Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.134
1.134 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.133 2014/12/05 16:17:56 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;
343: mpages_info.free = mlinks_info.free = hash_free;
1.47 schwarze 344:
345: mpages_info.key_offset = offsetof(struct mpage, inodev);
346: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 347:
348: progname = strrchr(argv[0], '/');
349: if (progname == NULL)
350: progname = argv[0];
351: else
352: ++progname;
353:
1.47 schwarze 354: /*
1.98 schwarze 355: * We accept a few different invocations.
1.47 schwarze 356: * The CHECKOP macro makes sure that invocation styles don't
357: * clobber each other.
358: */
359: #define CHECKOP(_op, _ch) do \
360: if (OP_DEFAULT != (_op)) { \
1.111 schwarze 361: fprintf(stderr, "%s: -%c: Conflicting option\n", \
362: progname, (_ch)); \
1.47 schwarze 363: goto usage; \
364: } while (/*CONSTCOND*/0)
1.10 schwarze 365:
1.47 schwarze 366: path_arg = NULL;
1.28 schwarze 367: op = OP_DEFAULT;
1.1 schwarze 368:
1.83 schwarze 369: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 370: switch (ch) {
1.98 schwarze 371: case 'a':
1.6 schwarze 372: use_all = 1;
373: break;
1.98 schwarze 374: case 'C':
1.47 schwarze 375: CHECKOP(op, ch);
376: path_arg = optarg;
1.28 schwarze 377: op = OP_CONFFILE;
1.25 schwarze 378: break;
1.98 schwarze 379: case 'D':
1.82 schwarze 380: debug++;
381: break;
1.98 schwarze 382: case 'd':
1.47 schwarze 383: CHECKOP(op, ch);
384: path_arg = optarg;
1.2 schwarze 385: op = OP_UPDATE;
1.1 schwarze 386: break;
1.98 schwarze 387: case 'n':
1.47 schwarze 388: nodb = 1;
389: break;
1.98 schwarze 390: case 'p':
1.83 schwarze 391: warnings = 1;
392: break;
1.98 schwarze 393: case 'Q':
1.73 schwarze 394: mparse_options |= MPARSE_QUICK;
1.59 schwarze 395: break;
1.98 schwarze 396: case 'T':
1.52 schwarze 397: if (strcmp(optarg, "utf8")) {
1.111 schwarze 398: fprintf(stderr, "%s: -T%s: "
399: "Unsupported output format\n",
400: progname, optarg);
1.52 schwarze 401: goto usage;
402: }
403: write_utf8 = 1;
404: break;
1.98 schwarze 405: case 't':
1.47 schwarze 406: CHECKOP(op, ch);
1.28 schwarze 407: dup2(STDOUT_FILENO, STDERR_FILENO);
408: op = OP_TEST;
1.47 schwarze 409: nodb = warnings = 1;
1.28 schwarze 410: break;
1.98 schwarze 411: case 'u':
1.47 schwarze 412: CHECKOP(op, ch);
413: path_arg = optarg;
1.1 schwarze 414: op = OP_DELETE;
415: break;
1.98 schwarze 416: case 'v':
1.83 schwarze 417: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 418: break;
1.1 schwarze 419: default:
1.28 schwarze 420: goto usage;
1.1 schwarze 421: }
422:
423: argc -= optind;
424: argv += optind;
425:
1.28 schwarze 426: if (OP_CONFFILE == op && argc > 0) {
1.111 schwarze 427: fprintf(stderr, "%s: -C: Too many arguments\n",
428: progname);
1.28 schwarze 429: goto usage;
430: }
431:
1.47 schwarze 432: exitcode = (int)MANDOCLEVEL_OK;
1.131 schwarze 433: mchars = mchars_alloc();
1.123 schwarze 434: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL,
1.131 schwarze 435: mchars, NULL);
1.47 schwarze 436: ohash_init(&mpages, 6, &mpages_info);
437: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 438:
1.47 schwarze 439: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.2 schwarze 440:
1.47 schwarze 441: /*
1.109 schwarze 442: * Most of these deal with a specific directory.
1.97 schwarze 443: * Jump into that directory first.
1.47 schwarze 444: */
1.120 schwarze 445: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.47 schwarze 446: goto out;
1.97 schwarze 447:
1.95 schwarze 448: if (dbopen(1)) {
1.97 schwarze 449: /*
450: * The existing database is usable. Process
451: * all files specified on the command-line.
452: */
453: use_all = 1;
1.95 schwarze 454: for (i = 0; i < argc; i++)
455: filescan(argv[i]);
456: if (OP_TEST != op)
457: dbprune();
458: } else {
459: /*
460: * Database missing or corrupt.
461: * Recreate from scratch.
462: */
1.97 schwarze 463: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 464: op = OP_DEFAULT;
465: if (0 == treescan())
466: goto out;
467: if (0 == dbopen(0))
468: goto out;
469: }
1.47 schwarze 470: if (OP_DELETE != op)
1.131 schwarze 471: mpages_merge(mp);
1.95 schwarze 472: dbclose(OP_DEFAULT == op ? 0 : 1);
1.47 schwarze 473: } else {
474: /*
475: * If we have arguments, use them as our manpaths.
476: * If we don't, grok from manpath(1) or however else
477: * manpath_parse() wants to do it.
478: */
479: if (argc > 0) {
1.101 schwarze 480: dirs.paths = mandoc_reallocarray(NULL,
481: argc, sizeof(char *));
1.47 schwarze 482: dirs.sz = (size_t)argc;
483: for (i = 0; i < argc; i++)
484: dirs.paths[i] = mandoc_strdup(argv[i]);
485: } else
486: manpath_parse(&dirs, path_arg, NULL, NULL);
1.84 schwarze 487:
488: if (0 == dirs.sz) {
489: exitcode = (int)MANDOCLEVEL_BADARG;
490: say("", "Empty manpath");
491: }
1.2 schwarze 492:
1.47 schwarze 493: /*
494: * First scan the tree rooted at a base directory, then
495: * build a new database and finally move it into place.
496: * Ignore zero-length directories and strip trailing
497: * slashes.
498: */
499: for (j = 0; j < dirs.sz; j++) {
500: sz = strlen(dirs.paths[j]);
501: if (sz && '/' == dirs.paths[j][sz - 1])
502: dirs.paths[j][--sz] = '\0';
503: if (0 == sz)
504: continue;
1.2 schwarze 505:
1.47 schwarze 506: if (j) {
507: ohash_init(&mpages, 6, &mpages_info);
508: ohash_init(&mlinks, 6, &mlinks_info);
509: }
1.1 schwarze 510:
1.120 schwarze 511: if (0 == set_basedir(dirs.paths[j], argc > 0))
512: continue;
1.47 schwarze 513: if (0 == treescan())
1.120 schwarze 514: continue;
1.47 schwarze 515: if (0 == dbopen(0))
1.120 schwarze 516: continue;
1.41 deraadt 517:
1.131 schwarze 518: mpages_merge(mp);
1.104 schwarze 519: if (warnings && !nodb &&
1.87 schwarze 520: ! (MPARSE_QUICK & mparse_options))
521: names_check();
1.47 schwarze 522: dbclose(0);
1.28 schwarze 523:
1.47 schwarze 524: if (j + 1 < dirs.sz) {
525: mpages_free();
526: ohash_delete(&mpages);
527: ohash_delete(&mlinks);
528: }
1.2 schwarze 529: }
1.47 schwarze 530: }
531: out:
532: manpath_free(&dirs);
1.123 schwarze 533: mparse_free(mp);
1.131 schwarze 534: mchars_free(mchars);
1.47 schwarze 535: mpages_free();
536: ohash_delete(&mpages);
537: ohash_delete(&mlinks);
538: return(exitcode);
539: usage:
1.83 schwarze 540: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
541: " %s [-aDnpQ] [-Tutf8] dir ...\n"
542: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
543: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 544: " %s [-Q] -t file ...\n",
1.98 schwarze 545: progname, progname, progname,
1.47 schwarze 546: progname, progname);
1.1 schwarze 547:
1.47 schwarze 548: return((int)MANDOCLEVEL_BADARG);
549: }
1.1 schwarze 550:
1.47 schwarze 551: /*
552: * Scan a directory tree rooted at "basedir" for manpages.
553: * We use fts(), scanning directory parts along the way for clues to our
554: * section and architecture.
555: *
556: * If use_all has been specified, grok all files.
557: * If not, sanitise paths to the following:
558: *
1.98 schwarze 559: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 560: * or
561: * [./]cat<section>[/<arch>]/<name>.0
562: *
563: * TODO: accomodate for multi-language directories.
564: */
565: static int
566: treescan(void)
567: {
1.96 schwarze 568: char buf[PATH_MAX];
1.47 schwarze 569: FTS *f;
570: FTSENT *ff;
571: struct mlink *mlink;
1.81 schwarze 572: int dform, gzip;
1.51 schwarze 573: char *dsec, *arch, *fsec, *cp;
574: const char *path;
1.47 schwarze 575: const char *argv[2];
1.1 schwarze 576:
1.47 schwarze 577: argv[0] = ".";
578: argv[1] = (char *)NULL;
1.1 schwarze 579:
1.96 schwarze 580: f = fts_open((char * const *)argv,
581: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.47 schwarze 582: if (NULL == f) {
583: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 584: say("", "&fts_open");
1.47 schwarze 585: return(0);
586: }
1.2 schwarze 587:
1.47 schwarze 588: dsec = arch = NULL;
589: dform = FORM_NONE;
1.2 schwarze 590:
1.47 schwarze 591: while (NULL != (ff = fts_read(f))) {
592: path = ff->fts_path + 2;
1.96 schwarze 593: switch (ff->fts_info) {
594:
595: /*
596: * Symbolic links require various sanity checks,
597: * then get handled just like regular files.
598: */
1.98 schwarze 599: case FTS_SL:
1.96 schwarze 600: if (NULL == realpath(path, buf)) {
601: if (warnings)
602: say(path, "&realpath");
603: continue;
604: }
605: if (strstr(buf, basedir) != buf) {
606: if (warnings) say("",
607: "%s: outside base directory", buf);
608: continue;
609: }
610: /* Use logical inode to avoid mpages dupe. */
611: if (-1 == stat(path, ff->fts_statp)) {
612: if (warnings)
613: say(path, "&stat");
614: continue;
615: }
616: /* FALLTHROUGH */
617:
1.15 schwarze 618: /*
1.47 schwarze 619: * If we're a regular file, add an mlink by using the
620: * stored directory data and handling the filename.
1.15 schwarze 621: */
1.98 schwarze 622: case FTS_F:
1.47 schwarze 623: if (0 == strcmp(path, MANDOC_DB))
624: continue;
625: if ( ! use_all && ff->fts_level < 2) {
626: if (warnings)
627: say(path, "Extraneous file");
628: continue;
1.81 schwarze 629: }
630: gzip = 0;
631: fsec = NULL;
632: while (NULL == fsec) {
633: fsec = strrchr(ff->fts_name, '.');
634: if (NULL == fsec || strcmp(fsec+1, "gz"))
635: break;
636: gzip = 1;
637: *fsec = '\0';
638: fsec = NULL;
639: }
640: if (NULL == fsec) {
1.47 schwarze 641: if ( ! use_all) {
642: if (warnings)
643: say(path,
644: "No filename suffix");
645: continue;
646: }
647: } else if (0 == strcmp(++fsec, "html")) {
648: if (warnings)
649: say(path, "Skip html");
650: continue;
651: } else if (0 == strcmp(fsec, "ps")) {
652: if (warnings)
653: say(path, "Skip ps");
654: continue;
655: } else if (0 == strcmp(fsec, "pdf")) {
656: if (warnings)
657: say(path, "Skip pdf");
658: continue;
659: } else if ( ! use_all &&
660: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
661: (FORM_CAT == dform && strcmp(fsec, "0")))) {
662: if (warnings)
663: say(path, "Wrong filename suffix");
664: continue;
665: } else
666: fsec[-1] = '\0';
1.51 schwarze 667:
1.47 schwarze 668: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 669: if (strlcpy(mlink->file, path,
670: sizeof(mlink->file)) >=
671: sizeof(mlink->file)) {
672: say(path, "Filename too long");
673: free(mlink);
674: continue;
675: }
1.47 schwarze 676: mlink->dform = dform;
1.51 schwarze 677: mlink->dsec = dsec;
678: mlink->arch = arch;
679: mlink->name = ff->fts_name;
680: mlink->fsec = fsec;
1.81 schwarze 681: mlink->gzip = gzip;
1.47 schwarze 682: mlink_add(mlink, ff->fts_statp);
683: continue;
1.96 schwarze 684:
1.98 schwarze 685: case FTS_D:
1.96 schwarze 686: /* FALLTHROUGH */
1.98 schwarze 687: case FTS_DP:
1.96 schwarze 688: break;
689:
690: default:
1.47 schwarze 691: if (warnings)
692: say(path, "Not a regular file");
693: continue;
694: }
695:
696: switch (ff->fts_level) {
1.98 schwarze 697: case 0:
1.47 schwarze 698: /* Ignore the root directory. */
699: break;
1.98 schwarze 700: case 1:
1.47 schwarze 701: /*
702: * This might contain manX/ or catX/.
703: * Try to infer this from the name.
704: * If we're not in use_all, enforce it.
705: */
706: cp = ff->fts_name;
707: if (FTS_DP == ff->fts_info)
708: break;
1.15 schwarze 709:
1.47 schwarze 710: if (0 == strncmp(cp, "man", 3)) {
711: dform = FORM_SRC;
712: dsec = cp + 3;
713: } else if (0 == strncmp(cp, "cat", 3)) {
714: dform = FORM_CAT;
715: dsec = cp + 3;
1.51 schwarze 716: } else {
717: dform = FORM_NONE;
718: dsec = NULL;
1.26 schwarze 719: }
1.47 schwarze 720:
1.98 schwarze 721: if (NULL != dsec || use_all)
1.47 schwarze 722: break;
723:
724: if (warnings)
725: say(path, "Unknown directory part");
726: fts_set(f, ff, FTS_SKIP);
727: break;
1.98 schwarze 728: case 2:
1.47 schwarze 729: /*
730: * Possibly our architecture.
731: * If we're descending, keep tabs on it.
732: */
733: if (FTS_DP != ff->fts_info && NULL != dsec)
734: arch = ff->fts_name;
1.51 schwarze 735: else
736: arch = NULL;
1.47 schwarze 737: break;
738: default:
739: if (FTS_DP == ff->fts_info || use_all)
740: break;
741: if (warnings)
742: say(path, "Extraneous directory part");
743: fts_set(f, ff, FTS_SKIP);
744: break;
1.14 schwarze 745: }
1.47 schwarze 746: }
747:
748: fts_close(f);
749: return(1);
750: }
1.1 schwarze 751:
1.47 schwarze 752: /*
753: * Add a file to the mlinks table.
754: * Do not verify that it's a "valid" looking manpage (we'll do that
755: * later).
756: *
757: * Try to infer the manual section, architecture, and page name from the
758: * path, assuming it looks like
759: *
1.98 schwarze 760: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 761: * or
762: * [./]cat<section>[/<arch>]/<name>.0
763: *
764: * See treescan() for the fts(3) version of this.
765: */
766: static void
767: filescan(const char *file)
768: {
769: char buf[PATH_MAX];
770: struct stat st;
771: struct mlink *mlink;
772: char *p, *start;
773:
774: assert(use_all);
775:
776: if (0 == strncmp(file, "./", 2))
777: file += 2;
778:
1.96 schwarze 779: /*
780: * We have to do lstat(2) before realpath(3) loses
781: * the information whether this is a symbolic link.
782: * We need to know that because for symbolic links,
783: * we want to use the orginal file name, while for
784: * regular files, we want to use the real path.
785: */
786: if (-1 == lstat(file, &st)) {
787: exitcode = (int)MANDOCLEVEL_BADARG;
788: say(file, "&lstat");
789: return;
790: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
791: exitcode = (int)MANDOCLEVEL_BADARG;
792: say(file, "Not a regular file");
793: return;
794: }
795:
796: /*
797: * We have to resolve the file name to the real path
798: * in any case for the base directory check.
799: */
1.47 schwarze 800: if (NULL == realpath(file, buf)) {
801: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 802: say(file, "&realpath");
1.47 schwarze 803: return;
1.63 schwarze 804: }
805:
1.109 schwarze 806: if (OP_TEST == op)
1.63 schwarze 807: start = buf;
1.109 schwarze 808: else if (strstr(buf, basedir) == buf)
809: start = buf + strlen(basedir);
1.63 schwarze 810: else {
1.47 schwarze 811: exitcode = (int)MANDOCLEVEL_BADARG;
812: say("", "%s: outside base directory", buf);
813: return;
1.63 schwarze 814: }
815:
1.96 schwarze 816: /*
817: * Now we are sure the file is inside our tree.
818: * If it is a symbolic link, ignore the real path
819: * and use the original name.
820: * This implies passing stuff like "cat1/../man1/foo.1"
821: * on the command line won't work. So don't do that.
822: * Note the stat(2) can still fail if the link target
823: * doesn't exist.
824: */
825: if (S_IFLNK & st.st_mode) {
826: if (-1 == stat(buf, &st)) {
827: exitcode = (int)MANDOCLEVEL_BADARG;
828: say(file, "&stat");
829: return;
830: }
1.100 schwarze 831: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
832: say(file, "Filename too long");
833: return;
834: }
1.109 schwarze 835: start = buf;
836: if (OP_TEST != op && strstr(buf, basedir) == buf)
837: start += strlen(basedir);
1.1 schwarze 838: }
1.63 schwarze 839:
1.47 schwarze 840: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 841: mlink->dform = FORM_NONE;
1.100 schwarze 842: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
843: sizeof(mlink->file)) {
844: say(start, "Filename too long");
1.134 ! schwarze 845: free(mlink);
1.100 schwarze 846: return;
847: }
1.1 schwarze 848:
1.10 schwarze 849: /*
1.47 schwarze 850: * First try to guess our directory structure.
851: * If we find a separator, try to look for man* or cat*.
852: * If we find one of these and what's underneath is a directory,
853: * assume it's an architecture.
1.10 schwarze 854: */
1.47 schwarze 855: if (NULL != (p = strchr(start, '/'))) {
856: *p++ = '\0';
857: if (0 == strncmp(start, "man", 3)) {
858: mlink->dform = FORM_SRC;
1.51 schwarze 859: mlink->dsec = start + 3;
1.47 schwarze 860: } else if (0 == strncmp(start, "cat", 3)) {
861: mlink->dform = FORM_CAT;
1.51 schwarze 862: mlink->dsec = start + 3;
1.47 schwarze 863: }
1.10 schwarze 864:
1.47 schwarze 865: start = p;
866: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
867: *p++ = '\0';
1.51 schwarze 868: mlink->arch = start;
1.47 schwarze 869: start = p;
1.41 deraadt 870: }
1.47 schwarze 871: }
1.7 schwarze 872:
1.47 schwarze 873: /*
874: * Now check the file suffix.
875: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
876: */
877: p = strrchr(start, '\0');
878: while (p-- > start && '/' != *p && '.' != *p)
879: /* Loop. */ ;
880:
881: if ('.' == *p) {
882: *p++ = '\0';
1.51 schwarze 883: mlink->fsec = p;
1.47 schwarze 884: }
1.41 deraadt 885:
1.47 schwarze 886: /*
887: * Now try to parse the name.
888: * Use the filename portion of the path.
889: */
890: mlink->name = start;
891: if (NULL != (p = strrchr(start, '/'))) {
892: mlink->name = p + 1;
893: *p = '\0';
894: }
895: mlink_add(mlink, &st);
896: }
1.2 schwarze 897:
1.47 schwarze 898: static void
899: mlink_add(struct mlink *mlink, const struct stat *st)
900: {
901: struct inodev inodev;
902: struct mpage *mpage;
903: unsigned int slot;
904:
905: assert(NULL != mlink->file);
906:
1.51 schwarze 907: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
908: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
909: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
910: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 911:
912: if ('0' == *mlink->fsec) {
913: free(mlink->fsec);
914: mlink->fsec = mandoc_strdup(mlink->dsec);
915: mlink->fform = FORM_CAT;
916: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
917: mlink->fform = FORM_SRC;
918: else
919: mlink->fform = FORM_NONE;
1.1 schwarze 920:
1.47 schwarze 921: slot = ohash_qlookup(&mlinks, mlink->file);
922: assert(NULL == ohash_find(&mlinks, slot));
923: ohash_insert(&mlinks, slot, mlink);
924:
925: inodev.st_ino = st->st_ino;
926: inodev.st_dev = st->st_dev;
927: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
928: sizeof(struct inodev), inodev.st_ino);
929: mpage = ohash_find(&mpages, slot);
930: if (NULL == mpage) {
931: mpage = mandoc_calloc(1, sizeof(struct mpage));
932: mpage->inodev.st_ino = inodev.st_ino;
933: mpage->inodev.st_dev = inodev.st_dev;
934: ohash_insert(&mpages, slot, mpage);
935: } else
936: mlink->next = mpage->mlinks;
937: mpage->mlinks = mlink;
1.75 schwarze 938: mlink->mpage = mpage;
1.47 schwarze 939: }
1.1 schwarze 940:
1.47 schwarze 941: static void
942: mlink_free(struct mlink *mlink)
943: {
1.1 schwarze 944:
1.47 schwarze 945: free(mlink->dsec);
946: free(mlink->arch);
947: free(mlink->name);
948: free(mlink->fsec);
949: free(mlink);
950: }
1.1 schwarze 951:
1.47 schwarze 952: static void
953: mpages_free(void)
954: {
955: struct mpage *mpage;
956: struct mlink *mlink;
957: unsigned int slot;
958:
959: mpage = ohash_first(&mpages, &slot);
960: while (NULL != mpage) {
961: while (NULL != (mlink = mpage->mlinks)) {
962: mpage->mlinks = mlink->next;
963: mlink_free(mlink);
964: }
965: free(mpage->sec);
966: free(mpage->arch);
967: free(mpage->title);
968: free(mpage->desc);
969: free(mpage);
970: mpage = ohash_next(&mpages, &slot);
971: }
972: }
1.1 schwarze 973:
1.47 schwarze 974: /*
975: * For each mlink to the mpage, check whether the path looks like
976: * it is formatted, and if it does, check whether a source manual
977: * exists by the same name, ignoring the suffix.
978: * If both conditions hold, drop the mlink.
979: */
980: static void
981: mlinks_undupe(struct mpage *mpage)
982: {
983: char buf[PATH_MAX];
984: struct mlink **prev;
985: struct mlink *mlink;
986: char *bufp;
987:
988: mpage->form = FORM_CAT;
989: prev = &mpage->mlinks;
990: while (NULL != (mlink = *prev)) {
991: if (FORM_CAT != mlink->dform) {
992: mpage->form = FORM_NONE;
993: goto nextlink;
994: }
1.100 schwarze 995: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 996: bufp = strstr(buf, "cat");
997: assert(NULL != bufp);
998: memcpy(bufp, "man", 3);
999: if (NULL != (bufp = strrchr(buf, '.')))
1000: *++bufp = '\0';
1.100 schwarze 1001: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 1002: if (NULL == ohash_find(&mlinks,
1.98 schwarze 1003: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 1004: goto nextlink;
1005: if (warnings)
1006: say(mlink->file, "Man source exists: %s", buf);
1007: if (use_all)
1008: goto nextlink;
1009: *prev = mlink->next;
1010: mlink_free(mlink);
1011: continue;
1012: nextlink:
1013: prev = &(*prev)->next;
1.1 schwarze 1014: }
1.47 schwarze 1015: }
1.1 schwarze 1016:
1.88 schwarze 1017: static void
1.50 schwarze 1018: mlink_check(struct mpage *mpage, struct mlink *mlink)
1019: {
1.88 schwarze 1020: struct str *str;
1021: unsigned int slot;
1.50 schwarze 1022:
1023: /*
1024: * Check whether the manual section given in a file
1025: * agrees with the directory where the file is located.
1026: * Some manuals have suffixes like (3p) on their
1027: * section number either inside the file or in the
1028: * directory name, some are linked into more than one
1029: * section, like encrypt(1) = makekey(8).
1030: */
1031:
1032: if (FORM_SRC == mpage->form &&
1.88 schwarze 1033: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1034: say(mlink->file, "Section \"%s\" manual in %s directory",
1035: mpage->sec, mlink->dsec);
1036:
1037: /*
1038: * Manual page directories exist for each kernel
1039: * architecture as returned by machine(1).
1040: * However, many manuals only depend on the
1041: * application architecture as returned by arch(1).
1042: * For example, some (2/ARM) manuals are shared
1043: * across the "armish" and "zaurus" kernel
1044: * architectures.
1045: * A few manuals are even shared across completely
1046: * different architectures, for example fdformat(1)
1047: * on amd64, i386, sparc, and sparc64.
1048: */
1049:
1.88 schwarze 1050: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1051: say(mlink->file, "Architecture \"%s\" manual in "
1052: "\"%s\" directory", mpage->arch, mlink->arch);
1053:
1.88 schwarze 1054: /*
1055: * XXX
1.90 schwarze 1056: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1057: */
1058:
1059: if (FORM_CAT == mpage->form)
1060: return;
1061:
1062: /*
1063: * Check whether this mlink
1064: * appears as a name in the NAME section.
1065: */
1.50 schwarze 1066:
1.90 schwarze 1067: slot = ohash_qlookup(&names, mlink->name);
1068: str = ohash_find(&names, slot);
1.88 schwarze 1069: assert(NULL != str);
1.90 schwarze 1070: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1071: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1072: }
1073:
1.47 schwarze 1074: /*
1075: * Run through the files in the global vector "mpages"
1076: * and add them to the database specified in "basedir".
1077: *
1078: * This handles the parsing scheme itself, using the cues of directory
1079: * and filename to determine whether the file is parsable or not.
1080: */
1081: static void
1.131 schwarze 1082: mpages_merge(struct mparse *mp)
1.47 schwarze 1083: {
1.70 schwarze 1084: char any[] = "any";
1.58 schwarze 1085: struct ohash_info str_info;
1.75 schwarze 1086: struct mpage *mpage, *mpage_dest;
1087: struct mlink *mlink, *mlink_dest;
1.47 schwarze 1088: struct mdoc *mdoc;
1089: struct man *man;
1.75 schwarze 1090: char *sodest;
1.69 schwarze 1091: char *cp;
1.117 schwarze 1092: int fd;
1.58 schwarze 1093: unsigned int pslot;
1.47 schwarze 1094: enum mandoclevel lvl;
1095:
1096: str_info.alloc = hash_alloc;
1.106 espie 1097: str_info.calloc = hash_calloc;
1098: str_info.free = hash_free;
1.47 schwarze 1099: str_info.key_offset = offsetof(struct str, key);
1100:
1.126 schwarze 1101: if ( ! nodb)
1.64 schwarze 1102: SQL_EXEC("BEGIN TRANSACTION");
1103:
1.47 schwarze 1104: mpage = ohash_first(&mpages, &pslot);
1.126 schwarze 1105: while (mpage != NULL) {
1.47 schwarze 1106: mlinks_undupe(mpage);
1.126 schwarze 1107: if (mpage->mlinks == NULL) {
1.47 schwarze 1108: mpage = ohash_next(&mpages, &pslot);
1109: continue;
1110: }
1.1 schwarze 1111:
1.90 schwarze 1112: name_mask = NAME_MASK;
1113: ohash_init(&names, 4, &str_info);
1.47 schwarze 1114: ohash_init(&strings, 6, &str_info);
1115: mparse_reset(mp);
1116: mdoc = NULL;
1117: man = NULL;
1.81 schwarze 1118: sodest = NULL;
1119:
1.125 schwarze 1120: mparse_open(mp, &fd, mpage->mlinks->file);
1.117 schwarze 1121: if (fd == -1) {
1122: say(mpage->mlinks->file, "&open");
1123: goto nextpage;
1.81 schwarze 1124: }
1.11 schwarze 1125:
1126: /*
1.24 schwarze 1127: * Try interpreting the file as mdoc(7) or man(7)
1128: * source code, unless it is already known to be
1129: * formatted. Fall back to formatted mode.
1.11 schwarze 1130: */
1.126 schwarze 1131: if (mpage->mlinks->dform != FORM_CAT ||
1132: mpage->mlinks->fform != FORM_CAT) {
1.117 schwarze 1133: lvl = mparse_readfd(mp, fd, mpage->mlinks->file);
1.47 schwarze 1134: if (lvl < MANDOCLEVEL_FATAL)
1.75 schwarze 1135: mparse_result(mp, &mdoc, &man, &sodest);
1.47 schwarze 1136: }
1.11 schwarze 1137:
1.126 schwarze 1138: if (sodest != NULL) {
1.75 schwarze 1139: mlink_dest = ohash_find(&mlinks,
1140: ohash_qlookup(&mlinks, sodest));
1.126 schwarze 1141: if (mlink_dest == NULL) {
1142: mandoc_asprintf(&cp, "%s.gz", sodest);
1143: mlink_dest = ohash_find(&mlinks,
1144: ohash_qlookup(&mlinks, cp));
1145: free(cp);
1146: }
1147: if (mlink_dest != NULL) {
1.75 schwarze 1148:
1149: /* The .so target exists. */
1150:
1151: mpage_dest = mlink_dest->mpage;
1152: mlink = mpage->mlinks;
1153: while (1) {
1154: mlink->mpage = mpage_dest;
1155:
1156: /*
1157: * If the target was already
1158: * processed, add the links
1159: * to the database now.
1160: * Otherwise, this will
1161: * happen when we come
1162: * to the target.
1163: */
1164:
1.94 schwarze 1165: if (mpage_dest->pageid)
1.124 schwarze 1166: dbadd_mlink_name(mlink);
1.75 schwarze 1167:
1.126 schwarze 1168: if (mlink->next == NULL)
1.75 schwarze 1169: break;
1170: mlink = mlink->next;
1171: }
1172:
1173: /* Move all links to the target. */
1174:
1175: mlink->next = mlink_dest->next;
1176: mlink_dest->next = mpage->mlinks;
1177: mpage->mlinks = NULL;
1178: }
1.81 schwarze 1179: goto nextpage;
1.126 schwarze 1180: } else if (mdoc != NULL) {
1.47 schwarze 1181: mpage->form = FORM_SRC;
1.112 schwarze 1182: mpage->sec = mdoc_meta(mdoc)->msec;
1183: mpage->sec = mandoc_strdup(
1.126 schwarze 1184: mpage->sec == NULL ? "" : mpage->sec);
1.47 schwarze 1185: mpage->arch = mdoc_meta(mdoc)->arch;
1186: mpage->arch = mandoc_strdup(
1.126 schwarze 1187: mpage->arch == NULL ? "" : mpage->arch);
1.47 schwarze 1188: mpage->title =
1189: mandoc_strdup(mdoc_meta(mdoc)->title);
1.126 schwarze 1190: } else if (man != NULL) {
1.47 schwarze 1191: mpage->form = FORM_SRC;
1192: mpage->sec =
1193: mandoc_strdup(man_meta(man)->msec);
1194: mpage->arch =
1195: mandoc_strdup(mpage->mlinks->arch);
1196: mpage->title =
1197: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1198: } else {
1.47 schwarze 1199: mpage->form = FORM_CAT;
1200: mpage->sec =
1201: mandoc_strdup(mpage->mlinks->dsec);
1202: mpage->arch =
1203: mandoc_strdup(mpage->mlinks->arch);
1204: mpage->title =
1205: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1206: }
1.54 schwarze 1207: putkey(mpage, mpage->sec, TYPE_sec);
1.119 schwarze 1208: if (*mpage->arch != '\0')
1209: putkey(mpage, mpage->arch, TYPE_arch);
1.1 schwarze 1210:
1.54 schwarze 1211: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1212: if ('\0' != *mlink->dsec)
1213: putkey(mpage, mlink->dsec, TYPE_sec);
1214: if ('\0' != *mlink->fsec)
1215: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1216: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1217: any : mlink->arch, TYPE_arch);
1.90 schwarze 1218: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1219: }
1.41 deraadt 1220:
1.129 schwarze 1221: assert(mpage->desc == NULL);
1222: if (mdoc != NULL)
1.127 schwarze 1223: parse_mdoc(mpage, mdoc_meta(mdoc), mdoc_node(mdoc));
1.129 schwarze 1224: else if (man != NULL)
1225: parse_man(mpage, man_meta(man), man_node(man));
1.47 schwarze 1226: else
1.117 schwarze 1227: parse_cat(mpage, fd);
1.129 schwarze 1228: if (mpage->desc == NULL)
1.92 schwarze 1229: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1230:
1231: if (warnings && !use_all)
1232: for (mlink = mpage->mlinks; mlink;
1233: mlink = mlink->next)
1234: mlink_check(mpage, mlink);
1.6 schwarze 1235:
1.131 schwarze 1236: dbadd(mpage);
1.81 schwarze 1237:
1238: nextpage:
1.125 schwarze 1239: if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1.117 schwarze 1240: exitcode = (int)MANDOCLEVEL_SYSERR;
1241: say(mpage->mlinks->file, "&wait gunzip");
1.81 schwarze 1242: }
1.47 schwarze 1243: ohash_delete(&strings);
1.90 schwarze 1244: ohash_delete(&names);
1.47 schwarze 1245: mpage = ohash_next(&mpages, &pslot);
1246: }
1.64 schwarze 1247:
1248: if (0 == nodb)
1249: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1250: }
1251:
1252: static void
1253: names_check(void)
1254: {
1255: sqlite3_stmt *stmt;
1256: const char *name, *sec, *arch, *key;
1257: int irc;
1258:
1259: sqlite3_prepare_v2(db,
1260: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1261: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1262: "WHERE bits & ? AND NOT EXISTS ("
1263: "SELECT pageid FROM mlinks "
1.90 schwarze 1264: "WHERE mlinks.pageid == names.pageid "
1265: "AND mlinks.name == names.name"
1.87 schwarze 1266: ")"
1267: ") JOIN ("
1.99 schwarze 1268: "SELECT sec, arch, name, pageid FROM mlinks "
1269: "GROUP BY pageid"
1.87 schwarze 1270: ") USING (pageid);",
1271: -1, &stmt, NULL);
1272:
1.91 schwarze 1273: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1274: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1275:
1276: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.113 schwarze 1277: name = (const char *)sqlite3_column_text(stmt, 0);
1278: sec = (const char *)sqlite3_column_text(stmt, 1);
1279: arch = (const char *)sqlite3_column_text(stmt, 2);
1280: key = (const char *)sqlite3_column_text(stmt, 3);
1.87 schwarze 1281: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1282: '\0' == *arch ? "" : "/",
1283: '\0' == *arch ? "" : arch, key);
1284: }
1285: sqlite3_finalize(stmt);
1.47 schwarze 1286: }
1.6 schwarze 1287:
1.47 schwarze 1288: static void
1.81 schwarze 1289: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1290: {
1291: FILE *stream;
1292: char *line, *p, *title;
1293: size_t len, plen, titlesz;
1.1 schwarze 1294:
1.81 schwarze 1295: stream = (-1 == fd) ?
1296: fopen(mpage->mlinks->file, "r") :
1297: fdopen(fd, "r");
1298: if (NULL == stream) {
1.114 doug 1299: if (-1 != fd)
1300: close(fd);
1.47 schwarze 1301: if (warnings)
1.80 schwarze 1302: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1303: return;
1304: }
1.1 schwarze 1305:
1.47 schwarze 1306: /* Skip to first blank line. */
1.1 schwarze 1307:
1.47 schwarze 1308: while (NULL != (line = fgetln(stream, &len)))
1309: if ('\n' == *line)
1310: break;
1.1 schwarze 1311:
1.47 schwarze 1312: /*
1313: * Assume the first line that is not indented
1314: * is the first section header. Skip to it.
1315: */
1.1 schwarze 1316:
1.47 schwarze 1317: while (NULL != (line = fgetln(stream, &len)))
1318: if ('\n' != *line && ' ' != *line)
1319: break;
1.98 schwarze 1320:
1.47 schwarze 1321: /*
1322: * Read up until the next section into a buffer.
1323: * Strip the leading and trailing newline from each read line,
1324: * appending a trailing space.
1325: * Ignore empty (whitespace-only) lines.
1326: */
1.28 schwarze 1327:
1.47 schwarze 1328: titlesz = 0;
1329: title = NULL;
1.38 schwarze 1330:
1.47 schwarze 1331: while (NULL != (line = fgetln(stream, &len))) {
1332: if (' ' != *line || '\n' != line[len - 1])
1333: break;
1334: while (len > 0 && isspace((unsigned char)*line)) {
1335: line++;
1336: len--;
1337: }
1338: if (1 == len)
1339: continue;
1340: title = mandoc_realloc(title, titlesz + len);
1341: memcpy(title + titlesz, line, len);
1342: titlesz += len;
1343: title[titlesz - 1] = ' ';
1344: }
1.28 schwarze 1345:
1.47 schwarze 1346: /*
1347: * If no page content can be found, or the input line
1348: * is already the next section header, or there is no
1349: * trailing newline, reuse the page title as the page
1350: * description.
1351: */
1.1 schwarze 1352:
1.47 schwarze 1353: if (NULL == title || '\0' == *title) {
1354: if (warnings)
1355: say(mpage->mlinks->file,
1356: "Cannot find NAME section");
1357: fclose(stream);
1358: free(title);
1359: return;
1360: }
1.24 schwarze 1361:
1.47 schwarze 1362: title = mandoc_realloc(title, titlesz + 1);
1363: title[titlesz] = '\0';
1.24 schwarze 1364:
1.47 schwarze 1365: /*
1366: * Skip to the first dash.
1367: * Use the remaining line as the description (no more than 70
1368: * bytes).
1369: */
1.28 schwarze 1370:
1.47 schwarze 1371: if (NULL != (p = strstr(title, "- "))) {
1372: for (p += 2; ' ' == *p || '\b' == *p; p++)
1373: /* Skip to next word. */ ;
1374: } else {
1375: if (warnings)
1376: say(mpage->mlinks->file,
1377: "No dash in title line");
1378: p = title;
1379: }
1.1 schwarze 1380:
1.47 schwarze 1381: plen = strlen(p);
1.1 schwarze 1382:
1.47 schwarze 1383: /* Strip backspace-encoding from line. */
1.1 schwarze 1384:
1.47 schwarze 1385: while (NULL != (line = memchr(p, '\b', plen))) {
1386: len = line - p;
1387: if (0 == len) {
1388: memmove(line, line + 1, plen--);
1389: continue;
1.98 schwarze 1390: }
1.47 schwarze 1391: memmove(line - 1, line + 1, plen - len);
1392: plen -= 2;
1393: }
1.1 schwarze 1394:
1.47 schwarze 1395: mpage->desc = mandoc_strdup(p);
1396: fclose(stream);
1397: free(title);
1398: }
1.16 schwarze 1399:
1.47 schwarze 1400: /*
1401: * Put a type/word pair into the word database for this particular file.
1402: */
1403: static void
1.69 schwarze 1404: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1405: {
1.69 schwarze 1406: char *cp;
1.37 schwarze 1407:
1.47 schwarze 1408: assert(NULL != value);
1.69 schwarze 1409: if (TYPE_arch == type)
1410: for (cp = value; *cp; cp++)
1411: if (isupper((unsigned char)*cp))
1412: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1413: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1414: }
1415:
1416: /*
1.47 schwarze 1417: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1418: */
1419: static void
1.47 schwarze 1420: putmdockey(const struct mpage *mpage,
1421: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1422: {
1.16 schwarze 1423:
1.47 schwarze 1424: for ( ; NULL != n; n = n->next) {
1425: if (NULL != n->child)
1426: putmdockey(mpage, n->child, m);
1427: if (MDOC_TEXT == n->type)
1428: putkey(mpage, n->string, m);
1429: }
1430: }
1.16 schwarze 1431:
1.47 schwarze 1432: static void
1.129 schwarze 1433: parse_man(struct mpage *mpage, const struct man_meta *meta,
1434: const struct man_node *n)
1.47 schwarze 1435: {
1436: const struct man_node *head, *body;
1.78 schwarze 1437: char *start, *title;
1.47 schwarze 1438: char byte;
1.78 schwarze 1439: size_t sz;
1.16 schwarze 1440:
1.47 schwarze 1441: if (NULL == n)
1442: return;
1.16 schwarze 1443:
1.47 schwarze 1444: /*
1445: * We're only searching for one thing: the first text child in
1446: * the BODY of a NAME section. Since we don't keep track of
1447: * sections in -man, run some hoops to find out whether we're in
1448: * the correct section or not.
1449: */
1.16 schwarze 1450:
1.47 schwarze 1451: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1452: body = n;
1453: assert(body->parent);
1454: if (NULL != (head = body->parent->head) &&
1.98 schwarze 1455: 1 == head->nchild &&
1456: NULL != (head = (head->child)) &&
1457: MAN_TEXT == head->type &&
1458: 0 == strcmp(head->string, "NAME") &&
1459: NULL != body->child) {
1.2 schwarze 1460:
1.47 schwarze 1461: /*
1462: * Suck the entire NAME section into memory.
1463: * Yes, we might run away.
1464: * But too many manuals have big, spread-out
1465: * NAME sections over many lines.
1466: */
1.2 schwarze 1467:
1.78 schwarze 1468: title = NULL;
1469: man_deroff(&title, body);
1.47 schwarze 1470: if (NULL == title)
1471: return;
1.16 schwarze 1472:
1.98 schwarze 1473: /*
1.47 schwarze 1474: * Go through a special heuristic dance here.
1475: * Conventionally, one or more manual names are
1476: * comma-specified prior to a whitespace, then a
1477: * dash, then a description. Try to puzzle out
1478: * the name parts here.
1479: */
1.16 schwarze 1480:
1.78 schwarze 1481: start = title;
1.47 schwarze 1482: for ( ;; ) {
1483: sz = strcspn(start, " ,");
1484: if ('\0' == start[sz])
1485: break;
1.1 schwarze 1486:
1.47 schwarze 1487: byte = start[sz];
1488: start[sz] = '\0';
1.67 schwarze 1489:
1490: /*
1491: * Assume a stray trailing comma in the
1492: * name list if a name begins with a dash.
1493: */
1494:
1495: if ('-' == start[0] ||
1496: ('\\' == start[0] && '-' == start[1]))
1497: break;
1.1 schwarze 1498:
1.90 schwarze 1499: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1500: if ( ! (mpage->name_head_done ||
1501: strcasecmp(start, meta->title))) {
1502: putkey(mpage, start, NAME_HEAD);
1503: mpage->name_head_done = 1;
1504: }
1.1 schwarze 1505:
1.47 schwarze 1506: if (' ' == byte) {
1507: start += sz + 1;
1508: break;
1509: }
1.1 schwarze 1510:
1.47 schwarze 1511: assert(',' == byte);
1512: start += sz + 1;
1513: while (' ' == *start)
1514: start++;
1515: }
1.1 schwarze 1516:
1.78 schwarze 1517: if (start == title) {
1.90 schwarze 1518: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1519: if ( ! (mpage->name_head_done ||
1520: strcasecmp(start, meta->title))) {
1521: putkey(mpage, start, NAME_HEAD);
1522: mpage->name_head_done = 1;
1523: }
1.47 schwarze 1524: free(title);
1525: return;
1526: }
1.1 schwarze 1527:
1.47 schwarze 1528: while (isspace((unsigned char)*start))
1529: start++;
1.1 schwarze 1530:
1.47 schwarze 1531: if (0 == strncmp(start, "-", 1))
1532: start += 1;
1533: else if (0 == strncmp(start, "\\-\\-", 4))
1534: start += 4;
1535: else if (0 == strncmp(start, "\\-", 2))
1536: start += 2;
1537: else if (0 == strncmp(start, "\\(en", 4))
1538: start += 4;
1539: else if (0 == strncmp(start, "\\(em", 4))
1540: start += 4;
1.1 schwarze 1541:
1.47 schwarze 1542: while (' ' == *start)
1543: start++;
1.1 schwarze 1544:
1.47 schwarze 1545: mpage->desc = mandoc_strdup(start);
1546: free(title);
1547: return;
1548: }
1549: }
1.1 schwarze 1550:
1.47 schwarze 1551: for (n = n->child; n; n = n->next) {
1552: if (NULL != mpage->desc)
1553: break;
1.129 schwarze 1554: parse_man(mpage, meta, n);
1.1 schwarze 1555: }
1556: }
1557:
1558: static void
1.127 schwarze 1559: parse_mdoc(struct mpage *mpage, const struct mdoc_meta *meta,
1560: const struct mdoc_node *n)
1.1 schwarze 1561: {
1562:
1.47 schwarze 1563: assert(NULL != n);
1564: for (n = n->child; NULL != n; n = n->next) {
1565: switch (n->type) {
1.98 schwarze 1566: case MDOC_ELEM:
1.47 schwarze 1567: /* FALLTHROUGH */
1.98 schwarze 1568: case MDOC_BLOCK:
1.47 schwarze 1569: /* FALLTHROUGH */
1.98 schwarze 1570: case MDOC_HEAD:
1.47 schwarze 1571: /* FALLTHROUGH */
1.98 schwarze 1572: case MDOC_BODY:
1.47 schwarze 1573: /* FALLTHROUGH */
1.98 schwarze 1574: case MDOC_TAIL:
1.47 schwarze 1575: if (NULL != mdocs[n->tok].fp)
1.127 schwarze 1576: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.47 schwarze 1577: break;
1578: if (mdocs[n->tok].mask)
1579: putmdockey(mpage, n->child,
1580: mdocs[n->tok].mask);
1581: break;
1582: default:
1583: assert(MDOC_ROOT != n->type);
1584: continue;
1585: }
1586: if (NULL != n->child)
1.127 schwarze 1587: parse_mdoc(mpage, meta, n);
1.1 schwarze 1588: }
1589: }
1590:
1.19 schwarze 1591: static int
1.127 schwarze 1592: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_meta *meta,
1593: const struct mdoc_node *n)
1.1 schwarze 1594: {
1.131 schwarze 1595: char *start, *end;
1.1 schwarze 1596: size_t sz;
1.19 schwarze 1597:
1.47 schwarze 1598: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1599: NULL == (n = n->child) ||
1600: MDOC_TEXT != n->type)
1.19 schwarze 1601: return(0);
1.1 schwarze 1602:
1603: /*
1604: * Only consider those `Fd' macro fields that begin with an
1605: * "inclusion" token (versus, e.g., #define).
1606: */
1.47 schwarze 1607:
1.1 schwarze 1608: if (strcmp("#include", n->string))
1.19 schwarze 1609: return(0);
1.1 schwarze 1610:
1611: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1612: return(0);
1.1 schwarze 1613:
1614: /*
1615: * Strip away the enclosing angle brackets and make sure we're
1616: * not zero-length.
1617: */
1618:
1619: start = n->string;
1620: if ('<' == *start || '"' == *start)
1621: start++;
1622:
1623: if (0 == (sz = strlen(start)))
1.19 schwarze 1624: return(0);
1.1 schwarze 1625:
1626: end = &start[(int)sz - 1];
1627: if ('>' == *end || '"' == *end)
1628: end--;
1629:
1.47 schwarze 1630: if (end > start)
1631: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1632: return(0);
1.1 schwarze 1633: }
1634:
1.133 schwarze 1635: static void
1636: parse_mdoc_fname(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1637: {
1.69 schwarze 1638: char *cp;
1.133 schwarze 1639: size_t sz;
1.1 schwarze 1640:
1.133 schwarze 1641: if (n->type != MDOC_TEXT)
1642: return;
1.19 schwarze 1643:
1.133 schwarze 1644: /* Skip function pointer punctuation. */
1.1 schwarze 1645:
1.133 schwarze 1646: cp = n->string;
1647: while (*cp == '(' || *cp == '*')
1.1 schwarze 1648: cp++;
1.133 schwarze 1649: sz = strcspn(cp, "()");
1.1 schwarze 1650:
1.133 schwarze 1651: putkeys(mpage, cp, sz, TYPE_Fn);
1.128 schwarze 1652: if (n->sec == SEC_SYNOPSIS)
1.133 schwarze 1653: putkeys(mpage, cp, sz, NAME_SYN);
1654: }
1.19 schwarze 1655:
1.133 schwarze 1656: static int
1657: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_meta *meta,
1658: const struct mdoc_node *n)
1659: {
1660:
1661: if (n->child == NULL)
1662: return(0);
1663:
1664: parse_mdoc_fname(mpage, n->child);
1.19 schwarze 1665:
1.133 schwarze 1666: for (n = n->child->next; n != NULL; n = n->next)
1667: if (n->type == MDOC_TEXT)
1.47 schwarze 1668: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1669:
1.128 schwarze 1670: return(0);
1671: }
1672:
1673: static int
1674: parse_mdoc_Fo(struct mpage *mpage, const struct mdoc_meta *meta,
1675: const struct mdoc_node *n)
1676: {
1.132 schwarze 1677:
1678: if (n->type != MDOC_HEAD)
1679: return(1);
1.128 schwarze 1680:
1.133 schwarze 1681: if (n->child != NULL)
1682: parse_mdoc_fname(mpage, n->child);
1683:
1.19 schwarze 1684: return(0);
1.1 schwarze 1685: }
1686:
1.19 schwarze 1687: static int
1.127 schwarze 1688: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_meta *meta,
1689: const struct mdoc_node *n)
1.1 schwarze 1690: {
1.47 schwarze 1691: char *cp;
1.1 schwarze 1692:
1693: if (NULL == (n = n->child))
1.19 schwarze 1694: return(0);
1.1 schwarze 1695:
1.47 schwarze 1696: if (NULL == n->next) {
1697: putkey(mpage, n->string, TYPE_Xr);
1698: return(0);
1699: }
1.1 schwarze 1700:
1.77 schwarze 1701: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1702: putkey(mpage, cp, TYPE_Xr);
1703: free(cp);
1704: return(0);
1.1 schwarze 1705: }
1706:
1.19 schwarze 1707: static int
1.127 schwarze 1708: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_meta *meta,
1709: const struct mdoc_node *n)
1.1 schwarze 1710: {
1711:
1.79 schwarze 1712: if (MDOC_BODY == n->type)
1713: mdoc_deroff(&mpage->desc, n);
1714: return(0);
1.1 schwarze 1715: }
1716:
1.19 schwarze 1717: static int
1.127 schwarze 1718: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_meta *meta,
1719: const struct mdoc_node *n)
1.1 schwarze 1720: {
1721:
1.86 schwarze 1722: if (SEC_NAME == n->sec)
1.90 schwarze 1723: putmdockey(mpage, n->child, NAME_TITLE);
1.127 schwarze 1724: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) {
1725: if (n->child == NULL)
1726: putkey(mpage, meta->name, NAME_SYN);
1727: else
1728: putmdockey(mpage, n->child, NAME_SYN);
1.129 schwarze 1729: }
1730: if ( ! (mpage->name_head_done ||
1731: n->child == NULL || n->child->string == NULL ||
1732: strcasecmp(n->child->string, meta->title))) {
1733: putkey(mpage, n->child->string, NAME_HEAD);
1734: mpage->name_head_done = 1;
1.127 schwarze 1735: }
1.86 schwarze 1736: return(0);
1.1 schwarze 1737: }
1738:
1.19 schwarze 1739: static int
1.127 schwarze 1740: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_meta *meta,
1741: const struct mdoc_node *n)
1.1 schwarze 1742: {
1743:
1.19 schwarze 1744: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1745: }
1746:
1.47 schwarze 1747: static int
1.127 schwarze 1748: parse_mdoc_head(struct mpage *mpage, const struct mdoc_meta *meta,
1749: const struct mdoc_node *n)
1.1 schwarze 1750: {
1751:
1.47 schwarze 1752: return(MDOC_HEAD == n->type);
1753: }
1.1 schwarze 1754:
1.47 schwarze 1755: static int
1.127 schwarze 1756: parse_mdoc_body(struct mpage *mpage, const struct mdoc_meta *meta,
1757: const struct mdoc_node *n)
1.47 schwarze 1758: {
1.1 schwarze 1759:
1.47 schwarze 1760: return(MDOC_BODY == n->type);
1.1 schwarze 1761: }
1762:
1.47 schwarze 1763: /*
1764: * Add a string to the hash table for the current manual.
1765: * Each string has a bitmask telling which macros it belongs to.
1766: * When we finish the manual, we'll dump the table.
1767: */
1.1 schwarze 1768: static void
1.131 schwarze 1769: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.1 schwarze 1770: {
1.90 schwarze 1771: struct ohash *htab;
1.47 schwarze 1772: struct str *s;
1.68 schwarze 1773: const char *end;
1.47 schwarze 1774: unsigned int slot;
1.131 schwarze 1775: int i, mustfree;
1.1 schwarze 1776:
1.47 schwarze 1777: if (0 == sz)
1778: return;
1.68 schwarze 1779:
1.131 schwarze 1780: mustfree = render_string(&cp, &sz);
1781:
1.90 schwarze 1782: if (TYPE_Nm & v) {
1783: htab = &names;
1784: v &= name_mask;
1.124 schwarze 1785: if (v & NAME_FIRST)
1786: name_mask &= ~NAME_FIRST;
1.90 schwarze 1787: if (debug > 1)
1788: say(mpage->mlinks->file,
1.130 schwarze 1789: "Adding name %*s, bits=%d", sz, cp, v);
1.90 schwarze 1790: } else {
1791: htab = &strings;
1792: if (debug > 1)
1793: for (i = 0; i < mansearch_keymax; i++)
1.118 schwarze 1794: if ((uint64_t)1 << i & v)
1.90 schwarze 1795: say(mpage->mlinks->file,
1796: "Adding key %s=%*s",
1797: mansearch_keynames[i], sz, cp);
1.68 schwarze 1798: }
1.47 schwarze 1799:
1800: end = cp + sz;
1.90 schwarze 1801: slot = ohash_qlookupi(htab, cp, &end);
1802: s = ohash_find(htab, slot);
1.1 schwarze 1803:
1.47 schwarze 1804: if (NULL != s && mpage == s->mpage) {
1805: s->mask |= v;
1.1 schwarze 1806: return;
1.47 schwarze 1807: } else if (NULL == s) {
1.101 schwarze 1808: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1809: memcpy(s->key, cp, sz);
1.90 schwarze 1810: ohash_insert(htab, slot, s);
1.47 schwarze 1811: }
1812: s->mpage = mpage;
1813: s->mask = v;
1.131 schwarze 1814:
1815: if (mustfree)
1816: free(cp);
1.1 schwarze 1817: }
1818:
1819: /*
1.47 schwarze 1820: * Take a Unicode codepoint and produce its UTF-8 encoding.
1821: * This isn't the best way to do this, but it works.
1822: * The magic numbers are from the UTF-8 packaging.
1823: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1824: */
1.47 schwarze 1825: static size_t
1826: utf8(unsigned int cp, char out[7])
1.1 schwarze 1827: {
1.47 schwarze 1828: size_t rc;
1.1 schwarze 1829:
1.47 schwarze 1830: rc = 0;
1831: if (cp <= 0x0000007F) {
1832: rc = 1;
1833: out[0] = (char)cp;
1834: } else if (cp <= 0x000007FF) {
1835: rc = 2;
1836: out[0] = (cp >> 6 & 31) | 192;
1837: out[1] = (cp & 63) | 128;
1838: } else if (cp <= 0x0000FFFF) {
1839: rc = 3;
1840: out[0] = (cp >> 12 & 15) | 224;
1841: out[1] = (cp >> 6 & 63) | 128;
1842: out[2] = (cp & 63) | 128;
1843: } else if (cp <= 0x001FFFFF) {
1844: rc = 4;
1845: out[0] = (cp >> 18 & 7) | 240;
1846: out[1] = (cp >> 12 & 63) | 128;
1847: out[2] = (cp >> 6 & 63) | 128;
1848: out[3] = (cp & 63) | 128;
1849: } else if (cp <= 0x03FFFFFF) {
1850: rc = 5;
1851: out[0] = (cp >> 24 & 3) | 248;
1852: out[1] = (cp >> 18 & 63) | 128;
1853: out[2] = (cp >> 12 & 63) | 128;
1854: out[3] = (cp >> 6 & 63) | 128;
1855: out[4] = (cp & 63) | 128;
1856: } else if (cp <= 0x7FFFFFFF) {
1857: rc = 6;
1858: out[0] = (cp >> 30 & 1) | 252;
1859: out[1] = (cp >> 24 & 63) | 128;
1860: out[2] = (cp >> 18 & 63) | 128;
1861: out[3] = (cp >> 12 & 63) | 128;
1862: out[4] = (cp >> 6 & 63) | 128;
1863: out[5] = (cp & 63) | 128;
1864: } else
1865: return(0);
1.19 schwarze 1866:
1.47 schwarze 1867: out[rc] = '\0';
1868: return(rc);
1.1 schwarze 1869: }
1870:
1.47 schwarze 1871: /*
1.131 schwarze 1872: * If the string contains escape sequences,
1873: * replace it with an allocated rendering and return 1,
1874: * such that the caller can free it after use.
1875: * Otherwise, do nothing and return 0.
1.47 schwarze 1876: */
1.131 schwarze 1877: static int
1878: render_string(char **public, size_t *psz)
1.1 schwarze 1879: {
1.131 schwarze 1880: const char *src, *scp, *addcp, *seq;
1881: char *dst;
1882: size_t ssz, dsz, addsz;
1.71 schwarze 1883: char utfbuf[7], res[6];
1.131 schwarze 1884: int seqlen, unicode;
1.47 schwarze 1885:
1886: res[0] = '\\';
1887: res[1] = '\t';
1888: res[2] = ASCII_NBRSP;
1889: res[3] = ASCII_HYPH;
1.71 schwarze 1890: res[4] = ASCII_BREAK;
1891: res[5] = '\0';
1.1 schwarze 1892:
1.131 schwarze 1893: src = scp = *public;
1894: ssz = *psz;
1895: dst = NULL;
1896: dsz = 0;
1897:
1898: while (scp < src + *psz) {
1899:
1900: /* Leave normal characters unchanged. */
1901:
1902: if (strchr(res, *scp) == NULL) {
1903: if (dst != NULL)
1904: dst[dsz++] = *scp;
1905: scp++;
1906: continue;
1907: }
1.1 schwarze 1908:
1.47 schwarze 1909: /*
1.131 schwarze 1910: * Found something that requires replacing,
1911: * make sure we have a destination buffer.
1.47 schwarze 1912: */
1.131 schwarze 1913:
1914: if (dst == NULL) {
1915: dst = mandoc_malloc(ssz + 1);
1916: dsz = scp - src;
1917: memcpy(dst, src, dsz);
1.47 schwarze 1918: }
1.39 schwarze 1919:
1.131 schwarze 1920: /* Handle single-char special characters. */
1921:
1922: switch (*scp) {
1923: case '\\':
1924: break;
1.98 schwarze 1925: case '\t':
1.71 schwarze 1926: /* FALLTHROUGH */
1.98 schwarze 1927: case ASCII_NBRSP:
1.131 schwarze 1928: dst[dsz++] = ' ';
1929: scp++;
1930: continue;
1931: case ASCII_HYPH:
1932: dst[dsz++] = '-';
1.71 schwarze 1933: /* FALLTHROUGH */
1.98 schwarze 1934: case ASCII_BREAK:
1.131 schwarze 1935: scp++;
1.47 schwarze 1936: continue;
1.71 schwarze 1937: default:
1.131 schwarze 1938: abort();
1.71 schwarze 1939: }
1.39 schwarze 1940:
1.47 schwarze 1941: /*
1.131 schwarze 1942: * Found an escape sequence.
1943: * Read past the slash, then parse it.
1944: * Ignore everything except characters.
1.47 schwarze 1945: */
1.52 schwarze 1946:
1.131 schwarze 1947: scp++;
1948: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.47 schwarze 1949: continue;
1.39 schwarze 1950:
1.47 schwarze 1951: /*
1.52 schwarze 1952: * Render the special character
1953: * as either UTF-8 or ASCII.
1.47 schwarze 1954: */
1.52 schwarze 1955:
1956: if (write_utf8) {
1.131 schwarze 1957: unicode = mchars_spec2cp(mchars, seq, seqlen);
1958: if (unicode <= 0)
1.52 schwarze 1959: continue;
1.131 schwarze 1960: addsz = utf8(unicode, utfbuf);
1961: if (addsz == 0)
1.52 schwarze 1962: continue;
1.131 schwarze 1963: addcp = utfbuf;
1.52 schwarze 1964: } else {
1.131 schwarze 1965: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1966: if (addcp == NULL)
1.52 schwarze 1967: continue;
1.131 schwarze 1968: if (*addcp == ASCII_NBRSP) {
1969: addcp = " ";
1970: addsz = 1;
1.52 schwarze 1971: }
1972: }
1.1 schwarze 1973:
1.47 schwarze 1974: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1975:
1.131 schwarze 1976: ssz += addsz;
1977: dst = mandoc_realloc(dst, ssz + 1);
1978: memcpy(dst + dsz, addcp, addsz);
1979: dsz += addsz;
1980: }
1981: if (dst != NULL) {
1982: *public = dst;
1983: *psz = dsz;
1984: }
1985:
1986: /* Trim trailing whitespace and NUL-terminate. */
1987:
1988: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1989: --*psz;
1990: if (dst != NULL) {
1991: (*public)[*psz] = '\0';
1992: return(1);
1993: } else
1994: return(0);
1.1 schwarze 1995: }
1996:
1.75 schwarze 1997: static void
1998: dbadd_mlink(const struct mlink *mlink)
1999: {
2000: size_t i;
2001:
2002: i = 1;
2003: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2004: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2005: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 2006: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 2007: SQL_STEP(stmts[STMT_INSERT_LINK]);
2008: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.124 schwarze 2009: }
2010:
2011: static void
2012: dbadd_mlink_name(const struct mlink *mlink)
2013: {
1.130 schwarze 2014: uint64_t bits;
1.124 schwarze 2015: size_t i;
2016:
2017: dbadd_mlink(mlink);
1.115 schwarze 2018:
2019: i = 1;
1.130 schwarze 2020: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2021: bits = NAME_FILE & NAME_MASK;
2022: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2023: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2024: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2025: }
2026:
2027: i = 1;
2028: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.115 schwarze 2029: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2030: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2031: SQL_STEP(stmts[STMT_INSERT_NAME]);
2032: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.75 schwarze 2033: }
2034:
1.11 schwarze 2035: /*
1.47 schwarze 2036: * Flush the current page's terms (and their bits) into the database.
2037: * Wrap the entire set of additions in a transaction to make sqlite be a
2038: * little faster.
1.53 schwarze 2039: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 2040: */
2041: static void
1.131 schwarze 2042: dbadd(struct mpage *mpage)
1.11 schwarze 2043: {
1.47 schwarze 2044: struct mlink *mlink;
2045: struct str *key;
1.131 schwarze 2046: char *cp;
1.47 schwarze 2047: size_t i;
2048: unsigned int slot;
1.131 schwarze 2049: int mustfree;
1.47 schwarze 2050:
1.85 schwarze 2051: mlink = mpage->mlinks;
1.11 schwarze 2052:
1.85 schwarze 2053: if (nodb) {
1.104 schwarze 2054: for (key = ohash_first(&names, &slot); NULL != key;
1.131 schwarze 2055: key = ohash_next(&names, &slot))
1.104 schwarze 2056: free(key);
2057: for (key = ohash_first(&strings, &slot); NULL != key;
1.131 schwarze 2058: key = ohash_next(&strings, &slot))
1.104 schwarze 2059: free(key);
1.102 schwarze 2060: if (0 == debug)
2061: return;
1.85 schwarze 2062: while (NULL != mlink) {
2063: fputs(mlink->name, stdout);
2064: if (NULL == mlink->next ||
2065: strcmp(mlink->dsec, mlink->next->dsec) ||
2066: strcmp(mlink->fsec, mlink->next->fsec) ||
2067: strcmp(mlink->arch, mlink->next->arch)) {
2068: putchar('(');
2069: if ('\0' == *mlink->dsec)
2070: fputs(mlink->fsec, stdout);
2071: else
2072: fputs(mlink->dsec, stdout);
2073: if ('\0' != *mlink->arch)
2074: printf("/%s", mlink->arch);
2075: putchar(')');
2076: }
2077: mlink = mlink->next;
2078: if (NULL != mlink)
2079: fputs(", ", stdout);
2080: }
1.89 schwarze 2081: printf(" - %s\n", mpage->desc);
1.11 schwarze 2082: return;
1.85 schwarze 2083: }
2084:
2085: if (debug)
2086: say(mlink->file, "Adding to database");
1.47 schwarze 2087:
1.131 schwarze 2088: cp = mpage->desc;
2089: i = strlen(cp);
2090: mustfree = render_string(&cp, &i);
1.47 schwarze 2091: i = 1;
1.131 schwarze 2092: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.116 schwarze 2093: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.47 schwarze 2094: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2095: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2096: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.131 schwarze 2097: if (mustfree)
2098: free(cp);
1.47 schwarze 2099:
1.85 schwarze 2100: while (NULL != mlink) {
1.75 schwarze 2101: dbadd_mlink(mlink);
1.85 schwarze 2102: mlink = mlink->next;
2103: }
1.91 schwarze 2104: mlink = mpage->mlinks;
1.47 schwarze 2105:
1.90 schwarze 2106: for (key = ohash_first(&names, &slot); NULL != key;
2107: key = ohash_next(&names, &slot)) {
2108: assert(key->mpage == mpage);
2109: i = 1;
2110: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.131 schwarze 2111: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.94 schwarze 2112: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2113: SQL_STEP(stmts[STMT_INSERT_NAME]);
2114: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2115: free(key);
2116: }
1.47 schwarze 2117: for (key = ohash_first(&strings, &slot); NULL != key;
2118: key = ohash_next(&strings, &slot)) {
2119: assert(key->mpage == mpage);
2120: i = 1;
2121: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.131 schwarze 2122: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.94 schwarze 2123: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2124: SQL_STEP(stmts[STMT_INSERT_KEY]);
2125: sqlite3_reset(stmts[STMT_INSERT_KEY]);
2126: free(key);
1.33 schwarze 2127: }
1.47 schwarze 2128: }
1.41 deraadt 2129:
1.47 schwarze 2130: static void
2131: dbprune(void)
2132: {
2133: struct mpage *mpage;
2134: struct mlink *mlink;
2135: size_t i;
2136: unsigned int slot;
1.11 schwarze 2137:
1.63 schwarze 2138: if (0 == nodb)
2139: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2140:
1.63 schwarze 2141: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2142: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2143: mlink = mpage->mlinks;
1.82 schwarze 2144: if (debug)
1.63 schwarze 2145: say(mlink->file, "Deleting from database");
2146: if (nodb)
2147: continue;
2148: for ( ; NULL != mlink; mlink = mlink->next) {
2149: i = 1;
2150: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2151: i, mlink->dsec);
2152: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2153: i, mlink->arch);
2154: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2155: i, mlink->name);
2156: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2157: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2158: }
1.11 schwarze 2159: }
1.63 schwarze 2160:
2161: if (0 == nodb)
2162: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2163: }
1.22 schwarze 2164:
1.47 schwarze 2165: /*
2166: * Close an existing database and its prepared statements.
2167: * If "real" is not set, rename the temporary file into the real one.
2168: */
2169: static void
2170: dbclose(int real)
2171: {
2172: size_t i;
1.72 schwarze 2173: int status;
2174: pid_t child;
1.11 schwarze 2175:
1.47 schwarze 2176: if (nodb)
2177: return;
1.11 schwarze 2178:
1.47 schwarze 2179: for (i = 0; i < STMT__MAX; i++) {
2180: sqlite3_finalize(stmts[i]);
2181: stmts[i] = NULL;
1.28 schwarze 2182: }
1.22 schwarze 2183:
1.47 schwarze 2184: sqlite3_close(db);
2185: db = NULL;
1.11 schwarze 2186:
1.47 schwarze 2187: if (real)
2188: return;
1.22 schwarze 2189:
1.72 schwarze 2190: if ('\0' == *tempfilename) {
2191: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2192: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2193: say(MANDOC_DB, "&rename");
1.72 schwarze 2194: }
2195: return;
2196: }
2197:
2198: switch (child = fork()) {
1.98 schwarze 2199: case -1:
1.72 schwarze 2200: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2201: say("", "&fork cmp");
1.72 schwarze 2202: return;
1.98 schwarze 2203: case 0:
1.72 schwarze 2204: execlp("cmp", "cmp", "-s",
2205: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2206: say("", "&exec cmp");
1.72 schwarze 2207: exit(0);
2208: default:
2209: break;
2210: }
2211: if (-1 == waitpid(child, &status, 0)) {
2212: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2213: say("", "&wait cmp");
1.72 schwarze 2214: } else if (WIFSIGNALED(status)) {
2215: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2216: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2217: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2218: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2219: say(MANDOC_DB,
2220: "Data changed, but cannot replace database");
2221: }
2222:
2223: *strrchr(tempfilename, '/') = '\0';
2224: switch (child = fork()) {
1.98 schwarze 2225: case -1:
1.72 schwarze 2226: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2227: say("", "&fork rm");
1.72 schwarze 2228: return;
1.98 schwarze 2229: case 0:
1.72 schwarze 2230: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2231: say("", "&exec rm");
1.72 schwarze 2232: exit((int)MANDOCLEVEL_SYSERR);
2233: default:
2234: break;
2235: }
2236: if (-1 == waitpid(child, &status, 0)) {
2237: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2238: say("", "&wait rm");
1.72 schwarze 2239: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2240: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2241: say("", "%s: Cannot remove temporary directory",
2242: tempfilename);
1.22 schwarze 2243: }
1.11 schwarze 2244: }
2245:
1.47 schwarze 2246: /*
2247: * This is straightforward stuff.
2248: * Open a database connection to a "temporary" database, then open a set
2249: * of prepared statements we'll use over and over again.
2250: * If "real" is set, we use the existing database; if not, we truncate a
2251: * temporary one.
2252: * Must be matched by dbclose().
2253: */
2254: static int
2255: dbopen(int real)
1.2 schwarze 2256: {
1.72 schwarze 2257: const char *sql;
1.47 schwarze 2258: int rc, ofl;
1.6 schwarze 2259:
1.98 schwarze 2260: if (nodb)
1.47 schwarze 2261: return(1);
1.6 schwarze 2262:
1.72 schwarze 2263: *tempfilename = '\0';
1.47 schwarze 2264: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2265:
2266: if (real) {
2267: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2268: if (SQLITE_OK != rc) {
1.47 schwarze 2269: exitcode = (int)MANDOCLEVEL_SYSERR;
1.108 sthen 2270: if (SQLITE_CANTOPEN != rc)
2271: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.47 schwarze 2272: return(0);
1.28 schwarze 2273: }
1.72 schwarze 2274: goto prepare_statements;
2275: }
2276:
2277: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2278:
1.72 schwarze 2279: remove(MANDOC_DB "~");
2280: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.98 schwarze 2281: if (SQLITE_OK == rc)
1.72 schwarze 2282: goto create_tables;
1.73 schwarze 2283: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2284: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2285: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.47 schwarze 2286: return(0);
2287: }
1.6 schwarze 2288:
1.100 schwarze 2289: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2290: sizeof(tempfilename));
1.72 schwarze 2291: if (NULL == mkdtemp(tempfilename)) {
2292: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2293: say("", "&%s", tempfilename);
1.72 schwarze 2294: return(0);
2295: }
1.100 schwarze 2296: (void)strlcat(tempfilename, "/" MANDOC_DB,
2297: sizeof(tempfilename));
1.72 schwarze 2298: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2299: if (SQLITE_OK != rc) {
1.47 schwarze 2300: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2301: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.47 schwarze 2302: return(0);
1.2 schwarze 2303: }
2304:
1.72 schwarze 2305: create_tables:
1.47 schwarze 2306: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2307: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2308: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2309: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2310: ");\n"
2311: "\n"
2312: "CREATE TABLE \"mlinks\" (\n"
2313: " \"sec\" TEXT NOT NULL,\n"
2314: " \"arch\" TEXT NOT NULL,\n"
2315: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2316: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2317: "ON DELETE CASCADE\n"
1.47 schwarze 2318: ");\n"
1.93 schwarze 2319: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2320: "\n"
1.90 schwarze 2321: "CREATE TABLE \"names\" (\n"
2322: " \"bits\" INTEGER NOT NULL,\n"
2323: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2324: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.130 schwarze 2325: "ON DELETE CASCADE,\n"
2326: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.90 schwarze 2327: ");\n"
2328: "\n"
1.47 schwarze 2329: "CREATE TABLE \"keys\" (\n"
2330: " \"bits\" INTEGER NOT NULL,\n"
2331: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2332: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2333: "ON DELETE CASCADE\n"
1.93 schwarze 2334: ");\n"
2335: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2336:
2337: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2338: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2339: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.103 schwarze 2340: sqlite3_close(db);
1.47 schwarze 2341: return(0);
1.2 schwarze 2342: }
2343:
1.47 schwarze 2344: prepare_statements:
1.103 schwarze 2345: if (SQLITE_OK != sqlite3_exec(db,
2346: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2347: exitcode = (int)MANDOCLEVEL_SYSERR;
2348: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2349: sqlite3_errmsg(db));
2350: sqlite3_close(db);
2351: return(0);
2352: }
2353:
1.94 schwarze 2354: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2355: "(SELECT pageid FROM mlinks WHERE "
2356: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2357: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2358: sql = "INSERT INTO mpages "
1.89 schwarze 2359: "(desc,form) VALUES (?,?)";
1.47 schwarze 2360: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2361: sql = "INSERT INTO mlinks "
1.61 schwarze 2362: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2363: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.130 schwarze 2364: sql = "SELECT bits FROM names where pageid = ?";
2365: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.90 schwarze 2366: sql = "INSERT INTO names "
2367: "(bits,name,pageid) VALUES (?,?,?)";
2368: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2369: sql = "INSERT INTO keys "
2370: "(bits,key,pageid) VALUES (?,?,?)";
2371: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2372:
1.47 schwarze 2373: /*
2374: * When opening a new database, we can turn off
2375: * synchronous mode for much better performance.
2376: */
1.6 schwarze 2377:
1.103 schwarze 2378: if (real && SQLITE_OK != sqlite3_exec(db,
2379: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2380: exitcode = (int)MANDOCLEVEL_SYSERR;
2381: say(MANDOC_DB, "PRAGMA synchronous: %s",
2382: sqlite3_errmsg(db));
2383: sqlite3_close(db);
2384: return(0);
2385: }
1.11 schwarze 2386:
1.47 schwarze 2387: return(1);
2388: }
1.6 schwarze 2389:
1.47 schwarze 2390: static void *
1.106 espie 2391: hash_calloc(size_t n, size_t sz, void *arg)
1.47 schwarze 2392: {
1.6 schwarze 2393:
1.106 espie 2394: return(mandoc_calloc(n, sz));
1.47 schwarze 2395: }
1.2 schwarze 2396:
1.47 schwarze 2397: static void *
2398: hash_alloc(size_t sz, void *arg)
2399: {
1.28 schwarze 2400:
1.47 schwarze 2401: return(mandoc_malloc(sz));
2402: }
1.6 schwarze 2403:
1.47 schwarze 2404: static void
1.106 espie 2405: hash_free(void *p, void *arg)
1.47 schwarze 2406: {
1.26 schwarze 2407:
1.47 schwarze 2408: free(p);
2409: }
1.6 schwarze 2410:
1.47 schwarze 2411: static int
1.120 schwarze 2412: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2413: {
2414: static char startdir[PATH_MAX];
1.110 schwarze 2415: static int getcwd_status; /* 1 = ok, 2 = failure */
2416: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2417: char *cp;
1.6 schwarze 2418:
1.47 schwarze 2419: /*
1.110 schwarze 2420: * Remember the original working directory, if possible.
2421: * This will be needed if the second or a later directory
2422: * on the command line is given as a relative path.
2423: * Do not error out if the current directory is not
2424: * searchable: Maybe it won't be needed after all.
2425: */
2426: if (0 == getcwd_status) {
2427: if (NULL == getcwd(startdir, sizeof(startdir))) {
2428: getcwd_status = 2;
2429: (void)strlcpy(startdir, strerror(errno),
2430: sizeof(startdir));
2431: } else
2432: getcwd_status = 1;
2433: }
2434:
2435: /*
2436: * We are leaving the old base directory.
2437: * Do not use it any longer, not even for messages.
2438: */
2439: *basedir = '\0';
2440:
2441: /*
2442: * If and only if the directory was changed earlier and
2443: * the next directory to process is given as a relative path,
2444: * first go back, or bail out if that is impossible.
1.47 schwarze 2445: */
1.110 schwarze 2446: if (chdir_status && '/' != *targetdir) {
2447: if (2 == getcwd_status) {
1.47 schwarze 2448: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2449: say("", "getcwd: %s", startdir);
1.47 schwarze 2450: return(0);
2451: }
1.110 schwarze 2452: if (-1 == chdir(startdir)) {
1.47 schwarze 2453: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2454: say("", "&chdir %s", startdir);
1.47 schwarze 2455: return(0);
1.2 schwarze 2456: }
2457: }
1.110 schwarze 2458:
2459: /*
2460: * Always resolve basedir to the canonicalized absolute
2461: * pathname and append a trailing slash, such that
2462: * we can reliably check whether files are inside.
2463: */
1.47 schwarze 2464: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2465: if (report_baddir || errno != ENOENT) {
2466: exitcode = (int)MANDOCLEVEL_BADARG;
2467: say("", "&%s: realpath", targetdir);
2468: }
1.47 schwarze 2469: return(0);
2470: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2471: if (report_baddir || errno != ENOENT) {
2472: exitcode = (int)MANDOCLEVEL_BADARG;
2473: say("", "&chdir");
2474: }
1.47 schwarze 2475: return(0);
1.109 schwarze 2476: }
1.110 schwarze 2477: chdir_status = 1;
1.109 schwarze 2478: cp = strchr(basedir, '\0');
2479: if ('/' != cp[-1]) {
2480: if (cp - basedir >= PATH_MAX - 1) {
2481: exitcode = (int)MANDOCLEVEL_SYSERR;
2482: say("", "Filename too long");
2483: return(0);
2484: }
2485: *cp++ = '/';
2486: *cp = '\0';
1.47 schwarze 2487: }
2488: return(1);
1.2 schwarze 2489: }
2490:
2491: static void
1.47 schwarze 2492: say(const char *file, const char *format, ...)
1.2 schwarze 2493: {
1.47 schwarze 2494: va_list ap;
1.80 schwarze 2495: int use_errno;
1.2 schwarze 2496:
1.47 schwarze 2497: if ('\0' != *basedir)
2498: fprintf(stderr, "%s", basedir);
2499: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2500: fputc('/', stderr);
1.47 schwarze 2501: if ('\0' != *file)
2502: fprintf(stderr, "%s", file);
1.31 schwarze 2503:
1.80 schwarze 2504: use_errno = 1;
2505: if (NULL != format) {
2506: switch (*format) {
1.98 schwarze 2507: case '&':
1.80 schwarze 2508: format++;
2509: break;
1.98 schwarze 2510: case '\0':
1.80 schwarze 2511: format = NULL;
2512: break;
2513: default:
2514: use_errno = 0;
2515: break;
2516: }
2517: }
2518: if (NULL != format) {
2519: if ('\0' != *basedir || '\0' != *file)
2520: fputs(": ", stderr);
2521: va_start(ap, format);
2522: vfprintf(stderr, format, ap);
2523: va_end(ap);
2524: }
2525: if (use_errno) {
2526: if ('\0' != *basedir || '\0' != *file || NULL != format)
2527: fputs(": ", stderr);
1.47 schwarze 2528: perror(NULL);
1.80 schwarze 2529: } else
2530: fputc('\n', stderr);
1.1 schwarze 2531: }