Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.135
1.135 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.134 2014/12/30 20:40:05 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 *);
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();
1.123 schwarze 435: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, 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: enum mandoclevel lvl;
1096:
1097: str_info.alloc = hash_alloc;
1.106 espie 1098: str_info.calloc = hash_calloc;
1099: str_info.free = hash_free;
1.135 ! schwarze 1100: str_info.data = NULL;
1.47 schwarze 1101: str_info.key_offset = offsetof(struct str, key);
1102:
1.126 schwarze 1103: if ( ! nodb)
1.64 schwarze 1104: SQL_EXEC("BEGIN TRANSACTION");
1105:
1.47 schwarze 1106: mpage = ohash_first(&mpages, &pslot);
1.126 schwarze 1107: while (mpage != NULL) {
1.47 schwarze 1108: mlinks_undupe(mpage);
1.126 schwarze 1109: if (mpage->mlinks == NULL) {
1.47 schwarze 1110: mpage = ohash_next(&mpages, &pslot);
1111: continue;
1112: }
1.1 schwarze 1113:
1.90 schwarze 1114: name_mask = NAME_MASK;
1115: ohash_init(&names, 4, &str_info);
1.47 schwarze 1116: ohash_init(&strings, 6, &str_info);
1117: mparse_reset(mp);
1118: mdoc = NULL;
1119: man = NULL;
1.81 schwarze 1120: sodest = NULL;
1121:
1.125 schwarze 1122: mparse_open(mp, &fd, mpage->mlinks->file);
1.117 schwarze 1123: if (fd == -1) {
1124: say(mpage->mlinks->file, "&open");
1125: goto nextpage;
1.81 schwarze 1126: }
1.11 schwarze 1127:
1128: /*
1.24 schwarze 1129: * Try interpreting the file as mdoc(7) or man(7)
1130: * source code, unless it is already known to be
1131: * formatted. Fall back to formatted mode.
1.11 schwarze 1132: */
1.126 schwarze 1133: if (mpage->mlinks->dform != FORM_CAT ||
1134: mpage->mlinks->fform != FORM_CAT) {
1.117 schwarze 1135: lvl = mparse_readfd(mp, fd, mpage->mlinks->file);
1.47 schwarze 1136: if (lvl < MANDOCLEVEL_FATAL)
1.75 schwarze 1137: mparse_result(mp, &mdoc, &man, &sodest);
1.47 schwarze 1138: }
1.11 schwarze 1139:
1.126 schwarze 1140: if (sodest != NULL) {
1.75 schwarze 1141: mlink_dest = ohash_find(&mlinks,
1142: ohash_qlookup(&mlinks, sodest));
1.126 schwarze 1143: if (mlink_dest == NULL) {
1144: mandoc_asprintf(&cp, "%s.gz", sodest);
1145: mlink_dest = ohash_find(&mlinks,
1146: ohash_qlookup(&mlinks, cp));
1147: free(cp);
1148: }
1149: if (mlink_dest != NULL) {
1.75 schwarze 1150:
1151: /* The .so target exists. */
1152:
1153: mpage_dest = mlink_dest->mpage;
1154: mlink = mpage->mlinks;
1155: while (1) {
1156: mlink->mpage = mpage_dest;
1157:
1158: /*
1159: * If the target was already
1160: * processed, add the links
1161: * to the database now.
1162: * Otherwise, this will
1163: * happen when we come
1164: * to the target.
1165: */
1166:
1.94 schwarze 1167: if (mpage_dest->pageid)
1.124 schwarze 1168: dbadd_mlink_name(mlink);
1.75 schwarze 1169:
1.126 schwarze 1170: if (mlink->next == NULL)
1.75 schwarze 1171: break;
1172: mlink = mlink->next;
1173: }
1174:
1175: /* Move all links to the target. */
1176:
1177: mlink->next = mlink_dest->next;
1178: mlink_dest->next = mpage->mlinks;
1179: mpage->mlinks = NULL;
1180: }
1.81 schwarze 1181: goto nextpage;
1.126 schwarze 1182: } else if (mdoc != NULL) {
1.47 schwarze 1183: mpage->form = FORM_SRC;
1.112 schwarze 1184: mpage->sec = mdoc_meta(mdoc)->msec;
1185: mpage->sec = mandoc_strdup(
1.126 schwarze 1186: mpage->sec == NULL ? "" : mpage->sec);
1.47 schwarze 1187: mpage->arch = mdoc_meta(mdoc)->arch;
1188: mpage->arch = mandoc_strdup(
1.126 schwarze 1189: mpage->arch == NULL ? "" : mpage->arch);
1.47 schwarze 1190: mpage->title =
1191: mandoc_strdup(mdoc_meta(mdoc)->title);
1.126 schwarze 1192: } else if (man != NULL) {
1.47 schwarze 1193: mpage->form = FORM_SRC;
1194: mpage->sec =
1195: mandoc_strdup(man_meta(man)->msec);
1196: mpage->arch =
1197: mandoc_strdup(mpage->mlinks->arch);
1198: mpage->title =
1199: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1200: } else {
1.47 schwarze 1201: mpage->form = FORM_CAT;
1202: mpage->sec =
1203: mandoc_strdup(mpage->mlinks->dsec);
1204: mpage->arch =
1205: mandoc_strdup(mpage->mlinks->arch);
1206: mpage->title =
1207: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1208: }
1.54 schwarze 1209: putkey(mpage, mpage->sec, TYPE_sec);
1.119 schwarze 1210: if (*mpage->arch != '\0')
1211: putkey(mpage, mpage->arch, TYPE_arch);
1.1 schwarze 1212:
1.54 schwarze 1213: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1214: if ('\0' != *mlink->dsec)
1215: putkey(mpage, mlink->dsec, TYPE_sec);
1216: if ('\0' != *mlink->fsec)
1217: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1218: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1219: any : mlink->arch, TYPE_arch);
1.90 schwarze 1220: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1221: }
1.41 deraadt 1222:
1.129 schwarze 1223: assert(mpage->desc == NULL);
1224: if (mdoc != NULL)
1.127 schwarze 1225: parse_mdoc(mpage, mdoc_meta(mdoc), mdoc_node(mdoc));
1.129 schwarze 1226: else if (man != NULL)
1227: parse_man(mpage, man_meta(man), man_node(man));
1.47 schwarze 1228: else
1.117 schwarze 1229: parse_cat(mpage, fd);
1.129 schwarze 1230: if (mpage->desc == NULL)
1.92 schwarze 1231: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1232:
1233: if (warnings && !use_all)
1234: for (mlink = mpage->mlinks; mlink;
1235: mlink = mlink->next)
1236: mlink_check(mpage, mlink);
1.6 schwarze 1237:
1.131 schwarze 1238: dbadd(mpage);
1.81 schwarze 1239:
1240: nextpage:
1.125 schwarze 1241: if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1.117 schwarze 1242: exitcode = (int)MANDOCLEVEL_SYSERR;
1243: say(mpage->mlinks->file, "&wait gunzip");
1.81 schwarze 1244: }
1.47 schwarze 1245: ohash_delete(&strings);
1.90 schwarze 1246: ohash_delete(&names);
1.47 schwarze 1247: mpage = ohash_next(&mpages, &pslot);
1248: }
1.64 schwarze 1249:
1250: if (0 == nodb)
1251: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1252: }
1253:
1254: static void
1255: names_check(void)
1256: {
1257: sqlite3_stmt *stmt;
1258: const char *name, *sec, *arch, *key;
1259: int irc;
1260:
1261: sqlite3_prepare_v2(db,
1262: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1263: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1264: "WHERE bits & ? AND NOT EXISTS ("
1265: "SELECT pageid FROM mlinks "
1.90 schwarze 1266: "WHERE mlinks.pageid == names.pageid "
1267: "AND mlinks.name == names.name"
1.87 schwarze 1268: ")"
1269: ") JOIN ("
1.99 schwarze 1270: "SELECT sec, arch, name, pageid FROM mlinks "
1271: "GROUP BY pageid"
1.87 schwarze 1272: ") USING (pageid);",
1273: -1, &stmt, NULL);
1274:
1.91 schwarze 1275: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1276: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1277:
1278: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.113 schwarze 1279: name = (const char *)sqlite3_column_text(stmt, 0);
1280: sec = (const char *)sqlite3_column_text(stmt, 1);
1281: arch = (const char *)sqlite3_column_text(stmt, 2);
1282: key = (const char *)sqlite3_column_text(stmt, 3);
1.87 schwarze 1283: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1284: '\0' == *arch ? "" : "/",
1285: '\0' == *arch ? "" : arch, key);
1286: }
1287: sqlite3_finalize(stmt);
1.47 schwarze 1288: }
1.6 schwarze 1289:
1.47 schwarze 1290: static void
1.81 schwarze 1291: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1292: {
1293: FILE *stream;
1294: char *line, *p, *title;
1295: size_t len, plen, titlesz;
1.1 schwarze 1296:
1.81 schwarze 1297: stream = (-1 == fd) ?
1298: fopen(mpage->mlinks->file, "r") :
1299: fdopen(fd, "r");
1300: if (NULL == stream) {
1.114 doug 1301: if (-1 != fd)
1302: close(fd);
1.47 schwarze 1303: if (warnings)
1.80 schwarze 1304: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1305: return;
1306: }
1.1 schwarze 1307:
1.47 schwarze 1308: /* Skip to first blank line. */
1.1 schwarze 1309:
1.47 schwarze 1310: while (NULL != (line = fgetln(stream, &len)))
1311: if ('\n' == *line)
1312: break;
1.1 schwarze 1313:
1.47 schwarze 1314: /*
1315: * Assume the first line that is not indented
1316: * is the first section header. Skip to it.
1317: */
1.1 schwarze 1318:
1.47 schwarze 1319: while (NULL != (line = fgetln(stream, &len)))
1320: if ('\n' != *line && ' ' != *line)
1321: break;
1.98 schwarze 1322:
1.47 schwarze 1323: /*
1324: * Read up until the next section into a buffer.
1325: * Strip the leading and trailing newline from each read line,
1326: * appending a trailing space.
1327: * Ignore empty (whitespace-only) lines.
1328: */
1.28 schwarze 1329:
1.47 schwarze 1330: titlesz = 0;
1331: title = NULL;
1.38 schwarze 1332:
1.47 schwarze 1333: while (NULL != (line = fgetln(stream, &len))) {
1334: if (' ' != *line || '\n' != line[len - 1])
1335: break;
1336: while (len > 0 && isspace((unsigned char)*line)) {
1337: line++;
1338: len--;
1339: }
1340: if (1 == len)
1341: continue;
1342: title = mandoc_realloc(title, titlesz + len);
1343: memcpy(title + titlesz, line, len);
1344: titlesz += len;
1345: title[titlesz - 1] = ' ';
1346: }
1.28 schwarze 1347:
1.47 schwarze 1348: /*
1349: * If no page content can be found, or the input line
1350: * is already the next section header, or there is no
1351: * trailing newline, reuse the page title as the page
1352: * description.
1353: */
1.1 schwarze 1354:
1.47 schwarze 1355: if (NULL == title || '\0' == *title) {
1356: if (warnings)
1357: say(mpage->mlinks->file,
1358: "Cannot find NAME section");
1359: fclose(stream);
1360: free(title);
1361: return;
1362: }
1.24 schwarze 1363:
1.47 schwarze 1364: title = mandoc_realloc(title, titlesz + 1);
1365: title[titlesz] = '\0';
1.24 schwarze 1366:
1.47 schwarze 1367: /*
1368: * Skip to the first dash.
1369: * Use the remaining line as the description (no more than 70
1370: * bytes).
1371: */
1.28 schwarze 1372:
1.47 schwarze 1373: if (NULL != (p = strstr(title, "- "))) {
1374: for (p += 2; ' ' == *p || '\b' == *p; p++)
1375: /* Skip to next word. */ ;
1376: } else {
1377: if (warnings)
1378: say(mpage->mlinks->file,
1379: "No dash in title line");
1380: p = title;
1381: }
1.1 schwarze 1382:
1.47 schwarze 1383: plen = strlen(p);
1.1 schwarze 1384:
1.47 schwarze 1385: /* Strip backspace-encoding from line. */
1.1 schwarze 1386:
1.47 schwarze 1387: while (NULL != (line = memchr(p, '\b', plen))) {
1388: len = line - p;
1389: if (0 == len) {
1390: memmove(line, line + 1, plen--);
1391: continue;
1.98 schwarze 1392: }
1.47 schwarze 1393: memmove(line - 1, line + 1, plen - len);
1394: plen -= 2;
1395: }
1.1 schwarze 1396:
1.47 schwarze 1397: mpage->desc = mandoc_strdup(p);
1398: fclose(stream);
1399: free(title);
1400: }
1.16 schwarze 1401:
1.47 schwarze 1402: /*
1403: * Put a type/word pair into the word database for this particular file.
1404: */
1405: static void
1.69 schwarze 1406: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1407: {
1.69 schwarze 1408: char *cp;
1.37 schwarze 1409:
1.47 schwarze 1410: assert(NULL != value);
1.69 schwarze 1411: if (TYPE_arch == type)
1412: for (cp = value; *cp; cp++)
1413: if (isupper((unsigned char)*cp))
1414: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1415: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1416: }
1417:
1418: /*
1.47 schwarze 1419: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1420: */
1421: static void
1.47 schwarze 1422: putmdockey(const struct mpage *mpage,
1423: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1424: {
1.16 schwarze 1425:
1.47 schwarze 1426: for ( ; NULL != n; n = n->next) {
1427: if (NULL != n->child)
1428: putmdockey(mpage, n->child, m);
1429: if (MDOC_TEXT == n->type)
1430: putkey(mpage, n->string, m);
1431: }
1432: }
1.16 schwarze 1433:
1.47 schwarze 1434: static void
1.129 schwarze 1435: parse_man(struct mpage *mpage, const struct man_meta *meta,
1436: const struct man_node *n)
1.47 schwarze 1437: {
1438: const struct man_node *head, *body;
1.78 schwarze 1439: char *start, *title;
1.47 schwarze 1440: char byte;
1.78 schwarze 1441: size_t sz;
1.16 schwarze 1442:
1.47 schwarze 1443: if (NULL == n)
1444: return;
1.16 schwarze 1445:
1.47 schwarze 1446: /*
1447: * We're only searching for one thing: the first text child in
1448: * the BODY of a NAME section. Since we don't keep track of
1449: * sections in -man, run some hoops to find out whether we're in
1450: * the correct section or not.
1451: */
1.16 schwarze 1452:
1.47 schwarze 1453: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1454: body = n;
1455: assert(body->parent);
1456: if (NULL != (head = body->parent->head) &&
1.98 schwarze 1457: 1 == head->nchild &&
1458: NULL != (head = (head->child)) &&
1459: MAN_TEXT == head->type &&
1460: 0 == strcmp(head->string, "NAME") &&
1461: NULL != body->child) {
1.2 schwarze 1462:
1.47 schwarze 1463: /*
1464: * Suck the entire NAME section into memory.
1465: * Yes, we might run away.
1466: * But too many manuals have big, spread-out
1467: * NAME sections over many lines.
1468: */
1.2 schwarze 1469:
1.78 schwarze 1470: title = NULL;
1471: man_deroff(&title, body);
1.47 schwarze 1472: if (NULL == title)
1473: return;
1.16 schwarze 1474:
1.98 schwarze 1475: /*
1.47 schwarze 1476: * Go through a special heuristic dance here.
1477: * Conventionally, one or more manual names are
1478: * comma-specified prior to a whitespace, then a
1479: * dash, then a description. Try to puzzle out
1480: * the name parts here.
1481: */
1.16 schwarze 1482:
1.78 schwarze 1483: start = title;
1.47 schwarze 1484: for ( ;; ) {
1485: sz = strcspn(start, " ,");
1486: if ('\0' == start[sz])
1487: break;
1.1 schwarze 1488:
1.47 schwarze 1489: byte = start[sz];
1490: start[sz] = '\0';
1.67 schwarze 1491:
1492: /*
1493: * Assume a stray trailing comma in the
1494: * name list if a name begins with a dash.
1495: */
1496:
1497: if ('-' == start[0] ||
1498: ('\\' == start[0] && '-' == start[1]))
1499: break;
1.1 schwarze 1500:
1.90 schwarze 1501: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1502: if ( ! (mpage->name_head_done ||
1503: strcasecmp(start, meta->title))) {
1504: putkey(mpage, start, NAME_HEAD);
1505: mpage->name_head_done = 1;
1506: }
1.1 schwarze 1507:
1.47 schwarze 1508: if (' ' == byte) {
1509: start += sz + 1;
1510: break;
1511: }
1.1 schwarze 1512:
1.47 schwarze 1513: assert(',' == byte);
1514: start += sz + 1;
1515: while (' ' == *start)
1516: start++;
1517: }
1.1 schwarze 1518:
1.78 schwarze 1519: if (start == title) {
1.90 schwarze 1520: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1521: if ( ! (mpage->name_head_done ||
1522: strcasecmp(start, meta->title))) {
1523: putkey(mpage, start, NAME_HEAD);
1524: mpage->name_head_done = 1;
1525: }
1.47 schwarze 1526: free(title);
1527: return;
1528: }
1.1 schwarze 1529:
1.47 schwarze 1530: while (isspace((unsigned char)*start))
1531: start++;
1.1 schwarze 1532:
1.47 schwarze 1533: if (0 == strncmp(start, "-", 1))
1534: start += 1;
1535: else if (0 == strncmp(start, "\\-\\-", 4))
1536: start += 4;
1537: else if (0 == strncmp(start, "\\-", 2))
1538: start += 2;
1539: else if (0 == strncmp(start, "\\(en", 4))
1540: start += 4;
1541: else if (0 == strncmp(start, "\\(em", 4))
1542: start += 4;
1.1 schwarze 1543:
1.47 schwarze 1544: while (' ' == *start)
1545: start++;
1.1 schwarze 1546:
1.47 schwarze 1547: mpage->desc = mandoc_strdup(start);
1548: free(title);
1549: return;
1550: }
1551: }
1.1 schwarze 1552:
1.47 schwarze 1553: for (n = n->child; n; n = n->next) {
1554: if (NULL != mpage->desc)
1555: break;
1.129 schwarze 1556: parse_man(mpage, meta, n);
1.1 schwarze 1557: }
1558: }
1559:
1560: static void
1.127 schwarze 1561: parse_mdoc(struct mpage *mpage, const struct mdoc_meta *meta,
1562: const struct mdoc_node *n)
1.1 schwarze 1563: {
1564:
1.47 schwarze 1565: assert(NULL != n);
1566: for (n = n->child; NULL != n; n = n->next) {
1567: switch (n->type) {
1.98 schwarze 1568: case MDOC_ELEM:
1.47 schwarze 1569: /* FALLTHROUGH */
1.98 schwarze 1570: case MDOC_BLOCK:
1.47 schwarze 1571: /* FALLTHROUGH */
1.98 schwarze 1572: case MDOC_HEAD:
1.47 schwarze 1573: /* FALLTHROUGH */
1.98 schwarze 1574: case MDOC_BODY:
1.47 schwarze 1575: /* FALLTHROUGH */
1.98 schwarze 1576: case MDOC_TAIL:
1.47 schwarze 1577: if (NULL != mdocs[n->tok].fp)
1.127 schwarze 1578: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.47 schwarze 1579: break;
1580: if (mdocs[n->tok].mask)
1581: putmdockey(mpage, n->child,
1582: mdocs[n->tok].mask);
1583: break;
1584: default:
1585: assert(MDOC_ROOT != n->type);
1586: continue;
1587: }
1588: if (NULL != n->child)
1.127 schwarze 1589: parse_mdoc(mpage, meta, n);
1.1 schwarze 1590: }
1591: }
1592:
1.19 schwarze 1593: static int
1.127 schwarze 1594: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_meta *meta,
1595: const struct mdoc_node *n)
1.1 schwarze 1596: {
1.131 schwarze 1597: char *start, *end;
1.1 schwarze 1598: size_t sz;
1.19 schwarze 1599:
1.47 schwarze 1600: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1601: NULL == (n = n->child) ||
1602: MDOC_TEXT != n->type)
1.19 schwarze 1603: return(0);
1.1 schwarze 1604:
1605: /*
1606: * Only consider those `Fd' macro fields that begin with an
1607: * "inclusion" token (versus, e.g., #define).
1608: */
1.47 schwarze 1609:
1.1 schwarze 1610: if (strcmp("#include", n->string))
1.19 schwarze 1611: return(0);
1.1 schwarze 1612:
1613: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1614: return(0);
1.1 schwarze 1615:
1616: /*
1617: * Strip away the enclosing angle brackets and make sure we're
1618: * not zero-length.
1619: */
1620:
1621: start = n->string;
1622: if ('<' == *start || '"' == *start)
1623: start++;
1624:
1625: if (0 == (sz = strlen(start)))
1.19 schwarze 1626: return(0);
1.1 schwarze 1627:
1628: end = &start[(int)sz - 1];
1629: if ('>' == *end || '"' == *end)
1630: end--;
1631:
1.47 schwarze 1632: if (end > start)
1633: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1634: return(0);
1.1 schwarze 1635: }
1636:
1.133 schwarze 1637: static void
1638: parse_mdoc_fname(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1639: {
1.69 schwarze 1640: char *cp;
1.133 schwarze 1641: size_t sz;
1.1 schwarze 1642:
1.133 schwarze 1643: if (n->type != MDOC_TEXT)
1644: return;
1.19 schwarze 1645:
1.133 schwarze 1646: /* Skip function pointer punctuation. */
1.1 schwarze 1647:
1.133 schwarze 1648: cp = n->string;
1649: while (*cp == '(' || *cp == '*')
1.1 schwarze 1650: cp++;
1.133 schwarze 1651: sz = strcspn(cp, "()");
1.1 schwarze 1652:
1.133 schwarze 1653: putkeys(mpage, cp, sz, TYPE_Fn);
1.128 schwarze 1654: if (n->sec == SEC_SYNOPSIS)
1.133 schwarze 1655: putkeys(mpage, cp, sz, NAME_SYN);
1656: }
1.19 schwarze 1657:
1.133 schwarze 1658: static int
1659: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_meta *meta,
1660: const struct mdoc_node *n)
1661: {
1662:
1663: if (n->child == NULL)
1664: return(0);
1665:
1666: parse_mdoc_fname(mpage, n->child);
1.19 schwarze 1667:
1.133 schwarze 1668: for (n = n->child->next; n != NULL; n = n->next)
1669: if (n->type == MDOC_TEXT)
1.47 schwarze 1670: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1671:
1.128 schwarze 1672: return(0);
1673: }
1674:
1675: static int
1676: parse_mdoc_Fo(struct mpage *mpage, const struct mdoc_meta *meta,
1677: const struct mdoc_node *n)
1678: {
1.132 schwarze 1679:
1680: if (n->type != MDOC_HEAD)
1681: return(1);
1.128 schwarze 1682:
1.133 schwarze 1683: if (n->child != NULL)
1684: parse_mdoc_fname(mpage, n->child);
1685:
1.19 schwarze 1686: return(0);
1.1 schwarze 1687: }
1688:
1.19 schwarze 1689: static int
1.127 schwarze 1690: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_meta *meta,
1691: const struct mdoc_node *n)
1.1 schwarze 1692: {
1.47 schwarze 1693: char *cp;
1.1 schwarze 1694:
1695: if (NULL == (n = n->child))
1.19 schwarze 1696: return(0);
1.1 schwarze 1697:
1.47 schwarze 1698: if (NULL == n->next) {
1699: putkey(mpage, n->string, TYPE_Xr);
1700: return(0);
1701: }
1.1 schwarze 1702:
1.77 schwarze 1703: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1704: putkey(mpage, cp, TYPE_Xr);
1705: free(cp);
1706: return(0);
1.1 schwarze 1707: }
1708:
1.19 schwarze 1709: static int
1.127 schwarze 1710: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_meta *meta,
1711: const struct mdoc_node *n)
1.1 schwarze 1712: {
1713:
1.79 schwarze 1714: if (MDOC_BODY == n->type)
1715: mdoc_deroff(&mpage->desc, n);
1716: return(0);
1.1 schwarze 1717: }
1718:
1.19 schwarze 1719: static int
1.127 schwarze 1720: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_meta *meta,
1721: const struct mdoc_node *n)
1.1 schwarze 1722: {
1723:
1.86 schwarze 1724: if (SEC_NAME == n->sec)
1.90 schwarze 1725: putmdockey(mpage, n->child, NAME_TITLE);
1.127 schwarze 1726: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) {
1727: if (n->child == NULL)
1728: putkey(mpage, meta->name, NAME_SYN);
1729: else
1730: putmdockey(mpage, n->child, NAME_SYN);
1.129 schwarze 1731: }
1732: if ( ! (mpage->name_head_done ||
1733: n->child == NULL || n->child->string == NULL ||
1734: strcasecmp(n->child->string, meta->title))) {
1735: putkey(mpage, n->child->string, NAME_HEAD);
1736: mpage->name_head_done = 1;
1.127 schwarze 1737: }
1.86 schwarze 1738: return(0);
1.1 schwarze 1739: }
1740:
1.19 schwarze 1741: static int
1.127 schwarze 1742: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_meta *meta,
1743: const struct mdoc_node *n)
1.1 schwarze 1744: {
1745:
1.19 schwarze 1746: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1747: }
1748:
1.47 schwarze 1749: static int
1.127 schwarze 1750: parse_mdoc_head(struct mpage *mpage, const struct mdoc_meta *meta,
1751: const struct mdoc_node *n)
1.1 schwarze 1752: {
1753:
1.47 schwarze 1754: return(MDOC_HEAD == n->type);
1755: }
1.1 schwarze 1756:
1.47 schwarze 1757: static int
1.127 schwarze 1758: parse_mdoc_body(struct mpage *mpage, const struct mdoc_meta *meta,
1759: const struct mdoc_node *n)
1.47 schwarze 1760: {
1.1 schwarze 1761:
1.47 schwarze 1762: return(MDOC_BODY == n->type);
1.1 schwarze 1763: }
1764:
1.47 schwarze 1765: /*
1766: * Add a string to the hash table for the current manual.
1767: * Each string has a bitmask telling which macros it belongs to.
1768: * When we finish the manual, we'll dump the table.
1769: */
1.1 schwarze 1770: static void
1.131 schwarze 1771: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.1 schwarze 1772: {
1.90 schwarze 1773: struct ohash *htab;
1.47 schwarze 1774: struct str *s;
1.68 schwarze 1775: const char *end;
1.47 schwarze 1776: unsigned int slot;
1.131 schwarze 1777: int i, mustfree;
1.1 schwarze 1778:
1.47 schwarze 1779: if (0 == sz)
1780: return;
1.68 schwarze 1781:
1.131 schwarze 1782: mustfree = render_string(&cp, &sz);
1783:
1.90 schwarze 1784: if (TYPE_Nm & v) {
1785: htab = &names;
1786: v &= name_mask;
1.124 schwarze 1787: if (v & NAME_FIRST)
1788: name_mask &= ~NAME_FIRST;
1.90 schwarze 1789: if (debug > 1)
1790: say(mpage->mlinks->file,
1.130 schwarze 1791: "Adding name %*s, bits=%d", sz, cp, v);
1.90 schwarze 1792: } else {
1793: htab = &strings;
1794: if (debug > 1)
1795: for (i = 0; i < mansearch_keymax; i++)
1.118 schwarze 1796: if ((uint64_t)1 << i & v)
1.90 schwarze 1797: say(mpage->mlinks->file,
1798: "Adding key %s=%*s",
1799: mansearch_keynames[i], sz, cp);
1.68 schwarze 1800: }
1.47 schwarze 1801:
1802: end = cp + sz;
1.90 schwarze 1803: slot = ohash_qlookupi(htab, cp, &end);
1804: s = ohash_find(htab, slot);
1.1 schwarze 1805:
1.47 schwarze 1806: if (NULL != s && mpage == s->mpage) {
1807: s->mask |= v;
1.1 schwarze 1808: return;
1.47 schwarze 1809: } else if (NULL == s) {
1.101 schwarze 1810: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1811: memcpy(s->key, cp, sz);
1.90 schwarze 1812: ohash_insert(htab, slot, s);
1.47 schwarze 1813: }
1814: s->mpage = mpage;
1815: s->mask = v;
1.131 schwarze 1816:
1817: if (mustfree)
1818: free(cp);
1.1 schwarze 1819: }
1820:
1821: /*
1.47 schwarze 1822: * Take a Unicode codepoint and produce its UTF-8 encoding.
1823: * This isn't the best way to do this, but it works.
1824: * The magic numbers are from the UTF-8 packaging.
1825: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1826: */
1.47 schwarze 1827: static size_t
1828: utf8(unsigned int cp, char out[7])
1.1 schwarze 1829: {
1.47 schwarze 1830: size_t rc;
1.1 schwarze 1831:
1.47 schwarze 1832: rc = 0;
1833: if (cp <= 0x0000007F) {
1834: rc = 1;
1835: out[0] = (char)cp;
1836: } else if (cp <= 0x000007FF) {
1837: rc = 2;
1838: out[0] = (cp >> 6 & 31) | 192;
1839: out[1] = (cp & 63) | 128;
1840: } else if (cp <= 0x0000FFFF) {
1841: rc = 3;
1842: out[0] = (cp >> 12 & 15) | 224;
1843: out[1] = (cp >> 6 & 63) | 128;
1844: out[2] = (cp & 63) | 128;
1845: } else if (cp <= 0x001FFFFF) {
1846: rc = 4;
1847: out[0] = (cp >> 18 & 7) | 240;
1848: out[1] = (cp >> 12 & 63) | 128;
1849: out[2] = (cp >> 6 & 63) | 128;
1850: out[3] = (cp & 63) | 128;
1851: } else if (cp <= 0x03FFFFFF) {
1852: rc = 5;
1853: out[0] = (cp >> 24 & 3) | 248;
1854: out[1] = (cp >> 18 & 63) | 128;
1855: out[2] = (cp >> 12 & 63) | 128;
1856: out[3] = (cp >> 6 & 63) | 128;
1857: out[4] = (cp & 63) | 128;
1858: } else if (cp <= 0x7FFFFFFF) {
1859: rc = 6;
1860: out[0] = (cp >> 30 & 1) | 252;
1861: out[1] = (cp >> 24 & 63) | 128;
1862: out[2] = (cp >> 18 & 63) | 128;
1863: out[3] = (cp >> 12 & 63) | 128;
1864: out[4] = (cp >> 6 & 63) | 128;
1865: out[5] = (cp & 63) | 128;
1866: } else
1867: return(0);
1.19 schwarze 1868:
1.47 schwarze 1869: out[rc] = '\0';
1870: return(rc);
1.1 schwarze 1871: }
1872:
1.47 schwarze 1873: /*
1.131 schwarze 1874: * If the string contains escape sequences,
1875: * replace it with an allocated rendering and return 1,
1876: * such that the caller can free it after use.
1877: * Otherwise, do nothing and return 0.
1.47 schwarze 1878: */
1.131 schwarze 1879: static int
1880: render_string(char **public, size_t *psz)
1.1 schwarze 1881: {
1.131 schwarze 1882: const char *src, *scp, *addcp, *seq;
1883: char *dst;
1884: size_t ssz, dsz, addsz;
1.71 schwarze 1885: char utfbuf[7], res[6];
1.131 schwarze 1886: int seqlen, unicode;
1.47 schwarze 1887:
1888: res[0] = '\\';
1889: res[1] = '\t';
1890: res[2] = ASCII_NBRSP;
1891: res[3] = ASCII_HYPH;
1.71 schwarze 1892: res[4] = ASCII_BREAK;
1893: res[5] = '\0';
1.1 schwarze 1894:
1.131 schwarze 1895: src = scp = *public;
1896: ssz = *psz;
1897: dst = NULL;
1898: dsz = 0;
1899:
1900: while (scp < src + *psz) {
1901:
1902: /* Leave normal characters unchanged. */
1903:
1904: if (strchr(res, *scp) == NULL) {
1905: if (dst != NULL)
1906: dst[dsz++] = *scp;
1907: scp++;
1908: continue;
1909: }
1.1 schwarze 1910:
1.47 schwarze 1911: /*
1.131 schwarze 1912: * Found something that requires replacing,
1913: * make sure we have a destination buffer.
1.47 schwarze 1914: */
1.131 schwarze 1915:
1916: if (dst == NULL) {
1917: dst = mandoc_malloc(ssz + 1);
1918: dsz = scp - src;
1919: memcpy(dst, src, dsz);
1.47 schwarze 1920: }
1.39 schwarze 1921:
1.131 schwarze 1922: /* Handle single-char special characters. */
1923:
1924: switch (*scp) {
1925: case '\\':
1926: break;
1.98 schwarze 1927: case '\t':
1.71 schwarze 1928: /* FALLTHROUGH */
1.98 schwarze 1929: case ASCII_NBRSP:
1.131 schwarze 1930: dst[dsz++] = ' ';
1931: scp++;
1932: continue;
1933: case ASCII_HYPH:
1934: dst[dsz++] = '-';
1.71 schwarze 1935: /* FALLTHROUGH */
1.98 schwarze 1936: case ASCII_BREAK:
1.131 schwarze 1937: scp++;
1.47 schwarze 1938: continue;
1.71 schwarze 1939: default:
1.131 schwarze 1940: abort();
1.71 schwarze 1941: }
1.39 schwarze 1942:
1.47 schwarze 1943: /*
1.131 schwarze 1944: * Found an escape sequence.
1945: * Read past the slash, then parse it.
1946: * Ignore everything except characters.
1.47 schwarze 1947: */
1.52 schwarze 1948:
1.131 schwarze 1949: scp++;
1950: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.47 schwarze 1951: continue;
1.39 schwarze 1952:
1.47 schwarze 1953: /*
1.52 schwarze 1954: * Render the special character
1955: * as either UTF-8 or ASCII.
1.47 schwarze 1956: */
1.52 schwarze 1957:
1958: if (write_utf8) {
1.131 schwarze 1959: unicode = mchars_spec2cp(mchars, seq, seqlen);
1960: if (unicode <= 0)
1.52 schwarze 1961: continue;
1.131 schwarze 1962: addsz = utf8(unicode, utfbuf);
1963: if (addsz == 0)
1.52 schwarze 1964: continue;
1.131 schwarze 1965: addcp = utfbuf;
1.52 schwarze 1966: } else {
1.131 schwarze 1967: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1968: if (addcp == NULL)
1.52 schwarze 1969: continue;
1.131 schwarze 1970: if (*addcp == ASCII_NBRSP) {
1971: addcp = " ";
1972: addsz = 1;
1.52 schwarze 1973: }
1974: }
1.1 schwarze 1975:
1.47 schwarze 1976: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1977:
1.131 schwarze 1978: ssz += addsz;
1979: dst = mandoc_realloc(dst, ssz + 1);
1980: memcpy(dst + dsz, addcp, addsz);
1981: dsz += addsz;
1982: }
1983: if (dst != NULL) {
1984: *public = dst;
1985: *psz = dsz;
1986: }
1987:
1988: /* Trim trailing whitespace and NUL-terminate. */
1989:
1990: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1991: --*psz;
1992: if (dst != NULL) {
1993: (*public)[*psz] = '\0';
1994: return(1);
1995: } else
1996: return(0);
1.1 schwarze 1997: }
1998:
1.75 schwarze 1999: static void
2000: dbadd_mlink(const struct mlink *mlink)
2001: {
2002: size_t i;
2003:
2004: i = 1;
2005: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2006: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2007: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 2008: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 2009: SQL_STEP(stmts[STMT_INSERT_LINK]);
2010: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.124 schwarze 2011: }
2012:
2013: static void
2014: dbadd_mlink_name(const struct mlink *mlink)
2015: {
1.130 schwarze 2016: uint64_t bits;
1.124 schwarze 2017: size_t i;
2018:
2019: dbadd_mlink(mlink);
1.115 schwarze 2020:
2021: i = 1;
1.130 schwarze 2022: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2023: bits = NAME_FILE & NAME_MASK;
2024: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2025: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2026: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2027: }
2028:
2029: i = 1;
2030: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.115 schwarze 2031: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2032: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2033: SQL_STEP(stmts[STMT_INSERT_NAME]);
2034: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.75 schwarze 2035: }
2036:
1.11 schwarze 2037: /*
1.47 schwarze 2038: * Flush the current page's terms (and their bits) into the database.
2039: * Wrap the entire set of additions in a transaction to make sqlite be a
2040: * little faster.
1.53 schwarze 2041: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 2042: */
2043: static void
1.131 schwarze 2044: dbadd(struct mpage *mpage)
1.11 schwarze 2045: {
1.47 schwarze 2046: struct mlink *mlink;
2047: struct str *key;
1.131 schwarze 2048: char *cp;
1.47 schwarze 2049: size_t i;
2050: unsigned int slot;
1.131 schwarze 2051: int mustfree;
1.47 schwarze 2052:
1.85 schwarze 2053: mlink = mpage->mlinks;
1.11 schwarze 2054:
1.85 schwarze 2055: if (nodb) {
1.104 schwarze 2056: for (key = ohash_first(&names, &slot); NULL != key;
1.131 schwarze 2057: key = ohash_next(&names, &slot))
1.104 schwarze 2058: free(key);
2059: for (key = ohash_first(&strings, &slot); NULL != key;
1.131 schwarze 2060: key = ohash_next(&strings, &slot))
1.104 schwarze 2061: free(key);
1.102 schwarze 2062: if (0 == debug)
2063: return;
1.85 schwarze 2064: while (NULL != mlink) {
2065: fputs(mlink->name, stdout);
2066: if (NULL == mlink->next ||
2067: strcmp(mlink->dsec, mlink->next->dsec) ||
2068: strcmp(mlink->fsec, mlink->next->fsec) ||
2069: strcmp(mlink->arch, mlink->next->arch)) {
2070: putchar('(');
2071: if ('\0' == *mlink->dsec)
2072: fputs(mlink->fsec, stdout);
2073: else
2074: fputs(mlink->dsec, stdout);
2075: if ('\0' != *mlink->arch)
2076: printf("/%s", mlink->arch);
2077: putchar(')');
2078: }
2079: mlink = mlink->next;
2080: if (NULL != mlink)
2081: fputs(", ", stdout);
2082: }
1.89 schwarze 2083: printf(" - %s\n", mpage->desc);
1.11 schwarze 2084: return;
1.85 schwarze 2085: }
2086:
2087: if (debug)
2088: say(mlink->file, "Adding to database");
1.47 schwarze 2089:
1.131 schwarze 2090: cp = mpage->desc;
2091: i = strlen(cp);
2092: mustfree = render_string(&cp, &i);
1.47 schwarze 2093: i = 1;
1.131 schwarze 2094: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.116 schwarze 2095: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.47 schwarze 2096: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2097: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2098: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.131 schwarze 2099: if (mustfree)
2100: free(cp);
1.47 schwarze 2101:
1.85 schwarze 2102: while (NULL != mlink) {
1.75 schwarze 2103: dbadd_mlink(mlink);
1.85 schwarze 2104: mlink = mlink->next;
2105: }
1.91 schwarze 2106: mlink = mpage->mlinks;
1.47 schwarze 2107:
1.90 schwarze 2108: for (key = ohash_first(&names, &slot); NULL != key;
2109: key = ohash_next(&names, &slot)) {
2110: assert(key->mpage == mpage);
2111: i = 1;
2112: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.131 schwarze 2113: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.94 schwarze 2114: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2115: SQL_STEP(stmts[STMT_INSERT_NAME]);
2116: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2117: free(key);
2118: }
1.47 schwarze 2119: for (key = ohash_first(&strings, &slot); NULL != key;
2120: key = ohash_next(&strings, &slot)) {
2121: assert(key->mpage == mpage);
2122: i = 1;
2123: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.131 schwarze 2124: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.94 schwarze 2125: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2126: SQL_STEP(stmts[STMT_INSERT_KEY]);
2127: sqlite3_reset(stmts[STMT_INSERT_KEY]);
2128: free(key);
1.33 schwarze 2129: }
1.47 schwarze 2130: }
1.41 deraadt 2131:
1.47 schwarze 2132: static void
2133: dbprune(void)
2134: {
2135: struct mpage *mpage;
2136: struct mlink *mlink;
2137: size_t i;
2138: unsigned int slot;
1.11 schwarze 2139:
1.63 schwarze 2140: if (0 == nodb)
2141: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2142:
1.63 schwarze 2143: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2144: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2145: mlink = mpage->mlinks;
1.82 schwarze 2146: if (debug)
1.63 schwarze 2147: say(mlink->file, "Deleting from database");
2148: if (nodb)
2149: continue;
2150: for ( ; NULL != mlink; mlink = mlink->next) {
2151: i = 1;
2152: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2153: i, mlink->dsec);
2154: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2155: i, mlink->arch);
2156: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2157: i, mlink->name);
2158: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2159: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2160: }
1.11 schwarze 2161: }
1.63 schwarze 2162:
2163: if (0 == nodb)
2164: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2165: }
1.22 schwarze 2166:
1.47 schwarze 2167: /*
2168: * Close an existing database and its prepared statements.
2169: * If "real" is not set, rename the temporary file into the real one.
2170: */
2171: static void
2172: dbclose(int real)
2173: {
2174: size_t i;
1.72 schwarze 2175: int status;
2176: pid_t child;
1.11 schwarze 2177:
1.47 schwarze 2178: if (nodb)
2179: return;
1.11 schwarze 2180:
1.47 schwarze 2181: for (i = 0; i < STMT__MAX; i++) {
2182: sqlite3_finalize(stmts[i]);
2183: stmts[i] = NULL;
1.28 schwarze 2184: }
1.22 schwarze 2185:
1.47 schwarze 2186: sqlite3_close(db);
2187: db = NULL;
1.11 schwarze 2188:
1.47 schwarze 2189: if (real)
2190: return;
1.22 schwarze 2191:
1.72 schwarze 2192: if ('\0' == *tempfilename) {
2193: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2194: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2195: say(MANDOC_DB, "&rename");
1.72 schwarze 2196: }
2197: return;
2198: }
2199:
2200: switch (child = fork()) {
1.98 schwarze 2201: case -1:
1.72 schwarze 2202: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2203: say("", "&fork cmp");
1.72 schwarze 2204: return;
1.98 schwarze 2205: case 0:
1.72 schwarze 2206: execlp("cmp", "cmp", "-s",
2207: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2208: say("", "&exec cmp");
1.72 schwarze 2209: exit(0);
2210: default:
2211: break;
2212: }
2213: if (-1 == waitpid(child, &status, 0)) {
2214: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2215: say("", "&wait cmp");
1.72 schwarze 2216: } else if (WIFSIGNALED(status)) {
2217: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2218: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2219: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2220: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2221: say(MANDOC_DB,
2222: "Data changed, but cannot replace database");
2223: }
2224:
2225: *strrchr(tempfilename, '/') = '\0';
2226: switch (child = fork()) {
1.98 schwarze 2227: case -1:
1.72 schwarze 2228: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2229: say("", "&fork rm");
1.72 schwarze 2230: return;
1.98 schwarze 2231: case 0:
1.72 schwarze 2232: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2233: say("", "&exec rm");
1.72 schwarze 2234: exit((int)MANDOCLEVEL_SYSERR);
2235: default:
2236: break;
2237: }
2238: if (-1 == waitpid(child, &status, 0)) {
2239: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2240: say("", "&wait rm");
1.72 schwarze 2241: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2242: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2243: say("", "%s: Cannot remove temporary directory",
2244: tempfilename);
1.22 schwarze 2245: }
1.11 schwarze 2246: }
2247:
1.47 schwarze 2248: /*
2249: * This is straightforward stuff.
2250: * Open a database connection to a "temporary" database, then open a set
2251: * of prepared statements we'll use over and over again.
2252: * If "real" is set, we use the existing database; if not, we truncate a
2253: * temporary one.
2254: * Must be matched by dbclose().
2255: */
2256: static int
2257: dbopen(int real)
1.2 schwarze 2258: {
1.72 schwarze 2259: const char *sql;
1.47 schwarze 2260: int rc, ofl;
1.6 schwarze 2261:
1.98 schwarze 2262: if (nodb)
1.47 schwarze 2263: return(1);
1.6 schwarze 2264:
1.72 schwarze 2265: *tempfilename = '\0';
1.47 schwarze 2266: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2267:
2268: if (real) {
2269: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2270: if (SQLITE_OK != rc) {
1.47 schwarze 2271: exitcode = (int)MANDOCLEVEL_SYSERR;
1.108 sthen 2272: if (SQLITE_CANTOPEN != rc)
2273: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.47 schwarze 2274: return(0);
1.28 schwarze 2275: }
1.72 schwarze 2276: goto prepare_statements;
2277: }
2278:
2279: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2280:
1.72 schwarze 2281: remove(MANDOC_DB "~");
2282: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.98 schwarze 2283: if (SQLITE_OK == rc)
1.72 schwarze 2284: goto create_tables;
1.73 schwarze 2285: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2286: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2287: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.47 schwarze 2288: return(0);
2289: }
1.6 schwarze 2290:
1.100 schwarze 2291: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2292: sizeof(tempfilename));
1.72 schwarze 2293: if (NULL == mkdtemp(tempfilename)) {
2294: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2295: say("", "&%s", tempfilename);
1.72 schwarze 2296: return(0);
2297: }
1.100 schwarze 2298: (void)strlcat(tempfilename, "/" MANDOC_DB,
2299: sizeof(tempfilename));
1.72 schwarze 2300: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2301: if (SQLITE_OK != rc) {
1.47 schwarze 2302: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2303: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.47 schwarze 2304: return(0);
1.2 schwarze 2305: }
2306:
1.72 schwarze 2307: create_tables:
1.47 schwarze 2308: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2309: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2310: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2311: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2312: ");\n"
2313: "\n"
2314: "CREATE TABLE \"mlinks\" (\n"
2315: " \"sec\" TEXT NOT NULL,\n"
2316: " \"arch\" TEXT NOT NULL,\n"
2317: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2318: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2319: "ON DELETE CASCADE\n"
1.47 schwarze 2320: ");\n"
1.93 schwarze 2321: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2322: "\n"
1.90 schwarze 2323: "CREATE TABLE \"names\" (\n"
2324: " \"bits\" INTEGER NOT NULL,\n"
2325: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2326: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.130 schwarze 2327: "ON DELETE CASCADE,\n"
2328: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.90 schwarze 2329: ");\n"
2330: "\n"
1.47 schwarze 2331: "CREATE TABLE \"keys\" (\n"
2332: " \"bits\" INTEGER NOT NULL,\n"
2333: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2334: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2335: "ON DELETE CASCADE\n"
1.93 schwarze 2336: ");\n"
2337: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2338:
2339: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2340: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2341: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.103 schwarze 2342: sqlite3_close(db);
1.47 schwarze 2343: return(0);
1.2 schwarze 2344: }
2345:
1.47 schwarze 2346: prepare_statements:
1.103 schwarze 2347: if (SQLITE_OK != sqlite3_exec(db,
2348: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2349: exitcode = (int)MANDOCLEVEL_SYSERR;
2350: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2351: sqlite3_errmsg(db));
2352: sqlite3_close(db);
2353: return(0);
2354: }
2355:
1.94 schwarze 2356: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2357: "(SELECT pageid FROM mlinks WHERE "
2358: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2359: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2360: sql = "INSERT INTO mpages "
1.89 schwarze 2361: "(desc,form) VALUES (?,?)";
1.47 schwarze 2362: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2363: sql = "INSERT INTO mlinks "
1.61 schwarze 2364: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2365: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.130 schwarze 2366: sql = "SELECT bits FROM names where pageid = ?";
2367: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.90 schwarze 2368: sql = "INSERT INTO names "
2369: "(bits,name,pageid) VALUES (?,?,?)";
2370: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2371: sql = "INSERT INTO keys "
2372: "(bits,key,pageid) VALUES (?,?,?)";
2373: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2374:
1.47 schwarze 2375: /*
2376: * When opening a new database, we can turn off
2377: * synchronous mode for much better performance.
2378: */
1.6 schwarze 2379:
1.103 schwarze 2380: if (real && SQLITE_OK != sqlite3_exec(db,
2381: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2382: exitcode = (int)MANDOCLEVEL_SYSERR;
2383: say(MANDOC_DB, "PRAGMA synchronous: %s",
2384: sqlite3_errmsg(db));
2385: sqlite3_close(db);
2386: return(0);
2387: }
1.11 schwarze 2388:
1.47 schwarze 2389: return(1);
2390: }
1.6 schwarze 2391:
1.47 schwarze 2392: static void *
1.106 espie 2393: hash_calloc(size_t n, size_t sz, void *arg)
1.47 schwarze 2394: {
1.6 schwarze 2395:
1.106 espie 2396: return(mandoc_calloc(n, sz));
1.47 schwarze 2397: }
1.2 schwarze 2398:
1.47 schwarze 2399: static void *
2400: hash_alloc(size_t sz, void *arg)
2401: {
1.28 schwarze 2402:
1.47 schwarze 2403: return(mandoc_malloc(sz));
2404: }
1.6 schwarze 2405:
1.47 schwarze 2406: static void
1.106 espie 2407: hash_free(void *p, void *arg)
1.47 schwarze 2408: {
1.26 schwarze 2409:
1.47 schwarze 2410: free(p);
2411: }
1.6 schwarze 2412:
1.47 schwarze 2413: static int
1.120 schwarze 2414: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2415: {
2416: static char startdir[PATH_MAX];
1.110 schwarze 2417: static int getcwd_status; /* 1 = ok, 2 = failure */
2418: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2419: char *cp;
1.6 schwarze 2420:
1.47 schwarze 2421: /*
1.110 schwarze 2422: * Remember the original working directory, if possible.
2423: * This will be needed if the second or a later directory
2424: * on the command line is given as a relative path.
2425: * Do not error out if the current directory is not
2426: * searchable: Maybe it won't be needed after all.
2427: */
2428: if (0 == getcwd_status) {
2429: if (NULL == getcwd(startdir, sizeof(startdir))) {
2430: getcwd_status = 2;
2431: (void)strlcpy(startdir, strerror(errno),
2432: sizeof(startdir));
2433: } else
2434: getcwd_status = 1;
2435: }
2436:
2437: /*
2438: * We are leaving the old base directory.
2439: * Do not use it any longer, not even for messages.
2440: */
2441: *basedir = '\0';
2442:
2443: /*
2444: * If and only if the directory was changed earlier and
2445: * the next directory to process is given as a relative path,
2446: * first go back, or bail out if that is impossible.
1.47 schwarze 2447: */
1.110 schwarze 2448: if (chdir_status && '/' != *targetdir) {
2449: if (2 == getcwd_status) {
1.47 schwarze 2450: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2451: say("", "getcwd: %s", startdir);
1.47 schwarze 2452: return(0);
2453: }
1.110 schwarze 2454: if (-1 == chdir(startdir)) {
1.47 schwarze 2455: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2456: say("", "&chdir %s", startdir);
1.47 schwarze 2457: return(0);
1.2 schwarze 2458: }
2459: }
1.110 schwarze 2460:
2461: /*
2462: * Always resolve basedir to the canonicalized absolute
2463: * pathname and append a trailing slash, such that
2464: * we can reliably check whether files are inside.
2465: */
1.47 schwarze 2466: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2467: if (report_baddir || errno != ENOENT) {
2468: exitcode = (int)MANDOCLEVEL_BADARG;
2469: say("", "&%s: realpath", targetdir);
2470: }
1.47 schwarze 2471: return(0);
2472: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2473: if (report_baddir || errno != ENOENT) {
2474: exitcode = (int)MANDOCLEVEL_BADARG;
2475: say("", "&chdir");
2476: }
1.47 schwarze 2477: return(0);
1.109 schwarze 2478: }
1.110 schwarze 2479: chdir_status = 1;
1.109 schwarze 2480: cp = strchr(basedir, '\0');
2481: if ('/' != cp[-1]) {
2482: if (cp - basedir >= PATH_MAX - 1) {
2483: exitcode = (int)MANDOCLEVEL_SYSERR;
2484: say("", "Filename too long");
2485: return(0);
2486: }
2487: *cp++ = '/';
2488: *cp = '\0';
1.47 schwarze 2489: }
2490: return(1);
1.2 schwarze 2491: }
2492:
2493: static void
1.47 schwarze 2494: say(const char *file, const char *format, ...)
1.2 schwarze 2495: {
1.47 schwarze 2496: va_list ap;
1.80 schwarze 2497: int use_errno;
1.2 schwarze 2498:
1.47 schwarze 2499: if ('\0' != *basedir)
2500: fprintf(stderr, "%s", basedir);
2501: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2502: fputc('/', stderr);
1.47 schwarze 2503: if ('\0' != *file)
2504: fprintf(stderr, "%s", file);
1.31 schwarze 2505:
1.80 schwarze 2506: use_errno = 1;
2507: if (NULL != format) {
2508: switch (*format) {
1.98 schwarze 2509: case '&':
1.80 schwarze 2510: format++;
2511: break;
1.98 schwarze 2512: case '\0':
1.80 schwarze 2513: format = NULL;
2514: break;
2515: default:
2516: use_errno = 0;
2517: break;
2518: }
2519: }
2520: if (NULL != format) {
2521: if ('\0' != *basedir || '\0' != *file)
2522: fputs(": ", stderr);
2523: va_start(ap, format);
2524: vfprintf(stderr, format, ap);
2525: va_end(ap);
2526: }
2527: if (use_errno) {
2528: if ('\0' != *basedir || '\0' != *file || NULL != format)
2529: fputs(": ", stderr);
1.47 schwarze 2530: perror(NULL);
1.80 schwarze 2531: } else
2532: fputc('\n', stderr);
1.1 schwarze 2533: }