Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.132
1.132 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.131 2014/12/05 14:26:23 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 {
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 *);
158: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_meta *,
159: const struct mdoc_node *);
1.128 schwarze 160: static int parse_mdoc_Fo(struct mpage *, const struct mdoc_meta *,
161: const struct mdoc_node *);
1.127 schwarze 162: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_meta *,
163: const struct mdoc_node *);
164: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_meta *,
165: const struct mdoc_node *);
166: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_meta *,
167: const struct mdoc_node *);
168: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_meta *,
169: const struct mdoc_node *);
1.69 schwarze 170: static void putkey(const struct mpage *, char *, uint64_t);
1.131 schwarze 171: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.47 schwarze 172: static void putmdockey(const struct mpage *,
173: const struct mdoc_node *, uint64_t);
1.131 schwarze 174: static int render_string(char **, size_t *);
1.47 schwarze 175: static void say(const char *, const char *, ...);
1.120 schwarze 176: static int set_basedir(const char *, int);
1.47 schwarze 177: static int treescan(void);
178: static size_t utf8(unsigned int, char [7]);
179:
1.72 schwarze 180: static char tempfilename[32];
1.47 schwarze 181: static char *progname;
1.59 schwarze 182: static int nodb; /* no database changes */
1.73 schwarze 183: static int mparse_options; /* abort the parse early */
1.98 schwarze 184: static int use_all; /* use all found files */
185: static int debug; /* print what we're doing */
186: static int warnings; /* warn about crap */
1.52 schwarze 187: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 188: static int exitcode; /* to be returned by main */
1.98 schwarze 189: static enum op op; /* operational mode */
1.47 schwarze 190: static char basedir[PATH_MAX]; /* current base directory */
1.131 schwarze 191: static struct mchars *mchars; /* table of named characters */
1.47 schwarze 192: static struct ohash mpages; /* table of distinct manual pages */
193: static struct ohash mlinks; /* table of directory entries */
1.90 schwarze 194: static struct ohash names; /* table of all names */
1.47 schwarze 195: static struct ohash strings; /* table of all strings */
196: static sqlite3 *db = NULL; /* current database */
197: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.90 schwarze 198: static uint64_t name_mask;
1.47 schwarze 199:
1.19 schwarze 200: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 201: { NULL, 0 }, /* Ap */
202: { NULL, 0 }, /* Dd */
203: { NULL, 0 }, /* Dt */
204: { NULL, 0 }, /* Os */
205: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
206: { parse_mdoc_head, TYPE_Ss }, /* Ss */
207: { NULL, 0 }, /* Pp */
208: { NULL, 0 }, /* D1 */
209: { NULL, 0 }, /* Dl */
210: { NULL, 0 }, /* Bd */
211: { NULL, 0 }, /* Ed */
212: { NULL, 0 }, /* Bl */
213: { NULL, 0 }, /* El */
214: { NULL, 0 }, /* It */
215: { NULL, 0 }, /* Ad */
216: { NULL, TYPE_An }, /* An */
217: { NULL, TYPE_Ar }, /* Ar */
218: { NULL, TYPE_Cd }, /* Cd */
219: { NULL, TYPE_Cm }, /* Cm */
220: { NULL, TYPE_Dv }, /* Dv */
221: { NULL, TYPE_Er }, /* Er */
222: { NULL, TYPE_Ev }, /* Ev */
223: { NULL, 0 }, /* Ex */
224: { NULL, TYPE_Fa }, /* Fa */
225: { parse_mdoc_Fd, 0 }, /* Fd */
226: { NULL, TYPE_Fl }, /* Fl */
227: { parse_mdoc_Fn, 0 }, /* Fn */
228: { NULL, TYPE_Ft }, /* Ft */
229: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 230: { NULL, TYPE_In }, /* In */
1.47 schwarze 231: { NULL, TYPE_Li }, /* Li */
1.89 schwarze 232: { parse_mdoc_Nd, 0 }, /* Nd */
1.86 schwarze 233: { parse_mdoc_Nm, 0 }, /* Nm */
1.47 schwarze 234: { NULL, 0 }, /* Op */
235: { NULL, 0 }, /* Ot */
236: { NULL, TYPE_Pa }, /* Pa */
237: { NULL, 0 }, /* Rv */
1.49 schwarze 238: { NULL, TYPE_St }, /* St */
1.47 schwarze 239: { NULL, TYPE_Va }, /* Va */
240: { parse_mdoc_body, TYPE_Va }, /* Vt */
241: { parse_mdoc_Xr, 0 }, /* Xr */
242: { NULL, 0 }, /* %A */
243: { NULL, 0 }, /* %B */
244: { NULL, 0 }, /* %D */
245: { NULL, 0 }, /* %I */
246: { NULL, 0 }, /* %J */
247: { NULL, 0 }, /* %N */
248: { NULL, 0 }, /* %O */
249: { NULL, 0 }, /* %P */
250: { NULL, 0 }, /* %R */
251: { NULL, 0 }, /* %T */
252: { NULL, 0 }, /* %V */
253: { NULL, 0 }, /* Ac */
254: { NULL, 0 }, /* Ao */
255: { NULL, 0 }, /* Aq */
256: { NULL, TYPE_At }, /* At */
257: { NULL, 0 }, /* Bc */
258: { NULL, 0 }, /* Bf */
259: { NULL, 0 }, /* Bo */
260: { NULL, 0 }, /* Bq */
261: { NULL, TYPE_Bsx }, /* Bsx */
262: { NULL, TYPE_Bx }, /* Bx */
263: { NULL, 0 }, /* Db */
264: { NULL, 0 }, /* Dc */
265: { NULL, 0 }, /* Do */
266: { NULL, 0 }, /* Dq */
267: { NULL, 0 }, /* Ec */
268: { NULL, 0 }, /* Ef */
269: { NULL, TYPE_Em }, /* Em */
270: { NULL, 0 }, /* Eo */
271: { NULL, TYPE_Fx }, /* Fx */
272: { NULL, TYPE_Ms }, /* Ms */
273: { NULL, 0 }, /* No */
274: { NULL, 0 }, /* Ns */
275: { NULL, TYPE_Nx }, /* Nx */
276: { NULL, TYPE_Ox }, /* Ox */
277: { NULL, 0 }, /* Pc */
278: { NULL, 0 }, /* Pf */
279: { NULL, 0 }, /* Po */
280: { NULL, 0 }, /* Pq */
281: { NULL, 0 }, /* Qc */
282: { NULL, 0 }, /* Ql */
283: { NULL, 0 }, /* Qo */
284: { NULL, 0 }, /* Qq */
285: { NULL, 0 }, /* Re */
286: { NULL, 0 }, /* Rs */
287: { NULL, 0 }, /* Sc */
288: { NULL, 0 }, /* So */
289: { NULL, 0 }, /* Sq */
290: { NULL, 0 }, /* Sm */
291: { NULL, 0 }, /* Sx */
292: { NULL, TYPE_Sy }, /* Sy */
293: { NULL, TYPE_Tn }, /* Tn */
294: { NULL, 0 }, /* Ux */
295: { NULL, 0 }, /* Xc */
296: { NULL, 0 }, /* Xo */
1.128 schwarze 297: { parse_mdoc_Fo, 0 }, /* Fo */
1.47 schwarze 298: { NULL, 0 }, /* Fc */
299: { NULL, 0 }, /* Oo */
300: { NULL, 0 }, /* Oc */
301: { NULL, 0 }, /* Bk */
302: { NULL, 0 }, /* Ek */
303: { NULL, 0 }, /* Bt */
304: { NULL, 0 }, /* Hf */
305: { NULL, 0 }, /* Fr */
306: { NULL, 0 }, /* Ud */
307: { NULL, TYPE_Lb }, /* Lb */
308: { NULL, 0 }, /* Lp */
309: { NULL, TYPE_Lk }, /* Lk */
310: { NULL, TYPE_Mt }, /* Mt */
311: { NULL, 0 }, /* Brq */
312: { NULL, 0 }, /* Bro */
313: { NULL, 0 }, /* Brc */
314: { NULL, 0 }, /* %C */
315: { NULL, 0 }, /* Es */
316: { NULL, 0 }, /* En */
317: { NULL, TYPE_Dx }, /* Dx */
318: { NULL, 0 }, /* %Q */
319: { NULL, 0 }, /* br */
320: { NULL, 0 }, /* sp */
321: { NULL, 0 }, /* %U */
322: { NULL, 0 }, /* Ta */
1.121 schwarze 323: { NULL, 0 }, /* ll */
1.1 schwarze 324: };
325:
1.98 schwarze 326:
1.1 schwarze 327: int
1.3 schwarze 328: mandocdb(int argc, char *argv[])
1.1 schwarze 329: {
1.47 schwarze 330: int ch, i;
331: size_t j, sz;
332: const char *path_arg;
333: struct manpaths dirs;
334: struct mparse *mp;
335: struct ohash_info mpages_info, mlinks_info;
336:
337: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
338: memset(&dirs, 0, sizeof(struct manpaths));
339:
340: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
1.106 espie 341: mpages_info.calloc = mlinks_info.calloc = hash_calloc;
342: mpages_info.free = mlinks_info.free = hash_free;
1.47 schwarze 343:
344: mpages_info.key_offset = offsetof(struct mpage, inodev);
345: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 346:
347: progname = strrchr(argv[0], '/');
348: if (progname == NULL)
349: progname = argv[0];
350: else
351: ++progname;
352:
1.47 schwarze 353: /*
1.98 schwarze 354: * We accept a few different invocations.
1.47 schwarze 355: * The CHECKOP macro makes sure that invocation styles don't
356: * clobber each other.
357: */
358: #define CHECKOP(_op, _ch) do \
359: if (OP_DEFAULT != (_op)) { \
1.111 schwarze 360: fprintf(stderr, "%s: -%c: Conflicting option\n", \
361: progname, (_ch)); \
1.47 schwarze 362: goto usage; \
363: } while (/*CONSTCOND*/0)
1.10 schwarze 364:
1.47 schwarze 365: path_arg = NULL;
1.28 schwarze 366: op = OP_DEFAULT;
1.1 schwarze 367:
1.83 schwarze 368: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 369: switch (ch) {
1.98 schwarze 370: case 'a':
1.6 schwarze 371: use_all = 1;
372: break;
1.98 schwarze 373: case 'C':
1.47 schwarze 374: CHECKOP(op, ch);
375: path_arg = optarg;
1.28 schwarze 376: op = OP_CONFFILE;
1.25 schwarze 377: break;
1.98 schwarze 378: case 'D':
1.82 schwarze 379: debug++;
380: break;
1.98 schwarze 381: case 'd':
1.47 schwarze 382: CHECKOP(op, ch);
383: path_arg = optarg;
1.2 schwarze 384: op = OP_UPDATE;
1.1 schwarze 385: break;
1.98 schwarze 386: case 'n':
1.47 schwarze 387: nodb = 1;
388: break;
1.98 schwarze 389: case 'p':
1.83 schwarze 390: warnings = 1;
391: break;
1.98 schwarze 392: case 'Q':
1.73 schwarze 393: mparse_options |= MPARSE_QUICK;
1.59 schwarze 394: break;
1.98 schwarze 395: case 'T':
1.52 schwarze 396: if (strcmp(optarg, "utf8")) {
1.111 schwarze 397: fprintf(stderr, "%s: -T%s: "
398: "Unsupported output format\n",
399: progname, optarg);
1.52 schwarze 400: goto usage;
401: }
402: write_utf8 = 1;
403: break;
1.98 schwarze 404: case 't':
1.47 schwarze 405: CHECKOP(op, ch);
1.28 schwarze 406: dup2(STDOUT_FILENO, STDERR_FILENO);
407: op = OP_TEST;
1.47 schwarze 408: nodb = warnings = 1;
1.28 schwarze 409: break;
1.98 schwarze 410: case 'u':
1.47 schwarze 411: CHECKOP(op, ch);
412: path_arg = optarg;
1.1 schwarze 413: op = OP_DELETE;
414: break;
1.98 schwarze 415: case 'v':
1.83 schwarze 416: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 417: break;
1.1 schwarze 418: default:
1.28 schwarze 419: goto usage;
1.1 schwarze 420: }
421:
422: argc -= optind;
423: argv += optind;
424:
1.28 schwarze 425: if (OP_CONFFILE == op && argc > 0) {
1.111 schwarze 426: fprintf(stderr, "%s: -C: Too many arguments\n",
427: progname);
1.28 schwarze 428: goto usage;
429: }
430:
1.47 schwarze 431: exitcode = (int)MANDOCLEVEL_OK;
1.131 schwarze 432: mchars = mchars_alloc();
1.123 schwarze 433: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL,
1.131 schwarze 434: mchars, NULL);
1.47 schwarze 435: ohash_init(&mpages, 6, &mpages_info);
436: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 437:
1.47 schwarze 438: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.2 schwarze 439:
1.47 schwarze 440: /*
1.109 schwarze 441: * Most of these deal with a specific directory.
1.97 schwarze 442: * Jump into that directory first.
1.47 schwarze 443: */
1.120 schwarze 444: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.47 schwarze 445: goto out;
1.97 schwarze 446:
1.95 schwarze 447: if (dbopen(1)) {
1.97 schwarze 448: /*
449: * The existing database is usable. Process
450: * all files specified on the command-line.
451: */
452: use_all = 1;
1.95 schwarze 453: for (i = 0; i < argc; i++)
454: filescan(argv[i]);
455: if (OP_TEST != op)
456: dbprune();
457: } else {
458: /*
459: * Database missing or corrupt.
460: * Recreate from scratch.
461: */
1.97 schwarze 462: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 463: op = OP_DEFAULT;
464: if (0 == treescan())
465: goto out;
466: if (0 == dbopen(0))
467: goto out;
468: }
1.47 schwarze 469: if (OP_DELETE != op)
1.131 schwarze 470: mpages_merge(mp);
1.95 schwarze 471: dbclose(OP_DEFAULT == op ? 0 : 1);
1.47 schwarze 472: } else {
473: /*
474: * If we have arguments, use them as our manpaths.
475: * If we don't, grok from manpath(1) or however else
476: * manpath_parse() wants to do it.
477: */
478: if (argc > 0) {
1.101 schwarze 479: dirs.paths = mandoc_reallocarray(NULL,
480: argc, sizeof(char *));
1.47 schwarze 481: dirs.sz = (size_t)argc;
482: for (i = 0; i < argc; i++)
483: dirs.paths[i] = mandoc_strdup(argv[i]);
484: } else
485: manpath_parse(&dirs, path_arg, NULL, NULL);
1.84 schwarze 486:
487: if (0 == dirs.sz) {
488: exitcode = (int)MANDOCLEVEL_BADARG;
489: say("", "Empty manpath");
490: }
1.2 schwarze 491:
1.47 schwarze 492: /*
493: * First scan the tree rooted at a base directory, then
494: * build a new database and finally move it into place.
495: * Ignore zero-length directories and strip trailing
496: * slashes.
497: */
498: for (j = 0; j < dirs.sz; j++) {
499: sz = strlen(dirs.paths[j]);
500: if (sz && '/' == dirs.paths[j][sz - 1])
501: dirs.paths[j][--sz] = '\0';
502: if (0 == sz)
503: continue;
1.2 schwarze 504:
1.47 schwarze 505: if (j) {
506: ohash_init(&mpages, 6, &mpages_info);
507: ohash_init(&mlinks, 6, &mlinks_info);
508: }
1.1 schwarze 509:
1.120 schwarze 510: if (0 == set_basedir(dirs.paths[j], argc > 0))
511: continue;
1.47 schwarze 512: if (0 == treescan())
1.120 schwarze 513: continue;
1.47 schwarze 514: if (0 == dbopen(0))
1.120 schwarze 515: continue;
1.41 deraadt 516:
1.131 schwarze 517: mpages_merge(mp);
1.104 schwarze 518: if (warnings && !nodb &&
1.87 schwarze 519: ! (MPARSE_QUICK & mparse_options))
520: names_check();
1.47 schwarze 521: dbclose(0);
1.28 schwarze 522:
1.47 schwarze 523: if (j + 1 < dirs.sz) {
524: mpages_free();
525: ohash_delete(&mpages);
526: ohash_delete(&mlinks);
527: }
1.2 schwarze 528: }
1.47 schwarze 529: }
530: out:
531: manpath_free(&dirs);
1.123 schwarze 532: mparse_free(mp);
1.131 schwarze 533: mchars_free(mchars);
1.47 schwarze 534: mpages_free();
535: ohash_delete(&mpages);
536: ohash_delete(&mlinks);
537: return(exitcode);
538: usage:
1.83 schwarze 539: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
540: " %s [-aDnpQ] [-Tutf8] dir ...\n"
541: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
542: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 543: " %s [-Q] -t file ...\n",
1.98 schwarze 544: progname, progname, progname,
1.47 schwarze 545: progname, progname);
1.1 schwarze 546:
1.47 schwarze 547: return((int)MANDOCLEVEL_BADARG);
548: }
1.1 schwarze 549:
1.47 schwarze 550: /*
551: * Scan a directory tree rooted at "basedir" for manpages.
552: * We use fts(), scanning directory parts along the way for clues to our
553: * section and architecture.
554: *
555: * If use_all has been specified, grok all files.
556: * If not, sanitise paths to the following:
557: *
1.98 schwarze 558: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 559: * or
560: * [./]cat<section>[/<arch>]/<name>.0
561: *
562: * TODO: accomodate for multi-language directories.
563: */
564: static int
565: treescan(void)
566: {
1.96 schwarze 567: char buf[PATH_MAX];
1.47 schwarze 568: FTS *f;
569: FTSENT *ff;
570: struct mlink *mlink;
1.81 schwarze 571: int dform, gzip;
1.51 schwarze 572: char *dsec, *arch, *fsec, *cp;
573: const char *path;
1.47 schwarze 574: const char *argv[2];
1.1 schwarze 575:
1.47 schwarze 576: argv[0] = ".";
577: argv[1] = (char *)NULL;
1.1 schwarze 578:
1.96 schwarze 579: f = fts_open((char * const *)argv,
580: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.47 schwarze 581: if (NULL == f) {
582: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 583: say("", "&fts_open");
1.47 schwarze 584: return(0);
585: }
1.2 schwarze 586:
1.47 schwarze 587: dsec = arch = NULL;
588: dform = FORM_NONE;
1.2 schwarze 589:
1.47 schwarze 590: while (NULL != (ff = fts_read(f))) {
591: path = ff->fts_path + 2;
1.96 schwarze 592: switch (ff->fts_info) {
593:
594: /*
595: * Symbolic links require various sanity checks,
596: * then get handled just like regular files.
597: */
1.98 schwarze 598: case FTS_SL:
1.96 schwarze 599: if (NULL == realpath(path, buf)) {
600: if (warnings)
601: say(path, "&realpath");
602: continue;
603: }
604: if (strstr(buf, basedir) != buf) {
605: if (warnings) say("",
606: "%s: outside base directory", buf);
607: continue;
608: }
609: /* Use logical inode to avoid mpages dupe. */
610: if (-1 == stat(path, ff->fts_statp)) {
611: if (warnings)
612: say(path, "&stat");
613: continue;
614: }
615: /* FALLTHROUGH */
616:
1.15 schwarze 617: /*
1.47 schwarze 618: * If we're a regular file, add an mlink by using the
619: * stored directory data and handling the filename.
1.15 schwarze 620: */
1.98 schwarze 621: case FTS_F:
1.47 schwarze 622: if (0 == strcmp(path, MANDOC_DB))
623: continue;
624: if ( ! use_all && ff->fts_level < 2) {
625: if (warnings)
626: say(path, "Extraneous file");
627: continue;
1.81 schwarze 628: }
629: gzip = 0;
630: fsec = NULL;
631: while (NULL == fsec) {
632: fsec = strrchr(ff->fts_name, '.');
633: if (NULL == fsec || strcmp(fsec+1, "gz"))
634: break;
635: gzip = 1;
636: *fsec = '\0';
637: fsec = NULL;
638: }
639: if (NULL == fsec) {
1.47 schwarze 640: if ( ! use_all) {
641: if (warnings)
642: say(path,
643: "No filename suffix");
644: continue;
645: }
646: } else if (0 == strcmp(++fsec, "html")) {
647: if (warnings)
648: say(path, "Skip html");
649: continue;
650: } else if (0 == strcmp(fsec, "ps")) {
651: if (warnings)
652: say(path, "Skip ps");
653: continue;
654: } else if (0 == strcmp(fsec, "pdf")) {
655: if (warnings)
656: say(path, "Skip pdf");
657: continue;
658: } else if ( ! use_all &&
659: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
660: (FORM_CAT == dform && strcmp(fsec, "0")))) {
661: if (warnings)
662: say(path, "Wrong filename suffix");
663: continue;
664: } else
665: fsec[-1] = '\0';
1.51 schwarze 666:
1.47 schwarze 667: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 668: if (strlcpy(mlink->file, path,
669: sizeof(mlink->file)) >=
670: sizeof(mlink->file)) {
671: say(path, "Filename too long");
672: free(mlink);
673: continue;
674: }
1.47 schwarze 675: mlink->dform = dform;
1.51 schwarze 676: mlink->dsec = dsec;
677: mlink->arch = arch;
678: mlink->name = ff->fts_name;
679: mlink->fsec = fsec;
1.81 schwarze 680: mlink->gzip = gzip;
1.47 schwarze 681: mlink_add(mlink, ff->fts_statp);
682: continue;
1.96 schwarze 683:
1.98 schwarze 684: case FTS_D:
1.96 schwarze 685: /* FALLTHROUGH */
1.98 schwarze 686: case FTS_DP:
1.96 schwarze 687: break;
688:
689: default:
1.47 schwarze 690: if (warnings)
691: say(path, "Not a regular file");
692: continue;
693: }
694:
695: switch (ff->fts_level) {
1.98 schwarze 696: case 0:
1.47 schwarze 697: /* Ignore the root directory. */
698: break;
1.98 schwarze 699: case 1:
1.47 schwarze 700: /*
701: * This might contain manX/ or catX/.
702: * Try to infer this from the name.
703: * If we're not in use_all, enforce it.
704: */
705: cp = ff->fts_name;
706: if (FTS_DP == ff->fts_info)
707: break;
1.15 schwarze 708:
1.47 schwarze 709: if (0 == strncmp(cp, "man", 3)) {
710: dform = FORM_SRC;
711: dsec = cp + 3;
712: } else if (0 == strncmp(cp, "cat", 3)) {
713: dform = FORM_CAT;
714: dsec = cp + 3;
1.51 schwarze 715: } else {
716: dform = FORM_NONE;
717: dsec = NULL;
1.26 schwarze 718: }
1.47 schwarze 719:
1.98 schwarze 720: if (NULL != dsec || use_all)
1.47 schwarze 721: break;
722:
723: if (warnings)
724: say(path, "Unknown directory part");
725: fts_set(f, ff, FTS_SKIP);
726: break;
1.98 schwarze 727: case 2:
1.47 schwarze 728: /*
729: * Possibly our architecture.
730: * If we're descending, keep tabs on it.
731: */
732: if (FTS_DP != ff->fts_info && NULL != dsec)
733: arch = ff->fts_name;
1.51 schwarze 734: else
735: arch = NULL;
1.47 schwarze 736: break;
737: default:
738: if (FTS_DP == ff->fts_info || use_all)
739: break;
740: if (warnings)
741: say(path, "Extraneous directory part");
742: fts_set(f, ff, FTS_SKIP);
743: break;
1.14 schwarze 744: }
1.47 schwarze 745: }
746:
747: fts_close(f);
748: return(1);
749: }
1.1 schwarze 750:
1.47 schwarze 751: /*
752: * Add a file to the mlinks table.
753: * Do not verify that it's a "valid" looking manpage (we'll do that
754: * later).
755: *
756: * Try to infer the manual section, architecture, and page name from the
757: * path, assuming it looks like
758: *
1.98 schwarze 759: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 760: * or
761: * [./]cat<section>[/<arch>]/<name>.0
762: *
763: * See treescan() for the fts(3) version of this.
764: */
765: static void
766: filescan(const char *file)
767: {
768: char buf[PATH_MAX];
769: struct stat st;
770: struct mlink *mlink;
771: char *p, *start;
772:
773: assert(use_all);
774:
775: if (0 == strncmp(file, "./", 2))
776: file += 2;
777:
1.96 schwarze 778: /*
779: * We have to do lstat(2) before realpath(3) loses
780: * the information whether this is a symbolic link.
781: * We need to know that because for symbolic links,
782: * we want to use the orginal file name, while for
783: * regular files, we want to use the real path.
784: */
785: if (-1 == lstat(file, &st)) {
786: exitcode = (int)MANDOCLEVEL_BADARG;
787: say(file, "&lstat");
788: return;
789: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
790: exitcode = (int)MANDOCLEVEL_BADARG;
791: say(file, "Not a regular file");
792: return;
793: }
794:
795: /*
796: * We have to resolve the file name to the real path
797: * in any case for the base directory check.
798: */
1.47 schwarze 799: if (NULL == realpath(file, buf)) {
800: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 801: say(file, "&realpath");
1.47 schwarze 802: return;
1.63 schwarze 803: }
804:
1.109 schwarze 805: if (OP_TEST == op)
1.63 schwarze 806: start = buf;
1.109 schwarze 807: else if (strstr(buf, basedir) == buf)
808: start = buf + strlen(basedir);
1.63 schwarze 809: else {
1.47 schwarze 810: exitcode = (int)MANDOCLEVEL_BADARG;
811: say("", "%s: outside base directory", buf);
812: return;
1.63 schwarze 813: }
814:
1.96 schwarze 815: /*
816: * Now we are sure the file is inside our tree.
817: * If it is a symbolic link, ignore the real path
818: * and use the original name.
819: * This implies passing stuff like "cat1/../man1/foo.1"
820: * on the command line won't work. So don't do that.
821: * Note the stat(2) can still fail if the link target
822: * doesn't exist.
823: */
824: if (S_IFLNK & st.st_mode) {
825: if (-1 == stat(buf, &st)) {
826: exitcode = (int)MANDOCLEVEL_BADARG;
827: say(file, "&stat");
828: return;
829: }
1.100 schwarze 830: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
831: say(file, "Filename too long");
832: return;
833: }
1.109 schwarze 834: start = buf;
835: if (OP_TEST != op && strstr(buf, basedir) == buf)
836: start += strlen(basedir);
1.1 schwarze 837: }
1.63 schwarze 838:
1.47 schwarze 839: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 840: mlink->dform = FORM_NONE;
1.100 schwarze 841: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
842: sizeof(mlink->file)) {
843: say(start, "Filename too long");
844: return;
845: }
1.1 schwarze 846:
1.10 schwarze 847: /*
1.47 schwarze 848: * First try to guess our directory structure.
849: * If we find a separator, try to look for man* or cat*.
850: * If we find one of these and what's underneath is a directory,
851: * assume it's an architecture.
1.10 schwarze 852: */
1.47 schwarze 853: if (NULL != (p = strchr(start, '/'))) {
854: *p++ = '\0';
855: if (0 == strncmp(start, "man", 3)) {
856: mlink->dform = FORM_SRC;
1.51 schwarze 857: mlink->dsec = start + 3;
1.47 schwarze 858: } else if (0 == strncmp(start, "cat", 3)) {
859: mlink->dform = FORM_CAT;
1.51 schwarze 860: mlink->dsec = start + 3;
1.47 schwarze 861: }
1.10 schwarze 862:
1.47 schwarze 863: start = p;
864: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
865: *p++ = '\0';
1.51 schwarze 866: mlink->arch = start;
1.47 schwarze 867: start = p;
1.41 deraadt 868: }
1.47 schwarze 869: }
1.7 schwarze 870:
1.47 schwarze 871: /*
872: * Now check the file suffix.
873: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
874: */
875: p = strrchr(start, '\0');
876: while (p-- > start && '/' != *p && '.' != *p)
877: /* Loop. */ ;
878:
879: if ('.' == *p) {
880: *p++ = '\0';
1.51 schwarze 881: mlink->fsec = p;
1.47 schwarze 882: }
1.41 deraadt 883:
1.47 schwarze 884: /*
885: * Now try to parse the name.
886: * Use the filename portion of the path.
887: */
888: mlink->name = start;
889: if (NULL != (p = strrchr(start, '/'))) {
890: mlink->name = p + 1;
891: *p = '\0';
892: }
893: mlink_add(mlink, &st);
894: }
1.2 schwarze 895:
1.47 schwarze 896: static void
897: mlink_add(struct mlink *mlink, const struct stat *st)
898: {
899: struct inodev inodev;
900: struct mpage *mpage;
901: unsigned int slot;
902:
903: assert(NULL != mlink->file);
904:
1.51 schwarze 905: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
906: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
907: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
908: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 909:
910: if ('0' == *mlink->fsec) {
911: free(mlink->fsec);
912: mlink->fsec = mandoc_strdup(mlink->dsec);
913: mlink->fform = FORM_CAT;
914: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
915: mlink->fform = FORM_SRC;
916: else
917: mlink->fform = FORM_NONE;
1.1 schwarze 918:
1.47 schwarze 919: slot = ohash_qlookup(&mlinks, mlink->file);
920: assert(NULL == ohash_find(&mlinks, slot));
921: ohash_insert(&mlinks, slot, mlink);
922:
923: inodev.st_ino = st->st_ino;
924: inodev.st_dev = st->st_dev;
925: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
926: sizeof(struct inodev), inodev.st_ino);
927: mpage = ohash_find(&mpages, slot);
928: if (NULL == mpage) {
929: mpage = mandoc_calloc(1, sizeof(struct mpage));
930: mpage->inodev.st_ino = inodev.st_ino;
931: mpage->inodev.st_dev = inodev.st_dev;
932: ohash_insert(&mpages, slot, mpage);
933: } else
934: mlink->next = mpage->mlinks;
935: mpage->mlinks = mlink;
1.75 schwarze 936: mlink->mpage = mpage;
1.47 schwarze 937: }
1.1 schwarze 938:
1.47 schwarze 939: static void
940: mlink_free(struct mlink *mlink)
941: {
1.1 schwarze 942:
1.47 schwarze 943: free(mlink->dsec);
944: free(mlink->arch);
945: free(mlink->name);
946: free(mlink->fsec);
947: free(mlink);
948: }
1.1 schwarze 949:
1.47 schwarze 950: static void
951: mpages_free(void)
952: {
953: struct mpage *mpage;
954: struct mlink *mlink;
955: unsigned int slot;
956:
957: mpage = ohash_first(&mpages, &slot);
958: while (NULL != mpage) {
959: while (NULL != (mlink = mpage->mlinks)) {
960: mpage->mlinks = mlink->next;
961: mlink_free(mlink);
962: }
963: free(mpage->sec);
964: free(mpage->arch);
965: free(mpage->title);
966: free(mpage->desc);
967: free(mpage);
968: mpage = ohash_next(&mpages, &slot);
969: }
970: }
1.1 schwarze 971:
1.47 schwarze 972: /*
973: * For each mlink to the mpage, check whether the path looks like
974: * it is formatted, and if it does, check whether a source manual
975: * exists by the same name, ignoring the suffix.
976: * If both conditions hold, drop the mlink.
977: */
978: static void
979: mlinks_undupe(struct mpage *mpage)
980: {
981: char buf[PATH_MAX];
982: struct mlink **prev;
983: struct mlink *mlink;
984: char *bufp;
985:
986: mpage->form = FORM_CAT;
987: prev = &mpage->mlinks;
988: while (NULL != (mlink = *prev)) {
989: if (FORM_CAT != mlink->dform) {
990: mpage->form = FORM_NONE;
991: goto nextlink;
992: }
1.100 schwarze 993: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 994: bufp = strstr(buf, "cat");
995: assert(NULL != bufp);
996: memcpy(bufp, "man", 3);
997: if (NULL != (bufp = strrchr(buf, '.')))
998: *++bufp = '\0';
1.100 schwarze 999: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 1000: if (NULL == ohash_find(&mlinks,
1.98 schwarze 1001: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 1002: goto nextlink;
1003: if (warnings)
1004: say(mlink->file, "Man source exists: %s", buf);
1005: if (use_all)
1006: goto nextlink;
1007: *prev = mlink->next;
1008: mlink_free(mlink);
1009: continue;
1010: nextlink:
1011: prev = &(*prev)->next;
1.1 schwarze 1012: }
1.47 schwarze 1013: }
1.1 schwarze 1014:
1.88 schwarze 1015: static void
1.50 schwarze 1016: mlink_check(struct mpage *mpage, struct mlink *mlink)
1017: {
1.88 schwarze 1018: struct str *str;
1019: unsigned int slot;
1.50 schwarze 1020:
1021: /*
1022: * Check whether the manual section given in a file
1023: * agrees with the directory where the file is located.
1024: * Some manuals have suffixes like (3p) on their
1025: * section number either inside the file or in the
1026: * directory name, some are linked into more than one
1027: * section, like encrypt(1) = makekey(8).
1028: */
1029:
1030: if (FORM_SRC == mpage->form &&
1.88 schwarze 1031: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1032: say(mlink->file, "Section \"%s\" manual in %s directory",
1033: mpage->sec, mlink->dsec);
1034:
1035: /*
1036: * Manual page directories exist for each kernel
1037: * architecture as returned by machine(1).
1038: * However, many manuals only depend on the
1039: * application architecture as returned by arch(1).
1040: * For example, some (2/ARM) manuals are shared
1041: * across the "armish" and "zaurus" kernel
1042: * architectures.
1043: * A few manuals are even shared across completely
1044: * different architectures, for example fdformat(1)
1045: * on amd64, i386, sparc, and sparc64.
1046: */
1047:
1.88 schwarze 1048: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1049: say(mlink->file, "Architecture \"%s\" manual in "
1050: "\"%s\" directory", mpage->arch, mlink->arch);
1051:
1.88 schwarze 1052: /*
1053: * XXX
1.90 schwarze 1054: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1055: */
1056:
1057: if (FORM_CAT == mpage->form)
1058: return;
1059:
1060: /*
1061: * Check whether this mlink
1062: * appears as a name in the NAME section.
1063: */
1.50 schwarze 1064:
1.90 schwarze 1065: slot = ohash_qlookup(&names, mlink->name);
1066: str = ohash_find(&names, slot);
1.88 schwarze 1067: assert(NULL != str);
1.90 schwarze 1068: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1069: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1070: }
1071:
1.47 schwarze 1072: /*
1073: * Run through the files in the global vector "mpages"
1074: * and add them to the database specified in "basedir".
1075: *
1076: * This handles the parsing scheme itself, using the cues of directory
1077: * and filename to determine whether the file is parsable or not.
1078: */
1079: static void
1.131 schwarze 1080: mpages_merge(struct mparse *mp)
1.47 schwarze 1081: {
1.70 schwarze 1082: char any[] = "any";
1.58 schwarze 1083: struct ohash_info str_info;
1.75 schwarze 1084: struct mpage *mpage, *mpage_dest;
1085: struct mlink *mlink, *mlink_dest;
1.47 schwarze 1086: struct mdoc *mdoc;
1087: struct man *man;
1.75 schwarze 1088: char *sodest;
1.69 schwarze 1089: char *cp;
1.117 schwarze 1090: int fd;
1.58 schwarze 1091: unsigned int pslot;
1.47 schwarze 1092: enum mandoclevel lvl;
1093:
1094: str_info.alloc = hash_alloc;
1.106 espie 1095: str_info.calloc = hash_calloc;
1096: str_info.free = hash_free;
1.47 schwarze 1097: str_info.key_offset = offsetof(struct str, key);
1098:
1.126 schwarze 1099: if ( ! nodb)
1.64 schwarze 1100: SQL_EXEC("BEGIN TRANSACTION");
1101:
1.47 schwarze 1102: mpage = ohash_first(&mpages, &pslot);
1.126 schwarze 1103: while (mpage != NULL) {
1.47 schwarze 1104: mlinks_undupe(mpage);
1.126 schwarze 1105: if (mpage->mlinks == NULL) {
1.47 schwarze 1106: mpage = ohash_next(&mpages, &pslot);
1107: continue;
1108: }
1.1 schwarze 1109:
1.90 schwarze 1110: name_mask = NAME_MASK;
1111: ohash_init(&names, 4, &str_info);
1.47 schwarze 1112: ohash_init(&strings, 6, &str_info);
1113: mparse_reset(mp);
1114: mdoc = NULL;
1115: man = NULL;
1.81 schwarze 1116: sodest = NULL;
1117:
1.125 schwarze 1118: mparse_open(mp, &fd, mpage->mlinks->file);
1.117 schwarze 1119: if (fd == -1) {
1120: say(mpage->mlinks->file, "&open");
1121: goto nextpage;
1.81 schwarze 1122: }
1.11 schwarze 1123:
1124: /*
1.24 schwarze 1125: * Try interpreting the file as mdoc(7) or man(7)
1126: * source code, unless it is already known to be
1127: * formatted. Fall back to formatted mode.
1.11 schwarze 1128: */
1.126 schwarze 1129: if (mpage->mlinks->dform != FORM_CAT ||
1130: mpage->mlinks->fform != FORM_CAT) {
1.117 schwarze 1131: lvl = mparse_readfd(mp, fd, mpage->mlinks->file);
1.47 schwarze 1132: if (lvl < MANDOCLEVEL_FATAL)
1.75 schwarze 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: mlink = mpage->mlinks;
1151: while (1) {
1152: mlink->mpage = mpage_dest;
1153:
1154: /*
1155: * If the target was already
1156: * processed, add the links
1157: * to the database now.
1158: * Otherwise, this will
1159: * happen when we come
1160: * to the target.
1161: */
1162:
1.94 schwarze 1163: if (mpage_dest->pageid)
1.124 schwarze 1164: dbadd_mlink_name(mlink);
1.75 schwarze 1165:
1.126 schwarze 1166: if (mlink->next == NULL)
1.75 schwarze 1167: break;
1168: mlink = mlink->next;
1169: }
1170:
1171: /* Move all links to the target. */
1172:
1173: mlink->next = mlink_dest->next;
1174: mlink_dest->next = mpage->mlinks;
1175: mpage->mlinks = NULL;
1176: }
1.81 schwarze 1177: goto nextpage;
1.126 schwarze 1178: } else if (mdoc != NULL) {
1.47 schwarze 1179: mpage->form = FORM_SRC;
1.112 schwarze 1180: mpage->sec = mdoc_meta(mdoc)->msec;
1181: mpage->sec = mandoc_strdup(
1.126 schwarze 1182: mpage->sec == NULL ? "" : mpage->sec);
1.47 schwarze 1183: mpage->arch = mdoc_meta(mdoc)->arch;
1184: mpage->arch = mandoc_strdup(
1.126 schwarze 1185: mpage->arch == NULL ? "" : mpage->arch);
1.47 schwarze 1186: mpage->title =
1187: mandoc_strdup(mdoc_meta(mdoc)->title);
1.126 schwarze 1188: } else if (man != NULL) {
1.47 schwarze 1189: mpage->form = FORM_SRC;
1190: mpage->sec =
1191: mandoc_strdup(man_meta(man)->msec);
1192: mpage->arch =
1193: mandoc_strdup(mpage->mlinks->arch);
1194: mpage->title =
1195: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1196: } else {
1.47 schwarze 1197: mpage->form = FORM_CAT;
1198: mpage->sec =
1199: mandoc_strdup(mpage->mlinks->dsec);
1200: mpage->arch =
1201: mandoc_strdup(mpage->mlinks->arch);
1202: mpage->title =
1203: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1204: }
1.54 schwarze 1205: putkey(mpage, mpage->sec, TYPE_sec);
1.119 schwarze 1206: if (*mpage->arch != '\0')
1207: putkey(mpage, mpage->arch, TYPE_arch);
1.1 schwarze 1208:
1.54 schwarze 1209: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1210: if ('\0' != *mlink->dsec)
1211: putkey(mpage, mlink->dsec, TYPE_sec);
1212: if ('\0' != *mlink->fsec)
1213: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1214: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1215: any : mlink->arch, TYPE_arch);
1.90 schwarze 1216: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1217: }
1.41 deraadt 1218:
1.129 schwarze 1219: assert(mpage->desc == NULL);
1220: if (mdoc != NULL)
1.127 schwarze 1221: parse_mdoc(mpage, mdoc_meta(mdoc), mdoc_node(mdoc));
1.129 schwarze 1222: else if (man != NULL)
1223: parse_man(mpage, man_meta(man), man_node(man));
1.47 schwarze 1224: else
1.117 schwarze 1225: parse_cat(mpage, fd);
1.129 schwarze 1226: if (mpage->desc == NULL)
1.92 schwarze 1227: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1228:
1229: if (warnings && !use_all)
1230: for (mlink = mpage->mlinks; mlink;
1231: mlink = mlink->next)
1232: mlink_check(mpage, mlink);
1.6 schwarze 1233:
1.131 schwarze 1234: dbadd(mpage);
1.81 schwarze 1235:
1236: nextpage:
1.125 schwarze 1237: if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1.117 schwarze 1238: exitcode = (int)MANDOCLEVEL_SYSERR;
1239: say(mpage->mlinks->file, "&wait gunzip");
1.81 schwarze 1240: }
1.47 schwarze 1241: ohash_delete(&strings);
1.90 schwarze 1242: ohash_delete(&names);
1.47 schwarze 1243: mpage = ohash_next(&mpages, &pslot);
1244: }
1.64 schwarze 1245:
1246: if (0 == nodb)
1247: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1248: }
1249:
1250: static void
1251: names_check(void)
1252: {
1253: sqlite3_stmt *stmt;
1254: const char *name, *sec, *arch, *key;
1255: int irc;
1256:
1257: sqlite3_prepare_v2(db,
1258: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1259: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1260: "WHERE bits & ? AND NOT EXISTS ("
1261: "SELECT pageid FROM mlinks "
1.90 schwarze 1262: "WHERE mlinks.pageid == names.pageid "
1263: "AND mlinks.name == names.name"
1.87 schwarze 1264: ")"
1265: ") JOIN ("
1.99 schwarze 1266: "SELECT sec, arch, name, pageid FROM mlinks "
1267: "GROUP BY pageid"
1.87 schwarze 1268: ") USING (pageid);",
1269: -1, &stmt, NULL);
1270:
1.91 schwarze 1271: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1272: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1273:
1274: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.113 schwarze 1275: name = (const char *)sqlite3_column_text(stmt, 0);
1276: sec = (const char *)sqlite3_column_text(stmt, 1);
1277: arch = (const char *)sqlite3_column_text(stmt, 2);
1278: key = (const char *)sqlite3_column_text(stmt, 3);
1.87 schwarze 1279: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1280: '\0' == *arch ? "" : "/",
1281: '\0' == *arch ? "" : arch, key);
1282: }
1283: sqlite3_finalize(stmt);
1.47 schwarze 1284: }
1.6 schwarze 1285:
1.47 schwarze 1286: static void
1.81 schwarze 1287: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1288: {
1289: FILE *stream;
1290: char *line, *p, *title;
1291: size_t len, plen, titlesz;
1.1 schwarze 1292:
1.81 schwarze 1293: stream = (-1 == fd) ?
1294: fopen(mpage->mlinks->file, "r") :
1295: fdopen(fd, "r");
1296: if (NULL == stream) {
1.114 doug 1297: if (-1 != fd)
1298: close(fd);
1.47 schwarze 1299: if (warnings)
1.80 schwarze 1300: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1301: return;
1302: }
1.1 schwarze 1303:
1.47 schwarze 1304: /* Skip to first blank line. */
1.1 schwarze 1305:
1.47 schwarze 1306: while (NULL != (line = fgetln(stream, &len)))
1307: if ('\n' == *line)
1308: break;
1.1 schwarze 1309:
1.47 schwarze 1310: /*
1311: * Assume the first line that is not indented
1312: * is the first section header. Skip to it.
1313: */
1.1 schwarze 1314:
1.47 schwarze 1315: while (NULL != (line = fgetln(stream, &len)))
1316: if ('\n' != *line && ' ' != *line)
1317: break;
1.98 schwarze 1318:
1.47 schwarze 1319: /*
1320: * Read up until the next section into a buffer.
1321: * Strip the leading and trailing newline from each read line,
1322: * appending a trailing space.
1323: * Ignore empty (whitespace-only) lines.
1324: */
1.28 schwarze 1325:
1.47 schwarze 1326: titlesz = 0;
1327: title = NULL;
1.38 schwarze 1328:
1.47 schwarze 1329: while (NULL != (line = fgetln(stream, &len))) {
1330: if (' ' != *line || '\n' != line[len - 1])
1331: break;
1332: while (len > 0 && isspace((unsigned char)*line)) {
1333: line++;
1334: len--;
1335: }
1336: if (1 == len)
1337: continue;
1338: title = mandoc_realloc(title, titlesz + len);
1339: memcpy(title + titlesz, line, len);
1340: titlesz += len;
1341: title[titlesz - 1] = ' ';
1342: }
1.28 schwarze 1343:
1.47 schwarze 1344: /*
1345: * If no page content can be found, or the input line
1346: * is already the next section header, or there is no
1347: * trailing newline, reuse the page title as the page
1348: * description.
1349: */
1.1 schwarze 1350:
1.47 schwarze 1351: if (NULL == title || '\0' == *title) {
1352: if (warnings)
1353: say(mpage->mlinks->file,
1354: "Cannot find NAME section");
1355: fclose(stream);
1356: free(title);
1357: return;
1358: }
1.24 schwarze 1359:
1.47 schwarze 1360: title = mandoc_realloc(title, titlesz + 1);
1361: title[titlesz] = '\0';
1.24 schwarze 1362:
1.47 schwarze 1363: /*
1364: * Skip to the first dash.
1365: * Use the remaining line as the description (no more than 70
1366: * bytes).
1367: */
1.28 schwarze 1368:
1.47 schwarze 1369: if (NULL != (p = strstr(title, "- "))) {
1370: for (p += 2; ' ' == *p || '\b' == *p; p++)
1371: /* Skip to next word. */ ;
1372: } else {
1373: if (warnings)
1374: say(mpage->mlinks->file,
1375: "No dash in title line");
1376: p = title;
1377: }
1.1 schwarze 1378:
1.47 schwarze 1379: plen = strlen(p);
1.1 schwarze 1380:
1.47 schwarze 1381: /* Strip backspace-encoding from line. */
1.1 schwarze 1382:
1.47 schwarze 1383: while (NULL != (line = memchr(p, '\b', plen))) {
1384: len = line - p;
1385: if (0 == len) {
1386: memmove(line, line + 1, plen--);
1387: continue;
1.98 schwarze 1388: }
1.47 schwarze 1389: memmove(line - 1, line + 1, plen - len);
1390: plen -= 2;
1391: }
1.1 schwarze 1392:
1.47 schwarze 1393: mpage->desc = mandoc_strdup(p);
1394: fclose(stream);
1395: free(title);
1396: }
1.16 schwarze 1397:
1.47 schwarze 1398: /*
1399: * Put a type/word pair into the word database for this particular file.
1400: */
1401: static void
1.69 schwarze 1402: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1403: {
1.69 schwarze 1404: char *cp;
1.37 schwarze 1405:
1.47 schwarze 1406: assert(NULL != value);
1.69 schwarze 1407: if (TYPE_arch == type)
1408: for (cp = value; *cp; cp++)
1409: if (isupper((unsigned char)*cp))
1410: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1411: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1412: }
1413:
1414: /*
1.47 schwarze 1415: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1416: */
1417: static void
1.47 schwarze 1418: putmdockey(const struct mpage *mpage,
1419: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1420: {
1.16 schwarze 1421:
1.47 schwarze 1422: for ( ; NULL != n; n = n->next) {
1423: if (NULL != n->child)
1424: putmdockey(mpage, n->child, m);
1425: if (MDOC_TEXT == n->type)
1426: putkey(mpage, n->string, m);
1427: }
1428: }
1.16 schwarze 1429:
1.47 schwarze 1430: static void
1.129 schwarze 1431: parse_man(struct mpage *mpage, const struct man_meta *meta,
1432: const struct man_node *n)
1.47 schwarze 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);
1.129 schwarze 1498: if ( ! (mpage->name_head_done ||
1499: strcasecmp(start, meta->title))) {
1500: putkey(mpage, start, NAME_HEAD);
1501: mpage->name_head_done = 1;
1502: }
1.1 schwarze 1503:
1.47 schwarze 1504: if (' ' == byte) {
1505: start += sz + 1;
1506: break;
1507: }
1.1 schwarze 1508:
1.47 schwarze 1509: assert(',' == byte);
1510: start += sz + 1;
1511: while (' ' == *start)
1512: start++;
1513: }
1.1 schwarze 1514:
1.78 schwarze 1515: if (start == title) {
1.90 schwarze 1516: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1517: if ( ! (mpage->name_head_done ||
1518: strcasecmp(start, meta->title))) {
1519: putkey(mpage, start, NAME_HEAD);
1520: mpage->name_head_done = 1;
1521: }
1.47 schwarze 1522: free(title);
1523: return;
1524: }
1.1 schwarze 1525:
1.47 schwarze 1526: while (isspace((unsigned char)*start))
1527: start++;
1.1 schwarze 1528:
1.47 schwarze 1529: if (0 == strncmp(start, "-", 1))
1530: start += 1;
1531: else if (0 == strncmp(start, "\\-\\-", 4))
1532: start += 4;
1533: else if (0 == strncmp(start, "\\-", 2))
1534: start += 2;
1535: else if (0 == strncmp(start, "\\(en", 4))
1536: start += 4;
1537: else if (0 == strncmp(start, "\\(em", 4))
1538: start += 4;
1.1 schwarze 1539:
1.47 schwarze 1540: while (' ' == *start)
1541: start++;
1.1 schwarze 1542:
1.47 schwarze 1543: mpage->desc = mandoc_strdup(start);
1544: free(title);
1545: return;
1546: }
1547: }
1.1 schwarze 1548:
1.47 schwarze 1549: for (n = n->child; n; n = n->next) {
1550: if (NULL != mpage->desc)
1551: break;
1.129 schwarze 1552: parse_man(mpage, meta, n);
1.1 schwarze 1553: }
1554: }
1555:
1556: static void
1.127 schwarze 1557: parse_mdoc(struct mpage *mpage, const struct mdoc_meta *meta,
1558: const struct mdoc_node *n)
1.1 schwarze 1559: {
1560:
1.47 schwarze 1561: assert(NULL != n);
1562: for (n = n->child; NULL != n; n = n->next) {
1563: switch (n->type) {
1.98 schwarze 1564: case MDOC_ELEM:
1.47 schwarze 1565: /* FALLTHROUGH */
1.98 schwarze 1566: case MDOC_BLOCK:
1.47 schwarze 1567: /* FALLTHROUGH */
1.98 schwarze 1568: case MDOC_HEAD:
1.47 schwarze 1569: /* FALLTHROUGH */
1.98 schwarze 1570: case MDOC_BODY:
1.47 schwarze 1571: /* FALLTHROUGH */
1.98 schwarze 1572: case MDOC_TAIL:
1.47 schwarze 1573: if (NULL != mdocs[n->tok].fp)
1.127 schwarze 1574: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.47 schwarze 1575: break;
1576: if (mdocs[n->tok].mask)
1577: putmdockey(mpage, n->child,
1578: mdocs[n->tok].mask);
1579: break;
1580: default:
1581: assert(MDOC_ROOT != n->type);
1582: continue;
1583: }
1584: if (NULL != n->child)
1.127 schwarze 1585: parse_mdoc(mpage, meta, n);
1.1 schwarze 1586: }
1587: }
1588:
1.19 schwarze 1589: static int
1.127 schwarze 1590: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_meta *meta,
1591: const struct mdoc_node *n)
1.1 schwarze 1592: {
1.131 schwarze 1593: char *start, *end;
1.1 schwarze 1594: size_t sz;
1.19 schwarze 1595:
1.47 schwarze 1596: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1597: NULL == (n = n->child) ||
1598: MDOC_TEXT != n->type)
1.19 schwarze 1599: return(0);
1.1 schwarze 1600:
1601: /*
1602: * Only consider those `Fd' macro fields that begin with an
1603: * "inclusion" token (versus, e.g., #define).
1604: */
1.47 schwarze 1605:
1.1 schwarze 1606: if (strcmp("#include", n->string))
1.19 schwarze 1607: return(0);
1.1 schwarze 1608:
1609: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1610: return(0);
1.1 schwarze 1611:
1612: /*
1613: * Strip away the enclosing angle brackets and make sure we're
1614: * not zero-length.
1615: */
1616:
1617: start = n->string;
1618: if ('<' == *start || '"' == *start)
1619: start++;
1620:
1621: if (0 == (sz = strlen(start)))
1.19 schwarze 1622: return(0);
1.1 schwarze 1623:
1624: end = &start[(int)sz - 1];
1625: if ('>' == *end || '"' == *end)
1626: end--;
1627:
1.47 schwarze 1628: if (end > start)
1629: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1630: return(0);
1.1 schwarze 1631: }
1632:
1.19 schwarze 1633: static int
1.127 schwarze 1634: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_meta *meta,
1635: const struct mdoc_node *n)
1.1 schwarze 1636: {
1.69 schwarze 1637: char *cp;
1.1 schwarze 1638:
1.47 schwarze 1639: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.19 schwarze 1640: return(0);
1641:
1.98 schwarze 1642: /*
1.47 schwarze 1643: * Parse: .Fn "struct type *name" "char *arg".
1.98 schwarze 1644: * First strip away pointer symbol.
1.47 schwarze 1645: * Then store the function name, then type.
1.98 schwarze 1646: * Finally, store the arguments.
1.47 schwarze 1647: */
1.1 schwarze 1648:
1.47 schwarze 1649: if (NULL == (cp = strrchr(n->string, ' ')))
1650: cp = n->string;
1.1 schwarze 1651:
1652: while ('*' == *cp)
1653: cp++;
1654:
1.47 schwarze 1655: putkey(mpage, cp, TYPE_Fn);
1.128 schwarze 1656: if (n->sec == SEC_SYNOPSIS)
1657: putkey(mpage, cp, NAME_SYN);
1.19 schwarze 1658:
1.47 schwarze 1659: if (n->string < cp)
1660: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.19 schwarze 1661:
1.47 schwarze 1662: for (n = n->next; NULL != n; n = n->next)
1663: if (MDOC_TEXT == n->type)
1664: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1665:
1.128 schwarze 1666: return(0);
1667: }
1668:
1669: static int
1670: parse_mdoc_Fo(struct mpage *mpage, const struct mdoc_meta *meta,
1671: const struct mdoc_node *n)
1672: {
1.132 ! schwarze 1673:
! 1674: if (n->type != MDOC_HEAD)
! 1675: return(1);
1.128 schwarze 1676:
1677: putmdockey(mpage, n->child, TYPE_Fn);
1678: if (n->sec == SEC_SYNOPSIS)
1679: putmdockey(mpage, n->child, NAME_SYN);
1.19 schwarze 1680: return(0);
1.1 schwarze 1681: }
1682:
1.19 schwarze 1683: static int
1.127 schwarze 1684: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_meta *meta,
1685: const struct mdoc_node *n)
1.1 schwarze 1686: {
1.47 schwarze 1687: char *cp;
1.1 schwarze 1688:
1689: if (NULL == (n = n->child))
1.19 schwarze 1690: return(0);
1.1 schwarze 1691:
1.47 schwarze 1692: if (NULL == n->next) {
1693: putkey(mpage, n->string, TYPE_Xr);
1694: return(0);
1695: }
1.1 schwarze 1696:
1.77 schwarze 1697: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1698: putkey(mpage, cp, TYPE_Xr);
1699: free(cp);
1700: return(0);
1.1 schwarze 1701: }
1702:
1.19 schwarze 1703: static int
1.127 schwarze 1704: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_meta *meta,
1705: const struct mdoc_node *n)
1.1 schwarze 1706: {
1707:
1.79 schwarze 1708: if (MDOC_BODY == n->type)
1709: mdoc_deroff(&mpage->desc, n);
1710: return(0);
1.1 schwarze 1711: }
1712:
1.19 schwarze 1713: static int
1.127 schwarze 1714: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_meta *meta,
1715: const struct mdoc_node *n)
1.1 schwarze 1716: {
1717:
1.86 schwarze 1718: if (SEC_NAME == n->sec)
1.90 schwarze 1719: putmdockey(mpage, n->child, NAME_TITLE);
1.127 schwarze 1720: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) {
1721: if (n->child == NULL)
1722: putkey(mpage, meta->name, NAME_SYN);
1723: else
1724: putmdockey(mpage, n->child, NAME_SYN);
1.129 schwarze 1725: }
1726: if ( ! (mpage->name_head_done ||
1727: n->child == NULL || n->child->string == NULL ||
1728: strcasecmp(n->child->string, meta->title))) {
1729: putkey(mpage, n->child->string, NAME_HEAD);
1730: mpage->name_head_done = 1;
1.127 schwarze 1731: }
1.86 schwarze 1732: return(0);
1.1 schwarze 1733: }
1734:
1.19 schwarze 1735: static int
1.127 schwarze 1736: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_meta *meta,
1737: const struct mdoc_node *n)
1.1 schwarze 1738: {
1739:
1.19 schwarze 1740: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1741: }
1742:
1.47 schwarze 1743: static int
1.127 schwarze 1744: parse_mdoc_head(struct mpage *mpage, const struct mdoc_meta *meta,
1745: const struct mdoc_node *n)
1.1 schwarze 1746: {
1747:
1.47 schwarze 1748: return(MDOC_HEAD == n->type);
1749: }
1.1 schwarze 1750:
1.47 schwarze 1751: static int
1.127 schwarze 1752: parse_mdoc_body(struct mpage *mpage, const struct mdoc_meta *meta,
1753: const struct mdoc_node *n)
1.47 schwarze 1754: {
1.1 schwarze 1755:
1.47 schwarze 1756: return(MDOC_BODY == n->type);
1.1 schwarze 1757: }
1758:
1.47 schwarze 1759: /*
1760: * Add a string to the hash table for the current manual.
1761: * Each string has a bitmask telling which macros it belongs to.
1762: * When we finish the manual, we'll dump the table.
1763: */
1.1 schwarze 1764: static void
1.131 schwarze 1765: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.1 schwarze 1766: {
1.90 schwarze 1767: struct ohash *htab;
1.47 schwarze 1768: struct str *s;
1.68 schwarze 1769: const char *end;
1.47 schwarze 1770: unsigned int slot;
1.131 schwarze 1771: int i, mustfree;
1.1 schwarze 1772:
1.47 schwarze 1773: if (0 == sz)
1774: return;
1.68 schwarze 1775:
1.131 schwarze 1776: mustfree = render_string(&cp, &sz);
1777:
1.90 schwarze 1778: if (TYPE_Nm & v) {
1779: htab = &names;
1780: v &= name_mask;
1.124 schwarze 1781: if (v & NAME_FIRST)
1782: name_mask &= ~NAME_FIRST;
1.90 schwarze 1783: if (debug > 1)
1784: say(mpage->mlinks->file,
1.130 schwarze 1785: "Adding name %*s, bits=%d", sz, cp, v);
1.90 schwarze 1786: } else {
1787: htab = &strings;
1788: if (debug > 1)
1789: for (i = 0; i < mansearch_keymax; i++)
1.118 schwarze 1790: if ((uint64_t)1 << i & v)
1.90 schwarze 1791: say(mpage->mlinks->file,
1792: "Adding key %s=%*s",
1793: mansearch_keynames[i], sz, cp);
1.68 schwarze 1794: }
1.47 schwarze 1795:
1796: end = cp + sz;
1.90 schwarze 1797: slot = ohash_qlookupi(htab, cp, &end);
1798: s = ohash_find(htab, slot);
1.1 schwarze 1799:
1.47 schwarze 1800: if (NULL != s && mpage == s->mpage) {
1801: s->mask |= v;
1.1 schwarze 1802: return;
1.47 schwarze 1803: } else if (NULL == s) {
1.101 schwarze 1804: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1805: memcpy(s->key, cp, sz);
1.90 schwarze 1806: ohash_insert(htab, slot, s);
1.47 schwarze 1807: }
1808: s->mpage = mpage;
1809: s->mask = v;
1.131 schwarze 1810:
1811: if (mustfree)
1812: free(cp);
1.1 schwarze 1813: }
1814:
1815: /*
1.47 schwarze 1816: * Take a Unicode codepoint and produce its UTF-8 encoding.
1817: * This isn't the best way to do this, but it works.
1818: * The magic numbers are from the UTF-8 packaging.
1819: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1820: */
1.47 schwarze 1821: static size_t
1822: utf8(unsigned int cp, char out[7])
1.1 schwarze 1823: {
1.47 schwarze 1824: size_t rc;
1.1 schwarze 1825:
1.47 schwarze 1826: rc = 0;
1827: if (cp <= 0x0000007F) {
1828: rc = 1;
1829: out[0] = (char)cp;
1830: } else if (cp <= 0x000007FF) {
1831: rc = 2;
1832: out[0] = (cp >> 6 & 31) | 192;
1833: out[1] = (cp & 63) | 128;
1834: } else if (cp <= 0x0000FFFF) {
1835: rc = 3;
1836: out[0] = (cp >> 12 & 15) | 224;
1837: out[1] = (cp >> 6 & 63) | 128;
1838: out[2] = (cp & 63) | 128;
1839: } else if (cp <= 0x001FFFFF) {
1840: rc = 4;
1841: out[0] = (cp >> 18 & 7) | 240;
1842: out[1] = (cp >> 12 & 63) | 128;
1843: out[2] = (cp >> 6 & 63) | 128;
1844: out[3] = (cp & 63) | 128;
1845: } else if (cp <= 0x03FFFFFF) {
1846: rc = 5;
1847: out[0] = (cp >> 24 & 3) | 248;
1848: out[1] = (cp >> 18 & 63) | 128;
1849: out[2] = (cp >> 12 & 63) | 128;
1850: out[3] = (cp >> 6 & 63) | 128;
1851: out[4] = (cp & 63) | 128;
1852: } else if (cp <= 0x7FFFFFFF) {
1853: rc = 6;
1854: out[0] = (cp >> 30 & 1) | 252;
1855: out[1] = (cp >> 24 & 63) | 128;
1856: out[2] = (cp >> 18 & 63) | 128;
1857: out[3] = (cp >> 12 & 63) | 128;
1858: out[4] = (cp >> 6 & 63) | 128;
1859: out[5] = (cp & 63) | 128;
1860: } else
1861: return(0);
1.19 schwarze 1862:
1.47 schwarze 1863: out[rc] = '\0';
1864: return(rc);
1.1 schwarze 1865: }
1866:
1.47 schwarze 1867: /*
1.131 schwarze 1868: * If the string contains escape sequences,
1869: * replace it with an allocated rendering and return 1,
1870: * such that the caller can free it after use.
1871: * Otherwise, do nothing and return 0.
1.47 schwarze 1872: */
1.131 schwarze 1873: static int
1874: render_string(char **public, size_t *psz)
1.1 schwarze 1875: {
1.131 schwarze 1876: const char *src, *scp, *addcp, *seq;
1877: char *dst;
1878: size_t ssz, dsz, addsz;
1.71 schwarze 1879: char utfbuf[7], res[6];
1.131 schwarze 1880: int seqlen, unicode;
1.47 schwarze 1881:
1882: res[0] = '\\';
1883: res[1] = '\t';
1884: res[2] = ASCII_NBRSP;
1885: res[3] = ASCII_HYPH;
1.71 schwarze 1886: res[4] = ASCII_BREAK;
1887: res[5] = '\0';
1.1 schwarze 1888:
1.131 schwarze 1889: src = scp = *public;
1890: ssz = *psz;
1891: dst = NULL;
1892: dsz = 0;
1893:
1894: while (scp < src + *psz) {
1895:
1896: /* Leave normal characters unchanged. */
1897:
1898: if (strchr(res, *scp) == NULL) {
1899: if (dst != NULL)
1900: dst[dsz++] = *scp;
1901: scp++;
1902: continue;
1903: }
1.1 schwarze 1904:
1.47 schwarze 1905: /*
1.131 schwarze 1906: * Found something that requires replacing,
1907: * make sure we have a destination buffer.
1.47 schwarze 1908: */
1.131 schwarze 1909:
1910: if (dst == NULL) {
1911: dst = mandoc_malloc(ssz + 1);
1912: dsz = scp - src;
1913: memcpy(dst, src, dsz);
1.47 schwarze 1914: }
1.39 schwarze 1915:
1.131 schwarze 1916: /* Handle single-char special characters. */
1917:
1918: switch (*scp) {
1919: case '\\':
1920: break;
1.98 schwarze 1921: case '\t':
1.71 schwarze 1922: /* FALLTHROUGH */
1.98 schwarze 1923: case ASCII_NBRSP:
1.131 schwarze 1924: dst[dsz++] = ' ';
1925: scp++;
1926: continue;
1927: case ASCII_HYPH:
1928: dst[dsz++] = '-';
1.71 schwarze 1929: /* FALLTHROUGH */
1.98 schwarze 1930: case ASCII_BREAK:
1.131 schwarze 1931: scp++;
1.47 schwarze 1932: continue;
1.71 schwarze 1933: default:
1.131 schwarze 1934: abort();
1.71 schwarze 1935: }
1.39 schwarze 1936:
1.47 schwarze 1937: /*
1.131 schwarze 1938: * Found an escape sequence.
1939: * Read past the slash, then parse it.
1940: * Ignore everything except characters.
1.47 schwarze 1941: */
1.52 schwarze 1942:
1.131 schwarze 1943: scp++;
1944: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.47 schwarze 1945: continue;
1.39 schwarze 1946:
1.47 schwarze 1947: /*
1.52 schwarze 1948: * Render the special character
1949: * as either UTF-8 or ASCII.
1.47 schwarze 1950: */
1.52 schwarze 1951:
1952: if (write_utf8) {
1.131 schwarze 1953: unicode = mchars_spec2cp(mchars, seq, seqlen);
1954: if (unicode <= 0)
1.52 schwarze 1955: continue;
1.131 schwarze 1956: addsz = utf8(unicode, utfbuf);
1957: if (addsz == 0)
1.52 schwarze 1958: continue;
1.131 schwarze 1959: addcp = utfbuf;
1.52 schwarze 1960: } else {
1.131 schwarze 1961: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1962: if (addcp == NULL)
1.52 schwarze 1963: continue;
1.131 schwarze 1964: if (*addcp == ASCII_NBRSP) {
1965: addcp = " ";
1966: addsz = 1;
1.52 schwarze 1967: }
1968: }
1.1 schwarze 1969:
1.47 schwarze 1970: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1971:
1.131 schwarze 1972: ssz += addsz;
1973: dst = mandoc_realloc(dst, ssz + 1);
1974: memcpy(dst + dsz, addcp, addsz);
1975: dsz += addsz;
1976: }
1977: if (dst != NULL) {
1978: *public = dst;
1979: *psz = dsz;
1980: }
1981:
1982: /* Trim trailing whitespace and NUL-terminate. */
1983:
1984: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1985: --*psz;
1986: if (dst != NULL) {
1987: (*public)[*psz] = '\0';
1988: return(1);
1989: } else
1990: return(0);
1.1 schwarze 1991: }
1992:
1.75 schwarze 1993: static void
1994: dbadd_mlink(const struct mlink *mlink)
1995: {
1996: size_t i;
1997:
1998: i = 1;
1999: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2000: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2001: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 2002: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 2003: SQL_STEP(stmts[STMT_INSERT_LINK]);
2004: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.124 schwarze 2005: }
2006:
2007: static void
2008: dbadd_mlink_name(const struct mlink *mlink)
2009: {
1.130 schwarze 2010: uint64_t bits;
1.124 schwarze 2011: size_t i;
2012:
2013: dbadd_mlink(mlink);
1.115 schwarze 2014:
2015: i = 1;
1.130 schwarze 2016: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2017: bits = NAME_FILE & NAME_MASK;
2018: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2019: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2020: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2021: }
2022:
2023: i = 1;
2024: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.115 schwarze 2025: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2026: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2027: SQL_STEP(stmts[STMT_INSERT_NAME]);
2028: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.75 schwarze 2029: }
2030:
1.11 schwarze 2031: /*
1.47 schwarze 2032: * Flush the current page's terms (and their bits) into the database.
2033: * Wrap the entire set of additions in a transaction to make sqlite be a
2034: * little faster.
1.53 schwarze 2035: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 2036: */
2037: static void
1.131 schwarze 2038: dbadd(struct mpage *mpage)
1.11 schwarze 2039: {
1.47 schwarze 2040: struct mlink *mlink;
2041: struct str *key;
1.131 schwarze 2042: char *cp;
1.47 schwarze 2043: size_t i;
2044: unsigned int slot;
1.131 schwarze 2045: int mustfree;
1.47 schwarze 2046:
1.85 schwarze 2047: mlink = mpage->mlinks;
1.11 schwarze 2048:
1.85 schwarze 2049: if (nodb) {
1.104 schwarze 2050: for (key = ohash_first(&names, &slot); NULL != key;
1.131 schwarze 2051: key = ohash_next(&names, &slot))
1.104 schwarze 2052: free(key);
2053: for (key = ohash_first(&strings, &slot); NULL != key;
1.131 schwarze 2054: key = ohash_next(&strings, &slot))
1.104 schwarze 2055: free(key);
1.102 schwarze 2056: if (0 == debug)
2057: return;
1.85 schwarze 2058: while (NULL != mlink) {
2059: fputs(mlink->name, stdout);
2060: if (NULL == mlink->next ||
2061: strcmp(mlink->dsec, mlink->next->dsec) ||
2062: strcmp(mlink->fsec, mlink->next->fsec) ||
2063: strcmp(mlink->arch, mlink->next->arch)) {
2064: putchar('(');
2065: if ('\0' == *mlink->dsec)
2066: fputs(mlink->fsec, stdout);
2067: else
2068: fputs(mlink->dsec, stdout);
2069: if ('\0' != *mlink->arch)
2070: printf("/%s", mlink->arch);
2071: putchar(')');
2072: }
2073: mlink = mlink->next;
2074: if (NULL != mlink)
2075: fputs(", ", stdout);
2076: }
1.89 schwarze 2077: printf(" - %s\n", mpage->desc);
1.11 schwarze 2078: return;
1.85 schwarze 2079: }
2080:
2081: if (debug)
2082: say(mlink->file, "Adding to database");
1.47 schwarze 2083:
1.131 schwarze 2084: cp = mpage->desc;
2085: i = strlen(cp);
2086: mustfree = render_string(&cp, &i);
1.47 schwarze 2087: i = 1;
1.131 schwarze 2088: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.116 schwarze 2089: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.47 schwarze 2090: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2091: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2092: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.131 schwarze 2093: if (mustfree)
2094: free(cp);
1.47 schwarze 2095:
1.85 schwarze 2096: while (NULL != mlink) {
1.75 schwarze 2097: dbadd_mlink(mlink);
1.85 schwarze 2098: mlink = mlink->next;
2099: }
1.91 schwarze 2100: mlink = mpage->mlinks;
1.47 schwarze 2101:
1.90 schwarze 2102: for (key = ohash_first(&names, &slot); NULL != key;
2103: key = ohash_next(&names, &slot)) {
2104: assert(key->mpage == mpage);
2105: i = 1;
2106: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.131 schwarze 2107: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.94 schwarze 2108: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2109: SQL_STEP(stmts[STMT_INSERT_NAME]);
2110: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2111: free(key);
2112: }
1.47 schwarze 2113: for (key = ohash_first(&strings, &slot); NULL != key;
2114: key = ohash_next(&strings, &slot)) {
2115: assert(key->mpage == mpage);
2116: i = 1;
2117: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.131 schwarze 2118: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.94 schwarze 2119: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2120: SQL_STEP(stmts[STMT_INSERT_KEY]);
2121: sqlite3_reset(stmts[STMT_INSERT_KEY]);
2122: free(key);
1.33 schwarze 2123: }
1.47 schwarze 2124: }
1.41 deraadt 2125:
1.47 schwarze 2126: static void
2127: dbprune(void)
2128: {
2129: struct mpage *mpage;
2130: struct mlink *mlink;
2131: size_t i;
2132: unsigned int slot;
1.11 schwarze 2133:
1.63 schwarze 2134: if (0 == nodb)
2135: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2136:
1.63 schwarze 2137: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2138: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2139: mlink = mpage->mlinks;
1.82 schwarze 2140: if (debug)
1.63 schwarze 2141: say(mlink->file, "Deleting from database");
2142: if (nodb)
2143: continue;
2144: for ( ; NULL != mlink; mlink = mlink->next) {
2145: i = 1;
2146: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2147: i, mlink->dsec);
2148: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2149: i, mlink->arch);
2150: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2151: i, mlink->name);
2152: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2153: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2154: }
1.11 schwarze 2155: }
1.63 schwarze 2156:
2157: if (0 == nodb)
2158: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2159: }
1.22 schwarze 2160:
1.47 schwarze 2161: /*
2162: * Close an existing database and its prepared statements.
2163: * If "real" is not set, rename the temporary file into the real one.
2164: */
2165: static void
2166: dbclose(int real)
2167: {
2168: size_t i;
1.72 schwarze 2169: int status;
2170: pid_t child;
1.11 schwarze 2171:
1.47 schwarze 2172: if (nodb)
2173: return;
1.11 schwarze 2174:
1.47 schwarze 2175: for (i = 0; i < STMT__MAX; i++) {
2176: sqlite3_finalize(stmts[i]);
2177: stmts[i] = NULL;
1.28 schwarze 2178: }
1.22 schwarze 2179:
1.47 schwarze 2180: sqlite3_close(db);
2181: db = NULL;
1.11 schwarze 2182:
1.47 schwarze 2183: if (real)
2184: return;
1.22 schwarze 2185:
1.72 schwarze 2186: if ('\0' == *tempfilename) {
2187: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2188: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2189: say(MANDOC_DB, "&rename");
1.72 schwarze 2190: }
2191: return;
2192: }
2193:
2194: switch (child = fork()) {
1.98 schwarze 2195: case -1:
1.72 schwarze 2196: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2197: say("", "&fork cmp");
1.72 schwarze 2198: return;
1.98 schwarze 2199: case 0:
1.72 schwarze 2200: execlp("cmp", "cmp", "-s",
2201: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2202: say("", "&exec cmp");
1.72 schwarze 2203: exit(0);
2204: default:
2205: break;
2206: }
2207: if (-1 == waitpid(child, &status, 0)) {
2208: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2209: say("", "&wait cmp");
1.72 schwarze 2210: } else if (WIFSIGNALED(status)) {
2211: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2212: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2213: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2214: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2215: say(MANDOC_DB,
2216: "Data changed, but cannot replace database");
2217: }
2218:
2219: *strrchr(tempfilename, '/') = '\0';
2220: switch (child = fork()) {
1.98 schwarze 2221: case -1:
1.72 schwarze 2222: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2223: say("", "&fork rm");
1.72 schwarze 2224: return;
1.98 schwarze 2225: case 0:
1.72 schwarze 2226: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2227: say("", "&exec rm");
1.72 schwarze 2228: exit((int)MANDOCLEVEL_SYSERR);
2229: default:
2230: break;
2231: }
2232: if (-1 == waitpid(child, &status, 0)) {
2233: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2234: say("", "&wait rm");
1.72 schwarze 2235: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2236: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2237: say("", "%s: Cannot remove temporary directory",
2238: tempfilename);
1.22 schwarze 2239: }
1.11 schwarze 2240: }
2241:
1.47 schwarze 2242: /*
2243: * This is straightforward stuff.
2244: * Open a database connection to a "temporary" database, then open a set
2245: * of prepared statements we'll use over and over again.
2246: * If "real" is set, we use the existing database; if not, we truncate a
2247: * temporary one.
2248: * Must be matched by dbclose().
2249: */
2250: static int
2251: dbopen(int real)
1.2 schwarze 2252: {
1.72 schwarze 2253: const char *sql;
1.47 schwarze 2254: int rc, ofl;
1.6 schwarze 2255:
1.98 schwarze 2256: if (nodb)
1.47 schwarze 2257: return(1);
1.6 schwarze 2258:
1.72 schwarze 2259: *tempfilename = '\0';
1.47 schwarze 2260: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2261:
2262: if (real) {
2263: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2264: if (SQLITE_OK != rc) {
1.47 schwarze 2265: exitcode = (int)MANDOCLEVEL_SYSERR;
1.108 sthen 2266: if (SQLITE_CANTOPEN != rc)
2267: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.47 schwarze 2268: return(0);
1.28 schwarze 2269: }
1.72 schwarze 2270: goto prepare_statements;
2271: }
2272:
2273: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2274:
1.72 schwarze 2275: remove(MANDOC_DB "~");
2276: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.98 schwarze 2277: if (SQLITE_OK == rc)
1.72 schwarze 2278: goto create_tables;
1.73 schwarze 2279: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2280: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2281: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.47 schwarze 2282: return(0);
2283: }
1.6 schwarze 2284:
1.100 schwarze 2285: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2286: sizeof(tempfilename));
1.72 schwarze 2287: if (NULL == mkdtemp(tempfilename)) {
2288: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2289: say("", "&%s", tempfilename);
1.72 schwarze 2290: return(0);
2291: }
1.100 schwarze 2292: (void)strlcat(tempfilename, "/" MANDOC_DB,
2293: sizeof(tempfilename));
1.72 schwarze 2294: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2295: if (SQLITE_OK != rc) {
1.47 schwarze 2296: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2297: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.47 schwarze 2298: return(0);
1.2 schwarze 2299: }
2300:
1.72 schwarze 2301: create_tables:
1.47 schwarze 2302: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2303: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2304: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2305: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2306: ");\n"
2307: "\n"
2308: "CREATE TABLE \"mlinks\" (\n"
2309: " \"sec\" TEXT NOT NULL,\n"
2310: " \"arch\" TEXT NOT NULL,\n"
2311: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2312: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2313: "ON DELETE CASCADE\n"
1.47 schwarze 2314: ");\n"
1.93 schwarze 2315: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2316: "\n"
1.90 schwarze 2317: "CREATE TABLE \"names\" (\n"
2318: " \"bits\" INTEGER NOT NULL,\n"
2319: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2320: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.130 schwarze 2321: "ON DELETE CASCADE,\n"
2322: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.90 schwarze 2323: ");\n"
2324: "\n"
1.47 schwarze 2325: "CREATE TABLE \"keys\" (\n"
2326: " \"bits\" INTEGER NOT NULL,\n"
2327: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2328: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2329: "ON DELETE CASCADE\n"
1.93 schwarze 2330: ");\n"
2331: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2332:
2333: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2334: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2335: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.103 schwarze 2336: sqlite3_close(db);
1.47 schwarze 2337: return(0);
1.2 schwarze 2338: }
2339:
1.47 schwarze 2340: prepare_statements:
1.103 schwarze 2341: if (SQLITE_OK != sqlite3_exec(db,
2342: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2343: exitcode = (int)MANDOCLEVEL_SYSERR;
2344: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2345: sqlite3_errmsg(db));
2346: sqlite3_close(db);
2347: return(0);
2348: }
2349:
1.94 schwarze 2350: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2351: "(SELECT pageid FROM mlinks WHERE "
2352: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2353: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2354: sql = "INSERT INTO mpages "
1.89 schwarze 2355: "(desc,form) VALUES (?,?)";
1.47 schwarze 2356: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2357: sql = "INSERT INTO mlinks "
1.61 schwarze 2358: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2359: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.130 schwarze 2360: sql = "SELECT bits FROM names where pageid = ?";
2361: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.90 schwarze 2362: sql = "INSERT INTO names "
2363: "(bits,name,pageid) VALUES (?,?,?)";
2364: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2365: sql = "INSERT INTO keys "
2366: "(bits,key,pageid) VALUES (?,?,?)";
2367: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2368:
1.47 schwarze 2369: /*
2370: * When opening a new database, we can turn off
2371: * synchronous mode for much better performance.
2372: */
1.6 schwarze 2373:
1.103 schwarze 2374: if (real && SQLITE_OK != sqlite3_exec(db,
2375: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2376: exitcode = (int)MANDOCLEVEL_SYSERR;
2377: say(MANDOC_DB, "PRAGMA synchronous: %s",
2378: sqlite3_errmsg(db));
2379: sqlite3_close(db);
2380: return(0);
2381: }
1.11 schwarze 2382:
1.47 schwarze 2383: return(1);
2384: }
1.6 schwarze 2385:
1.47 schwarze 2386: static void *
1.106 espie 2387: hash_calloc(size_t n, size_t sz, void *arg)
1.47 schwarze 2388: {
1.6 schwarze 2389:
1.106 espie 2390: return(mandoc_calloc(n, sz));
1.47 schwarze 2391: }
1.2 schwarze 2392:
1.47 schwarze 2393: static void *
2394: hash_alloc(size_t sz, void *arg)
2395: {
1.28 schwarze 2396:
1.47 schwarze 2397: return(mandoc_malloc(sz));
2398: }
1.6 schwarze 2399:
1.47 schwarze 2400: static void
1.106 espie 2401: hash_free(void *p, void *arg)
1.47 schwarze 2402: {
1.26 schwarze 2403:
1.47 schwarze 2404: free(p);
2405: }
1.6 schwarze 2406:
1.47 schwarze 2407: static int
1.120 schwarze 2408: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2409: {
2410: static char startdir[PATH_MAX];
1.110 schwarze 2411: static int getcwd_status; /* 1 = ok, 2 = failure */
2412: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2413: char *cp;
1.6 schwarze 2414:
1.47 schwarze 2415: /*
1.110 schwarze 2416: * Remember the original working directory, if possible.
2417: * This will be needed if the second or a later directory
2418: * on the command line is given as a relative path.
2419: * Do not error out if the current directory is not
2420: * searchable: Maybe it won't be needed after all.
2421: */
2422: if (0 == getcwd_status) {
2423: if (NULL == getcwd(startdir, sizeof(startdir))) {
2424: getcwd_status = 2;
2425: (void)strlcpy(startdir, strerror(errno),
2426: sizeof(startdir));
2427: } else
2428: getcwd_status = 1;
2429: }
2430:
2431: /*
2432: * We are leaving the old base directory.
2433: * Do not use it any longer, not even for messages.
2434: */
2435: *basedir = '\0';
2436:
2437: /*
2438: * If and only if the directory was changed earlier and
2439: * the next directory to process is given as a relative path,
2440: * first go back, or bail out if that is impossible.
1.47 schwarze 2441: */
1.110 schwarze 2442: if (chdir_status && '/' != *targetdir) {
2443: if (2 == getcwd_status) {
1.47 schwarze 2444: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2445: say("", "getcwd: %s", startdir);
1.47 schwarze 2446: return(0);
2447: }
1.110 schwarze 2448: if (-1 == chdir(startdir)) {
1.47 schwarze 2449: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2450: say("", "&chdir %s", startdir);
1.47 schwarze 2451: return(0);
1.2 schwarze 2452: }
2453: }
1.110 schwarze 2454:
2455: /*
2456: * Always resolve basedir to the canonicalized absolute
2457: * pathname and append a trailing slash, such that
2458: * we can reliably check whether files are inside.
2459: */
1.47 schwarze 2460: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2461: if (report_baddir || errno != ENOENT) {
2462: exitcode = (int)MANDOCLEVEL_BADARG;
2463: say("", "&%s: realpath", targetdir);
2464: }
1.47 schwarze 2465: return(0);
2466: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2467: if (report_baddir || errno != ENOENT) {
2468: exitcode = (int)MANDOCLEVEL_BADARG;
2469: say("", "&chdir");
2470: }
1.47 schwarze 2471: return(0);
1.109 schwarze 2472: }
1.110 schwarze 2473: chdir_status = 1;
1.109 schwarze 2474: cp = strchr(basedir, '\0');
2475: if ('/' != cp[-1]) {
2476: if (cp - basedir >= PATH_MAX - 1) {
2477: exitcode = (int)MANDOCLEVEL_SYSERR;
2478: say("", "Filename too long");
2479: return(0);
2480: }
2481: *cp++ = '/';
2482: *cp = '\0';
1.47 schwarze 2483: }
2484: return(1);
1.2 schwarze 2485: }
2486:
2487: static void
1.47 schwarze 2488: say(const char *file, const char *format, ...)
1.2 schwarze 2489: {
1.47 schwarze 2490: va_list ap;
1.80 schwarze 2491: int use_errno;
1.2 schwarze 2492:
1.47 schwarze 2493: if ('\0' != *basedir)
2494: fprintf(stderr, "%s", basedir);
2495: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2496: fputc('/', stderr);
1.47 schwarze 2497: if ('\0' != *file)
2498: fprintf(stderr, "%s", file);
1.31 schwarze 2499:
1.80 schwarze 2500: use_errno = 1;
2501: if (NULL != format) {
2502: switch (*format) {
1.98 schwarze 2503: case '&':
1.80 schwarze 2504: format++;
2505: break;
1.98 schwarze 2506: case '\0':
1.80 schwarze 2507: format = NULL;
2508: break;
2509: default:
2510: use_errno = 0;
2511: break;
2512: }
2513: }
2514: if (NULL != format) {
2515: if ('\0' != *basedir || '\0' != *file)
2516: fputs(": ", stderr);
2517: va_start(ap, format);
2518: vfprintf(stderr, format, ap);
2519: va_end(ap);
2520: }
2521: if (use_errno) {
2522: if ('\0' != *basedir || '\0' != *file || NULL != format)
2523: fputs(": ", stderr);
1.47 schwarze 2524: perror(NULL);
1.80 schwarze 2525: } else
2526: fputc('\n', stderr);
1.1 schwarze 2527: }