Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.137
1.137 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.136 2015/01/03 12:54:49 schwarze Exp $ */
1.1 schwarze 2: /*
1.47 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.136 schwarze 4: * Copyright (c) 2011-2015 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 {
77: const struct mpage *mpage; /* if set, the owning parse */
78: uint64_t mask; /* bitmask in sequence */
1.131 schwarze 79: char key[]; /* rendered text */
1.28 schwarze 80: };
81:
1.47 schwarze 82: struct inodev {
83: ino_t st_ino;
84: dev_t st_dev;
85: };
1.28 schwarze 86:
1.47 schwarze 87: struct mpage {
88: struct inodev inodev; /* used for hashing routine */
1.94 schwarze 89: int64_t pageid; /* pageid in mpages SQL table */
1.47 schwarze 90: char *sec; /* section from file content */
91: char *arch; /* architecture from file content */
92: char *title; /* title from file content */
93: char *desc; /* description from file content */
94: struct mlink *mlinks; /* singly linked list */
1.116 schwarze 95: int form; /* format from file content */
1.129 schwarze 96: int name_head_done;
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 */
1.130 schwarze 117: STMT_SELECT_NAME, /* retrieve existing name flags */
1.47 schwarze 118: STMT_INSERT_KEY, /* insert parsed key */
119: STMT__MAX
1.1 schwarze 120: };
121:
1.127 schwarze 122: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_meta *,
123: const struct mdoc_node *);
1.1 schwarze 124:
1.19 schwarze 125: struct mdoc_handler {
1.47 schwarze 126: mdoc_fp fp; /* optional handler */
127: uint64_t mask; /* set unless handler returns 0 */
1.19 schwarze 128: };
129:
1.47 schwarze 130: static void dbclose(int);
1.131 schwarze 131: static void dbadd(struct mpage *);
1.75 schwarze 132: static void dbadd_mlink(const struct mlink *mlink);
1.124 schwarze 133: static void dbadd_mlink_name(const struct mlink *mlink);
1.47 schwarze 134: static int dbopen(int);
135: static void dbprune(void);
136: static void filescan(const char *);
137: static void *hash_alloc(size_t, void *);
1.106 espie 138: static void hash_free(void *, void *);
139: static void *hash_calloc(size_t, size_t, void *);
1.47 schwarze 140: static void mlink_add(struct mlink *, const struct stat *);
1.88 schwarze 141: static void mlink_check(struct mpage *, struct mlink *);
1.47 schwarze 142: static void mlink_free(struct mlink *);
143: static void mlinks_undupe(struct mpage *);
144: static void mpages_free(void);
1.131 schwarze 145: static void mpages_merge(struct mparse *);
1.87 schwarze 146: static void names_check(void);
1.81 schwarze 147: static void parse_cat(struct mpage *, int);
1.129 schwarze 148: static void parse_man(struct mpage *, const struct man_meta *,
149: const struct man_node *);
1.127 schwarze 150: static void parse_mdoc(struct mpage *, const struct mdoc_meta *,
151: const struct mdoc_node *);
152: static int parse_mdoc_body(struct mpage *, const struct mdoc_meta *,
153: const struct mdoc_node *);
154: static int parse_mdoc_head(struct mpage *, const struct mdoc_meta *,
155: const struct mdoc_node *);
156: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_meta *,
157: const struct mdoc_node *);
1.133 schwarze 158: static void parse_mdoc_fname(struct mpage *, const struct mdoc_node *);
1.127 schwarze 159: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_meta *,
160: const struct mdoc_node *);
1.128 schwarze 161: static int parse_mdoc_Fo(struct mpage *, const struct mdoc_meta *,
162: const struct mdoc_node *);
1.127 schwarze 163: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_meta *,
164: const struct mdoc_node *);
165: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_meta *,
166: const struct mdoc_node *);
167: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_meta *,
168: const struct mdoc_node *);
169: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_meta *,
170: const struct mdoc_node *);
1.69 schwarze 171: static void putkey(const struct mpage *, char *, uint64_t);
1.131 schwarze 172: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.47 schwarze 173: static void putmdockey(const struct mpage *,
174: const struct mdoc_node *, uint64_t);
1.131 schwarze 175: static int render_string(char **, size_t *);
1.47 schwarze 176: static void say(const char *, const char *, ...);
1.120 schwarze 177: static int set_basedir(const char *, int);
1.47 schwarze 178: static int treescan(void);
179: static size_t utf8(unsigned int, char [7]);
180:
1.72 schwarze 181: static char tempfilename[32];
1.47 schwarze 182: static char *progname;
1.59 schwarze 183: static int nodb; /* no database changes */
1.73 schwarze 184: static int mparse_options; /* abort the parse early */
1.98 schwarze 185: static int use_all; /* use all found files */
186: static int debug; /* print what we're doing */
187: static int warnings; /* warn about crap */
1.52 schwarze 188: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 189: static int exitcode; /* to be returned by main */
1.98 schwarze 190: static enum op op; /* operational mode */
1.47 schwarze 191: static char basedir[PATH_MAX]; /* current base directory */
1.131 schwarze 192: static struct mchars *mchars; /* table of named characters */
1.47 schwarze 193: static struct ohash mpages; /* table of distinct manual pages */
194: static struct ohash mlinks; /* table of directory entries */
1.90 schwarze 195: static struct ohash names; /* table of all names */
1.47 schwarze 196: static struct ohash strings; /* table of all strings */
197: static sqlite3 *db = NULL; /* current database */
198: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.90 schwarze 199: static uint64_t name_mask;
1.47 schwarze 200:
1.19 schwarze 201: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 202: { NULL, 0 }, /* Ap */
203: { NULL, 0 }, /* Dd */
204: { NULL, 0 }, /* Dt */
205: { NULL, 0 }, /* Os */
206: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
207: { parse_mdoc_head, TYPE_Ss }, /* Ss */
208: { NULL, 0 }, /* Pp */
209: { NULL, 0 }, /* D1 */
210: { NULL, 0 }, /* Dl */
211: { NULL, 0 }, /* Bd */
212: { NULL, 0 }, /* Ed */
213: { NULL, 0 }, /* Bl */
214: { NULL, 0 }, /* El */
215: { NULL, 0 }, /* It */
216: { NULL, 0 }, /* Ad */
217: { NULL, TYPE_An }, /* An */
218: { NULL, TYPE_Ar }, /* Ar */
219: { NULL, TYPE_Cd }, /* Cd */
220: { NULL, TYPE_Cm }, /* Cm */
221: { NULL, TYPE_Dv }, /* Dv */
222: { NULL, TYPE_Er }, /* Er */
223: { NULL, TYPE_Ev }, /* Ev */
224: { NULL, 0 }, /* Ex */
225: { NULL, TYPE_Fa }, /* Fa */
226: { parse_mdoc_Fd, 0 }, /* Fd */
227: { NULL, TYPE_Fl }, /* Fl */
228: { parse_mdoc_Fn, 0 }, /* Fn */
229: { NULL, TYPE_Ft }, /* Ft */
230: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 231: { NULL, TYPE_In }, /* In */
1.47 schwarze 232: { NULL, TYPE_Li }, /* Li */
1.89 schwarze 233: { parse_mdoc_Nd, 0 }, /* Nd */
1.86 schwarze 234: { parse_mdoc_Nm, 0 }, /* Nm */
1.47 schwarze 235: { NULL, 0 }, /* Op */
236: { NULL, 0 }, /* Ot */
237: { NULL, TYPE_Pa }, /* Pa */
238: { NULL, 0 }, /* Rv */
1.49 schwarze 239: { NULL, TYPE_St }, /* St */
1.47 schwarze 240: { NULL, TYPE_Va }, /* Va */
241: { parse_mdoc_body, TYPE_Va }, /* Vt */
242: { parse_mdoc_Xr, 0 }, /* Xr */
243: { NULL, 0 }, /* %A */
244: { NULL, 0 }, /* %B */
245: { NULL, 0 }, /* %D */
246: { NULL, 0 }, /* %I */
247: { NULL, 0 }, /* %J */
248: { NULL, 0 }, /* %N */
249: { NULL, 0 }, /* %O */
250: { NULL, 0 }, /* %P */
251: { NULL, 0 }, /* %R */
252: { NULL, 0 }, /* %T */
253: { NULL, 0 }, /* %V */
254: { NULL, 0 }, /* Ac */
255: { NULL, 0 }, /* Ao */
256: { NULL, 0 }, /* Aq */
257: { NULL, TYPE_At }, /* At */
258: { NULL, 0 }, /* Bc */
259: { NULL, 0 }, /* Bf */
260: { NULL, 0 }, /* Bo */
261: { NULL, 0 }, /* Bq */
262: { NULL, TYPE_Bsx }, /* Bsx */
263: { NULL, TYPE_Bx }, /* Bx */
264: { NULL, 0 }, /* Db */
265: { NULL, 0 }, /* Dc */
266: { NULL, 0 }, /* Do */
267: { NULL, 0 }, /* Dq */
268: { NULL, 0 }, /* Ec */
269: { NULL, 0 }, /* Ef */
270: { NULL, TYPE_Em }, /* Em */
271: { NULL, 0 }, /* Eo */
272: { NULL, TYPE_Fx }, /* Fx */
273: { NULL, TYPE_Ms }, /* Ms */
274: { NULL, 0 }, /* No */
275: { NULL, 0 }, /* Ns */
276: { NULL, TYPE_Nx }, /* Nx */
277: { NULL, TYPE_Ox }, /* Ox */
278: { NULL, 0 }, /* Pc */
279: { NULL, 0 }, /* Pf */
280: { NULL, 0 }, /* Po */
281: { NULL, 0 }, /* Pq */
282: { NULL, 0 }, /* Qc */
283: { NULL, 0 }, /* Ql */
284: { NULL, 0 }, /* Qo */
285: { NULL, 0 }, /* Qq */
286: { NULL, 0 }, /* Re */
287: { NULL, 0 }, /* Rs */
288: { NULL, 0 }, /* Sc */
289: { NULL, 0 }, /* So */
290: { NULL, 0 }, /* Sq */
291: { NULL, 0 }, /* Sm */
292: { NULL, 0 }, /* Sx */
293: { NULL, TYPE_Sy }, /* Sy */
294: { NULL, TYPE_Tn }, /* Tn */
295: { NULL, 0 }, /* Ux */
296: { NULL, 0 }, /* Xc */
297: { NULL, 0 }, /* Xo */
1.128 schwarze 298: { parse_mdoc_Fo, 0 }, /* Fo */
1.47 schwarze 299: { NULL, 0 }, /* Fc */
300: { NULL, 0 }, /* Oo */
301: { NULL, 0 }, /* Oc */
302: { NULL, 0 }, /* Bk */
303: { NULL, 0 }, /* Ek */
304: { NULL, 0 }, /* Bt */
305: { NULL, 0 }, /* Hf */
306: { NULL, 0 }, /* Fr */
307: { NULL, 0 }, /* Ud */
308: { NULL, TYPE_Lb }, /* Lb */
309: { NULL, 0 }, /* Lp */
310: { NULL, TYPE_Lk }, /* Lk */
311: { NULL, TYPE_Mt }, /* Mt */
312: { NULL, 0 }, /* Brq */
313: { NULL, 0 }, /* Bro */
314: { NULL, 0 }, /* Brc */
315: { NULL, 0 }, /* %C */
316: { NULL, 0 }, /* Es */
317: { NULL, 0 }, /* En */
318: { NULL, TYPE_Dx }, /* Dx */
319: { NULL, 0 }, /* %Q */
320: { NULL, 0 }, /* br */
321: { NULL, 0 }, /* sp */
322: { NULL, 0 }, /* %U */
323: { NULL, 0 }, /* Ta */
1.121 schwarze 324: { NULL, 0 }, /* ll */
1.1 schwarze 325: };
326:
1.98 schwarze 327:
1.1 schwarze 328: int
1.3 schwarze 329: mandocdb(int argc, char *argv[])
1.1 schwarze 330: {
1.47 schwarze 331: int ch, i;
332: size_t j, sz;
333: const char *path_arg;
334: struct manpaths dirs;
335: struct mparse *mp;
336: struct ohash_info mpages_info, mlinks_info;
337:
338: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
339: memset(&dirs, 0, sizeof(struct manpaths));
340:
341: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
1.106 espie 342: mpages_info.calloc = mlinks_info.calloc = hash_calloc;
1.135 schwarze 343: mpages_info.free = mlinks_info.free = hash_free;
344: mpages_info.data = mlinks_info.data = NULL;
1.47 schwarze 345:
346: mpages_info.key_offset = offsetof(struct mpage, inodev);
347: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 348:
349: progname = strrchr(argv[0], '/');
350: if (progname == NULL)
351: progname = argv[0];
352: else
353: ++progname;
354:
1.47 schwarze 355: /*
1.98 schwarze 356: * We accept a few different invocations.
1.47 schwarze 357: * The CHECKOP macro makes sure that invocation styles don't
358: * clobber each other.
359: */
360: #define CHECKOP(_op, _ch) do \
361: if (OP_DEFAULT != (_op)) { \
1.111 schwarze 362: fprintf(stderr, "%s: -%c: Conflicting option\n", \
363: progname, (_ch)); \
1.47 schwarze 364: goto usage; \
365: } while (/*CONSTCOND*/0)
1.10 schwarze 366:
1.47 schwarze 367: path_arg = NULL;
1.28 schwarze 368: op = OP_DEFAULT;
1.1 schwarze 369:
1.83 schwarze 370: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 371: switch (ch) {
1.98 schwarze 372: case 'a':
1.6 schwarze 373: use_all = 1;
374: break;
1.98 schwarze 375: case 'C':
1.47 schwarze 376: CHECKOP(op, ch);
377: path_arg = optarg;
1.28 schwarze 378: op = OP_CONFFILE;
1.25 schwarze 379: break;
1.98 schwarze 380: case 'D':
1.82 schwarze 381: debug++;
382: break;
1.98 schwarze 383: case 'd':
1.47 schwarze 384: CHECKOP(op, ch);
385: path_arg = optarg;
1.2 schwarze 386: op = OP_UPDATE;
1.1 schwarze 387: break;
1.98 schwarze 388: case 'n':
1.47 schwarze 389: nodb = 1;
390: break;
1.98 schwarze 391: case 'p':
1.83 schwarze 392: warnings = 1;
393: break;
1.98 schwarze 394: case 'Q':
1.73 schwarze 395: mparse_options |= MPARSE_QUICK;
1.59 schwarze 396: break;
1.98 schwarze 397: case 'T':
1.52 schwarze 398: if (strcmp(optarg, "utf8")) {
1.111 schwarze 399: fprintf(stderr, "%s: -T%s: "
400: "Unsupported output format\n",
401: progname, optarg);
1.52 schwarze 402: goto usage;
403: }
404: write_utf8 = 1;
405: break;
1.98 schwarze 406: case 't':
1.47 schwarze 407: CHECKOP(op, ch);
1.28 schwarze 408: dup2(STDOUT_FILENO, STDERR_FILENO);
409: op = OP_TEST;
1.47 schwarze 410: nodb = warnings = 1;
1.28 schwarze 411: break;
1.98 schwarze 412: case 'u':
1.47 schwarze 413: CHECKOP(op, ch);
414: path_arg = optarg;
1.1 schwarze 415: op = OP_DELETE;
416: break;
1.98 schwarze 417: case 'v':
1.83 schwarze 418: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 419: break;
1.1 schwarze 420: default:
1.28 schwarze 421: goto usage;
1.1 schwarze 422: }
423:
424: argc -= optind;
425: argv += optind;
426:
1.28 schwarze 427: if (OP_CONFFILE == op && argc > 0) {
1.111 schwarze 428: fprintf(stderr, "%s: -C: Too many arguments\n",
429: progname);
1.28 schwarze 430: goto usage;
431: }
432:
1.47 schwarze 433: exitcode = (int)MANDOCLEVEL_OK;
1.131 schwarze 434: mchars = mchars_alloc();
1.137 ! schwarze 435: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL,
1.131 schwarze 436: mchars, NULL);
1.47 schwarze 437: ohash_init(&mpages, 6, &mpages_info);
438: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 439:
1.47 schwarze 440: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.2 schwarze 441:
1.47 schwarze 442: /*
1.109 schwarze 443: * Most of these deal with a specific directory.
1.97 schwarze 444: * Jump into that directory first.
1.47 schwarze 445: */
1.120 schwarze 446: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.47 schwarze 447: goto out;
1.97 schwarze 448:
1.95 schwarze 449: if (dbopen(1)) {
1.97 schwarze 450: /*
451: * The existing database is usable. Process
452: * all files specified on the command-line.
453: */
454: use_all = 1;
1.95 schwarze 455: for (i = 0; i < argc; i++)
456: filescan(argv[i]);
457: if (OP_TEST != op)
458: dbprune();
459: } else {
460: /*
461: * Database missing or corrupt.
462: * Recreate from scratch.
463: */
1.97 schwarze 464: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 465: op = OP_DEFAULT;
466: if (0 == treescan())
467: goto out;
468: if (0 == dbopen(0))
469: goto out;
470: }
1.47 schwarze 471: if (OP_DELETE != op)
1.131 schwarze 472: mpages_merge(mp);
1.95 schwarze 473: dbclose(OP_DEFAULT == op ? 0 : 1);
1.47 schwarze 474: } else {
475: /*
476: * If we have arguments, use them as our manpaths.
477: * If we don't, grok from manpath(1) or however else
478: * manpath_parse() wants to do it.
479: */
480: if (argc > 0) {
1.101 schwarze 481: dirs.paths = mandoc_reallocarray(NULL,
482: argc, sizeof(char *));
1.47 schwarze 483: dirs.sz = (size_t)argc;
484: for (i = 0; i < argc; i++)
485: dirs.paths[i] = mandoc_strdup(argv[i]);
486: } else
487: manpath_parse(&dirs, path_arg, NULL, NULL);
1.84 schwarze 488:
489: if (0 == dirs.sz) {
490: exitcode = (int)MANDOCLEVEL_BADARG;
491: say("", "Empty manpath");
492: }
1.2 schwarze 493:
1.47 schwarze 494: /*
495: * First scan the tree rooted at a base directory, then
496: * build a new database and finally move it into place.
497: * Ignore zero-length directories and strip trailing
498: * slashes.
499: */
500: for (j = 0; j < dirs.sz; j++) {
501: sz = strlen(dirs.paths[j]);
502: if (sz && '/' == dirs.paths[j][sz - 1])
503: dirs.paths[j][--sz] = '\0';
504: if (0 == sz)
505: continue;
1.2 schwarze 506:
1.47 schwarze 507: if (j) {
508: ohash_init(&mpages, 6, &mpages_info);
509: ohash_init(&mlinks, 6, &mlinks_info);
510: }
1.1 schwarze 511:
1.120 schwarze 512: if (0 == set_basedir(dirs.paths[j], argc > 0))
513: continue;
1.47 schwarze 514: if (0 == treescan())
1.120 schwarze 515: continue;
1.47 schwarze 516: if (0 == dbopen(0))
1.120 schwarze 517: continue;
1.41 deraadt 518:
1.131 schwarze 519: mpages_merge(mp);
1.104 schwarze 520: if (warnings && !nodb &&
1.87 schwarze 521: ! (MPARSE_QUICK & mparse_options))
522: names_check();
1.47 schwarze 523: dbclose(0);
1.28 schwarze 524:
1.47 schwarze 525: if (j + 1 < dirs.sz) {
526: mpages_free();
527: ohash_delete(&mpages);
528: ohash_delete(&mlinks);
529: }
1.2 schwarze 530: }
1.47 schwarze 531: }
532: out:
533: manpath_free(&dirs);
1.123 schwarze 534: mparse_free(mp);
1.131 schwarze 535: mchars_free(mchars);
1.47 schwarze 536: mpages_free();
537: ohash_delete(&mpages);
538: ohash_delete(&mlinks);
539: return(exitcode);
540: usage:
1.83 schwarze 541: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
542: " %s [-aDnpQ] [-Tutf8] dir ...\n"
543: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
544: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 545: " %s [-Q] -t file ...\n",
1.98 schwarze 546: progname, progname, progname,
1.47 schwarze 547: progname, progname);
1.1 schwarze 548:
1.47 schwarze 549: return((int)MANDOCLEVEL_BADARG);
550: }
1.1 schwarze 551:
1.47 schwarze 552: /*
553: * Scan a directory tree rooted at "basedir" for manpages.
554: * We use fts(), scanning directory parts along the way for clues to our
555: * section and architecture.
556: *
557: * If use_all has been specified, grok all files.
558: * If not, sanitise paths to the following:
559: *
1.98 schwarze 560: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 561: * or
562: * [./]cat<section>[/<arch>]/<name>.0
563: *
564: * TODO: accomodate for multi-language directories.
565: */
566: static int
567: treescan(void)
568: {
1.96 schwarze 569: char buf[PATH_MAX];
1.47 schwarze 570: FTS *f;
571: FTSENT *ff;
572: struct mlink *mlink;
1.81 schwarze 573: int dform, gzip;
1.51 schwarze 574: char *dsec, *arch, *fsec, *cp;
575: const char *path;
1.47 schwarze 576: const char *argv[2];
1.1 schwarze 577:
1.47 schwarze 578: argv[0] = ".";
579: argv[1] = (char *)NULL;
1.1 schwarze 580:
1.96 schwarze 581: f = fts_open((char * const *)argv,
582: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.47 schwarze 583: if (NULL == f) {
584: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 585: say("", "&fts_open");
1.47 schwarze 586: return(0);
587: }
1.2 schwarze 588:
1.47 schwarze 589: dsec = arch = NULL;
590: dform = FORM_NONE;
1.2 schwarze 591:
1.47 schwarze 592: while (NULL != (ff = fts_read(f))) {
593: path = ff->fts_path + 2;
1.96 schwarze 594: switch (ff->fts_info) {
595:
596: /*
597: * Symbolic links require various sanity checks,
598: * then get handled just like regular files.
599: */
1.98 schwarze 600: case FTS_SL:
1.96 schwarze 601: if (NULL == realpath(path, buf)) {
602: if (warnings)
603: say(path, "&realpath");
604: continue;
605: }
606: if (strstr(buf, basedir) != buf) {
607: if (warnings) say("",
608: "%s: outside base directory", buf);
609: continue;
610: }
611: /* Use logical inode to avoid mpages dupe. */
612: if (-1 == stat(path, ff->fts_statp)) {
613: if (warnings)
614: say(path, "&stat");
615: continue;
616: }
617: /* FALLTHROUGH */
618:
1.15 schwarze 619: /*
1.47 schwarze 620: * If we're a regular file, add an mlink by using the
621: * stored directory data and handling the filename.
1.15 schwarze 622: */
1.98 schwarze 623: case FTS_F:
1.47 schwarze 624: if (0 == strcmp(path, MANDOC_DB))
625: continue;
626: if ( ! use_all && ff->fts_level < 2) {
627: if (warnings)
628: say(path, "Extraneous file");
629: continue;
1.81 schwarze 630: }
631: gzip = 0;
632: fsec = NULL;
633: while (NULL == fsec) {
634: fsec = strrchr(ff->fts_name, '.');
635: if (NULL == fsec || strcmp(fsec+1, "gz"))
636: break;
637: gzip = 1;
638: *fsec = '\0';
639: fsec = NULL;
640: }
641: if (NULL == fsec) {
1.47 schwarze 642: if ( ! use_all) {
643: if (warnings)
644: say(path,
645: "No filename suffix");
646: continue;
647: }
648: } else if (0 == strcmp(++fsec, "html")) {
649: if (warnings)
650: say(path, "Skip html");
651: continue;
652: } else if (0 == strcmp(fsec, "ps")) {
653: if (warnings)
654: say(path, "Skip ps");
655: continue;
656: } else if (0 == strcmp(fsec, "pdf")) {
657: if (warnings)
658: say(path, "Skip pdf");
659: continue;
660: } else if ( ! use_all &&
661: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
662: (FORM_CAT == dform && strcmp(fsec, "0")))) {
663: if (warnings)
664: say(path, "Wrong filename suffix");
665: continue;
666: } else
667: fsec[-1] = '\0';
1.51 schwarze 668:
1.47 schwarze 669: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 670: if (strlcpy(mlink->file, path,
671: sizeof(mlink->file)) >=
672: sizeof(mlink->file)) {
673: say(path, "Filename too long");
674: free(mlink);
675: continue;
676: }
1.47 schwarze 677: mlink->dform = dform;
1.51 schwarze 678: mlink->dsec = dsec;
679: mlink->arch = arch;
680: mlink->name = ff->fts_name;
681: mlink->fsec = fsec;
1.81 schwarze 682: mlink->gzip = gzip;
1.47 schwarze 683: mlink_add(mlink, ff->fts_statp);
684: continue;
1.96 schwarze 685:
1.98 schwarze 686: case FTS_D:
1.96 schwarze 687: /* FALLTHROUGH */
1.98 schwarze 688: case FTS_DP:
1.96 schwarze 689: break;
690:
691: default:
1.47 schwarze 692: if (warnings)
693: say(path, "Not a regular file");
694: continue;
695: }
696:
697: switch (ff->fts_level) {
1.98 schwarze 698: case 0:
1.47 schwarze 699: /* Ignore the root directory. */
700: break;
1.98 schwarze 701: case 1:
1.47 schwarze 702: /*
703: * This might contain manX/ or catX/.
704: * Try to infer this from the name.
705: * If we're not in use_all, enforce it.
706: */
707: cp = ff->fts_name;
708: if (FTS_DP == ff->fts_info)
709: break;
1.15 schwarze 710:
1.47 schwarze 711: if (0 == strncmp(cp, "man", 3)) {
712: dform = FORM_SRC;
713: dsec = cp + 3;
714: } else if (0 == strncmp(cp, "cat", 3)) {
715: dform = FORM_CAT;
716: dsec = cp + 3;
1.51 schwarze 717: } else {
718: dform = FORM_NONE;
719: dsec = NULL;
1.26 schwarze 720: }
1.47 schwarze 721:
1.98 schwarze 722: if (NULL != dsec || use_all)
1.47 schwarze 723: break;
724:
725: if (warnings)
726: say(path, "Unknown directory part");
727: fts_set(f, ff, FTS_SKIP);
728: break;
1.98 schwarze 729: case 2:
1.47 schwarze 730: /*
731: * Possibly our architecture.
732: * If we're descending, keep tabs on it.
733: */
734: if (FTS_DP != ff->fts_info && NULL != dsec)
735: arch = ff->fts_name;
1.51 schwarze 736: else
737: arch = NULL;
1.47 schwarze 738: break;
739: default:
740: if (FTS_DP == ff->fts_info || use_all)
741: break;
742: if (warnings)
743: say(path, "Extraneous directory part");
744: fts_set(f, ff, FTS_SKIP);
745: break;
1.14 schwarze 746: }
1.47 schwarze 747: }
748:
749: fts_close(f);
750: return(1);
751: }
1.1 schwarze 752:
1.47 schwarze 753: /*
754: * Add a file to the mlinks table.
755: * Do not verify that it's a "valid" looking manpage (we'll do that
756: * later).
757: *
758: * Try to infer the manual section, architecture, and page name from the
759: * path, assuming it looks like
760: *
1.98 schwarze 761: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 762: * or
763: * [./]cat<section>[/<arch>]/<name>.0
764: *
765: * See treescan() for the fts(3) version of this.
766: */
767: static void
768: filescan(const char *file)
769: {
770: char buf[PATH_MAX];
771: struct stat st;
772: struct mlink *mlink;
773: char *p, *start;
774:
775: assert(use_all);
776:
777: if (0 == strncmp(file, "./", 2))
778: file += 2;
779:
1.96 schwarze 780: /*
781: * We have to do lstat(2) before realpath(3) loses
782: * the information whether this is a symbolic link.
783: * We need to know that because for symbolic links,
784: * we want to use the orginal file name, while for
785: * regular files, we want to use the real path.
786: */
787: if (-1 == lstat(file, &st)) {
788: exitcode = (int)MANDOCLEVEL_BADARG;
789: say(file, "&lstat");
790: return;
791: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
792: exitcode = (int)MANDOCLEVEL_BADARG;
793: say(file, "Not a regular file");
794: return;
795: }
796:
797: /*
798: * We have to resolve the file name to the real path
799: * in any case for the base directory check.
800: */
1.47 schwarze 801: if (NULL == realpath(file, buf)) {
802: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 803: say(file, "&realpath");
1.47 schwarze 804: return;
1.63 schwarze 805: }
806:
1.109 schwarze 807: if (OP_TEST == op)
1.63 schwarze 808: start = buf;
1.109 schwarze 809: else if (strstr(buf, basedir) == buf)
810: start = buf + strlen(basedir);
1.63 schwarze 811: else {
1.47 schwarze 812: exitcode = (int)MANDOCLEVEL_BADARG;
813: say("", "%s: outside base directory", buf);
814: return;
1.63 schwarze 815: }
816:
1.96 schwarze 817: /*
818: * Now we are sure the file is inside our tree.
819: * If it is a symbolic link, ignore the real path
820: * and use the original name.
821: * This implies passing stuff like "cat1/../man1/foo.1"
822: * on the command line won't work. So don't do that.
823: * Note the stat(2) can still fail if the link target
824: * doesn't exist.
825: */
826: if (S_IFLNK & st.st_mode) {
827: if (-1 == stat(buf, &st)) {
828: exitcode = (int)MANDOCLEVEL_BADARG;
829: say(file, "&stat");
830: return;
831: }
1.100 schwarze 832: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
833: say(file, "Filename too long");
834: return;
835: }
1.109 schwarze 836: start = buf;
837: if (OP_TEST != op && strstr(buf, basedir) == buf)
838: start += strlen(basedir);
1.1 schwarze 839: }
1.63 schwarze 840:
1.47 schwarze 841: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 842: mlink->dform = FORM_NONE;
1.100 schwarze 843: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
844: sizeof(mlink->file)) {
845: say(start, "Filename too long");
1.134 schwarze 846: free(mlink);
1.100 schwarze 847: return;
848: }
1.1 schwarze 849:
1.10 schwarze 850: /*
1.47 schwarze 851: * First try to guess our directory structure.
852: * If we find a separator, try to look for man* or cat*.
853: * If we find one of these and what's underneath is a directory,
854: * assume it's an architecture.
1.10 schwarze 855: */
1.47 schwarze 856: if (NULL != (p = strchr(start, '/'))) {
857: *p++ = '\0';
858: if (0 == strncmp(start, "man", 3)) {
859: mlink->dform = FORM_SRC;
1.51 schwarze 860: mlink->dsec = start + 3;
1.47 schwarze 861: } else if (0 == strncmp(start, "cat", 3)) {
862: mlink->dform = FORM_CAT;
1.51 schwarze 863: mlink->dsec = start + 3;
1.47 schwarze 864: }
1.10 schwarze 865:
1.47 schwarze 866: start = p;
867: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
868: *p++ = '\0';
1.51 schwarze 869: mlink->arch = start;
1.47 schwarze 870: start = p;
1.41 deraadt 871: }
1.47 schwarze 872: }
1.7 schwarze 873:
1.47 schwarze 874: /*
875: * Now check the file suffix.
876: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
877: */
878: p = strrchr(start, '\0');
879: while (p-- > start && '/' != *p && '.' != *p)
880: /* Loop. */ ;
881:
882: if ('.' == *p) {
883: *p++ = '\0';
1.51 schwarze 884: mlink->fsec = p;
1.47 schwarze 885: }
1.41 deraadt 886:
1.47 schwarze 887: /*
888: * Now try to parse the name.
889: * Use the filename portion of the path.
890: */
891: mlink->name = start;
892: if (NULL != (p = strrchr(start, '/'))) {
893: mlink->name = p + 1;
894: *p = '\0';
895: }
896: mlink_add(mlink, &st);
897: }
1.2 schwarze 898:
1.47 schwarze 899: static void
900: mlink_add(struct mlink *mlink, const struct stat *st)
901: {
902: struct inodev inodev;
903: struct mpage *mpage;
904: unsigned int slot;
905:
906: assert(NULL != mlink->file);
907:
1.51 schwarze 908: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
909: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
910: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
911: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 912:
913: if ('0' == *mlink->fsec) {
914: free(mlink->fsec);
915: mlink->fsec = mandoc_strdup(mlink->dsec);
916: mlink->fform = FORM_CAT;
917: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
918: mlink->fform = FORM_SRC;
919: else
920: mlink->fform = FORM_NONE;
1.1 schwarze 921:
1.47 schwarze 922: slot = ohash_qlookup(&mlinks, mlink->file);
923: assert(NULL == ohash_find(&mlinks, slot));
924: ohash_insert(&mlinks, slot, mlink);
925:
926: inodev.st_ino = st->st_ino;
927: inodev.st_dev = st->st_dev;
928: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
929: sizeof(struct inodev), inodev.st_ino);
930: mpage = ohash_find(&mpages, slot);
931: if (NULL == mpage) {
932: mpage = mandoc_calloc(1, sizeof(struct mpage));
933: mpage->inodev.st_ino = inodev.st_ino;
934: mpage->inodev.st_dev = inodev.st_dev;
935: ohash_insert(&mpages, slot, mpage);
936: } else
937: mlink->next = mpage->mlinks;
938: mpage->mlinks = mlink;
1.75 schwarze 939: mlink->mpage = mpage;
1.47 schwarze 940: }
1.1 schwarze 941:
1.47 schwarze 942: static void
943: mlink_free(struct mlink *mlink)
944: {
1.1 schwarze 945:
1.47 schwarze 946: free(mlink->dsec);
947: free(mlink->arch);
948: free(mlink->name);
949: free(mlink->fsec);
950: free(mlink);
951: }
1.1 schwarze 952:
1.47 schwarze 953: static void
954: mpages_free(void)
955: {
956: struct mpage *mpage;
957: struct mlink *mlink;
958: unsigned int slot;
959:
960: mpage = ohash_first(&mpages, &slot);
961: while (NULL != mpage) {
962: while (NULL != (mlink = mpage->mlinks)) {
963: mpage->mlinks = mlink->next;
964: mlink_free(mlink);
965: }
966: free(mpage->sec);
967: free(mpage->arch);
968: free(mpage->title);
969: free(mpage->desc);
970: free(mpage);
971: mpage = ohash_next(&mpages, &slot);
972: }
973: }
1.1 schwarze 974:
1.47 schwarze 975: /*
976: * For each mlink to the mpage, check whether the path looks like
977: * it is formatted, and if it does, check whether a source manual
978: * exists by the same name, ignoring the suffix.
979: * If both conditions hold, drop the mlink.
980: */
981: static void
982: mlinks_undupe(struct mpage *mpage)
983: {
984: char buf[PATH_MAX];
985: struct mlink **prev;
986: struct mlink *mlink;
987: char *bufp;
988:
989: mpage->form = FORM_CAT;
990: prev = &mpage->mlinks;
991: while (NULL != (mlink = *prev)) {
992: if (FORM_CAT != mlink->dform) {
993: mpage->form = FORM_NONE;
994: goto nextlink;
995: }
1.100 schwarze 996: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 997: bufp = strstr(buf, "cat");
998: assert(NULL != bufp);
999: memcpy(bufp, "man", 3);
1000: if (NULL != (bufp = strrchr(buf, '.')))
1001: *++bufp = '\0';
1.100 schwarze 1002: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 1003: if (NULL == ohash_find(&mlinks,
1.98 schwarze 1004: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 1005: goto nextlink;
1006: if (warnings)
1007: say(mlink->file, "Man source exists: %s", buf);
1008: if (use_all)
1009: goto nextlink;
1010: *prev = mlink->next;
1011: mlink_free(mlink);
1012: continue;
1013: nextlink:
1014: prev = &(*prev)->next;
1.1 schwarze 1015: }
1.47 schwarze 1016: }
1.1 schwarze 1017:
1.88 schwarze 1018: static void
1.50 schwarze 1019: mlink_check(struct mpage *mpage, struct mlink *mlink)
1020: {
1.88 schwarze 1021: struct str *str;
1022: unsigned int slot;
1.50 schwarze 1023:
1024: /*
1025: * Check whether the manual section given in a file
1026: * agrees with the directory where the file is located.
1027: * Some manuals have suffixes like (3p) on their
1028: * section number either inside the file or in the
1029: * directory name, some are linked into more than one
1030: * section, like encrypt(1) = makekey(8).
1031: */
1032:
1033: if (FORM_SRC == mpage->form &&
1.88 schwarze 1034: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1035: say(mlink->file, "Section \"%s\" manual in %s directory",
1036: mpage->sec, mlink->dsec);
1037:
1038: /*
1039: * Manual page directories exist for each kernel
1040: * architecture as returned by machine(1).
1041: * However, many manuals only depend on the
1042: * application architecture as returned by arch(1).
1043: * For example, some (2/ARM) manuals are shared
1044: * across the "armish" and "zaurus" kernel
1045: * architectures.
1046: * A few manuals are even shared across completely
1047: * different architectures, for example fdformat(1)
1048: * on amd64, i386, sparc, and sparc64.
1049: */
1050:
1.88 schwarze 1051: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1052: say(mlink->file, "Architecture \"%s\" manual in "
1053: "\"%s\" directory", mpage->arch, mlink->arch);
1054:
1.88 schwarze 1055: /*
1056: * XXX
1.90 schwarze 1057: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1058: */
1059:
1060: if (FORM_CAT == mpage->form)
1061: return;
1062:
1063: /*
1064: * Check whether this mlink
1065: * appears as a name in the NAME section.
1066: */
1.50 schwarze 1067:
1.90 schwarze 1068: slot = ohash_qlookup(&names, mlink->name);
1069: str = ohash_find(&names, slot);
1.88 schwarze 1070: assert(NULL != str);
1.90 schwarze 1071: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1072: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1073: }
1074:
1.47 schwarze 1075: /*
1076: * Run through the files in the global vector "mpages"
1077: * and add them to the database specified in "basedir".
1078: *
1079: * This handles the parsing scheme itself, using the cues of directory
1080: * and filename to determine whether the file is parsable or not.
1081: */
1082: static void
1.131 schwarze 1083: mpages_merge(struct mparse *mp)
1.47 schwarze 1084: {
1.70 schwarze 1085: char any[] = "any";
1.58 schwarze 1086: struct ohash_info str_info;
1.75 schwarze 1087: struct mpage *mpage, *mpage_dest;
1088: struct mlink *mlink, *mlink_dest;
1.47 schwarze 1089: struct mdoc *mdoc;
1090: struct man *man;
1.75 schwarze 1091: char *sodest;
1.69 schwarze 1092: char *cp;
1.117 schwarze 1093: int fd;
1.58 schwarze 1094: unsigned int pslot;
1.47 schwarze 1095:
1096: str_info.alloc = hash_alloc;
1.106 espie 1097: str_info.calloc = hash_calloc;
1098: str_info.free = hash_free;
1.135 schwarze 1099: str_info.data = NULL;
1.47 schwarze 1100: str_info.key_offset = offsetof(struct str, key);
1101:
1.126 schwarze 1102: if ( ! nodb)
1.64 schwarze 1103: SQL_EXEC("BEGIN TRANSACTION");
1104:
1.47 schwarze 1105: mpage = ohash_first(&mpages, &pslot);
1.126 schwarze 1106: while (mpage != NULL) {
1.47 schwarze 1107: mlinks_undupe(mpage);
1.136 schwarze 1108: if ((mlink = mpage->mlinks) == NULL) {
1.47 schwarze 1109: mpage = ohash_next(&mpages, &pslot);
1110: continue;
1111: }
1.1 schwarze 1112:
1.90 schwarze 1113: name_mask = NAME_MASK;
1114: ohash_init(&names, 4, &str_info);
1.47 schwarze 1115: ohash_init(&strings, 6, &str_info);
1116: mparse_reset(mp);
1117: mdoc = NULL;
1118: man = NULL;
1.81 schwarze 1119: sodest = NULL;
1120:
1.136 schwarze 1121: mparse_open(mp, &fd, mlink->file);
1.117 schwarze 1122: if (fd == -1) {
1.136 schwarze 1123: say(mlink->file, "&open");
1.117 schwarze 1124: goto nextpage;
1.81 schwarze 1125: }
1.11 schwarze 1126:
1127: /*
1.137 ! schwarze 1128: * Interpret the file as mdoc(7) or man(7) source
! 1129: * code, unless it is known to be formatted.
1.11 schwarze 1130: */
1.136 schwarze 1131: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.137 ! schwarze 1132: mparse_readfd(mp, fd, mlink->file);
! 1133: mparse_result(mp, &mdoc, &man, &sodest);
1.47 schwarze 1134: }
1.11 schwarze 1135:
1.126 schwarze 1136: if (sodest != NULL) {
1.75 schwarze 1137: mlink_dest = ohash_find(&mlinks,
1138: ohash_qlookup(&mlinks, sodest));
1.126 schwarze 1139: if (mlink_dest == NULL) {
1140: mandoc_asprintf(&cp, "%s.gz", sodest);
1141: mlink_dest = ohash_find(&mlinks,
1142: ohash_qlookup(&mlinks, cp));
1143: free(cp);
1144: }
1145: if (mlink_dest != NULL) {
1.75 schwarze 1146:
1147: /* The .so target exists. */
1148:
1149: mpage_dest = mlink_dest->mpage;
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;
1.136 schwarze 1189: mpage->sec = mandoc_strdup(man_meta(man)->msec);
1190: mpage->arch = mandoc_strdup(mlink->arch);
1191: mpage->title = mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1192: } else {
1.47 schwarze 1193: mpage->form = FORM_CAT;
1.136 schwarze 1194: mpage->sec = mandoc_strdup(mlink->dsec);
1195: mpage->arch = mandoc_strdup(mlink->arch);
1196: mpage->title = mandoc_strdup(mlink->name);
1.1 schwarze 1197: }
1.54 schwarze 1198: putkey(mpage, mpage->sec, TYPE_sec);
1.119 schwarze 1199: if (*mpage->arch != '\0')
1200: putkey(mpage, mpage->arch, TYPE_arch);
1.1 schwarze 1201:
1.136 schwarze 1202: for ( ; mlink != NULL; mlink = mlink->next) {
1.54 schwarze 1203: if ('\0' != *mlink->dsec)
1204: putkey(mpage, mlink->dsec, TYPE_sec);
1205: if ('\0' != *mlink->fsec)
1206: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1207: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1208: any : mlink->arch, TYPE_arch);
1.90 schwarze 1209: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1210: }
1.41 deraadt 1211:
1.129 schwarze 1212: assert(mpage->desc == NULL);
1213: if (mdoc != NULL)
1.127 schwarze 1214: parse_mdoc(mpage, mdoc_meta(mdoc), mdoc_node(mdoc));
1.129 schwarze 1215: else if (man != NULL)
1216: parse_man(mpage, man_meta(man), man_node(man));
1.47 schwarze 1217: else
1.117 schwarze 1218: parse_cat(mpage, fd);
1.129 schwarze 1219: if (mpage->desc == NULL)
1.92 schwarze 1220: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1221:
1222: if (warnings && !use_all)
1223: for (mlink = mpage->mlinks; mlink;
1224: mlink = mlink->next)
1225: mlink_check(mpage, mlink);
1.6 schwarze 1226:
1.131 schwarze 1227: dbadd(mpage);
1.136 schwarze 1228: mlink = mpage->mlinks;
1.81 schwarze 1229:
1230: nextpage:
1.125 schwarze 1231: if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1.117 schwarze 1232: exitcode = (int)MANDOCLEVEL_SYSERR;
1.136 schwarze 1233: say(mlink->file, "&wait gunzip");
1.81 schwarze 1234: }
1.47 schwarze 1235: ohash_delete(&strings);
1.90 schwarze 1236: ohash_delete(&names);
1.47 schwarze 1237: mpage = ohash_next(&mpages, &pslot);
1238: }
1.64 schwarze 1239:
1240: if (0 == nodb)
1241: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1242: }
1243:
1244: static void
1245: names_check(void)
1246: {
1247: sqlite3_stmt *stmt;
1248: const char *name, *sec, *arch, *key;
1249: int irc;
1250:
1251: sqlite3_prepare_v2(db,
1252: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1253: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1254: "WHERE bits & ? AND NOT EXISTS ("
1255: "SELECT pageid FROM mlinks "
1.90 schwarze 1256: "WHERE mlinks.pageid == names.pageid "
1257: "AND mlinks.name == names.name"
1.87 schwarze 1258: ")"
1259: ") JOIN ("
1.99 schwarze 1260: "SELECT sec, arch, name, pageid FROM mlinks "
1261: "GROUP BY pageid"
1.87 schwarze 1262: ") USING (pageid);",
1263: -1, &stmt, NULL);
1264:
1.91 schwarze 1265: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1266: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1267:
1268: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.113 schwarze 1269: name = (const char *)sqlite3_column_text(stmt, 0);
1270: sec = (const char *)sqlite3_column_text(stmt, 1);
1271: arch = (const char *)sqlite3_column_text(stmt, 2);
1272: key = (const char *)sqlite3_column_text(stmt, 3);
1.87 schwarze 1273: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1274: '\0' == *arch ? "" : "/",
1275: '\0' == *arch ? "" : arch, key);
1276: }
1277: sqlite3_finalize(stmt);
1.47 schwarze 1278: }
1.6 schwarze 1279:
1.47 schwarze 1280: static void
1.81 schwarze 1281: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1282: {
1283: FILE *stream;
1284: char *line, *p, *title;
1285: size_t len, plen, titlesz;
1.1 schwarze 1286:
1.81 schwarze 1287: stream = (-1 == fd) ?
1288: fopen(mpage->mlinks->file, "r") :
1289: fdopen(fd, "r");
1290: if (NULL == stream) {
1.114 doug 1291: if (-1 != fd)
1292: close(fd);
1.47 schwarze 1293: if (warnings)
1.80 schwarze 1294: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1295: return;
1296: }
1.1 schwarze 1297:
1.47 schwarze 1298: /* Skip to first blank line. */
1.1 schwarze 1299:
1.47 schwarze 1300: while (NULL != (line = fgetln(stream, &len)))
1301: if ('\n' == *line)
1302: break;
1.1 schwarze 1303:
1.47 schwarze 1304: /*
1305: * Assume the first line that is not indented
1306: * is the first section header. Skip to it.
1307: */
1.1 schwarze 1308:
1.47 schwarze 1309: while (NULL != (line = fgetln(stream, &len)))
1310: if ('\n' != *line && ' ' != *line)
1311: break;
1.98 schwarze 1312:
1.47 schwarze 1313: /*
1314: * Read up until the next section into a buffer.
1315: * Strip the leading and trailing newline from each read line,
1316: * appending a trailing space.
1317: * Ignore empty (whitespace-only) lines.
1318: */
1.28 schwarze 1319:
1.47 schwarze 1320: titlesz = 0;
1321: title = NULL;
1.38 schwarze 1322:
1.47 schwarze 1323: while (NULL != (line = fgetln(stream, &len))) {
1324: if (' ' != *line || '\n' != line[len - 1])
1325: break;
1326: while (len > 0 && isspace((unsigned char)*line)) {
1327: line++;
1328: len--;
1329: }
1330: if (1 == len)
1331: continue;
1332: title = mandoc_realloc(title, titlesz + len);
1333: memcpy(title + titlesz, line, len);
1334: titlesz += len;
1335: title[titlesz - 1] = ' ';
1336: }
1.28 schwarze 1337:
1.47 schwarze 1338: /*
1339: * If no page content can be found, or the input line
1340: * is already the next section header, or there is no
1341: * trailing newline, reuse the page title as the page
1342: * description.
1343: */
1.1 schwarze 1344:
1.47 schwarze 1345: if (NULL == title || '\0' == *title) {
1346: if (warnings)
1347: say(mpage->mlinks->file,
1348: "Cannot find NAME section");
1349: fclose(stream);
1350: free(title);
1351: return;
1352: }
1.24 schwarze 1353:
1.47 schwarze 1354: title = mandoc_realloc(title, titlesz + 1);
1355: title[titlesz] = '\0';
1.24 schwarze 1356:
1.47 schwarze 1357: /*
1358: * Skip to the first dash.
1359: * Use the remaining line as the description (no more than 70
1360: * bytes).
1361: */
1.28 schwarze 1362:
1.47 schwarze 1363: if (NULL != (p = strstr(title, "- "))) {
1364: for (p += 2; ' ' == *p || '\b' == *p; p++)
1365: /* Skip to next word. */ ;
1366: } else {
1367: if (warnings)
1368: say(mpage->mlinks->file,
1369: "No dash in title line");
1370: p = title;
1371: }
1.1 schwarze 1372:
1.47 schwarze 1373: plen = strlen(p);
1.1 schwarze 1374:
1.47 schwarze 1375: /* Strip backspace-encoding from line. */
1.1 schwarze 1376:
1.47 schwarze 1377: while (NULL != (line = memchr(p, '\b', plen))) {
1378: len = line - p;
1379: if (0 == len) {
1380: memmove(line, line + 1, plen--);
1381: continue;
1.98 schwarze 1382: }
1.47 schwarze 1383: memmove(line - 1, line + 1, plen - len);
1384: plen -= 2;
1385: }
1.1 schwarze 1386:
1.47 schwarze 1387: mpage->desc = mandoc_strdup(p);
1388: fclose(stream);
1389: free(title);
1390: }
1.16 schwarze 1391:
1.47 schwarze 1392: /*
1393: * Put a type/word pair into the word database for this particular file.
1394: */
1395: static void
1.69 schwarze 1396: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1397: {
1.69 schwarze 1398: char *cp;
1.37 schwarze 1399:
1.47 schwarze 1400: assert(NULL != value);
1.69 schwarze 1401: if (TYPE_arch == type)
1402: for (cp = value; *cp; cp++)
1403: if (isupper((unsigned char)*cp))
1404: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1405: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1406: }
1407:
1408: /*
1.47 schwarze 1409: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1410: */
1411: static void
1.47 schwarze 1412: putmdockey(const struct mpage *mpage,
1413: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1414: {
1.16 schwarze 1415:
1.47 schwarze 1416: for ( ; NULL != n; n = n->next) {
1417: if (NULL != n->child)
1418: putmdockey(mpage, n->child, m);
1419: if (MDOC_TEXT == n->type)
1420: putkey(mpage, n->string, m);
1421: }
1422: }
1.16 schwarze 1423:
1.47 schwarze 1424: static void
1.129 schwarze 1425: parse_man(struct mpage *mpage, const struct man_meta *meta,
1426: const struct man_node *n)
1.47 schwarze 1427: {
1428: const struct man_node *head, *body;
1.78 schwarze 1429: char *start, *title;
1.47 schwarze 1430: char byte;
1.78 schwarze 1431: size_t sz;
1.16 schwarze 1432:
1.47 schwarze 1433: if (NULL == n)
1434: return;
1.16 schwarze 1435:
1.47 schwarze 1436: /*
1437: * We're only searching for one thing: the first text child in
1438: * the BODY of a NAME section. Since we don't keep track of
1439: * sections in -man, run some hoops to find out whether we're in
1440: * the correct section or not.
1441: */
1.16 schwarze 1442:
1.47 schwarze 1443: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1444: body = n;
1445: assert(body->parent);
1446: if (NULL != (head = body->parent->head) &&
1.98 schwarze 1447: 1 == head->nchild &&
1448: NULL != (head = (head->child)) &&
1449: MAN_TEXT == head->type &&
1450: 0 == strcmp(head->string, "NAME") &&
1451: NULL != body->child) {
1.2 schwarze 1452:
1.47 schwarze 1453: /*
1454: * Suck the entire NAME section into memory.
1455: * Yes, we might run away.
1456: * But too many manuals have big, spread-out
1457: * NAME sections over many lines.
1458: */
1.2 schwarze 1459:
1.78 schwarze 1460: title = NULL;
1461: man_deroff(&title, body);
1.47 schwarze 1462: if (NULL == title)
1463: return;
1.16 schwarze 1464:
1.98 schwarze 1465: /*
1.47 schwarze 1466: * Go through a special heuristic dance here.
1467: * Conventionally, one or more manual names are
1468: * comma-specified prior to a whitespace, then a
1469: * dash, then a description. Try to puzzle out
1470: * the name parts here.
1471: */
1.16 schwarze 1472:
1.78 schwarze 1473: start = title;
1.47 schwarze 1474: for ( ;; ) {
1475: sz = strcspn(start, " ,");
1476: if ('\0' == start[sz])
1477: break;
1.1 schwarze 1478:
1.47 schwarze 1479: byte = start[sz];
1480: start[sz] = '\0';
1.67 schwarze 1481:
1482: /*
1483: * Assume a stray trailing comma in the
1484: * name list if a name begins with a dash.
1485: */
1486:
1487: if ('-' == start[0] ||
1488: ('\\' == start[0] && '-' == start[1]))
1489: break;
1.1 schwarze 1490:
1.90 schwarze 1491: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1492: if ( ! (mpage->name_head_done ||
1493: strcasecmp(start, meta->title))) {
1494: putkey(mpage, start, NAME_HEAD);
1495: mpage->name_head_done = 1;
1496: }
1.1 schwarze 1497:
1.47 schwarze 1498: if (' ' == byte) {
1499: start += sz + 1;
1500: break;
1501: }
1.1 schwarze 1502:
1.47 schwarze 1503: assert(',' == byte);
1504: start += sz + 1;
1505: while (' ' == *start)
1506: start++;
1507: }
1.1 schwarze 1508:
1.78 schwarze 1509: if (start == title) {
1.90 schwarze 1510: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1511: if ( ! (mpage->name_head_done ||
1512: strcasecmp(start, meta->title))) {
1513: putkey(mpage, start, NAME_HEAD);
1514: mpage->name_head_done = 1;
1515: }
1.47 schwarze 1516: free(title);
1517: return;
1518: }
1.1 schwarze 1519:
1.47 schwarze 1520: while (isspace((unsigned char)*start))
1521: start++;
1.1 schwarze 1522:
1.47 schwarze 1523: if (0 == strncmp(start, "-", 1))
1524: start += 1;
1525: else if (0 == strncmp(start, "\\-\\-", 4))
1526: start += 4;
1527: else if (0 == strncmp(start, "\\-", 2))
1528: start += 2;
1529: else if (0 == strncmp(start, "\\(en", 4))
1530: start += 4;
1531: else if (0 == strncmp(start, "\\(em", 4))
1532: start += 4;
1.1 schwarze 1533:
1.47 schwarze 1534: while (' ' == *start)
1535: start++;
1.1 schwarze 1536:
1.47 schwarze 1537: mpage->desc = mandoc_strdup(start);
1538: free(title);
1539: return;
1540: }
1541: }
1.1 schwarze 1542:
1.47 schwarze 1543: for (n = n->child; n; n = n->next) {
1544: if (NULL != mpage->desc)
1545: break;
1.129 schwarze 1546: parse_man(mpage, meta, n);
1.1 schwarze 1547: }
1548: }
1549:
1550: static void
1.127 schwarze 1551: parse_mdoc(struct mpage *mpage, const struct mdoc_meta *meta,
1552: const struct mdoc_node *n)
1.1 schwarze 1553: {
1554:
1.47 schwarze 1555: assert(NULL != n);
1556: for (n = n->child; NULL != n; n = n->next) {
1557: switch (n->type) {
1.98 schwarze 1558: case MDOC_ELEM:
1.47 schwarze 1559: /* FALLTHROUGH */
1.98 schwarze 1560: case MDOC_BLOCK:
1.47 schwarze 1561: /* FALLTHROUGH */
1.98 schwarze 1562: case MDOC_HEAD:
1.47 schwarze 1563: /* FALLTHROUGH */
1.98 schwarze 1564: case MDOC_BODY:
1.47 schwarze 1565: /* FALLTHROUGH */
1.98 schwarze 1566: case MDOC_TAIL:
1.47 schwarze 1567: if (NULL != mdocs[n->tok].fp)
1.127 schwarze 1568: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.47 schwarze 1569: break;
1570: if (mdocs[n->tok].mask)
1571: putmdockey(mpage, n->child,
1572: mdocs[n->tok].mask);
1573: break;
1574: default:
1575: assert(MDOC_ROOT != n->type);
1576: continue;
1577: }
1578: if (NULL != n->child)
1.127 schwarze 1579: parse_mdoc(mpage, meta, n);
1.1 schwarze 1580: }
1581: }
1582:
1.19 schwarze 1583: static int
1.127 schwarze 1584: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_meta *meta,
1585: const struct mdoc_node *n)
1.1 schwarze 1586: {
1.131 schwarze 1587: char *start, *end;
1.1 schwarze 1588: size_t sz;
1.19 schwarze 1589:
1.47 schwarze 1590: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1591: NULL == (n = n->child) ||
1592: MDOC_TEXT != n->type)
1.19 schwarze 1593: return(0);
1.1 schwarze 1594:
1595: /*
1596: * Only consider those `Fd' macro fields that begin with an
1597: * "inclusion" token (versus, e.g., #define).
1598: */
1.47 schwarze 1599:
1.1 schwarze 1600: if (strcmp("#include", n->string))
1.19 schwarze 1601: return(0);
1.1 schwarze 1602:
1603: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1604: return(0);
1.1 schwarze 1605:
1606: /*
1607: * Strip away the enclosing angle brackets and make sure we're
1608: * not zero-length.
1609: */
1610:
1611: start = n->string;
1612: if ('<' == *start || '"' == *start)
1613: start++;
1614:
1615: if (0 == (sz = strlen(start)))
1.19 schwarze 1616: return(0);
1.1 schwarze 1617:
1618: end = &start[(int)sz - 1];
1619: if ('>' == *end || '"' == *end)
1620: end--;
1621:
1.47 schwarze 1622: if (end > start)
1623: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1624: return(0);
1.1 schwarze 1625: }
1626:
1.133 schwarze 1627: static void
1628: parse_mdoc_fname(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1629: {
1.69 schwarze 1630: char *cp;
1.133 schwarze 1631: size_t sz;
1.1 schwarze 1632:
1.133 schwarze 1633: if (n->type != MDOC_TEXT)
1634: return;
1.19 schwarze 1635:
1.133 schwarze 1636: /* Skip function pointer punctuation. */
1.1 schwarze 1637:
1.133 schwarze 1638: cp = n->string;
1639: while (*cp == '(' || *cp == '*')
1.1 schwarze 1640: cp++;
1.133 schwarze 1641: sz = strcspn(cp, "()");
1.1 schwarze 1642:
1.133 schwarze 1643: putkeys(mpage, cp, sz, TYPE_Fn);
1.128 schwarze 1644: if (n->sec == SEC_SYNOPSIS)
1.133 schwarze 1645: putkeys(mpage, cp, sz, NAME_SYN);
1646: }
1.19 schwarze 1647:
1.133 schwarze 1648: static int
1649: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_meta *meta,
1650: const struct mdoc_node *n)
1651: {
1652:
1653: if (n->child == NULL)
1654: return(0);
1655:
1656: parse_mdoc_fname(mpage, n->child);
1.19 schwarze 1657:
1.133 schwarze 1658: for (n = n->child->next; n != NULL; n = n->next)
1659: if (n->type == MDOC_TEXT)
1.47 schwarze 1660: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1661:
1.128 schwarze 1662: return(0);
1663: }
1664:
1665: static int
1666: parse_mdoc_Fo(struct mpage *mpage, const struct mdoc_meta *meta,
1667: const struct mdoc_node *n)
1668: {
1.132 schwarze 1669:
1670: if (n->type != MDOC_HEAD)
1671: return(1);
1.128 schwarze 1672:
1.133 schwarze 1673: if (n->child != NULL)
1674: parse_mdoc_fname(mpage, n->child);
1675:
1.19 schwarze 1676: return(0);
1.1 schwarze 1677: }
1678:
1.19 schwarze 1679: static int
1.127 schwarze 1680: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_meta *meta,
1681: const struct mdoc_node *n)
1.1 schwarze 1682: {
1.47 schwarze 1683: char *cp;
1.1 schwarze 1684:
1685: if (NULL == (n = n->child))
1.19 schwarze 1686: return(0);
1.1 schwarze 1687:
1.47 schwarze 1688: if (NULL == n->next) {
1689: putkey(mpage, n->string, TYPE_Xr);
1690: return(0);
1691: }
1.1 schwarze 1692:
1.77 schwarze 1693: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1694: putkey(mpage, cp, TYPE_Xr);
1695: free(cp);
1696: return(0);
1.1 schwarze 1697: }
1698:
1.19 schwarze 1699: static int
1.127 schwarze 1700: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_meta *meta,
1701: const struct mdoc_node *n)
1.1 schwarze 1702: {
1703:
1.79 schwarze 1704: if (MDOC_BODY == n->type)
1705: mdoc_deroff(&mpage->desc, n);
1706: return(0);
1.1 schwarze 1707: }
1708:
1.19 schwarze 1709: static int
1.127 schwarze 1710: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_meta *meta,
1711: const struct mdoc_node *n)
1.1 schwarze 1712: {
1713:
1.86 schwarze 1714: if (SEC_NAME == n->sec)
1.90 schwarze 1715: putmdockey(mpage, n->child, NAME_TITLE);
1.127 schwarze 1716: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) {
1717: if (n->child == NULL)
1718: putkey(mpage, meta->name, NAME_SYN);
1719: else
1720: putmdockey(mpage, n->child, NAME_SYN);
1.129 schwarze 1721: }
1722: if ( ! (mpage->name_head_done ||
1723: n->child == NULL || n->child->string == NULL ||
1724: strcasecmp(n->child->string, meta->title))) {
1725: putkey(mpage, n->child->string, NAME_HEAD);
1726: mpage->name_head_done = 1;
1.127 schwarze 1727: }
1.86 schwarze 1728: return(0);
1.1 schwarze 1729: }
1730:
1.19 schwarze 1731: static int
1.127 schwarze 1732: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_meta *meta,
1733: const struct mdoc_node *n)
1.1 schwarze 1734: {
1735:
1.19 schwarze 1736: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1737: }
1738:
1.47 schwarze 1739: static int
1.127 schwarze 1740: parse_mdoc_head(struct mpage *mpage, const struct mdoc_meta *meta,
1741: const struct mdoc_node *n)
1.1 schwarze 1742: {
1743:
1.47 schwarze 1744: return(MDOC_HEAD == n->type);
1745: }
1.1 schwarze 1746:
1.47 schwarze 1747: static int
1.127 schwarze 1748: parse_mdoc_body(struct mpage *mpage, const struct mdoc_meta *meta,
1749: const struct mdoc_node *n)
1.47 schwarze 1750: {
1.1 schwarze 1751:
1.47 schwarze 1752: return(MDOC_BODY == n->type);
1.1 schwarze 1753: }
1754:
1.47 schwarze 1755: /*
1756: * Add a string to the hash table for the current manual.
1757: * Each string has a bitmask telling which macros it belongs to.
1758: * When we finish the manual, we'll dump the table.
1759: */
1.1 schwarze 1760: static void
1.131 schwarze 1761: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.1 schwarze 1762: {
1.90 schwarze 1763: struct ohash *htab;
1.47 schwarze 1764: struct str *s;
1.68 schwarze 1765: const char *end;
1.47 schwarze 1766: unsigned int slot;
1.131 schwarze 1767: int i, mustfree;
1.1 schwarze 1768:
1.47 schwarze 1769: if (0 == sz)
1770: return;
1.68 schwarze 1771:
1.131 schwarze 1772: mustfree = render_string(&cp, &sz);
1773:
1.90 schwarze 1774: if (TYPE_Nm & v) {
1775: htab = &names;
1776: v &= name_mask;
1.124 schwarze 1777: if (v & NAME_FIRST)
1778: name_mask &= ~NAME_FIRST;
1.90 schwarze 1779: if (debug > 1)
1780: say(mpage->mlinks->file,
1.130 schwarze 1781: "Adding name %*s, bits=%d", sz, cp, v);
1.90 schwarze 1782: } else {
1783: htab = &strings;
1784: if (debug > 1)
1785: for (i = 0; i < mansearch_keymax; i++)
1.118 schwarze 1786: if ((uint64_t)1 << i & v)
1.90 schwarze 1787: say(mpage->mlinks->file,
1788: "Adding key %s=%*s",
1789: mansearch_keynames[i], sz, cp);
1.68 schwarze 1790: }
1.47 schwarze 1791:
1792: end = cp + sz;
1.90 schwarze 1793: slot = ohash_qlookupi(htab, cp, &end);
1794: s = ohash_find(htab, slot);
1.1 schwarze 1795:
1.47 schwarze 1796: if (NULL != s && mpage == s->mpage) {
1797: s->mask |= v;
1.1 schwarze 1798: return;
1.47 schwarze 1799: } else if (NULL == s) {
1.101 schwarze 1800: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1801: memcpy(s->key, cp, sz);
1.90 schwarze 1802: ohash_insert(htab, slot, s);
1.47 schwarze 1803: }
1804: s->mpage = mpage;
1805: s->mask = v;
1.131 schwarze 1806:
1807: if (mustfree)
1808: free(cp);
1.1 schwarze 1809: }
1810:
1811: /*
1.47 schwarze 1812: * Take a Unicode codepoint and produce its UTF-8 encoding.
1813: * This isn't the best way to do this, but it works.
1814: * The magic numbers are from the UTF-8 packaging.
1815: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1816: */
1.47 schwarze 1817: static size_t
1818: utf8(unsigned int cp, char out[7])
1.1 schwarze 1819: {
1.47 schwarze 1820: size_t rc;
1.1 schwarze 1821:
1.47 schwarze 1822: rc = 0;
1823: if (cp <= 0x0000007F) {
1824: rc = 1;
1825: out[0] = (char)cp;
1826: } else if (cp <= 0x000007FF) {
1827: rc = 2;
1828: out[0] = (cp >> 6 & 31) | 192;
1829: out[1] = (cp & 63) | 128;
1830: } else if (cp <= 0x0000FFFF) {
1831: rc = 3;
1832: out[0] = (cp >> 12 & 15) | 224;
1833: out[1] = (cp >> 6 & 63) | 128;
1834: out[2] = (cp & 63) | 128;
1835: } else if (cp <= 0x001FFFFF) {
1836: rc = 4;
1837: out[0] = (cp >> 18 & 7) | 240;
1838: out[1] = (cp >> 12 & 63) | 128;
1839: out[2] = (cp >> 6 & 63) | 128;
1840: out[3] = (cp & 63) | 128;
1841: } else if (cp <= 0x03FFFFFF) {
1842: rc = 5;
1843: out[0] = (cp >> 24 & 3) | 248;
1844: out[1] = (cp >> 18 & 63) | 128;
1845: out[2] = (cp >> 12 & 63) | 128;
1846: out[3] = (cp >> 6 & 63) | 128;
1847: out[4] = (cp & 63) | 128;
1848: } else if (cp <= 0x7FFFFFFF) {
1849: rc = 6;
1850: out[0] = (cp >> 30 & 1) | 252;
1851: out[1] = (cp >> 24 & 63) | 128;
1852: out[2] = (cp >> 18 & 63) | 128;
1853: out[3] = (cp >> 12 & 63) | 128;
1854: out[4] = (cp >> 6 & 63) | 128;
1855: out[5] = (cp & 63) | 128;
1856: } else
1857: return(0);
1.19 schwarze 1858:
1.47 schwarze 1859: out[rc] = '\0';
1860: return(rc);
1.1 schwarze 1861: }
1862:
1.47 schwarze 1863: /*
1.131 schwarze 1864: * If the string contains escape sequences,
1865: * replace it with an allocated rendering and return 1,
1866: * such that the caller can free it after use.
1867: * Otherwise, do nothing and return 0.
1.47 schwarze 1868: */
1.131 schwarze 1869: static int
1870: render_string(char **public, size_t *psz)
1.1 schwarze 1871: {
1.131 schwarze 1872: const char *src, *scp, *addcp, *seq;
1873: char *dst;
1874: size_t ssz, dsz, addsz;
1.71 schwarze 1875: char utfbuf[7], res[6];
1.131 schwarze 1876: int seqlen, unicode;
1.47 schwarze 1877:
1878: res[0] = '\\';
1879: res[1] = '\t';
1880: res[2] = ASCII_NBRSP;
1881: res[3] = ASCII_HYPH;
1.71 schwarze 1882: res[4] = ASCII_BREAK;
1883: res[5] = '\0';
1.1 schwarze 1884:
1.131 schwarze 1885: src = scp = *public;
1886: ssz = *psz;
1887: dst = NULL;
1888: dsz = 0;
1889:
1890: while (scp < src + *psz) {
1891:
1892: /* Leave normal characters unchanged. */
1893:
1894: if (strchr(res, *scp) == NULL) {
1895: if (dst != NULL)
1896: dst[dsz++] = *scp;
1897: scp++;
1898: continue;
1899: }
1.1 schwarze 1900:
1.47 schwarze 1901: /*
1.131 schwarze 1902: * Found something that requires replacing,
1903: * make sure we have a destination buffer.
1.47 schwarze 1904: */
1.131 schwarze 1905:
1906: if (dst == NULL) {
1907: dst = mandoc_malloc(ssz + 1);
1908: dsz = scp - src;
1909: memcpy(dst, src, dsz);
1.47 schwarze 1910: }
1.39 schwarze 1911:
1.131 schwarze 1912: /* Handle single-char special characters. */
1913:
1914: switch (*scp) {
1915: case '\\':
1916: break;
1.98 schwarze 1917: case '\t':
1.71 schwarze 1918: /* FALLTHROUGH */
1.98 schwarze 1919: case ASCII_NBRSP:
1.131 schwarze 1920: dst[dsz++] = ' ';
1921: scp++;
1922: continue;
1923: case ASCII_HYPH:
1924: dst[dsz++] = '-';
1.71 schwarze 1925: /* FALLTHROUGH */
1.98 schwarze 1926: case ASCII_BREAK:
1.131 schwarze 1927: scp++;
1.47 schwarze 1928: continue;
1.71 schwarze 1929: default:
1.131 schwarze 1930: abort();
1.71 schwarze 1931: }
1.39 schwarze 1932:
1.47 schwarze 1933: /*
1.131 schwarze 1934: * Found an escape sequence.
1935: * Read past the slash, then parse it.
1936: * Ignore everything except characters.
1.47 schwarze 1937: */
1.52 schwarze 1938:
1.131 schwarze 1939: scp++;
1940: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.47 schwarze 1941: continue;
1.39 schwarze 1942:
1.47 schwarze 1943: /*
1.52 schwarze 1944: * Render the special character
1945: * as either UTF-8 or ASCII.
1.47 schwarze 1946: */
1.52 schwarze 1947:
1948: if (write_utf8) {
1.131 schwarze 1949: unicode = mchars_spec2cp(mchars, seq, seqlen);
1950: if (unicode <= 0)
1.52 schwarze 1951: continue;
1.131 schwarze 1952: addsz = utf8(unicode, utfbuf);
1953: if (addsz == 0)
1.52 schwarze 1954: continue;
1.131 schwarze 1955: addcp = utfbuf;
1.52 schwarze 1956: } else {
1.131 schwarze 1957: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1958: if (addcp == NULL)
1.52 schwarze 1959: continue;
1.131 schwarze 1960: if (*addcp == ASCII_NBRSP) {
1961: addcp = " ";
1962: addsz = 1;
1.52 schwarze 1963: }
1964: }
1.1 schwarze 1965:
1.47 schwarze 1966: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1967:
1.131 schwarze 1968: ssz += addsz;
1969: dst = mandoc_realloc(dst, ssz + 1);
1970: memcpy(dst + dsz, addcp, addsz);
1971: dsz += addsz;
1972: }
1973: if (dst != NULL) {
1974: *public = dst;
1975: *psz = dsz;
1976: }
1977:
1978: /* Trim trailing whitespace and NUL-terminate. */
1979:
1980: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1981: --*psz;
1982: if (dst != NULL) {
1983: (*public)[*psz] = '\0';
1984: return(1);
1985: } else
1986: return(0);
1.1 schwarze 1987: }
1988:
1.75 schwarze 1989: static void
1990: dbadd_mlink(const struct mlink *mlink)
1991: {
1992: size_t i;
1993:
1994: i = 1;
1995: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1996: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1997: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 1998: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 1999: SQL_STEP(stmts[STMT_INSERT_LINK]);
2000: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.124 schwarze 2001: }
2002:
2003: static void
2004: dbadd_mlink_name(const struct mlink *mlink)
2005: {
1.130 schwarze 2006: uint64_t bits;
1.124 schwarze 2007: size_t i;
2008:
2009: dbadd_mlink(mlink);
1.115 schwarze 2010:
2011: i = 1;
1.130 schwarze 2012: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2013: bits = NAME_FILE & NAME_MASK;
2014: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2015: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2016: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2017: }
2018:
2019: i = 1;
2020: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.115 schwarze 2021: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2022: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2023: SQL_STEP(stmts[STMT_INSERT_NAME]);
2024: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.75 schwarze 2025: }
2026:
1.11 schwarze 2027: /*
1.47 schwarze 2028: * Flush the current page's terms (and their bits) into the database.
2029: * Wrap the entire set of additions in a transaction to make sqlite be a
2030: * little faster.
1.53 schwarze 2031: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 2032: */
2033: static void
1.131 schwarze 2034: dbadd(struct mpage *mpage)
1.11 schwarze 2035: {
1.47 schwarze 2036: struct mlink *mlink;
2037: struct str *key;
1.131 schwarze 2038: char *cp;
1.47 schwarze 2039: size_t i;
2040: unsigned int slot;
1.131 schwarze 2041: int mustfree;
1.47 schwarze 2042:
1.85 schwarze 2043: mlink = mpage->mlinks;
1.11 schwarze 2044:
1.85 schwarze 2045: if (nodb) {
1.104 schwarze 2046: for (key = ohash_first(&names, &slot); NULL != key;
1.131 schwarze 2047: key = ohash_next(&names, &slot))
1.104 schwarze 2048: free(key);
2049: for (key = ohash_first(&strings, &slot); NULL != key;
1.131 schwarze 2050: key = ohash_next(&strings, &slot))
1.104 schwarze 2051: free(key);
1.102 schwarze 2052: if (0 == debug)
2053: return;
1.85 schwarze 2054: while (NULL != mlink) {
2055: fputs(mlink->name, stdout);
2056: if (NULL == mlink->next ||
2057: strcmp(mlink->dsec, mlink->next->dsec) ||
2058: strcmp(mlink->fsec, mlink->next->fsec) ||
2059: strcmp(mlink->arch, mlink->next->arch)) {
2060: putchar('(');
2061: if ('\0' == *mlink->dsec)
2062: fputs(mlink->fsec, stdout);
2063: else
2064: fputs(mlink->dsec, stdout);
2065: if ('\0' != *mlink->arch)
2066: printf("/%s", mlink->arch);
2067: putchar(')');
2068: }
2069: mlink = mlink->next;
2070: if (NULL != mlink)
2071: fputs(", ", stdout);
2072: }
1.89 schwarze 2073: printf(" - %s\n", mpage->desc);
1.11 schwarze 2074: return;
1.85 schwarze 2075: }
2076:
2077: if (debug)
2078: say(mlink->file, "Adding to database");
1.47 schwarze 2079:
1.131 schwarze 2080: cp = mpage->desc;
2081: i = strlen(cp);
2082: mustfree = render_string(&cp, &i);
1.47 schwarze 2083: i = 1;
1.131 schwarze 2084: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.116 schwarze 2085: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.47 schwarze 2086: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2087: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2088: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.131 schwarze 2089: if (mustfree)
2090: free(cp);
1.47 schwarze 2091:
1.85 schwarze 2092: while (NULL != mlink) {
1.75 schwarze 2093: dbadd_mlink(mlink);
1.85 schwarze 2094: mlink = mlink->next;
2095: }
1.91 schwarze 2096: mlink = mpage->mlinks;
1.47 schwarze 2097:
1.90 schwarze 2098: for (key = ohash_first(&names, &slot); NULL != key;
2099: key = ohash_next(&names, &slot)) {
2100: assert(key->mpage == mpage);
2101: i = 1;
2102: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.131 schwarze 2103: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.94 schwarze 2104: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2105: SQL_STEP(stmts[STMT_INSERT_NAME]);
2106: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2107: free(key);
2108: }
1.47 schwarze 2109: for (key = ohash_first(&strings, &slot); NULL != key;
2110: key = ohash_next(&strings, &slot)) {
2111: assert(key->mpage == mpage);
2112: i = 1;
2113: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.131 schwarze 2114: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.94 schwarze 2115: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2116: SQL_STEP(stmts[STMT_INSERT_KEY]);
2117: sqlite3_reset(stmts[STMT_INSERT_KEY]);
2118: free(key);
1.33 schwarze 2119: }
1.47 schwarze 2120: }
1.41 deraadt 2121:
1.47 schwarze 2122: static void
2123: dbprune(void)
2124: {
2125: struct mpage *mpage;
2126: struct mlink *mlink;
2127: size_t i;
2128: unsigned int slot;
1.11 schwarze 2129:
1.63 schwarze 2130: if (0 == nodb)
2131: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2132:
1.63 schwarze 2133: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2134: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2135: mlink = mpage->mlinks;
1.82 schwarze 2136: if (debug)
1.63 schwarze 2137: say(mlink->file, "Deleting from database");
2138: if (nodb)
2139: continue;
2140: for ( ; NULL != mlink; mlink = mlink->next) {
2141: i = 1;
2142: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2143: i, mlink->dsec);
2144: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2145: i, mlink->arch);
2146: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2147: i, mlink->name);
2148: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2149: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2150: }
1.11 schwarze 2151: }
1.63 schwarze 2152:
2153: if (0 == nodb)
2154: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2155: }
1.22 schwarze 2156:
1.47 schwarze 2157: /*
2158: * Close an existing database and its prepared statements.
2159: * If "real" is not set, rename the temporary file into the real one.
2160: */
2161: static void
2162: dbclose(int real)
2163: {
2164: size_t i;
1.72 schwarze 2165: int status;
2166: pid_t child;
1.11 schwarze 2167:
1.47 schwarze 2168: if (nodb)
2169: return;
1.11 schwarze 2170:
1.47 schwarze 2171: for (i = 0; i < STMT__MAX; i++) {
2172: sqlite3_finalize(stmts[i]);
2173: stmts[i] = NULL;
1.28 schwarze 2174: }
1.22 schwarze 2175:
1.47 schwarze 2176: sqlite3_close(db);
2177: db = NULL;
1.11 schwarze 2178:
1.47 schwarze 2179: if (real)
2180: return;
1.22 schwarze 2181:
1.72 schwarze 2182: if ('\0' == *tempfilename) {
2183: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2184: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2185: say(MANDOC_DB, "&rename");
1.72 schwarze 2186: }
2187: return;
2188: }
2189:
2190: switch (child = fork()) {
1.98 schwarze 2191: case -1:
1.72 schwarze 2192: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2193: say("", "&fork cmp");
1.72 schwarze 2194: return;
1.98 schwarze 2195: case 0:
1.72 schwarze 2196: execlp("cmp", "cmp", "-s",
2197: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2198: say("", "&exec cmp");
1.72 schwarze 2199: exit(0);
2200: default:
2201: break;
2202: }
2203: if (-1 == waitpid(child, &status, 0)) {
2204: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2205: say("", "&wait cmp");
1.72 schwarze 2206: } else if (WIFSIGNALED(status)) {
2207: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2208: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2209: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2210: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2211: say(MANDOC_DB,
2212: "Data changed, but cannot replace database");
2213: }
2214:
2215: *strrchr(tempfilename, '/') = '\0';
2216: switch (child = fork()) {
1.98 schwarze 2217: case -1:
1.72 schwarze 2218: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2219: say("", "&fork rm");
1.72 schwarze 2220: return;
1.98 schwarze 2221: case 0:
1.72 schwarze 2222: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2223: say("", "&exec rm");
1.72 schwarze 2224: exit((int)MANDOCLEVEL_SYSERR);
2225: default:
2226: break;
2227: }
2228: if (-1 == waitpid(child, &status, 0)) {
2229: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2230: say("", "&wait rm");
1.72 schwarze 2231: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2232: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2233: say("", "%s: Cannot remove temporary directory",
2234: tempfilename);
1.22 schwarze 2235: }
1.11 schwarze 2236: }
2237:
1.47 schwarze 2238: /*
2239: * This is straightforward stuff.
2240: * Open a database connection to a "temporary" database, then open a set
2241: * of prepared statements we'll use over and over again.
2242: * If "real" is set, we use the existing database; if not, we truncate a
2243: * temporary one.
2244: * Must be matched by dbclose().
2245: */
2246: static int
2247: dbopen(int real)
1.2 schwarze 2248: {
1.72 schwarze 2249: const char *sql;
1.47 schwarze 2250: int rc, ofl;
1.6 schwarze 2251:
1.98 schwarze 2252: if (nodb)
1.47 schwarze 2253: return(1);
1.6 schwarze 2254:
1.72 schwarze 2255: *tempfilename = '\0';
1.47 schwarze 2256: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2257:
2258: if (real) {
2259: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2260: if (SQLITE_OK != rc) {
1.47 schwarze 2261: exitcode = (int)MANDOCLEVEL_SYSERR;
1.108 sthen 2262: if (SQLITE_CANTOPEN != rc)
2263: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.47 schwarze 2264: return(0);
1.28 schwarze 2265: }
1.72 schwarze 2266: goto prepare_statements;
2267: }
2268:
2269: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2270:
1.72 schwarze 2271: remove(MANDOC_DB "~");
2272: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.98 schwarze 2273: if (SQLITE_OK == rc)
1.72 schwarze 2274: goto create_tables;
1.73 schwarze 2275: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2276: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2277: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.47 schwarze 2278: return(0);
2279: }
1.6 schwarze 2280:
1.100 schwarze 2281: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2282: sizeof(tempfilename));
1.72 schwarze 2283: if (NULL == mkdtemp(tempfilename)) {
2284: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2285: say("", "&%s", tempfilename);
1.72 schwarze 2286: return(0);
2287: }
1.100 schwarze 2288: (void)strlcat(tempfilename, "/" MANDOC_DB,
2289: sizeof(tempfilename));
1.72 schwarze 2290: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2291: if (SQLITE_OK != rc) {
1.47 schwarze 2292: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2293: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.47 schwarze 2294: return(0);
1.2 schwarze 2295: }
2296:
1.72 schwarze 2297: create_tables:
1.47 schwarze 2298: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2299: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2300: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2301: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2302: ");\n"
2303: "\n"
2304: "CREATE TABLE \"mlinks\" (\n"
2305: " \"sec\" TEXT NOT NULL,\n"
2306: " \"arch\" TEXT NOT NULL,\n"
2307: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2308: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2309: "ON DELETE CASCADE\n"
1.47 schwarze 2310: ");\n"
1.93 schwarze 2311: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2312: "\n"
1.90 schwarze 2313: "CREATE TABLE \"names\" (\n"
2314: " \"bits\" INTEGER NOT NULL,\n"
2315: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2316: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.130 schwarze 2317: "ON DELETE CASCADE,\n"
2318: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.90 schwarze 2319: ");\n"
2320: "\n"
1.47 schwarze 2321: "CREATE TABLE \"keys\" (\n"
2322: " \"bits\" INTEGER NOT NULL,\n"
2323: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2324: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2325: "ON DELETE CASCADE\n"
1.93 schwarze 2326: ");\n"
2327: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2328:
2329: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2330: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2331: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.103 schwarze 2332: sqlite3_close(db);
1.47 schwarze 2333: return(0);
1.2 schwarze 2334: }
2335:
1.47 schwarze 2336: prepare_statements:
1.103 schwarze 2337: if (SQLITE_OK != sqlite3_exec(db,
2338: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2339: exitcode = (int)MANDOCLEVEL_SYSERR;
2340: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2341: sqlite3_errmsg(db));
2342: sqlite3_close(db);
2343: return(0);
2344: }
2345:
1.94 schwarze 2346: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2347: "(SELECT pageid FROM mlinks WHERE "
2348: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2349: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2350: sql = "INSERT INTO mpages "
1.89 schwarze 2351: "(desc,form) VALUES (?,?)";
1.47 schwarze 2352: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2353: sql = "INSERT INTO mlinks "
1.61 schwarze 2354: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2355: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.130 schwarze 2356: sql = "SELECT bits FROM names where pageid = ?";
2357: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.90 schwarze 2358: sql = "INSERT INTO names "
2359: "(bits,name,pageid) VALUES (?,?,?)";
2360: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2361: sql = "INSERT INTO keys "
2362: "(bits,key,pageid) VALUES (?,?,?)";
2363: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2364:
1.47 schwarze 2365: /*
2366: * When opening a new database, we can turn off
2367: * synchronous mode for much better performance.
2368: */
1.6 schwarze 2369:
1.103 schwarze 2370: if (real && SQLITE_OK != sqlite3_exec(db,
2371: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2372: exitcode = (int)MANDOCLEVEL_SYSERR;
2373: say(MANDOC_DB, "PRAGMA synchronous: %s",
2374: sqlite3_errmsg(db));
2375: sqlite3_close(db);
2376: return(0);
2377: }
1.11 schwarze 2378:
1.47 schwarze 2379: return(1);
2380: }
1.6 schwarze 2381:
1.47 schwarze 2382: static void *
1.106 espie 2383: hash_calloc(size_t n, size_t sz, void *arg)
1.47 schwarze 2384: {
1.6 schwarze 2385:
1.106 espie 2386: return(mandoc_calloc(n, sz));
1.47 schwarze 2387: }
1.2 schwarze 2388:
1.47 schwarze 2389: static void *
2390: hash_alloc(size_t sz, void *arg)
2391: {
1.28 schwarze 2392:
1.47 schwarze 2393: return(mandoc_malloc(sz));
2394: }
1.6 schwarze 2395:
1.47 schwarze 2396: static void
1.106 espie 2397: hash_free(void *p, void *arg)
1.47 schwarze 2398: {
1.26 schwarze 2399:
1.47 schwarze 2400: free(p);
2401: }
1.6 schwarze 2402:
1.47 schwarze 2403: static int
1.120 schwarze 2404: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2405: {
2406: static char startdir[PATH_MAX];
1.110 schwarze 2407: static int getcwd_status; /* 1 = ok, 2 = failure */
2408: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2409: char *cp;
1.6 schwarze 2410:
1.47 schwarze 2411: /*
1.110 schwarze 2412: * Remember the original working directory, if possible.
2413: * This will be needed if the second or a later directory
2414: * on the command line is given as a relative path.
2415: * Do not error out if the current directory is not
2416: * searchable: Maybe it won't be needed after all.
2417: */
2418: if (0 == getcwd_status) {
2419: if (NULL == getcwd(startdir, sizeof(startdir))) {
2420: getcwd_status = 2;
2421: (void)strlcpy(startdir, strerror(errno),
2422: sizeof(startdir));
2423: } else
2424: getcwd_status = 1;
2425: }
2426:
2427: /*
2428: * We are leaving the old base directory.
2429: * Do not use it any longer, not even for messages.
2430: */
2431: *basedir = '\0';
2432:
2433: /*
2434: * If and only if the directory was changed earlier and
2435: * the next directory to process is given as a relative path,
2436: * first go back, or bail out if that is impossible.
1.47 schwarze 2437: */
1.110 schwarze 2438: if (chdir_status && '/' != *targetdir) {
2439: if (2 == getcwd_status) {
1.47 schwarze 2440: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2441: say("", "getcwd: %s", startdir);
1.47 schwarze 2442: return(0);
2443: }
1.110 schwarze 2444: if (-1 == chdir(startdir)) {
1.47 schwarze 2445: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2446: say("", "&chdir %s", startdir);
1.47 schwarze 2447: return(0);
1.2 schwarze 2448: }
2449: }
1.110 schwarze 2450:
2451: /*
2452: * Always resolve basedir to the canonicalized absolute
2453: * pathname and append a trailing slash, such that
2454: * we can reliably check whether files are inside.
2455: */
1.47 schwarze 2456: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2457: if (report_baddir || errno != ENOENT) {
2458: exitcode = (int)MANDOCLEVEL_BADARG;
2459: say("", "&%s: realpath", targetdir);
2460: }
1.47 schwarze 2461: return(0);
2462: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2463: if (report_baddir || errno != ENOENT) {
2464: exitcode = (int)MANDOCLEVEL_BADARG;
2465: say("", "&chdir");
2466: }
1.47 schwarze 2467: return(0);
1.109 schwarze 2468: }
1.110 schwarze 2469: chdir_status = 1;
1.109 schwarze 2470: cp = strchr(basedir, '\0');
2471: if ('/' != cp[-1]) {
2472: if (cp - basedir >= PATH_MAX - 1) {
2473: exitcode = (int)MANDOCLEVEL_SYSERR;
2474: say("", "Filename too long");
2475: return(0);
2476: }
2477: *cp++ = '/';
2478: *cp = '\0';
1.47 schwarze 2479: }
2480: return(1);
1.2 schwarze 2481: }
2482:
2483: static void
1.47 schwarze 2484: say(const char *file, const char *format, ...)
1.2 schwarze 2485: {
1.47 schwarze 2486: va_list ap;
1.80 schwarze 2487: int use_errno;
1.2 schwarze 2488:
1.47 schwarze 2489: if ('\0' != *basedir)
2490: fprintf(stderr, "%s", basedir);
2491: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2492: fputc('/', stderr);
1.47 schwarze 2493: if ('\0' != *file)
2494: fprintf(stderr, "%s", file);
1.31 schwarze 2495:
1.80 schwarze 2496: use_errno = 1;
2497: if (NULL != format) {
2498: switch (*format) {
1.98 schwarze 2499: case '&':
1.80 schwarze 2500: format++;
2501: break;
1.98 schwarze 2502: case '\0':
1.80 schwarze 2503: format = NULL;
2504: break;
2505: default:
2506: use_errno = 0;
2507: break;
2508: }
2509: }
2510: if (NULL != format) {
2511: if ('\0' != *basedir || '\0' != *file)
2512: fputs(": ", stderr);
2513: va_start(ap, format);
2514: vfprintf(stderr, format, ap);
2515: va_end(ap);
2516: }
2517: if (use_errno) {
2518: if ('\0' != *basedir || '\0' != *file || NULL != format)
2519: fputs(": ", stderr);
1.47 schwarze 2520: perror(NULL);
1.80 schwarze 2521: } else
2522: fputc('\n', stderr);
1.1 schwarze 2523: }