \
360: progname, (_ch)); \
1.47 schwarze 361: goto usage; \
362: } while (/*CONSTCOND*/0)
1.10 schwarze 363:
1.47 schwarze 364: path_arg = NULL;
1.28 schwarze 365: op = OP_DEFAULT;
1.1 schwarze 366:
1.83 schwarze 367: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 368: switch (ch) {
1.98 schwarze 369: case 'a':
1.6 schwarze 370: use_all = 1;
371: break;
1.98 schwarze 372: case 'C':
1.47 schwarze 373: CHECKOP(op, ch);
374: path_arg = optarg;
1.28 schwarze 375: op = OP_CONFFILE;
1.25 schwarze 376: break;
1.98 schwarze 377: case 'D':
1.82 schwarze 378: debug++;
379: break;
1.98 schwarze 380: case 'd':
1.47 schwarze 381: CHECKOP(op, ch);
382: path_arg = optarg;
1.2 schwarze 383: op = OP_UPDATE;
1.1 schwarze 384: break;
1.98 schwarze 385: case 'n':
1.47 schwarze 386: nodb = 1;
387: break;
1.98 schwarze 388: case 'p':
1.83 schwarze 389: warnings = 1;
390: break;
1.98 schwarze 391: case 'Q':
1.73 schwarze 392: mparse_options |= MPARSE_QUICK;
1.59 schwarze 393: break;
1.98 schwarze 394: case 'T':
1.52 schwarze 395: if (strcmp(optarg, "utf8")) {
1.111 schwarze 396: fprintf(stderr, "%s: -T%s: "
397: "Unsupported output format\n",
398: progname, optarg);
1.52 schwarze 399: goto usage;
400: }
401: write_utf8 = 1;
402: break;
1.98 schwarze 403: case 't':
1.47 schwarze 404: CHECKOP(op, ch);
1.28 schwarze 405: dup2(STDOUT_FILENO, STDERR_FILENO);
406: op = OP_TEST;
1.47 schwarze 407: nodb = warnings = 1;
1.28 schwarze 408: break;
1.98 schwarze 409: case 'u':
1.47 schwarze 410: CHECKOP(op, ch);
411: path_arg = optarg;
1.1 schwarze 412: op = OP_DELETE;
413: break;
1.98 schwarze 414: case 'v':
1.83 schwarze 415: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 416: break;
1.1 schwarze 417: default:
1.28 schwarze 418: goto usage;
1.1 schwarze 419: }
420:
421: argc -= optind;
422: argv += optind;
423:
1.28 schwarze 424: if (OP_CONFFILE == op && argc > 0) {
1.111 schwarze 425: fprintf(stderr, "%s: -C: Too many arguments\n",
426: progname);
1.28 schwarze 427: goto usage;
428: }
429:
1.47 schwarze 430: exitcode = (int)MANDOCLEVEL_OK;
431: mc = mchars_alloc();
1.123 schwarze 432: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL,
433: mc, NULL);
1.47 schwarze 434: ohash_init(&mpages, 6, &mpages_info);
435: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 436:
1.47 schwarze 437: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.2 schwarze 438:
1.47 schwarze 439: /*
1.109 schwarze 440: * Most of these deal with a specific directory.
1.97 schwarze 441: * Jump into that directory first.
1.47 schwarze 442: */
1.120 schwarze 443: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.47 schwarze 444: goto out;
1.97 schwarze 445:
1.95 schwarze 446: if (dbopen(1)) {
1.97 schwarze 447: /*
448: * The existing database is usable. Process
449: * all files specified on the command-line.
450: */
451: use_all = 1;
1.95 schwarze 452: for (i = 0; i < argc; i++)
453: filescan(argv[i]);
454: if (OP_TEST != op)
455: dbprune();
456: } else {
457: /*
458: * Database missing or corrupt.
459: * Recreate from scratch.
460: */
1.97 schwarze 461: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 462: op = OP_DEFAULT;
463: if (0 == treescan())
464: goto out;
465: if (0 == dbopen(0))
466: goto out;
467: }
1.47 schwarze 468: if (OP_DELETE != op)
1.58 schwarze 469: mpages_merge(mc, mp);
1.95 schwarze 470: dbclose(OP_DEFAULT == op ? 0 : 1);
1.47 schwarze 471: } else {
472: /*
473: * If we have arguments, use them as our manpaths.
474: * If we don't, grok from manpath(1) or however else
475: * manpath_parse() wants to do it.
476: */
477: if (argc > 0) {
1.101 schwarze 478: dirs.paths = mandoc_reallocarray(NULL,
479: argc, sizeof(char *));
1.47 schwarze 480: dirs.sz = (size_t)argc;
481: for (i = 0; i < argc; i++)
482: dirs.paths[i] = mandoc_strdup(argv[i]);
483: } else
484: manpath_parse(&dirs, path_arg, NULL, NULL);
1.84 schwarze 485:
486: if (0 == dirs.sz) {
487: exitcode = (int)MANDOCLEVEL_BADARG;
488: say("", "Empty manpath");
489: }
1.2 schwarze 490:
1.47 schwarze 491: /*
492: * First scan the tree rooted at a base directory, then
493: * build a new database and finally move it into place.
494: * Ignore zero-length directories and strip trailing
495: * slashes.
496: */
497: for (j = 0; j < dirs.sz; j++) {
498: sz = strlen(dirs.paths[j]);
499: if (sz && '/' == dirs.paths[j][sz - 1])
500: dirs.paths[j][--sz] = '\0';
501: if (0 == sz)
502: continue;
1.2 schwarze 503:
1.47 schwarze 504: if (j) {
505: ohash_init(&mpages, 6, &mpages_info);
506: ohash_init(&mlinks, 6, &mlinks_info);
507: }
1.1 schwarze 508:
1.120 schwarze 509: if (0 == set_basedir(dirs.paths[j], argc > 0))
510: continue;
1.47 schwarze 511: if (0 == treescan())
1.120 schwarze 512: continue;
1.47 schwarze 513: if (0 == dbopen(0))
1.120 schwarze 514: continue;
1.41 deraadt 515:
1.58 schwarze 516: mpages_merge(mc, mp);
1.104 schwarze 517: if (warnings && !nodb &&
1.87 schwarze 518: ! (MPARSE_QUICK & mparse_options))
519: names_check();
1.47 schwarze 520: dbclose(0);
1.28 schwarze 521:
1.47 schwarze 522: if (j + 1 < dirs.sz) {
523: mpages_free();
524: ohash_delete(&mpages);
525: ohash_delete(&mlinks);
526: }
1.2 schwarze 527: }
1.47 schwarze 528: }
529: out:
530: manpath_free(&dirs);
1.123 schwarze 531: mparse_free(mp);
1.47 schwarze 532: mchars_free(mc);
533: mpages_free();
534: ohash_delete(&mpages);
535: ohash_delete(&mlinks);
536: return(exitcode);
537: usage:
1.83 schwarze 538: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
539: " %s [-aDnpQ] [-Tutf8] dir ...\n"
540: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
541: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 542: " %s [-Q] -t file ...\n",
1.98 schwarze 543: progname, progname, progname,
1.47 schwarze 544: progname, progname);
1.1 schwarze 545:
1.47 schwarze 546: return((int)MANDOCLEVEL_BADARG);
547: }
1.1 schwarze 548:
1.47 schwarze 549: /*
550: * Scan a directory tree rooted at "basedir" for manpages.
551: * We use fts(), scanning directory parts along the way for clues to our
552: * section and architecture.
553: *
554: * If use_all has been specified, grok all files.
555: * If not, sanitise paths to the following:
556: *
1.98 schwarze 557: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 558: * or
559: * [./]cat<section>[/<arch>]/<name>.0
560: *
561: * TODO: accomodate for multi-language directories.
562: */
563: static int
564: treescan(void)
565: {
1.96 schwarze 566: char buf[PATH_MAX];
1.47 schwarze 567: FTS *f;
568: FTSENT *ff;
569: struct mlink *mlink;
1.81 schwarze 570: int dform, gzip;
1.51 schwarze 571: char *dsec, *arch, *fsec, *cp;
572: const char *path;
1.47 schwarze 573: const char *argv[2];
1.1 schwarze 574:
1.47 schwarze 575: argv[0] = ".";
576: argv[1] = (char *)NULL;
1.1 schwarze 577:
1.96 schwarze 578: f = fts_open((char * const *)argv,
579: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.47 schwarze 580: if (NULL == f) {
581: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 582: say("", "&fts_open");
1.47 schwarze 583: return(0);
584: }
1.2 schwarze 585:
1.47 schwarze 586: dsec = arch = NULL;
587: dform = FORM_NONE;
1.2 schwarze 588:
1.47 schwarze 589: while (NULL != (ff = fts_read(f))) {
590: path = ff->fts_path + 2;
1.96 schwarze 591: switch (ff->fts_info) {
592:
593: /*
594: * Symbolic links require various sanity checks,
595: * then get handled just like regular files.
596: */
1.98 schwarze 597: case FTS_SL:
1.96 schwarze 598: if (NULL == realpath(path, buf)) {
599: if (warnings)
600: say(path, "&realpath");
601: continue;
602: }
603: if (strstr(buf, basedir) != buf) {
604: if (warnings) say("",
605: "%s: outside base directory", buf);
606: continue;
607: }
608: /* Use logical inode to avoid mpages dupe. */
609: if (-1 == stat(path, ff->fts_statp)) {
610: if (warnings)
611: say(path, "&stat");
612: continue;
613: }
614: /* FALLTHROUGH */
615:
1.15 schwarze 616: /*
1.47 schwarze 617: * If we're a regular file, add an mlink by using the
618: * stored directory data and handling the filename.
1.15 schwarze 619: */
1.98 schwarze 620: case FTS_F:
1.47 schwarze 621: if (0 == strcmp(path, MANDOC_DB))
622: continue;
623: if ( ! use_all && ff->fts_level < 2) {
624: if (warnings)
625: say(path, "Extraneous file");
626: continue;
1.81 schwarze 627: }
628: gzip = 0;
629: fsec = NULL;
630: while (NULL == fsec) {
631: fsec = strrchr(ff->fts_name, '.');
632: if (NULL == fsec || strcmp(fsec+1, "gz"))
633: break;
634: gzip = 1;
635: *fsec = '\0';
636: fsec = NULL;
637: }
638: if (NULL == fsec) {
1.47 schwarze 639: if ( ! use_all) {
640: if (warnings)
641: say(path,
642: "No filename suffix");
643: continue;
644: }
645: } else if (0 == strcmp(++fsec, "html")) {
646: if (warnings)
647: say(path, "Skip html");
648: continue;
649: } else if (0 == strcmp(fsec, "ps")) {
650: if (warnings)
651: say(path, "Skip ps");
652: continue;
653: } else if (0 == strcmp(fsec, "pdf")) {
654: if (warnings)
655: say(path, "Skip pdf");
656: continue;
657: } else if ( ! use_all &&
658: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
659: (FORM_CAT == dform && strcmp(fsec, "0")))) {
660: if (warnings)
661: say(path, "Wrong filename suffix");
662: continue;
663: } else
664: fsec[-1] = '\0';
1.51 schwarze 665:
1.47 schwarze 666: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 667: if (strlcpy(mlink->file, path,
668: sizeof(mlink->file)) >=
669: sizeof(mlink->file)) {
670: say(path, "Filename too long");
671: free(mlink);
672: continue;
673: }
1.47 schwarze 674: mlink->dform = dform;
1.51 schwarze 675: mlink->dsec = dsec;
676: mlink->arch = arch;
677: mlink->name = ff->fts_name;
678: mlink->fsec = fsec;
1.81 schwarze 679: mlink->gzip = gzip;
1.47 schwarze 680: mlink_add(mlink, ff->fts_statp);
681: continue;
1.96 schwarze 682:
1.98 schwarze 683: case FTS_D:
1.96 schwarze 684: /* FALLTHROUGH */
1.98 schwarze 685: case FTS_DP:
1.96 schwarze 686: break;
687:
688: default:
1.47 schwarze 689: if (warnings)
690: say(path, "Not a regular file");
691: continue;
692: }
693:
694: switch (ff->fts_level) {
1.98 schwarze 695: case 0:
1.47 schwarze 696: /* Ignore the root directory. */
697: break;
1.98 schwarze 698: case 1:
1.47 schwarze 699: /*
700: * This might contain manX/ or catX/.
701: * Try to infer this from the name.
702: * If we're not in use_all, enforce it.
703: */
704: cp = ff->fts_name;
705: if (FTS_DP == ff->fts_info)
706: break;
1.15 schwarze 707:
1.47 schwarze 708: if (0 == strncmp(cp, "man", 3)) {
709: dform = FORM_SRC;
710: dsec = cp + 3;
711: } else if (0 == strncmp(cp, "cat", 3)) {
712: dform = FORM_CAT;
713: dsec = cp + 3;
1.51 schwarze 714: } else {
715: dform = FORM_NONE;
716: dsec = NULL;
1.26 schwarze 717: }
1.47 schwarze 718:
1.98 schwarze 719: if (NULL != dsec || use_all)
1.47 schwarze 720: break;
721:
722: if (warnings)
723: say(path, "Unknown directory part");
724: fts_set(f, ff, FTS_SKIP);
725: break;
1.98 schwarze 726: case 2:
1.47 schwarze 727: /*
728: * Possibly our architecture.
729: * If we're descending, keep tabs on it.
730: */
731: if (FTS_DP != ff->fts_info && NULL != dsec)
732: arch = ff->fts_name;
1.51 schwarze 733: else
734: arch = NULL;
1.47 schwarze 735: break;
736: default:
737: if (FTS_DP == ff->fts_info || use_all)
738: break;
739: if (warnings)
740: say(path, "Extraneous directory part");
741: fts_set(f, ff, FTS_SKIP);
742: break;
1.14 schwarze 743: }
1.47 schwarze 744: }
745:
746: fts_close(f);
747: return(1);
748: }
1.1 schwarze 749:
1.47 schwarze 750: /*
751: * Add a file to the mlinks table.
752: * Do not verify that it's a "valid" looking manpage (we'll do that
753: * later).
754: *
755: * Try to infer the manual section, architecture, and page name from the
756: * path, assuming it looks like
757: *
1.98 schwarze 758: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 759: * or
760: * [./]cat<section>[/<arch>]/<name>.0
761: *
762: * See treescan() for the fts(3) version of this.
763: */
764: static void
765: filescan(const char *file)
766: {
767: char buf[PATH_MAX];
768: struct stat st;
769: struct mlink *mlink;
770: char *p, *start;
771:
772: assert(use_all);
773:
774: if (0 == strncmp(file, "./", 2))
775: file += 2;
776:
1.96 schwarze 777: /*
778: * We have to do lstat(2) before realpath(3) loses
779: * the information whether this is a symbolic link.
780: * We need to know that because for symbolic links,
781: * we want to use the orginal file name, while for
782: * regular files, we want to use the real path.
783: */
784: if (-1 == lstat(file, &st)) {
785: exitcode = (int)MANDOCLEVEL_BADARG;
786: say(file, "&lstat");
787: return;
788: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
789: exitcode = (int)MANDOCLEVEL_BADARG;
790: say(file, "Not a regular file");
791: return;
792: }
793:
794: /*
795: * We have to resolve the file name to the real path
796: * in any case for the base directory check.
797: */
1.47 schwarze 798: if (NULL == realpath(file, buf)) {
799: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 800: say(file, "&realpath");
1.47 schwarze 801: return;
1.63 schwarze 802: }
803:
1.109 schwarze 804: if (OP_TEST == op)
1.63 schwarze 805: start = buf;
1.109 schwarze 806: else if (strstr(buf, basedir) == buf)
807: start = buf + strlen(basedir);
1.63 schwarze 808: else {
1.47 schwarze 809: exitcode = (int)MANDOCLEVEL_BADARG;
810: say("", "%s: outside base directory", buf);
811: return;
1.63 schwarze 812: }
813:
1.96 schwarze 814: /*
815: * Now we are sure the file is inside our tree.
816: * If it is a symbolic link, ignore the real path
817: * and use the original name.
818: * This implies passing stuff like "cat1/../man1/foo.1"
819: * on the command line won't work. So don't do that.
820: * Note the stat(2) can still fail if the link target
821: * doesn't exist.
822: */
823: if (S_IFLNK & st.st_mode) {
824: if (-1 == stat(buf, &st)) {
825: exitcode = (int)MANDOCLEVEL_BADARG;
826: say(file, "&stat");
827: return;
828: }
1.100 schwarze 829: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
830: say(file, "Filename too long");
831: return;
832: }
1.109 schwarze 833: start = buf;
834: if (OP_TEST != op && strstr(buf, basedir) == buf)
835: start += strlen(basedir);
1.1 schwarze 836: }
1.63 schwarze 837:
1.47 schwarze 838: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 839: mlink->dform = FORM_NONE;
1.100 schwarze 840: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
841: sizeof(mlink->file)) {
842: say(start, "Filename too long");
843: return;
844: }
1.1 schwarze 845:
1.10 schwarze 846: /*
1.47 schwarze 847: * First try to guess our directory structure.
848: * If we find a separator, try to look for man* or cat*.
849: * If we find one of these and what's underneath is a directory,
850: * assume it's an architecture.
1.10 schwarze 851: */
1.47 schwarze 852: if (NULL != (p = strchr(start, '/'))) {
853: *p++ = '\0';
854: if (0 == strncmp(start, "man", 3)) {
855: mlink->dform = FORM_SRC;
1.51 schwarze 856: mlink->dsec = start + 3;
1.47 schwarze 857: } else if (0 == strncmp(start, "cat", 3)) {
858: mlink->dform = FORM_CAT;
1.51 schwarze 859: mlink->dsec = start + 3;
1.47 schwarze 860: }
1.10 schwarze 861:
1.47 schwarze 862: start = p;
863: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
864: *p++ = '\0';
1.51 schwarze 865: mlink->arch = start;
1.47 schwarze 866: start = p;
1.41 deraadt 867: }
1.47 schwarze 868: }
1.7 schwarze 869:
1.47 schwarze 870: /*
871: * Now check the file suffix.
872: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
873: */
874: p = strrchr(start, '\0');
875: while (p-- > start && '/' != *p && '.' != *p)
876: /* Loop. */ ;
877:
878: if ('.' == *p) {
879: *p++ = '\0';
1.51 schwarze 880: mlink->fsec = p;
1.47 schwarze 881: }
1.41 deraadt 882:
1.47 schwarze 883: /*
884: * Now try to parse the name.
885: * Use the filename portion of the path.
886: */
887: mlink->name = start;
888: if (NULL != (p = strrchr(start, '/'))) {
889: mlink->name = p + 1;
890: *p = '\0';
891: }
892: mlink_add(mlink, &st);
893: }
1.2 schwarze 894:
1.47 schwarze 895: static void
896: mlink_add(struct mlink *mlink, const struct stat *st)
897: {
898: struct inodev inodev;
899: struct mpage *mpage;
900: unsigned int slot;
901:
902: assert(NULL != mlink->file);
903:
1.51 schwarze 904: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
905: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
906: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
907: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 908:
909: if ('0' == *mlink->fsec) {
910: free(mlink->fsec);
911: mlink->fsec = mandoc_strdup(mlink->dsec);
912: mlink->fform = FORM_CAT;
913: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
914: mlink->fform = FORM_SRC;
915: else
916: mlink->fform = FORM_NONE;
1.1 schwarze 917:
1.47 schwarze 918: slot = ohash_qlookup(&mlinks, mlink->file);
919: assert(NULL == ohash_find(&mlinks, slot));
920: ohash_insert(&mlinks, slot, mlink);
921:
922: inodev.st_ino = st->st_ino;
923: inodev.st_dev = st->st_dev;
924: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
925: sizeof(struct inodev), inodev.st_ino);
926: mpage = ohash_find(&mpages, slot);
927: if (NULL == mpage) {
928: mpage = mandoc_calloc(1, sizeof(struct mpage));
929: mpage->inodev.st_ino = inodev.st_ino;
930: mpage->inodev.st_dev = inodev.st_dev;
931: ohash_insert(&mpages, slot, mpage);
932: } else
933: mlink->next = mpage->mlinks;
934: mpage->mlinks = mlink;
1.75 schwarze 935: mlink->mpage = mpage;
1.47 schwarze 936: }
1.1 schwarze 937:
1.47 schwarze 938: static void
939: mlink_free(struct mlink *mlink)
940: {
1.1 schwarze 941:
1.47 schwarze 942: free(mlink->dsec);
943: free(mlink->arch);
944: free(mlink->name);
945: free(mlink->fsec);
946: free(mlink);
947: }
1.1 schwarze 948:
1.47 schwarze 949: static void
950: mpages_free(void)
951: {
952: struct mpage *mpage;
953: struct mlink *mlink;
954: unsigned int slot;
955:
956: mpage = ohash_first(&mpages, &slot);
957: while (NULL != mpage) {
958: while (NULL != (mlink = mpage->mlinks)) {
959: mpage->mlinks = mlink->next;
960: mlink_free(mlink);
961: }
962: free(mpage->sec);
963: free(mpage->arch);
964: free(mpage->title);
965: free(mpage->desc);
966: free(mpage);
967: mpage = ohash_next(&mpages, &slot);
968: }
969: }
1.1 schwarze 970:
1.47 schwarze 971: /*
972: * For each mlink to the mpage, check whether the path looks like
973: * it is formatted, and if it does, check whether a source manual
974: * exists by the same name, ignoring the suffix.
975: * If both conditions hold, drop the mlink.
976: */
977: static void
978: mlinks_undupe(struct mpage *mpage)
979: {
980: char buf[PATH_MAX];
981: struct mlink **prev;
982: struct mlink *mlink;
983: char *bufp;
984:
985: mpage->form = FORM_CAT;
986: prev = &mpage->mlinks;
987: while (NULL != (mlink = *prev)) {
988: if (FORM_CAT != mlink->dform) {
989: mpage->form = FORM_NONE;
990: goto nextlink;
991: }
1.100 schwarze 992: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 993: bufp = strstr(buf, "cat");
994: assert(NULL != bufp);
995: memcpy(bufp, "man", 3);
996: if (NULL != (bufp = strrchr(buf, '.')))
997: *++bufp = '\0';
1.100 schwarze 998: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 999: if (NULL == ohash_find(&mlinks,
1.98 schwarze 1000: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 1001: goto nextlink;
1002: if (warnings)
1003: say(mlink->file, "Man source exists: %s", buf);
1004: if (use_all)
1005: goto nextlink;
1006: *prev = mlink->next;
1007: mlink_free(mlink);
1008: continue;
1009: nextlink:
1010: prev = &(*prev)->next;
1.1 schwarze 1011: }
1.47 schwarze 1012: }
1.1 schwarze 1013:
1.88 schwarze 1014: static void
1.50 schwarze 1015: mlink_check(struct mpage *mpage, struct mlink *mlink)
1016: {
1.88 schwarze 1017: struct str *str;
1018: unsigned int slot;
1.50 schwarze 1019:
1020: /*
1021: * Check whether the manual section given in a file
1022: * agrees with the directory where the file is located.
1023: * Some manuals have suffixes like (3p) on their
1024: * section number either inside the file or in the
1025: * directory name, some are linked into more than one
1026: * section, like encrypt(1) = makekey(8).
1027: */
1028:
1029: if (FORM_SRC == mpage->form &&
1.88 schwarze 1030: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1031: say(mlink->file, "Section \"%s\" manual in %s directory",
1032: mpage->sec, mlink->dsec);
1033:
1034: /*
1035: * Manual page directories exist for each kernel
1036: * architecture as returned by machine(1).
1037: * However, many manuals only depend on the
1038: * application architecture as returned by arch(1).
1039: * For example, some (2/ARM) manuals are shared
1040: * across the "armish" and "zaurus" kernel
1041: * architectures.
1042: * A few manuals are even shared across completely
1043: * different architectures, for example fdformat(1)
1044: * on amd64, i386, sparc, and sparc64.
1045: */
1046:
1.88 schwarze 1047: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1048: say(mlink->file, "Architecture \"%s\" manual in "
1049: "\"%s\" directory", mpage->arch, mlink->arch);
1050:
1.88 schwarze 1051: /*
1052: * XXX
1.90 schwarze 1053: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1054: */
1055:
1056: if (FORM_CAT == mpage->form)
1057: return;
1058:
1059: /*
1060: * Check whether this mlink
1061: * appears as a name in the NAME section.
1062: */
1.50 schwarze 1063:
1.90 schwarze 1064: slot = ohash_qlookup(&names, mlink->name);
1065: str = ohash_find(&names, slot);
1.88 schwarze 1066: assert(NULL != str);
1.90 schwarze 1067: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1068: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1069: }
1070:
1.47 schwarze 1071: /*
1072: * Run through the files in the global vector "mpages"
1073: * and add them to the database specified in "basedir".
1074: *
1075: * This handles the parsing scheme itself, using the cues of directory
1076: * and filename to determine whether the file is parsable or not.
1077: */
1078: static void
1.58 schwarze 1079: mpages_merge(struct mchars *mc, struct mparse *mp)
1.47 schwarze 1080: {
1.70 schwarze 1081: char any[] = "any";
1.58 schwarze 1082: struct ohash_info str_info;
1.75 schwarze 1083: struct mpage *mpage, *mpage_dest;
1084: struct mlink *mlink, *mlink_dest;
1.47 schwarze 1085: struct mdoc *mdoc;
1086: struct man *man;
1.75 schwarze 1087: char *sodest;
1.69 schwarze 1088: char *cp;
1.117 schwarze 1089: int fd;
1.58 schwarze 1090: unsigned int pslot;
1.47 schwarze 1091: enum mandoclevel lvl;
1092:
1093: str_info.alloc = hash_alloc;
1.106 espie 1094: str_info.calloc = hash_calloc;
1095: str_info.free = hash_free;
1.47 schwarze 1096: str_info.key_offset = offsetof(struct str, key);
1097:
1.126 schwarze 1098: if ( ! nodb)
1.64 schwarze 1099: SQL_EXEC("BEGIN TRANSACTION");
1100:
1.47 schwarze 1101: mpage = ohash_first(&mpages, &pslot);
1.126 schwarze 1102: while (mpage != NULL) {
1.47 schwarze 1103: mlinks_undupe(mpage);
1.126 schwarze 1104: if (mpage->mlinks == NULL) {
1.47 schwarze 1105: mpage = ohash_next(&mpages, &pslot);
1106: continue;
1107: }
1.1 schwarze 1108:
1.90 schwarze 1109: name_mask = NAME_MASK;
1110: ohash_init(&names, 4, &str_info);
1.47 schwarze 1111: ohash_init(&strings, 6, &str_info);
1112: mparse_reset(mp);
1113: mdoc = NULL;
1114: man = NULL;
1.81 schwarze 1115: sodest = NULL;
1116:
1.125 schwarze 1117: mparse_open(mp, &fd, mpage->mlinks->file);
1.117 schwarze 1118: if (fd == -1) {
1119: say(mpage->mlinks->file, "&open");
1120: goto nextpage;
1.81 schwarze 1121: }
1.11 schwarze 1122:
1123: /*
1.24 schwarze 1124: * Try interpreting the file as mdoc(7) or man(7)
1125: * source code, unless it is already known to be
1126: * formatted. Fall back to formatted mode.
1.11 schwarze 1127: */
1.126 schwarze 1128: if (mpage->mlinks->dform != FORM_CAT ||
1129: mpage->mlinks->fform != FORM_CAT) {
1.117 schwarze 1130: lvl = mparse_readfd(mp, fd, mpage->mlinks->file);
1.47 schwarze 1131: if (lvl < MANDOCLEVEL_FATAL)
1.75 schwarze 1132: mparse_result(mp, &mdoc, &man, &sodest);
1.47 schwarze 1133: }
1.11 schwarze 1134:
1.126 schwarze 1135: if (sodest != NULL) {
1.75 schwarze 1136: mlink_dest = ohash_find(&mlinks,
1137: ohash_qlookup(&mlinks, sodest));
1.126 schwarze 1138: if (mlink_dest == NULL) {
1139: mandoc_asprintf(&cp, "%s.gz", sodest);
1140: mlink_dest = ohash_find(&mlinks,
1141: ohash_qlookup(&mlinks, cp));
1142: free(cp);
1143: }
1144: if (mlink_dest != NULL) {
1.75 schwarze 1145:
1146: /* The .so target exists. */
1147:
1148: mpage_dest = mlink_dest->mpage;
1149: mlink = mpage->mlinks;
1150: while (1) {
1151: mlink->mpage = mpage_dest;
1152:
1153: /*
1154: * If the target was already
1155: * processed, add the links
1156: * to the database now.
1157: * Otherwise, this will
1158: * happen when we come
1159: * to the target.
1160: */
1161:
1.94 schwarze 1162: if (mpage_dest->pageid)
1.124 schwarze 1163: dbadd_mlink_name(mlink);
1.75 schwarze 1164:
1.126 schwarze 1165: if (mlink->next == NULL)
1.75 schwarze 1166: break;
1167: mlink = mlink->next;
1168: }
1169:
1170: /* Move all links to the target. */
1171:
1172: mlink->next = mlink_dest->next;
1173: mlink_dest->next = mpage->mlinks;
1174: mpage->mlinks = NULL;
1175: }
1.81 schwarze 1176: goto nextpage;
1.126 schwarze 1177: } else if (mdoc != NULL) {
1.47 schwarze 1178: mpage->form = FORM_SRC;
1.112 schwarze 1179: mpage->sec = mdoc_meta(mdoc)->msec;
1180: mpage->sec = mandoc_strdup(
1.126 schwarze 1181: mpage->sec == NULL ? "" : mpage->sec);
1.47 schwarze 1182: mpage->arch = mdoc_meta(mdoc)->arch;
1183: mpage->arch = mandoc_strdup(
1.126 schwarze 1184: mpage->arch == NULL ? "" : mpage->arch);
1.47 schwarze 1185: mpage->title =
1186: mandoc_strdup(mdoc_meta(mdoc)->title);
1.126 schwarze 1187: } else if (man != NULL) {
1.47 schwarze 1188: mpage->form = FORM_SRC;
1189: mpage->sec =
1190: mandoc_strdup(man_meta(man)->msec);
1191: mpage->arch =
1192: mandoc_strdup(mpage->mlinks->arch);
1193: mpage->title =
1194: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1195: } else {
1.47 schwarze 1196: mpage->form = FORM_CAT;
1197: mpage->sec =
1198: mandoc_strdup(mpage->mlinks->dsec);
1199: mpage->arch =
1200: mandoc_strdup(mpage->mlinks->arch);
1201: mpage->title =
1202: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1203: }
1.54 schwarze 1204: putkey(mpage, mpage->sec, TYPE_sec);
1.119 schwarze 1205: if (*mpage->arch != '\0')
1206: putkey(mpage, mpage->arch, TYPE_arch);
1.1 schwarze 1207:
1.54 schwarze 1208: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1209: if ('\0' != *mlink->dsec)
1210: putkey(mpage, mlink->dsec, TYPE_sec);
1211: if ('\0' != *mlink->fsec)
1212: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1213: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1214: any : mlink->arch, TYPE_arch);
1.90 schwarze 1215: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1216: }
1.41 deraadt 1217:
1.92 schwarze 1218: assert(NULL == mpage->desc);
1.47 schwarze 1219: if (NULL != mdoc) {
1220: if (NULL != (cp = mdoc_meta(mdoc)->name))
1.90 schwarze 1221: putkey(mpage, cp, NAME_HEAD);
1.127 schwarze 1222: parse_mdoc(mpage, mdoc_meta(mdoc), mdoc_node(mdoc));
1.47 schwarze 1223: } else if (NULL != man)
1224: parse_man(mpage, man_node(man));
1225: else
1.117 schwarze 1226: parse_cat(mpage, fd);
1.92 schwarze 1227: if (NULL == mpage->desc)
1228: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1229:
1230: if (warnings && !use_all)
1231: for (mlink = mpage->mlinks; mlink;
1232: mlink = mlink->next)
1233: mlink_check(mpage, mlink);
1.6 schwarze 1234:
1.62 schwarze 1235: dbadd(mpage, mc);
1.81 schwarze 1236:
1237: nextpage:
1.125 schwarze 1238: if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1.117 schwarze 1239: exitcode = (int)MANDOCLEVEL_SYSERR;
1240: say(mpage->mlinks->file, "&wait gunzip");
1.81 schwarze 1241: }
1.47 schwarze 1242: ohash_delete(&strings);
1.90 schwarze 1243: ohash_delete(&names);
1.47 schwarze 1244: mpage = ohash_next(&mpages, &pslot);
1245: }
1.64 schwarze 1246:
1247: if (0 == nodb)
1248: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1249: }
1250:
1251: static void
1252: names_check(void)
1253: {
1254: sqlite3_stmt *stmt;
1255: const char *name, *sec, *arch, *key;
1256: int irc;
1257:
1258: sqlite3_prepare_v2(db,
1259: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1260: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1261: "WHERE bits & ? AND NOT EXISTS ("
1262: "SELECT pageid FROM mlinks "
1.90 schwarze 1263: "WHERE mlinks.pageid == names.pageid "
1264: "AND mlinks.name == names.name"
1.87 schwarze 1265: ")"
1266: ") JOIN ("
1.99 schwarze 1267: "SELECT sec, arch, name, pageid FROM mlinks "
1268: "GROUP BY pageid"
1.87 schwarze 1269: ") USING (pageid);",
1270: -1, &stmt, NULL);
1271:
1.91 schwarze 1272: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1273: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1274:
1275: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.113 schwarze 1276: name = (const char *)sqlite3_column_text(stmt, 0);
1277: sec = (const char *)sqlite3_column_text(stmt, 1);
1278: arch = (const char *)sqlite3_column_text(stmt, 2);
1279: key = (const char *)sqlite3_column_text(stmt, 3);
1.87 schwarze 1280: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1281: '\0' == *arch ? "" : "/",
1282: '\0' == *arch ? "" : arch, key);
1283: }
1284: sqlite3_finalize(stmt);
1.47 schwarze 1285: }
1.6 schwarze 1286:
1.47 schwarze 1287: static void
1.81 schwarze 1288: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1289: {
1290: FILE *stream;
1291: char *line, *p, *title;
1292: size_t len, plen, titlesz;
1.1 schwarze 1293:
1.81 schwarze 1294: stream = (-1 == fd) ?
1295: fopen(mpage->mlinks->file, "r") :
1296: fdopen(fd, "r");
1297: if (NULL == stream) {
1.114 doug 1298: if (-1 != fd)
1299: close(fd);
1.47 schwarze 1300: if (warnings)
1.80 schwarze 1301: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1302: return;
1303: }
1.1 schwarze 1304:
1.47 schwarze 1305: /* Skip to first blank line. */
1.1 schwarze 1306:
1.47 schwarze 1307: while (NULL != (line = fgetln(stream, &len)))
1308: if ('\n' == *line)
1309: break;
1.1 schwarze 1310:
1.47 schwarze 1311: /*
1312: * Assume the first line that is not indented
1313: * is the first section header. Skip to it.
1314: */
1.1 schwarze 1315:
1.47 schwarze 1316: while (NULL != (line = fgetln(stream, &len)))
1317: if ('\n' != *line && ' ' != *line)
1318: break;
1.98 schwarze 1319:
1.47 schwarze 1320: /*
1321: * Read up until the next section into a buffer.
1322: * Strip the leading and trailing newline from each read line,
1323: * appending a trailing space.
1324: * Ignore empty (whitespace-only) lines.
1325: */
1.28 schwarze 1326:
1.47 schwarze 1327: titlesz = 0;
1328: title = NULL;
1.38 schwarze 1329:
1.47 schwarze 1330: while (NULL != (line = fgetln(stream, &len))) {
1331: if (' ' != *line || '\n' != line[len - 1])
1332: break;
1333: while (len > 0 && isspace((unsigned char)*line)) {
1334: line++;
1335: len--;
1336: }
1337: if (1 == len)
1338: continue;
1339: title = mandoc_realloc(title, titlesz + len);
1340: memcpy(title + titlesz, line, len);
1341: titlesz += len;
1342: title[titlesz - 1] = ' ';
1343: }
1.28 schwarze 1344:
1.47 schwarze 1345: /*
1346: * If no page content can be found, or the input line
1347: * is already the next section header, or there is no
1348: * trailing newline, reuse the page title as the page
1349: * description.
1350: */
1.1 schwarze 1351:
1.47 schwarze 1352: if (NULL == title || '\0' == *title) {
1353: if (warnings)
1354: say(mpage->mlinks->file,
1355: "Cannot find NAME section");
1356: fclose(stream);
1357: free(title);
1358: return;
1359: }
1.24 schwarze 1360:
1.47 schwarze 1361: title = mandoc_realloc(title, titlesz + 1);
1362: title[titlesz] = '\0';
1.24 schwarze 1363:
1.47 schwarze 1364: /*
1365: * Skip to the first dash.
1366: * Use the remaining line as the description (no more than 70
1367: * bytes).
1368: */
1.28 schwarze 1369:
1.47 schwarze 1370: if (NULL != (p = strstr(title, "- "))) {
1371: for (p += 2; ' ' == *p || '\b' == *p; p++)
1372: /* Skip to next word. */ ;
1373: } else {
1374: if (warnings)
1375: say(mpage->mlinks->file,
1376: "No dash in title line");
1377: p = title;
1378: }
1.1 schwarze 1379:
1.47 schwarze 1380: plen = strlen(p);
1.1 schwarze 1381:
1.47 schwarze 1382: /* Strip backspace-encoding from line. */
1.1 schwarze 1383:
1.47 schwarze 1384: while (NULL != (line = memchr(p, '\b', plen))) {
1385: len = line - p;
1386: if (0 == len) {
1387: memmove(line, line + 1, plen--);
1388: continue;
1.98 schwarze 1389: }
1.47 schwarze 1390: memmove(line - 1, line + 1, plen - len);
1391: plen -= 2;
1392: }
1.1 schwarze 1393:
1.47 schwarze 1394: mpage->desc = mandoc_strdup(p);
1395: fclose(stream);
1396: free(title);
1397: }
1.16 schwarze 1398:
1.47 schwarze 1399: /*
1400: * Put a type/word pair into the word database for this particular file.
1401: */
1402: static void
1.69 schwarze 1403: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1404: {
1.69 schwarze 1405: char *cp;
1.37 schwarze 1406:
1.47 schwarze 1407: assert(NULL != value);
1.69 schwarze 1408: if (TYPE_arch == type)
1409: for (cp = value; *cp; cp++)
1410: if (isupper((unsigned char)*cp))
1411: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1412: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1413: }
1414:
1415: /*
1.47 schwarze 1416: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1417: */
1418: static void
1.47 schwarze 1419: putmdockey(const struct mpage *mpage,
1420: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1421: {
1.16 schwarze 1422:
1.47 schwarze 1423: for ( ; NULL != n; n = n->next) {
1424: if (NULL != n->child)
1425: putmdockey(mpage, n->child, m);
1426: if (MDOC_TEXT == n->type)
1427: putkey(mpage, n->string, m);
1428: }
1429: }
1.16 schwarze 1430:
1.47 schwarze 1431: static void
1432: parse_man(struct mpage *mpage, const struct man_node *n)
1433: {
1434: const struct man_node *head, *body;
1.78 schwarze 1435: char *start, *title;
1.47 schwarze 1436: char byte;
1.78 schwarze 1437: size_t sz;
1.16 schwarze 1438:
1.47 schwarze 1439: if (NULL == n)
1440: return;
1.16 schwarze 1441:
1.47 schwarze 1442: /*
1443: * We're only searching for one thing: the first text child in
1444: * the BODY of a NAME section. Since we don't keep track of
1445: * sections in -man, run some hoops to find out whether we're in
1446: * the correct section or not.
1447: */
1.16 schwarze 1448:
1.47 schwarze 1449: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1450: body = n;
1451: assert(body->parent);
1452: if (NULL != (head = body->parent->head) &&
1.98 schwarze 1453: 1 == head->nchild &&
1454: NULL != (head = (head->child)) &&
1455: MAN_TEXT == head->type &&
1456: 0 == strcmp(head->string, "NAME") &&
1457: NULL != body->child) {
1.2 schwarze 1458:
1.47 schwarze 1459: /*
1460: * Suck the entire NAME section into memory.
1461: * Yes, we might run away.
1462: * But too many manuals have big, spread-out
1463: * NAME sections over many lines.
1464: */
1.2 schwarze 1465:
1.78 schwarze 1466: title = NULL;
1467: man_deroff(&title, body);
1.47 schwarze 1468: if (NULL == title)
1469: return;
1.16 schwarze 1470:
1.98 schwarze 1471: /*
1.47 schwarze 1472: * Go through a special heuristic dance here.
1473: * Conventionally, one or more manual names are
1474: * comma-specified prior to a whitespace, then a
1475: * dash, then a description. Try to puzzle out
1476: * the name parts here.
1477: */
1.16 schwarze 1478:
1.78 schwarze 1479: start = title;
1.47 schwarze 1480: for ( ;; ) {
1481: sz = strcspn(start, " ,");
1482: if ('\0' == start[sz])
1483: break;
1.1 schwarze 1484:
1.47 schwarze 1485: byte = start[sz];
1486: start[sz] = '\0';
1.67 schwarze 1487:
1488: /*
1489: * Assume a stray trailing comma in the
1490: * name list if a name begins with a dash.
1491: */
1492:
1493: if ('-' == start[0] ||
1494: ('\\' == start[0] && '-' == start[1]))
1495: break;
1.1 schwarze 1496:
1.90 schwarze 1497: putkey(mpage, start, NAME_TITLE);
1.1 schwarze 1498:
1.47 schwarze 1499: if (' ' == byte) {
1500: start += sz + 1;
1501: break;
1502: }
1.1 schwarze 1503:
1.47 schwarze 1504: assert(',' == byte);
1505: start += sz + 1;
1506: while (' ' == *start)
1507: start++;
1508: }
1.1 schwarze 1509:
1.78 schwarze 1510: if (start == title) {
1.90 schwarze 1511: putkey(mpage, start, NAME_TITLE);
1.47 schwarze 1512: free(title);
1513: return;
1514: }
1.1 schwarze 1515:
1.47 schwarze 1516: while (isspace((unsigned char)*start))
1517: start++;
1.1 schwarze 1518:
1.47 schwarze 1519: if (0 == strncmp(start, "-", 1))
1520: start += 1;
1521: else if (0 == strncmp(start, "\\-\\-", 4))
1522: start += 4;
1523: else if (0 == strncmp(start, "\\-", 2))
1524: start += 2;
1525: else if (0 == strncmp(start, "\\(en", 4))
1526: start += 4;
1527: else if (0 == strncmp(start, "\\(em", 4))
1528: start += 4;
1.1 schwarze 1529:
1.47 schwarze 1530: while (' ' == *start)
1531: start++;
1.1 schwarze 1532:
1.47 schwarze 1533: mpage->desc = mandoc_strdup(start);
1534: free(title);
1535: return;
1536: }
1537: }
1.1 schwarze 1538:
1.47 schwarze 1539: for (n = n->child; n; n = n->next) {
1540: if (NULL != mpage->desc)
1541: break;
1542: parse_man(mpage, n);
1.1 schwarze 1543: }
1544: }
1545:
1546: static void
1.127 schwarze 1547: parse_mdoc(struct mpage *mpage, const struct mdoc_meta *meta,
1548: const struct mdoc_node *n)
1.1 schwarze 1549: {
1550:
1.47 schwarze 1551: assert(NULL != n);
1552: for (n = n->child; NULL != n; n = n->next) {
1553: switch (n->type) {
1.98 schwarze 1554: case MDOC_ELEM:
1.47 schwarze 1555: /* FALLTHROUGH */
1.98 schwarze 1556: case MDOC_BLOCK:
1.47 schwarze 1557: /* FALLTHROUGH */
1.98 schwarze 1558: case MDOC_HEAD:
1.47 schwarze 1559: /* FALLTHROUGH */
1.98 schwarze 1560: case MDOC_BODY:
1.47 schwarze 1561: /* FALLTHROUGH */
1.98 schwarze 1562: case MDOC_TAIL:
1.47 schwarze 1563: if (NULL != mdocs[n->tok].fp)
1.127 schwarze 1564: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.47 schwarze 1565: break;
1566: if (mdocs[n->tok].mask)
1567: putmdockey(mpage, n->child,
1568: mdocs[n->tok].mask);
1569: break;
1570: default:
1571: assert(MDOC_ROOT != n->type);
1572: continue;
1573: }
1574: if (NULL != n->child)
1.127 schwarze 1575: parse_mdoc(mpage, meta, n);
1.1 schwarze 1576: }
1577: }
1578:
1.19 schwarze 1579: static int
1.127 schwarze 1580: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_meta *meta,
1581: const struct mdoc_node *n)
1.1 schwarze 1582: {
1583: const char *start, *end;
1584: size_t sz;
1.19 schwarze 1585:
1.47 schwarze 1586: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1587: NULL == (n = n->child) ||
1588: MDOC_TEXT != n->type)
1.19 schwarze 1589: return(0);
1.1 schwarze 1590:
1591: /*
1592: * Only consider those `Fd' macro fields that begin with an
1593: * "inclusion" token (versus, e.g., #define).
1594: */
1.47 schwarze 1595:
1.1 schwarze 1596: if (strcmp("#include", n->string))
1.19 schwarze 1597: return(0);
1.1 schwarze 1598:
1599: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1600: return(0);
1.1 schwarze 1601:
1602: /*
1603: * Strip away the enclosing angle brackets and make sure we're
1604: * not zero-length.
1605: */
1606:
1607: start = n->string;
1608: if ('<' == *start || '"' == *start)
1609: start++;
1610:
1611: if (0 == (sz = strlen(start)))
1.19 schwarze 1612: return(0);
1.1 schwarze 1613:
1614: end = &start[(int)sz - 1];
1615: if ('>' == *end || '"' == *end)
1616: end--;
1617:
1.47 schwarze 1618: if (end > start)
1619: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1620: return(0);
1.1 schwarze 1621: }
1622:
1.19 schwarze 1623: static int
1.127 schwarze 1624: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_meta *meta,
1625: const struct mdoc_node *n)
1.1 schwarze 1626: {
1.69 schwarze 1627: char *cp;
1.1 schwarze 1628:
1.47 schwarze 1629: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.19 schwarze 1630: return(0);
1631:
1.98 schwarze 1632: /*
1.47 schwarze 1633: * Parse: .Fn "struct type *name" "char *arg".
1.98 schwarze 1634: * First strip away pointer symbol.
1.47 schwarze 1635: * Then store the function name, then type.
1.98 schwarze 1636: * Finally, store the arguments.
1.47 schwarze 1637: */
1.1 schwarze 1638:
1.47 schwarze 1639: if (NULL == (cp = strrchr(n->string, ' ')))
1640: cp = n->string;
1.1 schwarze 1641:
1642: while ('*' == *cp)
1643: cp++;
1644:
1.47 schwarze 1645: putkey(mpage, cp, TYPE_Fn);
1.128 ! schwarze 1646: if (n->sec == SEC_SYNOPSIS)
! 1647: putkey(mpage, cp, NAME_SYN);
1.19 schwarze 1648:
1.47 schwarze 1649: if (n->string < cp)
1650: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.19 schwarze 1651:
1.47 schwarze 1652: for (n = n->next; NULL != n; n = n->next)
1653: if (MDOC_TEXT == n->type)
1654: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1655:
1.128 ! schwarze 1656: return(0);
! 1657: }
! 1658:
! 1659: static int
! 1660: parse_mdoc_Fo(struct mpage *mpage, const struct mdoc_meta *meta,
! 1661: const struct mdoc_node *n)
! 1662: {
! 1663:
! 1664: putmdockey(mpage, n->child, TYPE_Fn);
! 1665: if (n->sec == SEC_SYNOPSIS)
! 1666: putmdockey(mpage, n->child, NAME_SYN);
1.19 schwarze 1667: return(0);
1.1 schwarze 1668: }
1669:
1.19 schwarze 1670: static int
1.127 schwarze 1671: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_meta *meta,
1672: const struct mdoc_node *n)
1.1 schwarze 1673: {
1.47 schwarze 1674: char *cp;
1.1 schwarze 1675:
1676: if (NULL == (n = n->child))
1.19 schwarze 1677: return(0);
1.1 schwarze 1678:
1.47 schwarze 1679: if (NULL == n->next) {
1680: putkey(mpage, n->string, TYPE_Xr);
1681: return(0);
1682: }
1.1 schwarze 1683:
1.77 schwarze 1684: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1685: putkey(mpage, cp, TYPE_Xr);
1686: free(cp);
1687: return(0);
1.1 schwarze 1688: }
1689:
1.19 schwarze 1690: static int
1.127 schwarze 1691: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_meta *meta,
1692: const struct mdoc_node *n)
1.1 schwarze 1693: {
1694:
1.79 schwarze 1695: if (MDOC_BODY == n->type)
1696: mdoc_deroff(&mpage->desc, n);
1697: return(0);
1.1 schwarze 1698: }
1699:
1.19 schwarze 1700: static int
1.127 schwarze 1701: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_meta *meta,
1702: const struct mdoc_node *n)
1.1 schwarze 1703: {
1704:
1.86 schwarze 1705: if (SEC_NAME == n->sec)
1.90 schwarze 1706: putmdockey(mpage, n->child, NAME_TITLE);
1.127 schwarze 1707: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) {
1708: if (n->child == NULL)
1709: putkey(mpage, meta->name, NAME_SYN);
1710: else
1711: putmdockey(mpage, n->child, NAME_SYN);
1712: }
1.86 schwarze 1713: return(0);
1.1 schwarze 1714: }
1715:
1.19 schwarze 1716: static int
1.127 schwarze 1717: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_meta *meta,
1718: const struct mdoc_node *n)
1.1 schwarze 1719: {
1720:
1.19 schwarze 1721: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1722: }
1723:
1.47 schwarze 1724: static int
1.127 schwarze 1725: parse_mdoc_head(struct mpage *mpage, const struct mdoc_meta *meta,
1726: const struct mdoc_node *n)
1.1 schwarze 1727: {
1728:
1.47 schwarze 1729: return(MDOC_HEAD == n->type);
1730: }
1.1 schwarze 1731:
1.47 schwarze 1732: static int
1.127 schwarze 1733: parse_mdoc_body(struct mpage *mpage, const struct mdoc_meta *meta,
1734: const struct mdoc_node *n)
1.47 schwarze 1735: {
1.1 schwarze 1736:
1.47 schwarze 1737: return(MDOC_BODY == n->type);
1.1 schwarze 1738: }
1739:
1.47 schwarze 1740: /*
1741: * Add a string to the hash table for the current manual.
1742: * Each string has a bitmask telling which macros it belongs to.
1743: * When we finish the manual, we'll dump the table.
1744: */
1.1 schwarze 1745: static void
1.47 schwarze 1746: putkeys(const struct mpage *mpage,
1747: const char *cp, size_t sz, uint64_t v)
1.1 schwarze 1748: {
1.90 schwarze 1749: struct ohash *htab;
1.47 schwarze 1750: struct str *s;
1.68 schwarze 1751: const char *end;
1.47 schwarze 1752: unsigned int slot;
1.68 schwarze 1753: int i;
1.1 schwarze 1754:
1.47 schwarze 1755: if (0 == sz)
1756: return;
1.68 schwarze 1757:
1.90 schwarze 1758: if (TYPE_Nm & v) {
1759: htab = &names;
1760: v &= name_mask;
1.124 schwarze 1761: if (v & NAME_FIRST)
1762: name_mask &= ~NAME_FIRST;
1.90 schwarze 1763: if (debug > 1)
1764: say(mpage->mlinks->file,
1765: "Adding name %*s", sz, cp);
1766: } else {
1767: htab = &strings;
1768: if (debug > 1)
1769: for (i = 0; i < mansearch_keymax; i++)
1.118 schwarze 1770: if ((uint64_t)1 << i & v)
1.90 schwarze 1771: say(mpage->mlinks->file,
1772: "Adding key %s=%*s",
1773: mansearch_keynames[i], sz, cp);
1.68 schwarze 1774: }
1.47 schwarze 1775:
1776: end = cp + sz;
1.90 schwarze 1777: slot = ohash_qlookupi(htab, cp, &end);
1778: s = ohash_find(htab, slot);
1.1 schwarze 1779:
1.47 schwarze 1780: if (NULL != s && mpage == s->mpage) {
1781: s->mask |= v;
1.1 schwarze 1782: return;
1.47 schwarze 1783: } else if (NULL == s) {
1.101 schwarze 1784: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1785: memcpy(s->key, cp, sz);
1.90 schwarze 1786: ohash_insert(htab, slot, s);
1.47 schwarze 1787: }
1788: s->mpage = mpage;
1789: s->mask = v;
1.1 schwarze 1790: }
1791:
1792: /*
1.47 schwarze 1793: * Take a Unicode codepoint and produce its UTF-8 encoding.
1794: * This isn't the best way to do this, but it works.
1795: * The magic numbers are from the UTF-8 packaging.
1796: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1797: */
1.47 schwarze 1798: static size_t
1799: utf8(unsigned int cp, char out[7])
1.1 schwarze 1800: {
1.47 schwarze 1801: size_t rc;
1.1 schwarze 1802:
1.47 schwarze 1803: rc = 0;
1804: if (cp <= 0x0000007F) {
1805: rc = 1;
1806: out[0] = (char)cp;
1807: } else if (cp <= 0x000007FF) {
1808: rc = 2;
1809: out[0] = (cp >> 6 & 31) | 192;
1810: out[1] = (cp & 63) | 128;
1811: } else if (cp <= 0x0000FFFF) {
1812: rc = 3;
1813: out[0] = (cp >> 12 & 15) | 224;
1814: out[1] = (cp >> 6 & 63) | 128;
1815: out[2] = (cp & 63) | 128;
1816: } else if (cp <= 0x001FFFFF) {
1817: rc = 4;
1818: out[0] = (cp >> 18 & 7) | 240;
1819: out[1] = (cp >> 12 & 63) | 128;
1820: out[2] = (cp >> 6 & 63) | 128;
1821: out[3] = (cp & 63) | 128;
1822: } else if (cp <= 0x03FFFFFF) {
1823: rc = 5;
1824: out[0] = (cp >> 24 & 3) | 248;
1825: out[1] = (cp >> 18 & 63) | 128;
1826: out[2] = (cp >> 12 & 63) | 128;
1827: out[3] = (cp >> 6 & 63) | 128;
1828: out[4] = (cp & 63) | 128;
1829: } else if (cp <= 0x7FFFFFFF) {
1830: rc = 6;
1831: out[0] = (cp >> 30 & 1) | 252;
1832: out[1] = (cp >> 24 & 63) | 128;
1833: out[2] = (cp >> 18 & 63) | 128;
1834: out[3] = (cp >> 12 & 63) | 128;
1835: out[4] = (cp >> 6 & 63) | 128;
1836: out[5] = (cp & 63) | 128;
1837: } else
1838: return(0);
1.19 schwarze 1839:
1.47 schwarze 1840: out[rc] = '\0';
1841: return(rc);
1.1 schwarze 1842: }
1843:
1.47 schwarze 1844: /*
1.53 schwarze 1845: * Store the rendered version of a key, or alias the pointer
1846: * if the key contains no escape sequences.
1.47 schwarze 1847: */
1848: static void
1.53 schwarze 1849: render_key(struct mchars *mc, struct str *key)
1.1 schwarze 1850: {
1.47 schwarze 1851: size_t sz, bsz, pos;
1.71 schwarze 1852: char utfbuf[7], res[6];
1.47 schwarze 1853: char *buf;
1854: const char *seq, *cpp, *val;
1855: int len, u;
1856: enum mandoc_esc esc;
1857:
1.53 schwarze 1858: assert(NULL == key->rendered);
1.47 schwarze 1859:
1860: res[0] = '\\';
1861: res[1] = '\t';
1862: res[2] = ASCII_NBRSP;
1863: res[3] = ASCII_HYPH;
1.71 schwarze 1864: res[4] = ASCII_BREAK;
1865: res[5] = '\0';
1.1 schwarze 1866:
1.47 schwarze 1867: val = key->key;
1868: bsz = strlen(val);
1.1 schwarze 1869:
1870: /*
1.47 schwarze 1871: * Pre-check: if we have no stop-characters, then set the
1872: * pointer as ourselvse and get out of here.
1.1 schwarze 1873: */
1.47 schwarze 1874: if (strcspn(val, res) == bsz) {
1.53 schwarze 1875: key->rendered = key->key;
1.47 schwarze 1876: return;
1.98 schwarze 1877: }
1.1 schwarze 1878:
1.47 schwarze 1879: /* Pre-allocate by the length of the input */
1.39 schwarze 1880:
1.47 schwarze 1881: buf = mandoc_malloc(++bsz);
1882: pos = 0;
1.39 schwarze 1883:
1.47 schwarze 1884: while ('\0' != *val) {
1885: /*
1886: * Halt on the first escape sequence.
1887: * This also halts on the end of string, in which case
1888: * we just copy, fallthrough, and exit the loop.
1889: */
1890: if ((sz = strcspn(val, res)) > 0) {
1891: memcpy(&buf[pos], val, sz);
1892: pos += sz;
1893: val += sz;
1894: }
1.39 schwarze 1895:
1.71 schwarze 1896: switch (*val) {
1.98 schwarze 1897: case ASCII_HYPH:
1.47 schwarze 1898: buf[pos++] = '-';
1899: val++;
1900: continue;
1.98 schwarze 1901: case '\t':
1.71 schwarze 1902: /* FALLTHROUGH */
1.98 schwarze 1903: case ASCII_NBRSP:
1.47 schwarze 1904: buf[pos++] = ' ';
1905: val++;
1.71 schwarze 1906: /* FALLTHROUGH */
1.98 schwarze 1907: case ASCII_BREAK:
1.47 schwarze 1908: continue;
1.71 schwarze 1909: default:
1910: break;
1911: }
1912: if ('\\' != *val)
1.47 schwarze 1913: break;
1.39 schwarze 1914:
1.47 schwarze 1915: /* Read past the slash. */
1.39 schwarze 1916:
1.47 schwarze 1917: val++;
1.39 schwarze 1918:
1.47 schwarze 1919: /*
1920: * Parse the escape sequence and see if it's a
1921: * predefined character or special character.
1922: */
1.52 schwarze 1923:
1.98 schwarze 1924: esc = mandoc_escape((const char **)&val,
1925: &seq, &len);
1.47 schwarze 1926: if (ESCAPE_ERROR == esc)
1927: break;
1928: if (ESCAPE_SPECIAL != esc)
1929: continue;
1.39 schwarze 1930:
1.47 schwarze 1931: /*
1.52 schwarze 1932: * Render the special character
1933: * as either UTF-8 or ASCII.
1.47 schwarze 1934: */
1.52 schwarze 1935:
1936: if (write_utf8) {
1.122 schwarze 1937: if ((u = mchars_spec2cp(mc, seq, len)) <= 0)
1.52 schwarze 1938: continue;
1939: cpp = utfbuf;
1940: if (0 == (sz = utf8(u, utfbuf)))
1941: continue;
1942: sz = strlen(cpp);
1943: } else {
1944: cpp = mchars_spec2str(mc, seq, len, &sz);
1945: if (NULL == cpp)
1946: continue;
1947: if (ASCII_NBRSP == *cpp) {
1948: cpp = " ";
1949: sz = 1;
1950: }
1951: }
1.1 schwarze 1952:
1.47 schwarze 1953: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1954:
1.47 schwarze 1955: bsz += sz;
1956: buf = mandoc_realloc(buf, bsz);
1957: memcpy(&buf[pos], cpp, sz);
1958: pos += sz;
1.1 schwarze 1959: }
1960:
1.47 schwarze 1961: buf[pos] = '\0';
1.53 schwarze 1962: key->rendered = buf;
1.1 schwarze 1963: }
1964:
1.75 schwarze 1965: static void
1966: dbadd_mlink(const struct mlink *mlink)
1967: {
1968: size_t i;
1969:
1970: i = 1;
1971: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1972: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1973: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 1974: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 1975: SQL_STEP(stmts[STMT_INSERT_LINK]);
1976: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.124 schwarze 1977: }
1978:
1979: static void
1980: dbadd_mlink_name(const struct mlink *mlink)
1981: {
1982: size_t i;
1983:
1984: dbadd_mlink(mlink);
1.115 schwarze 1985:
1986: i = 1;
1.124 schwarze 1987: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, NAME_FILE & NAME_MASK);
1.115 schwarze 1988: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
1989: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
1990: SQL_STEP(stmts[STMT_INSERT_NAME]);
1991: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.75 schwarze 1992: }
1993:
1.11 schwarze 1994: /*
1.47 schwarze 1995: * Flush the current page's terms (and their bits) into the database.
1996: * Wrap the entire set of additions in a transaction to make sqlite be a
1997: * little faster.
1.53 schwarze 1998: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 1999: */
2000: static void
1.75 schwarze 2001: dbadd(struct mpage *mpage, struct mchars *mc)
1.11 schwarze 2002: {
1.47 schwarze 2003: struct mlink *mlink;
2004: struct str *key;
2005: size_t i;
2006: unsigned int slot;
2007:
1.85 schwarze 2008: mlink = mpage->mlinks;
1.11 schwarze 2009:
1.85 schwarze 2010: if (nodb) {
1.104 schwarze 2011: for (key = ohash_first(&names, &slot); NULL != key;
2012: key = ohash_next(&names, &slot)) {
2013: if (key->rendered != key->key)
2014: free(key->rendered);
2015: free(key);
2016: }
2017: for (key = ohash_first(&strings, &slot); NULL != key;
2018: key = ohash_next(&strings, &slot)) {
2019: if (key->rendered != key->key)
2020: free(key->rendered);
2021: free(key);
2022: }
1.102 schwarze 2023: if (0 == debug)
2024: return;
1.85 schwarze 2025: while (NULL != mlink) {
2026: fputs(mlink->name, stdout);
2027: if (NULL == mlink->next ||
2028: strcmp(mlink->dsec, mlink->next->dsec) ||
2029: strcmp(mlink->fsec, mlink->next->fsec) ||
2030: strcmp(mlink->arch, mlink->next->arch)) {
2031: putchar('(');
2032: if ('\0' == *mlink->dsec)
2033: fputs(mlink->fsec, stdout);
2034: else
2035: fputs(mlink->dsec, stdout);
2036: if ('\0' != *mlink->arch)
2037: printf("/%s", mlink->arch);
2038: putchar(')');
2039: }
2040: mlink = mlink->next;
2041: if (NULL != mlink)
2042: fputs(", ", stdout);
2043: }
1.89 schwarze 2044: printf(" - %s\n", mpage->desc);
1.11 schwarze 2045: return;
1.85 schwarze 2046: }
2047:
2048: if (debug)
2049: say(mlink->file, "Adding to database");
1.47 schwarze 2050:
1.105 schwarze 2051: i = strlen(mpage->desc) + 1;
2052: key = mandoc_calloc(1, sizeof(struct str) + i);
2053: memcpy(key->key, mpage->desc, i);
2054: render_key(mc, key);
2055:
1.47 schwarze 2056: i = 1;
1.105 schwarze 2057: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, key->rendered);
1.116 schwarze 2058: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.47 schwarze 2059: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2060: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2061: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.105 schwarze 2062:
2063: if (key->rendered != key->key)
2064: free(key->rendered);
2065: free(key);
1.47 schwarze 2066:
1.85 schwarze 2067: while (NULL != mlink) {
1.75 schwarze 2068: dbadd_mlink(mlink);
1.85 schwarze 2069: mlink = mlink->next;
2070: }
1.91 schwarze 2071: mlink = mpage->mlinks;
1.47 schwarze 2072:
1.90 schwarze 2073: for (key = ohash_first(&names, &slot); NULL != key;
2074: key = ohash_next(&names, &slot)) {
2075: assert(key->mpage == mpage);
2076: if (NULL == key->rendered)
2077: render_key(mc, key);
2078: i = 1;
2079: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2080: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered);
1.94 schwarze 2081: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2082: SQL_STEP(stmts[STMT_INSERT_NAME]);
2083: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2084: if (key->rendered != key->key)
2085: free(key->rendered);
2086: free(key);
2087: }
1.47 schwarze 2088: for (key = ohash_first(&strings, &slot); NULL != key;
2089: key = ohash_next(&strings, &slot)) {
2090: assert(key->mpage == mpage);
1.53 schwarze 2091: if (NULL == key->rendered)
2092: render_key(mc, key);
1.47 schwarze 2093: i = 1;
2094: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.53 schwarze 2095: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.94 schwarze 2096: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2097: SQL_STEP(stmts[STMT_INSERT_KEY]);
2098: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.53 schwarze 2099: if (key->rendered != key->key)
2100: free(key->rendered);
1.47 schwarze 2101: free(key);
1.33 schwarze 2102: }
1.47 schwarze 2103: }
1.41 deraadt 2104:
1.47 schwarze 2105: static void
2106: dbprune(void)
2107: {
2108: struct mpage *mpage;
2109: struct mlink *mlink;
2110: size_t i;
2111: unsigned int slot;
1.11 schwarze 2112:
1.63 schwarze 2113: if (0 == nodb)
2114: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2115:
1.63 schwarze 2116: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2117: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2118: mlink = mpage->mlinks;
1.82 schwarze 2119: if (debug)
1.63 schwarze 2120: say(mlink->file, "Deleting from database");
2121: if (nodb)
2122: continue;
2123: for ( ; NULL != mlink; mlink = mlink->next) {
2124: i = 1;
2125: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2126: i, mlink->dsec);
2127: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2128: i, mlink->arch);
2129: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2130: i, mlink->name);
2131: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2132: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2133: }
1.11 schwarze 2134: }
1.63 schwarze 2135:
2136: if (0 == nodb)
2137: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2138: }
1.22 schwarze 2139:
1.47 schwarze 2140: /*
2141: * Close an existing database and its prepared statements.
2142: * If "real" is not set, rename the temporary file into the real one.
2143: */
2144: static void
2145: dbclose(int real)
2146: {
2147: size_t i;
1.72 schwarze 2148: int status;
2149: pid_t child;
1.11 schwarze 2150:
1.47 schwarze 2151: if (nodb)
2152: return;
1.11 schwarze 2153:
1.47 schwarze 2154: for (i = 0; i < STMT__MAX; i++) {
2155: sqlite3_finalize(stmts[i]);
2156: stmts[i] = NULL;
1.28 schwarze 2157: }
1.22 schwarze 2158:
1.47 schwarze 2159: sqlite3_close(db);
2160: db = NULL;
1.11 schwarze 2161:
1.47 schwarze 2162: if (real)
2163: return;
1.22 schwarze 2164:
1.72 schwarze 2165: if ('\0' == *tempfilename) {
2166: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2167: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2168: say(MANDOC_DB, "&rename");
1.72 schwarze 2169: }
2170: return;
2171: }
2172:
2173: switch (child = fork()) {
1.98 schwarze 2174: case -1:
1.72 schwarze 2175: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2176: say("", "&fork cmp");
1.72 schwarze 2177: return;
1.98 schwarze 2178: case 0:
1.72 schwarze 2179: execlp("cmp", "cmp", "-s",
2180: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2181: say("", "&exec cmp");
1.72 schwarze 2182: exit(0);
2183: default:
2184: break;
2185: }
2186: if (-1 == waitpid(child, &status, 0)) {
2187: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2188: say("", "&wait cmp");
1.72 schwarze 2189: } else if (WIFSIGNALED(status)) {
2190: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2191: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2192: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2193: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2194: say(MANDOC_DB,
2195: "Data changed, but cannot replace database");
2196: }
2197:
2198: *strrchr(tempfilename, '/') = '\0';
2199: switch (child = fork()) {
1.98 schwarze 2200: case -1:
1.72 schwarze 2201: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2202: say("", "&fork rm");
1.72 schwarze 2203: return;
1.98 schwarze 2204: case 0:
1.72 schwarze 2205: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2206: say("", "&exec rm");
1.72 schwarze 2207: exit((int)MANDOCLEVEL_SYSERR);
2208: default:
2209: break;
2210: }
2211: if (-1 == waitpid(child, &status, 0)) {
2212: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2213: say("", "&wait rm");
1.72 schwarze 2214: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2215: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2216: say("", "%s: Cannot remove temporary directory",
2217: tempfilename);
1.22 schwarze 2218: }
1.11 schwarze 2219: }
2220:
1.47 schwarze 2221: /*
2222: * This is straightforward stuff.
2223: * Open a database connection to a "temporary" database, then open a set
2224: * of prepared statements we'll use over and over again.
2225: * If "real" is set, we use the existing database; if not, we truncate a
2226: * temporary one.
2227: * Must be matched by dbclose().
2228: */
2229: static int
2230: dbopen(int real)
1.2 schwarze 2231: {
1.72 schwarze 2232: const char *sql;
1.47 schwarze 2233: int rc, ofl;
1.6 schwarze 2234:
1.98 schwarze 2235: if (nodb)
1.47 schwarze 2236: return(1);
1.6 schwarze 2237:
1.72 schwarze 2238: *tempfilename = '\0';
1.47 schwarze 2239: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2240:
2241: if (real) {
2242: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2243: if (SQLITE_OK != rc) {
1.47 schwarze 2244: exitcode = (int)MANDOCLEVEL_SYSERR;
1.108 sthen 2245: if (SQLITE_CANTOPEN != rc)
2246: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.47 schwarze 2247: return(0);
1.28 schwarze 2248: }
1.72 schwarze 2249: goto prepare_statements;
2250: }
2251:
2252: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2253:
1.72 schwarze 2254: remove(MANDOC_DB "~");
2255: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.98 schwarze 2256: if (SQLITE_OK == rc)
1.72 schwarze 2257: goto create_tables;
1.73 schwarze 2258: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2259: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2260: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.47 schwarze 2261: return(0);
2262: }
1.6 schwarze 2263:
1.100 schwarze 2264: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2265: sizeof(tempfilename));
1.72 schwarze 2266: if (NULL == mkdtemp(tempfilename)) {
2267: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2268: say("", "&%s", tempfilename);
1.72 schwarze 2269: return(0);
2270: }
1.100 schwarze 2271: (void)strlcat(tempfilename, "/" MANDOC_DB,
2272: sizeof(tempfilename));
1.72 schwarze 2273: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2274: if (SQLITE_OK != rc) {
1.47 schwarze 2275: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2276: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.47 schwarze 2277: return(0);
1.2 schwarze 2278: }
2279:
1.72 schwarze 2280: create_tables:
1.47 schwarze 2281: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2282: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2283: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2284: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2285: ");\n"
2286: "\n"
2287: "CREATE TABLE \"mlinks\" (\n"
2288: " \"sec\" TEXT NOT NULL,\n"
2289: " \"arch\" TEXT NOT NULL,\n"
2290: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2291: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2292: "ON DELETE CASCADE\n"
1.47 schwarze 2293: ");\n"
1.93 schwarze 2294: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2295: "\n"
1.90 schwarze 2296: "CREATE TABLE \"names\" (\n"
2297: " \"bits\" INTEGER NOT NULL,\n"
2298: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2299: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.90 schwarze 2300: "ON DELETE CASCADE\n"
2301: ");\n"
2302: "\n"
1.47 schwarze 2303: "CREATE TABLE \"keys\" (\n"
2304: " \"bits\" INTEGER NOT NULL,\n"
2305: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2306: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2307: "ON DELETE CASCADE\n"
1.93 schwarze 2308: ");\n"
2309: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2310:
2311: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2312: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2313: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.103 schwarze 2314: sqlite3_close(db);
1.47 schwarze 2315: return(0);
1.2 schwarze 2316: }
2317:
1.47 schwarze 2318: prepare_statements:
1.103 schwarze 2319: if (SQLITE_OK != sqlite3_exec(db,
2320: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2321: exitcode = (int)MANDOCLEVEL_SYSERR;
2322: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2323: sqlite3_errmsg(db));
2324: sqlite3_close(db);
2325: return(0);
2326: }
2327:
1.94 schwarze 2328: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2329: "(SELECT pageid FROM mlinks WHERE "
2330: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2331: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2332: sql = "INSERT INTO mpages "
1.89 schwarze 2333: "(desc,form) VALUES (?,?)";
1.47 schwarze 2334: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2335: sql = "INSERT INTO mlinks "
1.61 schwarze 2336: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2337: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.90 schwarze 2338: sql = "INSERT INTO names "
2339: "(bits,name,pageid) VALUES (?,?,?)";
2340: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2341: sql = "INSERT INTO keys "
2342: "(bits,key,pageid) VALUES (?,?,?)";
2343: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2344:
1.47 schwarze 2345: /*
2346: * When opening a new database, we can turn off
2347: * synchronous mode for much better performance.
2348: */
1.6 schwarze 2349:
1.103 schwarze 2350: if (real && SQLITE_OK != sqlite3_exec(db,
2351: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2352: exitcode = (int)MANDOCLEVEL_SYSERR;
2353: say(MANDOC_DB, "PRAGMA synchronous: %s",
2354: sqlite3_errmsg(db));
2355: sqlite3_close(db);
2356: return(0);
2357: }
1.11 schwarze 2358:
1.47 schwarze 2359: return(1);
2360: }
1.6 schwarze 2361:
1.47 schwarze 2362: static void *
1.106 espie 2363: hash_calloc(size_t n, size_t sz, void *arg)
1.47 schwarze 2364: {
1.6 schwarze 2365:
1.106 espie 2366: return(mandoc_calloc(n, sz));
1.47 schwarze 2367: }
1.2 schwarze 2368:
1.47 schwarze 2369: static void *
2370: hash_alloc(size_t sz, void *arg)
2371: {
1.28 schwarze 2372:
1.47 schwarze 2373: return(mandoc_malloc(sz));
2374: }
1.6 schwarze 2375:
1.47 schwarze 2376: static void
1.106 espie 2377: hash_free(void *p, void *arg)
1.47 schwarze 2378: {
1.26 schwarze 2379:
1.47 schwarze 2380: free(p);
2381: }
1.6 schwarze 2382:
1.47 schwarze 2383: static int
1.120 schwarze 2384: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2385: {
2386: static char startdir[PATH_MAX];
1.110 schwarze 2387: static int getcwd_status; /* 1 = ok, 2 = failure */
2388: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2389: char *cp;
1.6 schwarze 2390:
1.47 schwarze 2391: /*
1.110 schwarze 2392: * Remember the original working directory, if possible.
2393: * This will be needed if the second or a later directory
2394: * on the command line is given as a relative path.
2395: * Do not error out if the current directory is not
2396: * searchable: Maybe it won't be needed after all.
2397: */
2398: if (0 == getcwd_status) {
2399: if (NULL == getcwd(startdir, sizeof(startdir))) {
2400: getcwd_status = 2;
2401: (void)strlcpy(startdir, strerror(errno),
2402: sizeof(startdir));
2403: } else
2404: getcwd_status = 1;
2405: }
2406:
2407: /*
2408: * We are leaving the old base directory.
2409: * Do not use it any longer, not even for messages.
2410: */
2411: *basedir = '\0';
2412:
2413: /*
2414: * If and only if the directory was changed earlier and
2415: * the next directory to process is given as a relative path,
2416: * first go back, or bail out if that is impossible.
1.47 schwarze 2417: */
1.110 schwarze 2418: if (chdir_status && '/' != *targetdir) {
2419: if (2 == getcwd_status) {
1.47 schwarze 2420: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2421: say("", "getcwd: %s", startdir);
1.47 schwarze 2422: return(0);
2423: }
1.110 schwarze 2424: if (-1 == chdir(startdir)) {
1.47 schwarze 2425: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2426: say("", "&chdir %s", startdir);
1.47 schwarze 2427: return(0);
1.2 schwarze 2428: }
2429: }
1.110 schwarze 2430:
2431: /*
2432: * Always resolve basedir to the canonicalized absolute
2433: * pathname and append a trailing slash, such that
2434: * we can reliably check whether files are inside.
2435: */
1.47 schwarze 2436: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2437: if (report_baddir || errno != ENOENT) {
2438: exitcode = (int)MANDOCLEVEL_BADARG;
2439: say("", "&%s: realpath", targetdir);
2440: }
1.47 schwarze 2441: return(0);
2442: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2443: if (report_baddir || errno != ENOENT) {
2444: exitcode = (int)MANDOCLEVEL_BADARG;
2445: say("", "&chdir");
2446: }
1.47 schwarze 2447: return(0);
1.109 schwarze 2448: }
1.110 schwarze 2449: chdir_status = 1;
1.109 schwarze 2450: cp = strchr(basedir, '\0');
2451: if ('/' != cp[-1]) {
2452: if (cp - basedir >= PATH_MAX - 1) {
2453: exitcode = (int)MANDOCLEVEL_SYSERR;
2454: say("", "Filename too long");
2455: return(0);
2456: }
2457: *cp++ = '/';
2458: *cp = '\0';
1.47 schwarze 2459: }
2460: return(1);
1.2 schwarze 2461: }
2462:
2463: static void
1.47 schwarze 2464: say(const char *file, const char *format, ...)
1.2 schwarze 2465: {
1.47 schwarze 2466: va_list ap;
1.80 schwarze 2467: int use_errno;
1.2 schwarze 2468:
1.47 schwarze 2469: if ('\0' != *basedir)
2470: fprintf(stderr, "%s", basedir);
2471: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2472: fputc('/', stderr);
1.47 schwarze 2473: if ('\0' != *file)
2474: fprintf(stderr, "%s", file);
1.31 schwarze 2475:
1.80 schwarze 2476: use_errno = 1;
2477: if (NULL != format) {
2478: switch (*format) {
1.98 schwarze 2479: case '&':
1.80 schwarze 2480: format++;
2481: break;
1.98 schwarze 2482: case '\0':
1.80 schwarze 2483: format = NULL;
2484: break;
2485: default:
2486: use_errno = 0;
2487: break;
2488: }
2489: }
2490: if (NULL != format) {
2491: if ('\0' != *basedir || '\0' != *file)
2492: fputs(": ", stderr);
2493: va_start(ap, format);
2494: vfprintf(stderr, format, ap);
2495: va_end(ap);
2496: }
2497: if (use_errno) {
2498: if ('\0' != *basedir || '\0' != *file || NULL != format)
2499: fputs(": ", stderr);
1.47 schwarze 2500: perror(NULL);
1.80 schwarze 2501: } else
2502: fputc('\n', stderr);
1.1 schwarze 2503: }