Annotation of mandoc/mandocdb.c, Revision 1.217
1.217 ! schwarze 1: /* $Id: mandocdb.c,v 1.216 2016/03/17 21:26:26 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.211 schwarze 4: * Copyright (c) 2011-2016 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 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: *
1.187 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.187 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 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: */
18: #include "config.h"
19:
1.162 schwarze 20: #include <sys/types.h>
1.50 kristaps 21: #include <sys/stat.h>
1.115 schwarze 22: #include <sys/wait.h>
1.1 kristaps 23:
24: #include <assert.h>
1.43 kristaps 25: #include <ctype.h>
1.208 schwarze 26: #if HAVE_ERR
1.198 schwarze 27: #include <err.h>
1.208 schwarze 28: #endif
1.63 schwarze 29: #include <errno.h>
1.1 kristaps 30: #include <fcntl.h>
1.158 schwarze 31: #if HAVE_FTS
1.50 kristaps 32: #include <fts.h>
1.157 schwarze 33: #else
34: #include "compat_fts.h"
35: #endif
1.58 schwarze 36: #include <limits.h>
1.50 kristaps 37: #include <stddef.h>
1.59 schwarze 38: #include <stdio.h>
1.1 kristaps 39: #include <stdint.h>
40: #include <stdlib.h>
41: #include <string.h>
1.17 schwarze 42: #include <unistd.h>
1.1 kristaps 43:
1.50 kristaps 44: #include <sqlite3.h>
1.1 kristaps 45:
1.188 schwarze 46: #include "mandoc_aux.h"
1.202 schwarze 47: #include "mandoc_ohash.h"
1.188 schwarze 48: #include "mandoc.h"
49: #include "roff.h"
1.50 kristaps 50: #include "mdoc.h"
1.1 kristaps 51: #include "man.h"
1.187 schwarze 52: #include "manconf.h"
1.55 kristaps 53: #include "mansearch.h"
1.1 kristaps 54:
1.111 schwarze 55: extern int mansearch_keymax;
56: extern const char *const mansearch_keynames[];
57:
1.52 kristaps 58: #define SQL_EXEC(_v) \
59: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.134 schwarze 60: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.52 kristaps 61: #define SQL_BIND_TEXT(_s, _i, _v) \
62: if (SQLITE_OK != sqlite3_bind_text \
63: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.134 schwarze 64: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 65: #define SQL_BIND_INT(_s, _i, _v) \
66: if (SQLITE_OK != sqlite3_bind_int \
67: ((_s), (_i)++, (_v))) \
1.134 schwarze 68: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 69: #define SQL_BIND_INT64(_s, _i, _v) \
70: if (SQLITE_OK != sqlite3_bind_int64 \
71: ((_s), (_i)++, (_v))) \
1.134 schwarze 72: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 73: #define SQL_STEP(_s) \
74: if (SQLITE_DONE != sqlite3_step((_s))) \
1.134 schwarze 75: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 76:
1.50 kristaps 77: enum op {
78: OP_DEFAULT = 0, /* new dbs from dir list or default config */
79: OP_CONFFILE, /* new databases from custom config file */
80: OP_UPDATE, /* delete/add entries in existing database */
81: OP_DELETE, /* delete entries from existing database */
82: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 83: };
84:
1.50 kristaps 85: struct str {
1.78 schwarze 86: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 87: uint64_t mask; /* bitmask in sequence */
1.176 schwarze 88: char key[]; /* rendered text */
1.38 schwarze 89: };
90:
1.79 schwarze 91: struct inodev {
92: ino_t st_ino;
93: dev_t st_dev;
1.50 kristaps 94: };
1.5 kristaps 95:
1.78 schwarze 96: struct mpage {
1.79 schwarze 97: struct inodev inodev; /* used for hashing routine */
1.137 schwarze 98: int64_t pageid; /* pageid in mpages SQL table */
1.82 schwarze 99: char *sec; /* section from file content */
100: char *arch; /* architecture from file content */
101: char *title; /* title from file content */
102: char *desc; /* description from file content */
103: struct mlink *mlinks; /* singly linked list */
1.161 schwarze 104: int form; /* format from file content */
1.174 schwarze 105: int name_head_done;
1.82 schwarze 106: };
107:
108: struct mlink {
1.58 schwarze 109: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 110: char *dsec; /* section from directory */
111: char *arch; /* architecture from directory */
112: char *name; /* name from file name (not empty) */
113: char *fsec; /* section from file name suffix */
1.85 schwarze 114: struct mlink *next; /* singly linked list */
1.118 schwarze 115: struct mpage *mpage; /* parent */
1.161 schwarze 116: int dform; /* format from directory */
117: int fform; /* format from file name suffix */
1.124 schwarze 118: int gzip; /* filename has a .gz suffix */
1.3 kristaps 119: };
120:
1.50 kristaps 121: enum stmt {
1.81 schwarze 122: STMT_DELETE_PAGE = 0, /* delete mpage */
123: STMT_INSERT_PAGE, /* insert mpage */
124: STMT_INSERT_LINK, /* insert mlink */
1.133 schwarze 125: STMT_INSERT_NAME, /* insert name */
1.175 schwarze 126: STMT_SELECT_NAME, /* retrieve existing name flags */
1.81 schwarze 127: STMT_INSERT_KEY, /* insert parsed key */
1.50 kristaps 128: STMT__MAX
1.1 kristaps 129: };
130:
1.190 schwarze 131: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.189 schwarze 132: const struct roff_node *);
1.1 kristaps 133:
1.50 kristaps 134: struct mdoc_handler {
135: mdoc_fp fp; /* optional handler */
136: uint64_t mask; /* set unless handler returns 0 */
1.1 kristaps 137: };
138:
1.59 schwarze 139: static void dbclose(int);
1.176 schwarze 140: static void dbadd(struct mpage *);
1.118 schwarze 141: static void dbadd_mlink(const struct mlink *mlink);
1.169 schwarze 142: static void dbadd_mlink_name(const struct mlink *mlink);
1.59 schwarze 143: static int dbopen(int);
144: static void dbprune(void);
145: static void filescan(const char *);
1.84 schwarze 146: static void mlink_add(struct mlink *, const struct stat *);
1.131 schwarze 147: static void mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 148: static void mlink_free(struct mlink *);
1.89 schwarze 149: static void mlinks_undupe(struct mpage *);
1.78 schwarze 150: static void mpages_free(void);
1.176 schwarze 151: static void mpages_merge(struct mparse *);
1.130 schwarze 152: static void names_check(void);
1.124 schwarze 153: static void parse_cat(struct mpage *, int);
1.190 schwarze 154: static void parse_man(struct mpage *, const struct roff_meta *,
1.189 schwarze 155: const struct roff_node *);
1.190 schwarze 156: static void parse_mdoc(struct mpage *, const struct roff_meta *,
1.189 schwarze 157: const struct roff_node *);
1.190 schwarze 158: static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
1.189 schwarze 159: const struct roff_node *);
1.190 schwarze 160: static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
1.189 schwarze 161: const struct roff_node *);
162: static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
1.190 schwarze 163: static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
1.189 schwarze 164: const struct roff_node *);
1.190 schwarze 165: static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
1.189 schwarze 166: const struct roff_node *);
1.190 schwarze 167: static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
1.189 schwarze 168: const struct roff_node *);
1.190 schwarze 169: static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
1.189 schwarze 170: const struct roff_node *);
1.190 schwarze 171: static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
1.189 schwarze 172: const struct roff_node *);
1.211 schwarze 173: static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
174: const struct roff_node *);
1.190 schwarze 175: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.189 schwarze 176: const struct roff_node *);
1.112 schwarze 177: static void putkey(const struct mpage *, char *, uint64_t);
1.176 schwarze 178: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.78 schwarze 179: static void putmdockey(const struct mpage *,
1.189 schwarze 180: const struct roff_node *, uint64_t);
1.176 schwarze 181: static int render_string(char **, size_t *);
1.59 schwarze 182: static void say(const char *, const char *, ...);
1.165 schwarze 183: static int set_basedir(const char *, int);
1.59 schwarze 184: static int treescan(void);
1.50 kristaps 185: static size_t utf8(unsigned int, char [7]);
186:
1.115 schwarze 187: static char tempfilename[32];
1.102 schwarze 188: static int nodb; /* no database changes */
1.116 schwarze 189: static int mparse_options; /* abort the parse early */
1.141 schwarze 190: static int use_all; /* use all found files */
191: static int debug; /* print what we're doing */
192: static int warnings; /* warn about crap */
1.95 schwarze 193: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 194: static int exitcode; /* to be returned by main */
1.141 schwarze 195: static enum op op; /* operational mode */
1.59 schwarze 196: static char basedir[PATH_MAX]; /* current base directory */
1.78 schwarze 197: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 198: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 199: static struct ohash names; /* table of all names */
1.50 kristaps 200: static struct ohash strings; /* table of all strings */
201: static sqlite3 *db = NULL; /* current database */
202: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.133 schwarze 203: static uint64_t name_mask;
1.25 schwarze 204:
205: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 206: { NULL, 0 }, /* Ap */
207: { NULL, 0 }, /* Dd */
208: { NULL, 0 }, /* Dt */
209: { NULL, 0 }, /* Os */
210: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
211: { parse_mdoc_head, TYPE_Ss }, /* Ss */
212: { NULL, 0 }, /* Pp */
213: { NULL, 0 }, /* D1 */
214: { NULL, 0 }, /* Dl */
215: { NULL, 0 }, /* Bd */
216: { NULL, 0 }, /* Ed */
217: { NULL, 0 }, /* Bl */
218: { NULL, 0 }, /* El */
219: { NULL, 0 }, /* It */
220: { NULL, 0 }, /* Ad */
221: { NULL, TYPE_An }, /* An */
222: { NULL, TYPE_Ar }, /* Ar */
223: { NULL, TYPE_Cd }, /* Cd */
224: { NULL, TYPE_Cm }, /* Cm */
225: { NULL, TYPE_Dv }, /* Dv */
226: { NULL, TYPE_Er }, /* Er */
227: { NULL, TYPE_Ev }, /* Ev */
228: { NULL, 0 }, /* Ex */
229: { NULL, TYPE_Fa }, /* Fa */
230: { parse_mdoc_Fd, 0 }, /* Fd */
231: { NULL, TYPE_Fl }, /* Fl */
232: { parse_mdoc_Fn, 0 }, /* Fn */
233: { NULL, TYPE_Ft }, /* Ft */
234: { NULL, TYPE_Ic }, /* Ic */
1.92 schwarze 235: { NULL, TYPE_In }, /* In */
1.68 schwarze 236: { NULL, TYPE_Li }, /* Li */
1.132 schwarze 237: { parse_mdoc_Nd, 0 }, /* Nd */
1.129 schwarze 238: { parse_mdoc_Nm, 0 }, /* Nm */
1.68 schwarze 239: { NULL, 0 }, /* Op */
240: { NULL, 0 }, /* Ot */
241: { NULL, TYPE_Pa }, /* Pa */
242: { NULL, 0 }, /* Rv */
1.92 schwarze 243: { NULL, TYPE_St }, /* St */
1.211 schwarze 244: { parse_mdoc_Va, TYPE_Va }, /* Va */
245: { parse_mdoc_Va, TYPE_Vt }, /* Vt */
1.68 schwarze 246: { parse_mdoc_Xr, 0 }, /* Xr */
247: { NULL, 0 }, /* %A */
248: { NULL, 0 }, /* %B */
249: { NULL, 0 }, /* %D */
250: { NULL, 0 }, /* %I */
251: { NULL, 0 }, /* %J */
252: { NULL, 0 }, /* %N */
253: { NULL, 0 }, /* %O */
254: { NULL, 0 }, /* %P */
255: { NULL, 0 }, /* %R */
256: { NULL, 0 }, /* %T */
257: { NULL, 0 }, /* %V */
258: { NULL, 0 }, /* Ac */
259: { NULL, 0 }, /* Ao */
260: { NULL, 0 }, /* Aq */
261: { NULL, TYPE_At }, /* At */
262: { NULL, 0 }, /* Bc */
263: { NULL, 0 }, /* Bf */
264: { NULL, 0 }, /* Bo */
265: { NULL, 0 }, /* Bq */
266: { NULL, TYPE_Bsx }, /* Bsx */
267: { NULL, TYPE_Bx }, /* Bx */
268: { NULL, 0 }, /* Db */
269: { NULL, 0 }, /* Dc */
270: { NULL, 0 }, /* Do */
271: { NULL, 0 }, /* Dq */
272: { NULL, 0 }, /* Ec */
273: { NULL, 0 }, /* Ef */
274: { NULL, TYPE_Em }, /* Em */
275: { NULL, 0 }, /* Eo */
276: { NULL, TYPE_Fx }, /* Fx */
277: { NULL, TYPE_Ms }, /* Ms */
278: { NULL, 0 }, /* No */
279: { NULL, 0 }, /* Ns */
280: { NULL, TYPE_Nx }, /* Nx */
281: { NULL, TYPE_Ox }, /* Ox */
282: { NULL, 0 }, /* Pc */
283: { NULL, 0 }, /* Pf */
284: { NULL, 0 }, /* Po */
285: { NULL, 0 }, /* Pq */
286: { NULL, 0 }, /* Qc */
287: { NULL, 0 }, /* Ql */
288: { NULL, 0 }, /* Qo */
289: { NULL, 0 }, /* Qq */
290: { NULL, 0 }, /* Re */
291: { NULL, 0 }, /* Rs */
292: { NULL, 0 }, /* Sc */
293: { NULL, 0 }, /* So */
294: { NULL, 0 }, /* Sq */
295: { NULL, 0 }, /* Sm */
296: { NULL, 0 }, /* Sx */
297: { NULL, TYPE_Sy }, /* Sy */
298: { NULL, TYPE_Tn }, /* Tn */
299: { NULL, 0 }, /* Ux */
300: { NULL, 0 }, /* Xc */
301: { NULL, 0 }, /* Xo */
1.173 schwarze 302: { parse_mdoc_Fo, 0 }, /* Fo */
1.68 schwarze 303: { NULL, 0 }, /* Fc */
304: { NULL, 0 }, /* Oo */
305: { NULL, 0 }, /* Oc */
306: { NULL, 0 }, /* Bk */
307: { NULL, 0 }, /* Ek */
308: { NULL, 0 }, /* Bt */
309: { NULL, 0 }, /* Hf */
310: { NULL, 0 }, /* Fr */
311: { NULL, 0 }, /* Ud */
312: { NULL, TYPE_Lb }, /* Lb */
313: { NULL, 0 }, /* Lp */
314: { NULL, TYPE_Lk }, /* Lk */
315: { NULL, TYPE_Mt }, /* Mt */
316: { NULL, 0 }, /* Brq */
317: { NULL, 0 }, /* Bro */
318: { NULL, 0 }, /* Brc */
319: { NULL, 0 }, /* %C */
320: { NULL, 0 }, /* Es */
321: { NULL, 0 }, /* En */
322: { NULL, TYPE_Dx }, /* Dx */
323: { NULL, 0 }, /* %Q */
324: { NULL, 0 }, /* br */
325: { NULL, 0 }, /* sp */
326: { NULL, 0 }, /* %U */
327: { NULL, 0 }, /* Ta */
1.166 schwarze 328: { NULL, 0 }, /* ll */
1.1 kristaps 329: };
330:
1.141 schwarze 331:
1.1 kristaps 332: int
1.179 schwarze 333: mandocdb(int argc, char *argv[])
1.1 kristaps 334: {
1.187 schwarze 335: struct manconf conf;
336: struct mparse *mp;
1.206 schwarze 337: const char *path_arg, *progname;
1.187 schwarze 338: size_t j, sz;
1.59 schwarze 339: int ch, i;
1.50 kristaps 340:
1.207 schwarze 341: #if HAVE_PLEDGE
342: if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
1.214 schwarze 343: warn("pledge");
1.207 schwarze 344: return (int)MANDOCLEVEL_SYSERR;
345: }
346: #endif
347:
1.187 schwarze 348: memset(&conf, 0, sizeof(conf));
1.50 kristaps 349: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
350:
351: /*
1.141 schwarze 352: * We accept a few different invocations.
1.50 kristaps 353: * The CHECKOP macro makes sure that invocation styles don't
354: * clobber each other.
355: */
356: #define CHECKOP(_op, _ch) do \
357: if (OP_DEFAULT != (_op)) { \
1.198 schwarze 358: warnx("-%c: Conflicting option", (_ch)); \
1.50 kristaps 359: goto usage; \
360: } while (/*CONSTCOND*/0)
1.10 kristaps 361:
1.59 schwarze 362: path_arg = NULL;
1.38 schwarze 363: op = OP_DEFAULT;
1.1 kristaps 364:
1.126 schwarze 365: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 366: switch (ch) {
1.141 schwarze 367: case 'a':
1.12 schwarze 368: use_all = 1;
369: break;
1.141 schwarze 370: case 'C':
1.50 kristaps 371: CHECKOP(op, ch);
1.59 schwarze 372: path_arg = optarg;
1.38 schwarze 373: op = OP_CONFFILE;
1.34 schwarze 374: break;
1.141 schwarze 375: case 'D':
1.125 schwarze 376: debug++;
377: break;
1.141 schwarze 378: case 'd':
1.50 kristaps 379: CHECKOP(op, ch);
1.59 schwarze 380: path_arg = optarg;
1.5 kristaps 381: op = OP_UPDATE;
382: break;
1.141 schwarze 383: case 'n':
1.50 kristaps 384: nodb = 1;
385: break;
1.141 schwarze 386: case 'p':
1.126 schwarze 387: warnings = 1;
388: break;
1.141 schwarze 389: case 'Q':
1.116 schwarze 390: mparse_options |= MPARSE_QUICK;
1.102 schwarze 391: break;
1.141 schwarze 392: case 'T':
1.95 schwarze 393: if (strcmp(optarg, "utf8")) {
1.198 schwarze 394: warnx("-T%s: Unsupported output format",
395: optarg);
1.95 schwarze 396: goto usage;
397: }
398: write_utf8 = 1;
399: break;
1.141 schwarze 400: case 't':
1.50 kristaps 401: CHECKOP(op, ch);
1.38 schwarze 402: dup2(STDOUT_FILENO, STDERR_FILENO);
403: op = OP_TEST;
1.50 kristaps 404: nodb = warnings = 1;
1.38 schwarze 405: break;
1.141 schwarze 406: case 'u':
1.50 kristaps 407: CHECKOP(op, ch);
1.59 schwarze 408: path_arg = optarg;
1.5 kristaps 409: op = OP_DELETE;
410: break;
1.141 schwarze 411: case 'v':
1.126 schwarze 412: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 413: break;
1.1 kristaps 414: default:
1.38 schwarze 415: goto usage;
1.1 kristaps 416: }
417:
418: argc -= optind;
419: argv += optind;
420:
1.207 schwarze 421: #if HAVE_PLEDGE
1.210 schwarze 422: if (nodb) {
423: if (pledge("stdio rpath", NULL) == -1) {
1.214 schwarze 424: warn("pledge");
1.210 schwarze 425: return (int)MANDOCLEVEL_SYSERR;
426: }
1.207 schwarze 427: }
428: #endif
429:
1.38 schwarze 430: if (OP_CONFFILE == op && argc > 0) {
1.198 schwarze 431: warnx("-C: Too many arguments");
1.38 schwarze 432: goto usage;
433: }
434:
1.59 schwarze 435: exitcode = (int)MANDOCLEVEL_OK;
1.203 schwarze 436: mchars_alloc();
437: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
1.202 schwarze 438: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
439: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.50 kristaps 440:
441: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 442:
1.50 kristaps 443: /*
1.150 schwarze 444: * Most of these deal with a specific directory.
1.140 schwarze 445: * Jump into that directory first.
1.50 kristaps 446: */
1.165 schwarze 447: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.50 kristaps 448: goto out;
1.140 schwarze 449:
1.138 schwarze 450: if (dbopen(1)) {
1.140 schwarze 451: /*
452: * The existing database is usable. Process
453: * all files specified on the command-line.
454: */
1.207 schwarze 455: #if HAVE_PLEDGE
1.210 schwarze 456: if (!nodb) {
457: if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
1.214 schwarze 458: warn("pledge");
1.210 schwarze 459: exitcode = (int)MANDOCLEVEL_SYSERR;
460: goto out;
461: }
1.207 schwarze 462: }
463: #endif
1.140 schwarze 464: use_all = 1;
1.138 schwarze 465: for (i = 0; i < argc; i++)
466: filescan(argv[i]);
467: if (OP_TEST != op)
468: dbprune();
469: } else {
470: /*
471: * Database missing or corrupt.
472: * Recreate from scratch.
473: */
1.140 schwarze 474: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 475: op = OP_DEFAULT;
476: if (0 == treescan())
477: goto out;
478: if (0 == dbopen(0))
479: goto out;
480: }
1.50 kristaps 481: if (OP_DELETE != op)
1.176 schwarze 482: mpages_merge(mp);
1.138 schwarze 483: dbclose(OP_DEFAULT == op ? 0 : 1);
1.50 kristaps 484: } else {
485: /*
486: * If we have arguments, use them as our manpaths.
487: * If we don't, grok from manpath(1) or however else
1.187 schwarze 488: * manconf_parse() wants to do it.
1.50 kristaps 489: */
490: if (argc > 0) {
1.187 schwarze 491: conf.manpath.paths = mandoc_reallocarray(NULL,
1.144 schwarze 492: argc, sizeof(char *));
1.187 schwarze 493: conf.manpath.sz = (size_t)argc;
1.50 kristaps 494: for (i = 0; i < argc; i++)
1.187 schwarze 495: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.50 kristaps 496: } else
1.187 schwarze 497: manconf_parse(&conf, path_arg, NULL, NULL);
1.127 schwarze 498:
1.187 schwarze 499: if (conf.manpath.sz == 0) {
1.127 schwarze 500: exitcode = (int)MANDOCLEVEL_BADARG;
501: say("", "Empty manpath");
502: }
1.50 kristaps 503:
504: /*
1.69 schwarze 505: * First scan the tree rooted at a base directory, then
506: * build a new database and finally move it into place.
1.50 kristaps 507: * Ignore zero-length directories and strip trailing
508: * slashes.
509: */
1.187 schwarze 510: for (j = 0; j < conf.manpath.sz; j++) {
511: sz = strlen(conf.manpath.paths[j]);
512: if (sz && conf.manpath.paths[j][sz - 1] == '/')
513: conf.manpath.paths[j][--sz] = '\0';
1.50 kristaps 514: if (0 == sz)
515: continue;
1.66 schwarze 516:
517: if (j) {
1.202 schwarze 518: mandoc_ohash_init(&mpages, 6,
519: offsetof(struct mpage, inodev));
520: mandoc_ohash_init(&mlinks, 6,
521: offsetof(struct mlink, file));
1.66 schwarze 522: }
523:
1.187 schwarze 524: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.165 schwarze 525: continue;
1.59 schwarze 526: if (0 == treescan())
1.165 schwarze 527: continue;
1.59 schwarze 528: if (0 == dbopen(0))
1.165 schwarze 529: continue;
1.52 kristaps 530:
1.176 schwarze 531: mpages_merge(mp);
1.147 schwarze 532: if (warnings && !nodb &&
1.130 schwarze 533: ! (MPARSE_QUICK & mparse_options))
534: names_check();
1.59 schwarze 535: dbclose(0);
1.66 schwarze 536:
1.187 schwarze 537: if (j + 1 < conf.manpath.sz) {
1.80 schwarze 538: mpages_free();
1.78 schwarze 539: ohash_delete(&mpages);
1.83 schwarze 540: ohash_delete(&mlinks);
1.66 schwarze 541: }
1.50 kristaps 542: }
543: }
544: out:
1.187 schwarze 545: manconf_free(&conf);
1.168 schwarze 546: mparse_free(mp);
1.203 schwarze 547: mchars_free();
1.80 schwarze 548: mpages_free();
1.78 schwarze 549: ohash_delete(&mpages);
1.83 schwarze 550: ohash_delete(&mlinks);
1.197 schwarze 551: return exitcode;
1.50 kristaps 552: usage:
1.206 schwarze 553: progname = getprogname();
1.126 schwarze 554: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
555: " %s [-aDnpQ] [-Tutf8] dir ...\n"
556: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
557: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 558: " %s [-Q] -t file ...\n",
1.206 schwarze 559: progname, progname, progname, progname, progname);
1.50 kristaps 560:
1.197 schwarze 561: return (int)MANDOCLEVEL_BADARG;
1.50 kristaps 562: }
563:
564: /*
1.59 schwarze 565: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 566: * We use fts(), scanning directory parts along the way for clues to our
567: * section and architecture.
568: *
569: * If use_all has been specified, grok all files.
570: * If not, sanitise paths to the following:
571: *
1.141 schwarze 572: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 573: * or
574: * [./]cat<section>[/<arch>]/<name>.0
575: *
1.216 schwarze 576: * TODO: accommodate for multi-language directories.
1.50 kristaps 577: */
578: static int
1.59 schwarze 579: treescan(void)
1.50 kristaps 580: {
1.139 schwarze 581: char buf[PATH_MAX];
1.50 kristaps 582: FTS *f;
583: FTSENT *ff;
1.84 schwarze 584: struct mlink *mlink;
1.124 schwarze 585: int dform, gzip;
1.94 schwarze 586: char *dsec, *arch, *fsec, *cp;
587: const char *path;
1.50 kristaps 588: const char *argv[2];
589:
590: argv[0] = ".";
591: argv[1] = (char *)NULL;
592:
1.139 schwarze 593: f = fts_open((char * const *)argv,
594: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.200 schwarze 595: if (f == NULL) {
1.59 schwarze 596: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 597: say("", "&fts_open");
1.197 schwarze 598: return 0;
1.50 kristaps 599: }
600:
601: dsec = arch = NULL;
602: dform = FORM_NONE;
603:
1.200 schwarze 604: while ((ff = fts_read(f)) != NULL) {
1.50 kristaps 605: path = ff->fts_path + 2;
1.139 schwarze 606: switch (ff->fts_info) {
607:
608: /*
609: * Symbolic links require various sanity checks,
610: * then get handled just like regular files.
611: */
1.141 schwarze 612: case FTS_SL:
1.200 schwarze 613: if (realpath(path, buf) == NULL) {
1.139 schwarze 614: if (warnings)
615: say(path, "&realpath");
616: continue;
617: }
1.184 schwarze 618: if (strstr(buf, basedir) != buf
619: #ifdef HOMEBREWDIR
620: && strstr(buf, HOMEBREWDIR) != buf
621: #endif
622: ) {
1.139 schwarze 623: if (warnings) say("",
624: "%s: outside base directory", buf);
625: continue;
626: }
627: /* Use logical inode to avoid mpages dupe. */
1.200 schwarze 628: if (stat(path, ff->fts_statp) == -1) {
1.139 schwarze 629: if (warnings)
630: say(path, "&stat");
631: continue;
632: }
633: /* FALLTHROUGH */
634:
1.50 kristaps 635: /*
1.83 schwarze 636: * If we're a regular file, add an mlink by using the
1.50 kristaps 637: * stored directory data and handling the filename.
638: */
1.141 schwarze 639: case FTS_F:
1.200 schwarze 640: if ( ! strcmp(path, MANDOC_DB))
1.60 schwarze 641: continue;
1.50 kristaps 642: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 643: if (warnings)
1.59 schwarze 644: say(path, "Extraneous file");
1.50 kristaps 645: continue;
1.124 schwarze 646: }
647: gzip = 0;
648: fsec = NULL;
1.200 schwarze 649: while (fsec == NULL) {
1.124 schwarze 650: fsec = strrchr(ff->fts_name, '.');
1.200 schwarze 651: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.124 schwarze 652: break;
653: gzip = 1;
654: *fsec = '\0';
655: fsec = NULL;
656: }
1.200 schwarze 657: if (fsec == NULL) {
1.60 schwarze 658: if ( ! use_all) {
1.56 schwarze 659: if (warnings)
1.60 schwarze 660: say(path,
661: "No filename suffix");
1.50 kristaps 662: continue;
663: }
1.200 schwarze 664: } else if ( ! strcmp(++fsec, "html")) {
1.60 schwarze 665: if (warnings)
666: say(path, "Skip html");
667: continue;
1.200 schwarze 668: } else if ( ! strcmp(fsec, "ps")) {
1.60 schwarze 669: if (warnings)
670: say(path, "Skip ps");
671: continue;
1.200 schwarze 672: } else if ( ! strcmp(fsec, "pdf")) {
1.60 schwarze 673: if (warnings)
674: say(path, "Skip pdf");
675: continue;
676: } else if ( ! use_all &&
1.200 schwarze 677: ((dform == FORM_SRC &&
1.185 schwarze 678: strncmp(fsec, dsec, strlen(dsec))) ||
1.200 schwarze 679: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.60 schwarze 680: if (warnings)
681: say(path, "Wrong filename suffix");
682: continue;
1.66 schwarze 683: } else
1.84 schwarze 684: fsec[-1] = '\0';
1.94 schwarze 685:
1.84 schwarze 686: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 687: if (strlcpy(mlink->file, path,
688: sizeof(mlink->file)) >=
689: sizeof(mlink->file)) {
690: say(path, "Filename too long");
691: free(mlink);
692: continue;
693: }
1.84 schwarze 694: mlink->dform = dform;
1.94 schwarze 695: mlink->dsec = dsec;
696: mlink->arch = arch;
697: mlink->name = ff->fts_name;
698: mlink->fsec = fsec;
1.124 schwarze 699: mlink->gzip = gzip;
1.84 schwarze 700: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 701: continue;
1.139 schwarze 702:
1.141 schwarze 703: case FTS_D:
704: case FTS_DP:
1.139 schwarze 705: break;
706:
707: default:
1.60 schwarze 708: if (warnings)
709: say(path, "Not a regular file");
1.50 kristaps 710: continue;
1.60 schwarze 711: }
1.1 kristaps 712:
1.50 kristaps 713: switch (ff->fts_level) {
1.141 schwarze 714: case 0:
1.50 kristaps 715: /* Ignore the root directory. */
716: break;
1.141 schwarze 717: case 1:
1.50 kristaps 718: /*
719: * This might contain manX/ or catX/.
720: * Try to infer this from the name.
721: * If we're not in use_all, enforce it.
722: */
723: cp = ff->fts_name;
1.200 schwarze 724: if (ff->fts_info == FTS_DP) {
725: dform = FORM_NONE;
726: dsec = NULL;
1.50 kristaps 727: break;
1.200 schwarze 728: }
1.1 kristaps 729:
1.200 schwarze 730: if ( ! strncmp(cp, "man", 3)) {
1.50 kristaps 731: dform = FORM_SRC;
1.66 schwarze 732: dsec = cp + 3;
1.200 schwarze 733: } else if ( ! strncmp(cp, "cat", 3)) {
1.50 kristaps 734: dform = FORM_CAT;
1.66 schwarze 735: dsec = cp + 3;
1.94 schwarze 736: } else {
737: dform = FORM_NONE;
738: dsec = NULL;
1.50 kristaps 739: }
1.1 kristaps 740:
1.200 schwarze 741: if (dsec != NULL || use_all)
1.50 kristaps 742: break;
1.1 kristaps 743:
1.56 schwarze 744: if (warnings)
1.59 schwarze 745: say(path, "Unknown directory part");
1.50 kristaps 746: fts_set(f, ff, FTS_SKIP);
747: break;
1.141 schwarze 748: case 2:
1.50 kristaps 749: /*
750: * Possibly our architecture.
751: * If we're descending, keep tabs on it.
752: */
1.200 schwarze 753: if (ff->fts_info != FTS_DP && dsec != NULL)
1.66 schwarze 754: arch = ff->fts_name;
1.94 schwarze 755: else
756: arch = NULL;
1.50 kristaps 757: break;
758: default:
1.200 schwarze 759: if (ff->fts_info == FTS_DP || use_all)
1.50 kristaps 760: break;
1.56 schwarze 761: if (warnings)
1.59 schwarze 762: say(path, "Extraneous directory part");
1.50 kristaps 763: fts_set(f, ff, FTS_SKIP);
764: break;
1.5 kristaps 765: }
1.50 kristaps 766: }
767:
768: fts_close(f);
1.197 schwarze 769: return 1;
1.50 kristaps 770: }
1.5 kristaps 771:
1.50 kristaps 772: /*
1.89 schwarze 773: * Add a file to the mlinks table.
1.50 kristaps 774: * Do not verify that it's a "valid" looking manpage (we'll do that
775: * later).
776: *
777: * Try to infer the manual section, architecture, and page name from the
778: * path, assuming it looks like
779: *
1.141 schwarze 780: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 781: * or
782: * [./]cat<section>[/<arch>]/<name>.0
783: *
784: * See treescan() for the fts(3) version of this.
785: */
786: static void
1.59 schwarze 787: filescan(const char *file)
1.50 kristaps 788: {
1.59 schwarze 789: char buf[PATH_MAX];
1.84 schwarze 790: struct stat st;
791: struct mlink *mlink;
1.59 schwarze 792: char *p, *start;
1.5 kristaps 793:
1.50 kristaps 794: assert(use_all);
1.5 kristaps 795:
1.50 kristaps 796: if (0 == strncmp(file, "./", 2))
797: file += 2;
1.5 kristaps 798:
1.139 schwarze 799: /*
800: * We have to do lstat(2) before realpath(3) loses
801: * the information whether this is a symbolic link.
802: * We need to know that because for symbolic links,
803: * we want to use the orginal file name, while for
804: * regular files, we want to use the real path.
805: */
806: if (-1 == lstat(file, &st)) {
807: exitcode = (int)MANDOCLEVEL_BADARG;
808: say(file, "&lstat");
809: return;
810: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
811: exitcode = (int)MANDOCLEVEL_BADARG;
812: say(file, "Not a regular file");
813: return;
814: }
815:
816: /*
817: * We have to resolve the file name to the real path
818: * in any case for the base directory check.
819: */
1.59 schwarze 820: if (NULL == realpath(file, buf)) {
821: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 822: say(file, "&realpath");
1.59 schwarze 823: return;
1.106 schwarze 824: }
825:
1.150 schwarze 826: if (OP_TEST == op)
1.106 schwarze 827: start = buf;
1.150 schwarze 828: else if (strstr(buf, basedir) == buf)
829: start = buf + strlen(basedir);
1.184 schwarze 830: #ifdef HOMEBREWDIR
831: else if (strstr(buf, HOMEBREWDIR) == buf)
832: start = buf;
833: #endif
1.106 schwarze 834: else {
1.59 schwarze 835: exitcode = (int)MANDOCLEVEL_BADARG;
836: say("", "%s: outside base directory", buf);
837: return;
1.106 schwarze 838: }
839:
1.139 schwarze 840: /*
841: * Now we are sure the file is inside our tree.
842: * If it is a symbolic link, ignore the real path
843: * and use the original name.
844: * This implies passing stuff like "cat1/../man1/foo.1"
845: * on the command line won't work. So don't do that.
846: * Note the stat(2) can still fail if the link target
847: * doesn't exist.
848: */
849: if (S_IFLNK & st.st_mode) {
850: if (-1 == stat(buf, &st)) {
851: exitcode = (int)MANDOCLEVEL_BADARG;
852: say(file, "&stat");
853: return;
854: }
1.143 schwarze 855: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
856: say(file, "Filename too long");
857: return;
858: }
1.150 schwarze 859: start = buf;
860: if (OP_TEST != op && strstr(buf, basedir) == buf)
861: start += strlen(basedir);
1.50 kristaps 862: }
1.106 schwarze 863:
1.84 schwarze 864: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 865: mlink->dform = FORM_NONE;
1.143 schwarze 866: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
867: sizeof(mlink->file)) {
868: say(start, "Filename too long");
1.180 schwarze 869: free(mlink);
1.143 schwarze 870: return;
871: }
1.17 schwarze 872:
1.50 kristaps 873: /*
874: * First try to guess our directory structure.
875: * If we find a separator, try to look for man* or cat*.
876: * If we find one of these and what's underneath is a directory,
877: * assume it's an architecture.
878: */
879: if (NULL != (p = strchr(start, '/'))) {
880: *p++ = '\0';
881: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 882: mlink->dform = FORM_SRC;
1.94 schwarze 883: mlink->dsec = start + 3;
1.50 kristaps 884: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 885: mlink->dform = FORM_CAT;
1.94 schwarze 886: mlink->dsec = start + 3;
1.17 schwarze 887: }
1.5 kristaps 888:
1.50 kristaps 889: start = p;
1.84 schwarze 890: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 891: *p++ = '\0';
1.94 schwarze 892: mlink->arch = start;
1.50 kristaps 893: start = p;
1.84 schwarze 894: }
1.50 kristaps 895: }
896:
897: /*
898: * Now check the file suffix.
899: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
900: */
901: p = strrchr(start, '\0');
902: while (p-- > start && '/' != *p && '.' != *p)
903: /* Loop. */ ;
904:
905: if ('.' == *p) {
906: *p++ = '\0';
1.94 schwarze 907: mlink->fsec = p;
1.5 kristaps 908: }
909:
1.10 kristaps 910: /*
1.50 kristaps 911: * Now try to parse the name.
912: * Use the filename portion of the path.
1.10 kristaps 913: */
1.84 schwarze 914: mlink->name = start;
1.50 kristaps 915: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 916: mlink->name = p + 1;
1.50 kristaps 917: *p = '\0';
1.84 schwarze 918: }
919: mlink_add(mlink, &st);
1.50 kristaps 920: }
1.5 kristaps 921:
1.50 kristaps 922: static void
1.84 schwarze 923: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 924: {
1.83 schwarze 925: struct inodev inodev;
1.78 schwarze 926: struct mpage *mpage;
1.83 schwarze 927: unsigned int slot;
1.50 kristaps 928:
1.84 schwarze 929: assert(NULL != mlink->file);
1.50 kristaps 930:
1.94 schwarze 931: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
932: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
933: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
934: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 935:
936: if ('0' == *mlink->fsec) {
937: free(mlink->fsec);
938: mlink->fsec = mandoc_strdup(mlink->dsec);
939: mlink->fform = FORM_CAT;
940: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
941: mlink->fform = FORM_SRC;
1.74 schwarze 942: else
1.84 schwarze 943: mlink->fform = FORM_NONE;
1.82 schwarze 944:
1.83 schwarze 945: slot = ohash_qlookup(&mlinks, mlink->file);
946: assert(NULL == ohash_find(&mlinks, slot));
947: ohash_insert(&mlinks, slot, mlink);
948:
1.186 schwarze 949: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 950: inodev.st_ino = st->st_ino;
951: inodev.st_dev = st->st_dev;
952: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
953: sizeof(struct inodev), inodev.st_ino);
954: mpage = ohash_find(&mpages, slot);
955: if (NULL == mpage) {
956: mpage = mandoc_calloc(1, sizeof(struct mpage));
957: mpage->inodev.st_ino = inodev.st_ino;
958: mpage->inodev.st_dev = inodev.st_dev;
959: ohash_insert(&mpages, slot, mpage);
960: } else
1.85 schwarze 961: mlink->next = mpage->mlinks;
1.82 schwarze 962: mpage->mlinks = mlink;
1.118 schwarze 963: mlink->mpage = mpage;
1.82 schwarze 964: }
965:
966: static void
967: mlink_free(struct mlink *mlink)
968: {
969:
970: free(mlink->dsec);
971: free(mlink->arch);
972: free(mlink->name);
973: free(mlink->fsec);
974: free(mlink);
1.50 kristaps 975: }
1.3 kristaps 976:
1.50 kristaps 977: static void
1.78 schwarze 978: mpages_free(void)
1.50 kristaps 979: {
1.78 schwarze 980: struct mpage *mpage;
1.82 schwarze 981: struct mlink *mlink;
1.80 schwarze 982: unsigned int slot;
1.3 kristaps 983:
1.80 schwarze 984: mpage = ohash_first(&mpages, &slot);
985: while (NULL != mpage) {
1.82 schwarze 986: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 987: mpage->mlinks = mlink->next;
1.82 schwarze 988: mlink_free(mlink);
989: }
1.78 schwarze 990: free(mpage->sec);
991: free(mpage->arch);
1.82 schwarze 992: free(mpage->title);
993: free(mpage->desc);
1.78 schwarze 994: free(mpage);
1.80 schwarze 995: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 996: }
997: }
1.38 schwarze 998:
1.50 kristaps 999: /*
1.89 schwarze 1000: * For each mlink to the mpage, check whether the path looks like
1001: * it is formatted, and if it does, check whether a source manual
1002: * exists by the same name, ignoring the suffix.
1003: * If both conditions hold, drop the mlink.
1004: */
1005: static void
1006: mlinks_undupe(struct mpage *mpage)
1007: {
1008: char buf[PATH_MAX];
1009: struct mlink **prev;
1010: struct mlink *mlink;
1011: char *bufp;
1012:
1013: mpage->form = FORM_CAT;
1.90 schwarze 1014: prev = &mpage->mlinks;
1015: while (NULL != (mlink = *prev)) {
1.89 schwarze 1016: if (FORM_CAT != mlink->dform) {
1017: mpage->form = FORM_NONE;
1.90 schwarze 1018: goto nextlink;
1.89 schwarze 1019: }
1.143 schwarze 1020: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1021: bufp = strstr(buf, "cat");
1022: assert(NULL != bufp);
1023: memcpy(bufp, "man", 3);
1024: if (NULL != (bufp = strrchr(buf, '.')))
1025: *++bufp = '\0';
1.143 schwarze 1026: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1027: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1028: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1029: goto nextlink;
1.89 schwarze 1030: if (warnings)
1031: say(mlink->file, "Man source exists: %s", buf);
1032: if (use_all)
1.90 schwarze 1033: goto nextlink;
1.89 schwarze 1034: *prev = mlink->next;
1035: mlink_free(mlink);
1.90 schwarze 1036: continue;
1037: nextlink:
1038: prev = &(*prev)->next;
1.89 schwarze 1039: }
1040: }
1041:
1.131 schwarze 1042: static void
1.93 schwarze 1043: mlink_check(struct mpage *mpage, struct mlink *mlink)
1044: {
1.131 schwarze 1045: struct str *str;
1046: unsigned int slot;
1.93 schwarze 1047:
1048: /*
1049: * Check whether the manual section given in a file
1050: * agrees with the directory where the file is located.
1051: * Some manuals have suffixes like (3p) on their
1052: * section number either inside the file or in the
1053: * directory name, some are linked into more than one
1054: * section, like encrypt(1) = makekey(8).
1055: */
1056:
1057: if (FORM_SRC == mpage->form &&
1.131 schwarze 1058: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1059: say(mlink->file, "Section \"%s\" manual in %s directory",
1060: mpage->sec, mlink->dsec);
1061:
1062: /*
1063: * Manual page directories exist for each kernel
1064: * architecture as returned by machine(1).
1065: * However, many manuals only depend on the
1066: * application architecture as returned by arch(1).
1067: * For example, some (2/ARM) manuals are shared
1068: * across the "armish" and "zaurus" kernel
1069: * architectures.
1070: * A few manuals are even shared across completely
1071: * different architectures, for example fdformat(1)
1072: * on amd64, i386, sparc, and sparc64.
1073: */
1074:
1.131 schwarze 1075: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1076: say(mlink->file, "Architecture \"%s\" manual in "
1077: "\"%s\" directory", mpage->arch, mlink->arch);
1078:
1.131 schwarze 1079: /*
1080: * XXX
1.133 schwarze 1081: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1082: */
1083:
1084: if (FORM_CAT == mpage->form)
1085: return;
1086:
1087: /*
1088: * Check whether this mlink
1089: * appears as a name in the NAME section.
1090: */
1.93 schwarze 1091:
1.133 schwarze 1092: slot = ohash_qlookup(&names, mlink->name);
1093: str = ohash_find(&names, slot);
1.131 schwarze 1094: assert(NULL != str);
1.133 schwarze 1095: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1096: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1097: }
1098:
1.89 schwarze 1099: /*
1.80 schwarze 1100: * Run through the files in the global vector "mpages"
1101: * and add them to the database specified in "basedir".
1.50 kristaps 1102: *
1103: * This handles the parsing scheme itself, using the cues of directory
1104: * and filename to determine whether the file is parsable or not.
1105: */
1.59 schwarze 1106: static void
1.176 schwarze 1107: mpages_merge(struct mparse *mp)
1.50 kristaps 1108: {
1.113 schwarze 1109: char any[] = "any";
1.118 schwarze 1110: struct mpage *mpage, *mpage_dest;
1111: struct mlink *mlink, *mlink_dest;
1.191 schwarze 1112: struct roff_man *man;
1.118 schwarze 1113: char *sodest;
1.112 schwarze 1114: char *cp;
1.162 schwarze 1115: int fd;
1.101 schwarze 1116: unsigned int pslot;
1.69 schwarze 1117:
1.171 schwarze 1118: if ( ! nodb)
1.107 schwarze 1119: SQL_EXEC("BEGIN TRANSACTION");
1120:
1.80 schwarze 1121: mpage = ohash_first(&mpages, &pslot);
1.171 schwarze 1122: while (mpage != NULL) {
1.89 schwarze 1123: mlinks_undupe(mpage);
1.182 schwarze 1124: if ((mlink = mpage->mlinks) == NULL) {
1.89 schwarze 1125: mpage = ohash_next(&mpages, &pslot);
1126: continue;
1.50 kristaps 1127: }
1128:
1.133 schwarze 1129: name_mask = NAME_MASK;
1.202 schwarze 1130: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1131: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.50 kristaps 1132: mparse_reset(mp);
1133: man = NULL;
1.124 schwarze 1134: sodest = NULL;
1135:
1.213 schwarze 1136: if ((fd = mparse_open(mp, mlink->file)) == -1) {
1.182 schwarze 1137: say(mlink->file, "&open");
1.162 schwarze 1138: goto nextpage;
1.124 schwarze 1139: }
1.14 schwarze 1140:
1141: /*
1.183 schwarze 1142: * Interpret the file as mdoc(7) or man(7) source
1143: * code, unless it is known to be formatted.
1.14 schwarze 1144: */
1.182 schwarze 1145: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1146: mparse_readfd(mp, fd, mlink->file);
1.212 schwarze 1147: close(fd);
1.192 schwarze 1148: mparse_result(mp, &man, &sodest);
1.76 schwarze 1149: }
1.14 schwarze 1150:
1.171 schwarze 1151: if (sodest != NULL) {
1.118 schwarze 1152: mlink_dest = ohash_find(&mlinks,
1153: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1154: if (mlink_dest == NULL) {
1155: mandoc_asprintf(&cp, "%s.gz", sodest);
1156: mlink_dest = ohash_find(&mlinks,
1157: ohash_qlookup(&mlinks, cp));
1158: free(cp);
1159: }
1160: if (mlink_dest != NULL) {
1.118 schwarze 1161:
1162: /* The .so target exists. */
1163:
1164: mpage_dest = mlink_dest->mpage;
1165: while (1) {
1166: mlink->mpage = mpage_dest;
1167:
1168: /*
1169: * If the target was already
1170: * processed, add the links
1171: * to the database now.
1172: * Otherwise, this will
1173: * happen when we come
1174: * to the target.
1175: */
1176:
1.137 schwarze 1177: if (mpage_dest->pageid)
1.169 schwarze 1178: dbadd_mlink_name(mlink);
1.118 schwarze 1179:
1.171 schwarze 1180: if (mlink->next == NULL)
1.118 schwarze 1181: break;
1182: mlink = mlink->next;
1183: }
1184:
1185: /* Move all links to the target. */
1186:
1187: mlink->next = mlink_dest->next;
1188: mlink_dest->next = mpage->mlinks;
1189: mpage->mlinks = NULL;
1190: }
1.124 schwarze 1191: goto nextpage;
1.192 schwarze 1192: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.204 schwarze 1193: mdoc_validate(man);
1.82 schwarze 1194: mpage->form = FORM_SRC;
1.193 schwarze 1195: mpage->sec = man->meta.msec;
1.155 schwarze 1196: mpage->sec = mandoc_strdup(
1.171 schwarze 1197: mpage->sec == NULL ? "" : mpage->sec);
1.193 schwarze 1198: mpage->arch = man->meta.arch;
1.84 schwarze 1199: mpage->arch = mandoc_strdup(
1.171 schwarze 1200: mpage->arch == NULL ? "" : mpage->arch);
1.193 schwarze 1201: mpage->title = mandoc_strdup(man->meta.title);
1.192 schwarze 1202: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.205 schwarze 1203: man_validate(man);
1.82 schwarze 1204: mpage->form = FORM_SRC;
1.193 schwarze 1205: mpage->sec = mandoc_strdup(man->meta.msec);
1.182 schwarze 1206: mpage->arch = mandoc_strdup(mlink->arch);
1.193 schwarze 1207: mpage->title = mandoc_strdup(man->meta.title);
1.76 schwarze 1208: } else {
1.82 schwarze 1209: mpage->form = FORM_CAT;
1.182 schwarze 1210: mpage->sec = mandoc_strdup(mlink->dsec);
1211: mpage->arch = mandoc_strdup(mlink->arch);
1212: mpage->title = mandoc_strdup(mlink->name);
1.76 schwarze 1213: }
1.97 schwarze 1214: putkey(mpage, mpage->sec, TYPE_sec);
1.164 schwarze 1215: if (*mpage->arch != '\0')
1216: putkey(mpage, mpage->arch, TYPE_arch);
1.50 kristaps 1217:
1.182 schwarze 1218: for ( ; mlink != NULL; mlink = mlink->next) {
1.97 schwarze 1219: if ('\0' != *mlink->dsec)
1220: putkey(mpage, mlink->dsec, TYPE_sec);
1221: if ('\0' != *mlink->fsec)
1222: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1223: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1224: any : mlink->arch, TYPE_arch);
1.133 schwarze 1225: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1226: }
1.12 schwarze 1227:
1.174 schwarze 1228: assert(mpage->desc == NULL);
1.192 schwarze 1229: if (man != NULL && man->macroset == MACROSET_MDOC)
1.193 schwarze 1230: parse_mdoc(mpage, &man->meta, man->first);
1.174 schwarze 1231: else if (man != NULL)
1.193 schwarze 1232: parse_man(mpage, &man->meta, man->first);
1.50 kristaps 1233: else
1.162 schwarze 1234: parse_cat(mpage, fd);
1.174 schwarze 1235: if (mpage->desc == NULL)
1.135 schwarze 1236: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1237:
1238: if (warnings && !use_all)
1239: for (mlink = mpage->mlinks; mlink;
1240: mlink = mlink->next)
1241: mlink_check(mpage, mlink);
1.44 kristaps 1242:
1.176 schwarze 1243: dbadd(mpage);
1.182 schwarze 1244: mlink = mpage->mlinks;
1.124 schwarze 1245:
1246: nextpage:
1.66 schwarze 1247: ohash_delete(&strings);
1.133 schwarze 1248: ohash_delete(&names);
1.80 schwarze 1249: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1250: }
1.107 schwarze 1251:
1252: if (0 == nodb)
1253: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1254: }
1255:
1256: static void
1257: names_check(void)
1258: {
1259: sqlite3_stmt *stmt;
1260: const char *name, *sec, *arch, *key;
1261:
1262: sqlite3_prepare_v2(db,
1263: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1264: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1265: "WHERE bits & ? AND NOT EXISTS ("
1266: "SELECT pageid FROM mlinks "
1.133 schwarze 1267: "WHERE mlinks.pageid == names.pageid "
1268: "AND mlinks.name == names.name"
1.130 schwarze 1269: ")"
1270: ") JOIN ("
1.142 schwarze 1271: "SELECT sec, arch, name, pageid FROM mlinks "
1272: "GROUP BY pageid"
1.130 schwarze 1273: ") USING (pageid);",
1274: -1, &stmt, NULL);
1275:
1.201 schwarze 1276: if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK)
1.134 schwarze 1277: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1278:
1.201 schwarze 1279: while (sqlite3_step(stmt) == SQLITE_ROW) {
1.154 schwarze 1280: name = (const char *)sqlite3_column_text(stmt, 0);
1281: sec = (const char *)sqlite3_column_text(stmt, 1);
1282: arch = (const char *)sqlite3_column_text(stmt, 2);
1283: key = (const char *)sqlite3_column_text(stmt, 3);
1.130 schwarze 1284: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1285: '\0' == *arch ? "" : "/",
1286: '\0' == *arch ? "" : arch, key);
1287: }
1288: sqlite3_finalize(stmt);
1.50 kristaps 1289: }
1.12 schwarze 1290:
1.50 kristaps 1291: static void
1.124 schwarze 1292: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1293: {
1294: FILE *stream;
1295: char *line, *p, *title;
1.209 schwarze 1296: size_t linesz, plen, titlesz;
1297: ssize_t len;
1298: int offs;
1.12 schwarze 1299:
1.124 schwarze 1300: stream = (-1 == fd) ?
1301: fopen(mpage->mlinks->file, "r") :
1302: fdopen(fd, "r");
1303: if (NULL == stream) {
1.159 schwarze 1304: if (-1 != fd)
1305: close(fd);
1.56 schwarze 1306: if (warnings)
1.123 schwarze 1307: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1308: return;
1309: }
1.1 kristaps 1310:
1.209 schwarze 1311: line = NULL;
1312: linesz = 0;
1313:
1.50 kristaps 1314: /* Skip to first blank line. */
1.1 kristaps 1315:
1.209 schwarze 1316: while (getline(&line, &linesz, stream) != -1)
1317: if (*line == '\n')
1.50 kristaps 1318: break;
1.1 kristaps 1319:
1.50 kristaps 1320: /*
1321: * Assume the first line that is not indented
1322: * is the first section header. Skip to it.
1323: */
1.1 kristaps 1324:
1.209 schwarze 1325: while (getline(&line, &linesz, stream) != -1)
1326: if (*line != '\n' && *line != ' ')
1.50 kristaps 1327: break;
1.141 schwarze 1328:
1.50 kristaps 1329: /*
1330: * Read up until the next section into a buffer.
1331: * Strip the leading and trailing newline from each read line,
1332: * appending a trailing space.
1333: * Ignore empty (whitespace-only) lines.
1334: */
1.1 kristaps 1335:
1.50 kristaps 1336: titlesz = 0;
1337: title = NULL;
1.38 schwarze 1338:
1.209 schwarze 1339: while ((len = getline(&line, &linesz, stream)) != -1) {
1340: if (*line != ' ')
1.50 kristaps 1341: break;
1.209 schwarze 1342: offs = 0;
1343: while (isspace((unsigned char)line[offs]))
1344: offs++;
1345: if (line[offs] == '\0')
1.38 schwarze 1346: continue;
1.209 schwarze 1347: title = mandoc_realloc(title, titlesz + len - offs);
1348: memcpy(title + titlesz, line + offs, len - offs);
1349: titlesz += len - offs;
1.50 kristaps 1350: title[titlesz - 1] = ' ';
1351: }
1.209 schwarze 1352: free(line);
1.38 schwarze 1353:
1.50 kristaps 1354: /*
1355: * If no page content can be found, or the input line
1356: * is already the next section header, or there is no
1357: * trailing newline, reuse the page title as the page
1358: * description.
1359: */
1.44 kristaps 1360:
1.50 kristaps 1361: if (NULL == title || '\0' == *title) {
1.56 schwarze 1362: if (warnings)
1.82 schwarze 1363: say(mpage->mlinks->file,
1364: "Cannot find NAME section");
1.50 kristaps 1365: fclose(stream);
1366: free(title);
1367: return;
1368: }
1.1 kristaps 1369:
1.209 schwarze 1370: title[titlesz - 1] = '\0';
1.33 schwarze 1371:
1.50 kristaps 1372: /*
1373: * Skip to the first dash.
1374: * Use the remaining line as the description (no more than 70
1375: * bytes).
1376: */
1.33 schwarze 1377:
1.50 kristaps 1378: if (NULL != (p = strstr(title, "- "))) {
1379: for (p += 2; ' ' == *p || '\b' == *p; p++)
1380: /* Skip to next word. */ ;
1381: } else {
1.56 schwarze 1382: if (warnings)
1.82 schwarze 1383: say(mpage->mlinks->file,
1384: "No dash in title line");
1.50 kristaps 1385: p = title;
1386: }
1.38 schwarze 1387:
1.50 kristaps 1388: plen = strlen(p);
1.1 kristaps 1389:
1.50 kristaps 1390: /* Strip backspace-encoding from line. */
1.1 kristaps 1391:
1.50 kristaps 1392: while (NULL != (line = memchr(p, '\b', plen))) {
1393: len = line - p;
1394: if (0 == len) {
1395: memmove(line, line + 1, plen--);
1396: continue;
1.141 schwarze 1397: }
1.50 kristaps 1398: memmove(line - 1, line + 1, plen - len);
1399: plen -= 2;
1400: }
1.1 kristaps 1401:
1.78 schwarze 1402: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1403: fclose(stream);
1404: free(title);
1405: }
1.1 kristaps 1406:
1.50 kristaps 1407: /*
1408: * Put a type/word pair into the word database for this particular file.
1409: */
1410: static void
1.112 schwarze 1411: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1412: {
1.112 schwarze 1413: char *cp;
1.18 kristaps 1414:
1.50 kristaps 1415: assert(NULL != value);
1.112 schwarze 1416: if (TYPE_arch == type)
1417: for (cp = value; *cp; cp++)
1418: if (isupper((unsigned char)*cp))
1419: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1420: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1421: }
1422:
1423: /*
1.50 kristaps 1424: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1425: */
1426: static void
1.78 schwarze 1427: putmdockey(const struct mpage *mpage,
1.189 schwarze 1428: const struct roff_node *n, uint64_t m)
1.3 kristaps 1429: {
1.18 kristaps 1430:
1.50 kristaps 1431: for ( ; NULL != n; n = n->next) {
1432: if (NULL != n->child)
1.78 schwarze 1433: putmdockey(mpage, n->child, m);
1.188 schwarze 1434: if (n->type == ROFFT_TEXT)
1.78 schwarze 1435: putkey(mpage, n->string, m);
1.50 kristaps 1436: }
1437: }
1.18 kristaps 1438:
1.61 schwarze 1439: static void
1.190 schwarze 1440: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1441: const struct roff_node *n)
1.50 kristaps 1442: {
1.189 schwarze 1443: const struct roff_node *head, *body;
1.121 schwarze 1444: char *start, *title;
1.50 kristaps 1445: char byte;
1.121 schwarze 1446: size_t sz;
1.18 kristaps 1447:
1.215 schwarze 1448: if (n == NULL)
1.61 schwarze 1449: return;
1.18 kristaps 1450:
1.50 kristaps 1451: /*
1452: * We're only searching for one thing: the first text child in
1453: * the BODY of a NAME section. Since we don't keep track of
1454: * sections in -man, run some hoops to find out whether we're in
1455: * the correct section or not.
1456: */
1.18 kristaps 1457:
1.188 schwarze 1458: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.50 kristaps 1459: body = n;
1.215 schwarze 1460: if ((head = body->parent->head) != NULL &&
1461: (head = head->child) != NULL &&
1462: head->next == NULL &&
1.188 schwarze 1463: head->type == ROFFT_TEXT &&
1.215 schwarze 1464: strcmp(head->string, "NAME") == 0 &&
1465: body->child != NULL) {
1.3 kristaps 1466:
1.50 kristaps 1467: /*
1468: * Suck the entire NAME section into memory.
1469: * Yes, we might run away.
1470: * But too many manuals have big, spread-out
1471: * NAME sections over many lines.
1472: */
1.3 kristaps 1473:
1.121 schwarze 1474: title = NULL;
1.194 schwarze 1475: deroff(&title, body);
1.50 kristaps 1476: if (NULL == title)
1.61 schwarze 1477: return;
1.18 kristaps 1478:
1.141 schwarze 1479: /*
1.50 kristaps 1480: * Go through a special heuristic dance here.
1481: * Conventionally, one or more manual names are
1482: * comma-specified prior to a whitespace, then a
1483: * dash, then a description. Try to puzzle out
1484: * the name parts here.
1485: */
1.18 kristaps 1486:
1.121 schwarze 1487: start = title;
1.50 kristaps 1488: for ( ;; ) {
1489: sz = strcspn(start, " ,");
1490: if ('\0' == start[sz])
1491: break;
1.1 kristaps 1492:
1.50 kristaps 1493: byte = start[sz];
1494: start[sz] = '\0';
1.110 schwarze 1495:
1496: /*
1497: * Assume a stray trailing comma in the
1498: * name list if a name begins with a dash.
1499: */
1500:
1501: if ('-' == start[0] ||
1502: ('\\' == start[0] && '-' == start[1]))
1503: break;
1.1 kristaps 1504:
1.133 schwarze 1505: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1506: if ( ! (mpage->name_head_done ||
1507: strcasecmp(start, meta->title))) {
1508: putkey(mpage, start, NAME_HEAD);
1509: mpage->name_head_done = 1;
1510: }
1.1 kristaps 1511:
1.50 kristaps 1512: if (' ' == byte) {
1513: start += sz + 1;
1514: break;
1515: }
1.1 kristaps 1516:
1.50 kristaps 1517: assert(',' == byte);
1518: start += sz + 1;
1519: while (' ' == *start)
1520: start++;
1521: }
1.1 kristaps 1522:
1.121 schwarze 1523: if (start == title) {
1.133 schwarze 1524: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1525: if ( ! (mpage->name_head_done ||
1526: strcasecmp(start, meta->title))) {
1527: putkey(mpage, start, NAME_HEAD);
1528: mpage->name_head_done = 1;
1529: }
1.50 kristaps 1530: free(title);
1.61 schwarze 1531: return;
1.50 kristaps 1532: }
1.1 kristaps 1533:
1.50 kristaps 1534: while (isspace((unsigned char)*start))
1535: start++;
1.1 kristaps 1536:
1.50 kristaps 1537: if (0 == strncmp(start, "-", 1))
1538: start += 1;
1539: else if (0 == strncmp(start, "\\-\\-", 4))
1540: start += 4;
1541: else if (0 == strncmp(start, "\\-", 2))
1542: start += 2;
1543: else if (0 == strncmp(start, "\\(en", 4))
1544: start += 4;
1545: else if (0 == strncmp(start, "\\(em", 4))
1546: start += 4;
1.1 kristaps 1547:
1.50 kristaps 1548: while (' ' == *start)
1549: start++;
1.1 kristaps 1550:
1.78 schwarze 1551: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1552: free(title);
1.61 schwarze 1553: return;
1.50 kristaps 1554: }
1555: }
1.1 kristaps 1556:
1.77 schwarze 1557: for (n = n->child; n; n = n->next) {
1.78 schwarze 1558: if (NULL != mpage->desc)
1.77 schwarze 1559: break;
1.174 schwarze 1560: parse_man(mpage, meta, n);
1.77 schwarze 1561: }
1.1 kristaps 1562: }
1563:
1564: static void
1.190 schwarze 1565: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1566: const struct roff_node *n)
1.1 kristaps 1567: {
1568:
1.50 kristaps 1569: assert(NULL != n);
1570: for (n = n->child; NULL != n; n = n->next) {
1571: switch (n->type) {
1.188 schwarze 1572: case ROFFT_ELEM:
1573: case ROFFT_BLOCK:
1574: case ROFFT_HEAD:
1575: case ROFFT_BODY:
1576: case ROFFT_TAIL:
1.50 kristaps 1577: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1578: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1579: break;
1.68 schwarze 1580: if (mdocs[n->tok].mask)
1.78 schwarze 1581: putmdockey(mpage, n->child,
1582: mdocs[n->tok].mask);
1.50 kristaps 1583: break;
1584: default:
1.188 schwarze 1585: assert(n->type != ROFFT_ROOT);
1.50 kristaps 1586: continue;
1587: }
1588: if (NULL != n->child)
1.172 schwarze 1589: parse_mdoc(mpage, meta, n);
1.1 kristaps 1590: }
1591: }
1592:
1.25 schwarze 1593: static int
1.190 schwarze 1594: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1595: const struct roff_node *n)
1.1 kristaps 1596: {
1.176 schwarze 1597: char *start, *end;
1.1 kristaps 1598: size_t sz;
1.25 schwarze 1599:
1.50 kristaps 1600: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1601: NULL == (n = n->child) ||
1.188 schwarze 1602: n->type != ROFFT_TEXT)
1.197 schwarze 1603: return 0;
1.1 kristaps 1604:
1605: /*
1606: * Only consider those `Fd' macro fields that begin with an
1607: * "inclusion" token (versus, e.g., #define).
1608: */
1.50 kristaps 1609:
1.1 kristaps 1610: if (strcmp("#include", n->string))
1.197 schwarze 1611: return 0;
1.1 kristaps 1612:
1.188 schwarze 1613: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1614: return 0;
1.1 kristaps 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)))
1.197 schwarze 1626: return 0;
1.1 kristaps 1627:
1628: end = &start[(int)sz - 1];
1629: if ('>' == *end || '"' == *end)
1630: end--;
1631:
1.50 kristaps 1632: if (end > start)
1.78 schwarze 1633: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1634: return 0;
1.1 kristaps 1635: }
1636:
1.178 schwarze 1637: static void
1.189 schwarze 1638: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1639: {
1.112 schwarze 1640: char *cp;
1.178 schwarze 1641: size_t sz;
1.1 kristaps 1642:
1.188 schwarze 1643: if (n->type != ROFFT_TEXT)
1.178 schwarze 1644: return;
1.25 schwarze 1645:
1.178 schwarze 1646: /* Skip function pointer punctuation. */
1.1 kristaps 1647:
1.178 schwarze 1648: cp = n->string;
1649: while (*cp == '(' || *cp == '*')
1.1 kristaps 1650: cp++;
1.178 schwarze 1651: sz = strcspn(cp, "()");
1.1 kristaps 1652:
1.178 schwarze 1653: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1654: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1655: putkeys(mpage, cp, sz, NAME_SYN);
1656: }
1.25 schwarze 1657:
1.178 schwarze 1658: static int
1.190 schwarze 1659: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1660: const struct roff_node *n)
1.178 schwarze 1661: {
1662:
1663: if (n->child == NULL)
1.197 schwarze 1664: return 0;
1.178 schwarze 1665:
1666: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1667:
1.178 schwarze 1668: for (n = n->child->next; n != NULL; n = n->next)
1.188 schwarze 1669: if (n->type == ROFFT_TEXT)
1.78 schwarze 1670: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1671:
1.197 schwarze 1672: return 0;
1.173 schwarze 1673: }
1674:
1675: static int
1.190 schwarze 1676: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1677: const struct roff_node *n)
1.173 schwarze 1678: {
1.177 schwarze 1679:
1.188 schwarze 1680: if (n->type != ROFFT_HEAD)
1.197 schwarze 1681: return 1;
1.173 schwarze 1682:
1.178 schwarze 1683: if (n->child != NULL)
1684: parse_mdoc_fname(mpage, n->child);
1685:
1.197 schwarze 1686: return 0;
1.1 kristaps 1687: }
1688:
1.25 schwarze 1689: static int
1.211 schwarze 1690: parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1691: const struct roff_node *n)
1692: {
1693: char *cp;
1694:
1695: if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1696: return 0;
1697:
1.215 schwarze 1698: if (n->child != NULL &&
1699: n->child->next == NULL &&
1700: n->child->type == ROFFT_TEXT)
1.211 schwarze 1701: return 1;
1702:
1703: cp = NULL;
1704: deroff(&cp, n);
1705: if (cp != NULL) {
1706: putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1707: n->type == ROFFT_BODY ? TYPE_Va : 0));
1708: free(cp);
1709: }
1710:
1711: return 0;
1712: }
1713:
1714: static int
1.190 schwarze 1715: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1716: const struct roff_node *n)
1.1 kristaps 1717: {
1.67 schwarze 1718: char *cp;
1.1 kristaps 1719:
1720: if (NULL == (n = n->child))
1.197 schwarze 1721: return 0;
1.1 kristaps 1722:
1.67 schwarze 1723: if (NULL == n->next) {
1.78 schwarze 1724: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1725: return 0;
1.67 schwarze 1726: }
1727:
1.120 schwarze 1728: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1729: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1730: free(cp);
1.197 schwarze 1731: return 0;
1.1 kristaps 1732: }
1733:
1.25 schwarze 1734: static int
1.190 schwarze 1735: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1736: const struct roff_node *n)
1.1 kristaps 1737: {
1738:
1.188 schwarze 1739: if (n->type == ROFFT_BODY)
1.194 schwarze 1740: deroff(&mpage->desc, n);
1.197 schwarze 1741: return 0;
1.1 kristaps 1742: }
1743:
1.25 schwarze 1744: static int
1.190 schwarze 1745: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1746: const struct roff_node *n)
1.1 kristaps 1747: {
1748:
1.129 schwarze 1749: if (SEC_NAME == n->sec)
1.133 schwarze 1750: putmdockey(mpage, n->child, NAME_TITLE);
1.188 schwarze 1751: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1752: if (n->child == NULL)
1753: putkey(mpage, meta->name, NAME_SYN);
1754: else
1755: putmdockey(mpage, n->child, NAME_SYN);
1.174 schwarze 1756: }
1757: if ( ! (mpage->name_head_done ||
1758: n->child == NULL || n->child->string == NULL ||
1759: strcasecmp(n->child->string, meta->title))) {
1.188 schwarze 1760: putkey(mpage, n->child->string, ROFFT_HEAD);
1.174 schwarze 1761: mpage->name_head_done = 1;
1.172 schwarze 1762: }
1.197 schwarze 1763: return 0;
1.1 kristaps 1764: }
1765:
1.25 schwarze 1766: static int
1.190 schwarze 1767: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1768: const struct roff_node *n)
1.1 kristaps 1769: {
1770:
1.197 schwarze 1771: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1772: }
1773:
1.50 kristaps 1774: static int
1.190 schwarze 1775: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1776: const struct roff_node *n)
1.1 kristaps 1777: {
1778:
1.197 schwarze 1779: return n->type == ROFFT_HEAD;
1.1 kristaps 1780: }
1781:
1.50 kristaps 1782: /*
1.66 schwarze 1783: * Add a string to the hash table for the current manual.
1784: * Each string has a bitmask telling which macros it belongs to.
1785: * When we finish the manual, we'll dump the table.
1.50 kristaps 1786: */
1787: static void
1.176 schwarze 1788: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1789: {
1.133 schwarze 1790: struct ohash *htab;
1.50 kristaps 1791: struct str *s;
1.111 schwarze 1792: const char *end;
1.65 schwarze 1793: unsigned int slot;
1.176 schwarze 1794: int i, mustfree;
1.25 schwarze 1795:
1.50 kristaps 1796: if (0 == sz)
1797: return;
1.111 schwarze 1798:
1.176 schwarze 1799: mustfree = render_string(&cp, &sz);
1800:
1.133 schwarze 1801: if (TYPE_Nm & v) {
1802: htab = &names;
1803: v &= name_mask;
1.169 schwarze 1804: if (v & NAME_FIRST)
1805: name_mask &= ~NAME_FIRST;
1.133 schwarze 1806: if (debug > 1)
1807: say(mpage->mlinks->file,
1.175 schwarze 1808: "Adding name %*s, bits=%d", sz, cp, v);
1.133 schwarze 1809: } else {
1810: htab = &strings;
1811: if (debug > 1)
1812: for (i = 0; i < mansearch_keymax; i++)
1.163 schwarze 1813: if ((uint64_t)1 << i & v)
1.133 schwarze 1814: say(mpage->mlinks->file,
1815: "Adding key %s=%*s",
1816: mansearch_keynames[i], sz, cp);
1.111 schwarze 1817: }
1.25 schwarze 1818:
1.65 schwarze 1819: end = cp + sz;
1.133 schwarze 1820: slot = ohash_qlookupi(htab, cp, &end);
1821: s = ohash_find(htab, slot);
1.25 schwarze 1822:
1.78 schwarze 1823: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1824: s->mask |= v;
1825: return;
1826: } else if (NULL == s) {
1.144 schwarze 1827: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1828: memcpy(s->key, cp, sz);
1.133 schwarze 1829: ohash_insert(htab, slot, s);
1.1 kristaps 1830: }
1.78 schwarze 1831: s->mpage = mpage;
1.50 kristaps 1832: s->mask = v;
1.176 schwarze 1833:
1834: if (mustfree)
1835: free(cp);
1.1 kristaps 1836: }
1837:
1.50 kristaps 1838: /*
1839: * Take a Unicode codepoint and produce its UTF-8 encoding.
1840: * This isn't the best way to do this, but it works.
1841: * The magic numbers are from the UTF-8 packaging.
1842: * They're not as scary as they seem: read the UTF-8 spec for details.
1843: */
1844: static size_t
1845: utf8(unsigned int cp, char out[7])
1.1 kristaps 1846: {
1.50 kristaps 1847: size_t rc;
1.1 kristaps 1848:
1.50 kristaps 1849: rc = 0;
1850: if (cp <= 0x0000007F) {
1851: rc = 1;
1852: out[0] = (char)cp;
1853: } else if (cp <= 0x000007FF) {
1854: rc = 2;
1855: out[0] = (cp >> 6 & 31) | 192;
1856: out[1] = (cp & 63) | 128;
1857: } else if (cp <= 0x0000FFFF) {
1858: rc = 3;
1859: out[0] = (cp >> 12 & 15) | 224;
1860: out[1] = (cp >> 6 & 63) | 128;
1861: out[2] = (cp & 63) | 128;
1862: } else if (cp <= 0x001FFFFF) {
1863: rc = 4;
1864: out[0] = (cp >> 18 & 7) | 240;
1865: out[1] = (cp >> 12 & 63) | 128;
1866: out[2] = (cp >> 6 & 63) | 128;
1867: out[3] = (cp & 63) | 128;
1868: } else if (cp <= 0x03FFFFFF) {
1869: rc = 5;
1870: out[0] = (cp >> 24 & 3) | 248;
1871: out[1] = (cp >> 18 & 63) | 128;
1872: out[2] = (cp >> 12 & 63) | 128;
1873: out[3] = (cp >> 6 & 63) | 128;
1874: out[4] = (cp & 63) | 128;
1875: } else if (cp <= 0x7FFFFFFF) {
1876: rc = 6;
1877: out[0] = (cp >> 30 & 1) | 252;
1878: out[1] = (cp >> 24 & 63) | 128;
1879: out[2] = (cp >> 18 & 63) | 128;
1880: out[3] = (cp >> 12 & 63) | 128;
1881: out[4] = (cp >> 6 & 63) | 128;
1882: out[5] = (cp & 63) | 128;
1883: } else
1.197 schwarze 1884: return 0;
1.1 kristaps 1885:
1.50 kristaps 1886: out[rc] = '\0';
1.197 schwarze 1887: return rc;
1.50 kristaps 1888: }
1.1 kristaps 1889:
1.50 kristaps 1890: /*
1.176 schwarze 1891: * If the string contains escape sequences,
1892: * replace it with an allocated rendering and return 1,
1893: * such that the caller can free it after use.
1894: * Otherwise, do nothing and return 0.
1.50 kristaps 1895: */
1.176 schwarze 1896: static int
1897: render_string(char **public, size_t *psz)
1.50 kristaps 1898: {
1.176 schwarze 1899: const char *src, *scp, *addcp, *seq;
1900: char *dst;
1901: size_t ssz, dsz, addsz;
1.114 schwarze 1902: char utfbuf[7], res[6];
1.176 schwarze 1903: int seqlen, unicode;
1.50 kristaps 1904:
1905: res[0] = '\\';
1906: res[1] = '\t';
1907: res[2] = ASCII_NBRSP;
1908: res[3] = ASCII_HYPH;
1.114 schwarze 1909: res[4] = ASCII_BREAK;
1910: res[5] = '\0';
1.1 kristaps 1911:
1.176 schwarze 1912: src = scp = *public;
1913: ssz = *psz;
1914: dst = NULL;
1915: dsz = 0;
1916:
1917: while (scp < src + *psz) {
1918:
1919: /* Leave normal characters unchanged. */
1920:
1921: if (strchr(res, *scp) == NULL) {
1922: if (dst != NULL)
1923: dst[dsz++] = *scp;
1924: scp++;
1925: continue;
1926: }
1.46 kristaps 1927:
1.50 kristaps 1928: /*
1.176 schwarze 1929: * Found something that requires replacing,
1930: * make sure we have a destination buffer.
1.50 kristaps 1931: */
1.176 schwarze 1932:
1933: if (dst == NULL) {
1934: dst = mandoc_malloc(ssz + 1);
1935: dsz = scp - src;
1936: memcpy(dst, src, dsz);
1.50 kristaps 1937: }
1.46 kristaps 1938:
1.176 schwarze 1939: /* Handle single-char special characters. */
1940:
1941: switch (*scp) {
1942: case '\\':
1943: break;
1.141 schwarze 1944: case '\t':
1945: case ASCII_NBRSP:
1.176 schwarze 1946: dst[dsz++] = ' ';
1947: scp++;
1948: continue;
1949: case ASCII_HYPH:
1950: dst[dsz++] = '-';
1.114 schwarze 1951: /* FALLTHROUGH */
1.141 schwarze 1952: case ASCII_BREAK:
1.176 schwarze 1953: scp++;
1.50 kristaps 1954: continue;
1.114 schwarze 1955: default:
1.176 schwarze 1956: abort();
1.114 schwarze 1957: }
1.46 kristaps 1958:
1.50 kristaps 1959: /*
1.176 schwarze 1960: * Found an escape sequence.
1961: * Read past the slash, then parse it.
1962: * Ignore everything except characters.
1.50 kristaps 1963: */
1.95 schwarze 1964:
1.176 schwarze 1965: scp++;
1966: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1967: continue;
1.1 kristaps 1968:
1.50 kristaps 1969: /*
1.95 schwarze 1970: * Render the special character
1971: * as either UTF-8 or ASCII.
1.50 kristaps 1972: */
1.95 schwarze 1973:
1974: if (write_utf8) {
1.203 schwarze 1975: unicode = mchars_spec2cp(seq, seqlen);
1.176 schwarze 1976: if (unicode <= 0)
1.95 schwarze 1977: continue;
1.176 schwarze 1978: addsz = utf8(unicode, utfbuf);
1979: if (addsz == 0)
1.95 schwarze 1980: continue;
1.176 schwarze 1981: addcp = utfbuf;
1.95 schwarze 1982: } else {
1.203 schwarze 1983: addcp = mchars_spec2str(seq, seqlen, &addsz);
1.176 schwarze 1984: if (addcp == NULL)
1.95 schwarze 1985: continue;
1.176 schwarze 1986: if (*addcp == ASCII_NBRSP) {
1987: addcp = " ";
1988: addsz = 1;
1.95 schwarze 1989: }
1990: }
1.1 kristaps 1991:
1.50 kristaps 1992: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1993:
1.176 schwarze 1994: ssz += addsz;
1995: dst = mandoc_realloc(dst, ssz + 1);
1996: memcpy(dst + dsz, addcp, addsz);
1997: dsz += addsz;
1998: }
1999: if (dst != NULL) {
2000: *public = dst;
2001: *psz = dsz;
2002: }
2003:
2004: /* Trim trailing whitespace and NUL-terminate. */
2005:
2006: while (*psz > 0 && (*public)[*psz - 1] == ' ')
2007: --*psz;
2008: if (dst != NULL) {
2009: (*public)[*psz] = '\0';
1.197 schwarze 2010: return 1;
1.176 schwarze 2011: } else
1.197 schwarze 2012: return 0;
1.1 kristaps 2013: }
2014:
1.118 schwarze 2015: static void
2016: dbadd_mlink(const struct mlink *mlink)
2017: {
2018: size_t i;
2019:
2020: i = 1;
2021: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2022: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2023: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 2024: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 2025: SQL_STEP(stmts[STMT_INSERT_LINK]);
2026: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.169 schwarze 2027: }
2028:
2029: static void
2030: dbadd_mlink_name(const struct mlink *mlink)
2031: {
1.175 schwarze 2032: uint64_t bits;
1.169 schwarze 2033: size_t i;
2034:
2035: dbadd_mlink(mlink);
1.160 schwarze 2036:
2037: i = 1;
1.175 schwarze 2038: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2039: bits = NAME_FILE & NAME_MASK;
2040: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2041: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2042: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2043: }
2044:
2045: i = 1;
2046: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.160 schwarze 2047: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2048: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2049: SQL_STEP(stmts[STMT_INSERT_NAME]);
2050: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.118 schwarze 2051: }
2052:
1.14 schwarze 2053: /*
1.50 kristaps 2054: * Flush the current page's terms (and their bits) into the database.
2055: * Wrap the entire set of additions in a transaction to make sqlite be a
2056: * little faster.
1.96 schwarze 2057: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2058: */
2059: static void
1.176 schwarze 2060: dbadd(struct mpage *mpage)
1.14 schwarze 2061: {
1.87 schwarze 2062: struct mlink *mlink;
1.50 kristaps 2063: struct str *key;
1.176 schwarze 2064: char *cp;
1.52 kristaps 2065: size_t i;
1.66 schwarze 2066: unsigned int slot;
1.176 schwarze 2067: int mustfree;
1.14 schwarze 2068:
1.128 schwarze 2069: mlink = mpage->mlinks;
1.43 kristaps 2070:
1.128 schwarze 2071: if (nodb) {
1.147 schwarze 2072: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2073: key = ohash_next(&names, &slot))
1.147 schwarze 2074: free(key);
2075: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2076: key = ohash_next(&strings, &slot))
1.147 schwarze 2077: free(key);
1.145 schwarze 2078: if (0 == debug)
2079: return;
1.128 schwarze 2080: while (NULL != mlink) {
2081: fputs(mlink->name, stdout);
2082: if (NULL == mlink->next ||
2083: strcmp(mlink->dsec, mlink->next->dsec) ||
2084: strcmp(mlink->fsec, mlink->next->fsec) ||
2085: strcmp(mlink->arch, mlink->next->arch)) {
2086: putchar('(');
2087: if ('\0' == *mlink->dsec)
2088: fputs(mlink->fsec, stdout);
2089: else
2090: fputs(mlink->dsec, stdout);
2091: if ('\0' != *mlink->arch)
2092: printf("/%s", mlink->arch);
2093: putchar(')');
2094: }
2095: mlink = mlink->next;
2096: if (NULL != mlink)
2097: fputs(", ", stdout);
2098: }
1.132 schwarze 2099: printf(" - %s\n", mpage->desc);
1.14 schwarze 2100: return;
1.128 schwarze 2101: }
2102:
2103: if (debug)
2104: say(mlink->file, "Adding to database");
1.28 kristaps 2105:
1.176 schwarze 2106: cp = mpage->desc;
2107: i = strlen(cp);
2108: mustfree = render_string(&cp, &i);
1.52 kristaps 2109: i = 1;
1.176 schwarze 2110: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.161 schwarze 2111: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.81 schwarze 2112: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2113: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2114: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.176 schwarze 2115: if (mustfree)
2116: free(cp);
1.81 schwarze 2117:
1.128 schwarze 2118: while (NULL != mlink) {
1.118 schwarze 2119: dbadd_mlink(mlink);
1.128 schwarze 2120: mlink = mlink->next;
2121: }
1.134 schwarze 2122: mlink = mpage->mlinks;
1.50 kristaps 2123:
1.133 schwarze 2124: for (key = ohash_first(&names, &slot); NULL != key;
2125: key = ohash_next(&names, &slot)) {
2126: assert(key->mpage == mpage);
2127: i = 1;
2128: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.176 schwarze 2129: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.137 schwarze 2130: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2131: SQL_STEP(stmts[STMT_INSERT_NAME]);
2132: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2133: free(key);
2134: }
1.66 schwarze 2135: for (key = ohash_first(&strings, &slot); NULL != key;
2136: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2137: assert(key->mpage == mpage);
1.52 kristaps 2138: i = 1;
2139: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.176 schwarze 2140: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.137 schwarze 2141: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2142: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2143: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 2144: free(key);
1.38 schwarze 2145: }
1.14 schwarze 2146: }
2147:
1.5 kristaps 2148: static void
1.59 schwarze 2149: dbprune(void)
1.5 kristaps 2150: {
1.78 schwarze 2151: struct mpage *mpage;
1.82 schwarze 2152: struct mlink *mlink;
1.52 kristaps 2153: size_t i;
1.80 schwarze 2154: unsigned int slot;
1.5 kristaps 2155:
1.106 schwarze 2156: if (0 == nodb)
2157: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2158:
1.106 schwarze 2159: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2160: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2161: mlink = mpage->mlinks;
1.125 schwarze 2162: if (debug)
1.106 schwarze 2163: say(mlink->file, "Deleting from database");
2164: if (nodb)
2165: continue;
2166: for ( ; NULL != mlink; mlink = mlink->next) {
2167: i = 1;
2168: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2169: i, mlink->dsec);
2170: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2171: i, mlink->arch);
2172: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2173: i, mlink->name);
2174: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2175: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2176: }
1.5 kristaps 2177: }
1.106 schwarze 2178:
2179: if (0 == nodb)
2180: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2181: }
2182:
1.4 kristaps 2183: /*
1.50 kristaps 2184: * Close an existing database and its prepared statements.
2185: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2186: */
1.35 kristaps 2187: static void
1.59 schwarze 2188: dbclose(int real)
1.4 kristaps 2189: {
1.50 kristaps 2190: size_t i;
1.115 schwarze 2191: int status;
2192: pid_t child;
1.4 kristaps 2193:
1.50 kristaps 2194: if (nodb)
1.38 schwarze 2195: return;
1.50 kristaps 2196:
2197: for (i = 0; i < STMT__MAX; i++) {
2198: sqlite3_finalize(stmts[i]);
2199: stmts[i] = NULL;
1.4 kristaps 2200: }
2201:
1.50 kristaps 2202: sqlite3_close(db);
2203: db = NULL;
1.12 schwarze 2204:
1.50 kristaps 2205: if (real)
2206: return;
1.12 schwarze 2207:
1.115 schwarze 2208: if ('\0' == *tempfilename) {
2209: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2210: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2211: say(MANDOC_DB, "&rename");
1.115 schwarze 2212: }
2213: return;
2214: }
2215:
2216: switch (child = fork()) {
1.141 schwarze 2217: case -1:
1.115 schwarze 2218: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2219: say("", "&fork cmp");
1.115 schwarze 2220: return;
1.141 schwarze 2221: case 0:
1.115 schwarze 2222: execlp("cmp", "cmp", "-s",
1.196 schwarze 2223: tempfilename, MANDOC_DB, (char *)NULL);
1.123 schwarze 2224: say("", "&exec cmp");
1.115 schwarze 2225: exit(0);
2226: default:
2227: break;
2228: }
2229: if (-1 == waitpid(child, &status, 0)) {
2230: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2231: say("", "&wait cmp");
1.115 schwarze 2232: } else if (WIFSIGNALED(status)) {
2233: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2234: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2235: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2236: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2237: say(MANDOC_DB,
2238: "Data changed, but cannot replace database");
2239: }
2240:
2241: *strrchr(tempfilename, '/') = '\0';
2242: switch (child = fork()) {
1.141 schwarze 2243: case -1:
1.115 schwarze 2244: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2245: say("", "&fork rm");
1.115 schwarze 2246: return;
1.141 schwarze 2247: case 0:
1.196 schwarze 2248: execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
1.123 schwarze 2249: say("", "&exec rm");
1.115 schwarze 2250: exit((int)MANDOCLEVEL_SYSERR);
2251: default:
2252: break;
2253: }
2254: if (-1 == waitpid(child, &status, 0)) {
2255: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2256: say("", "&wait rm");
1.115 schwarze 2257: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2258: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2259: say("", "%s: Cannot remove temporary directory",
2260: tempfilename);
1.59 schwarze 2261: }
1.50 kristaps 2262: }
1.14 schwarze 2263:
1.50 kristaps 2264: /*
2265: * This is straightforward stuff.
2266: * Open a database connection to a "temporary" database, then open a set
2267: * of prepared statements we'll use over and over again.
2268: * If "real" is set, we use the existing database; if not, we truncate a
2269: * temporary one.
2270: * Must be matched by dbclose().
2271: */
2272: static int
1.59 schwarze 2273: dbopen(int real)
1.50 kristaps 2274: {
1.115 schwarze 2275: const char *sql;
1.50 kristaps 2276: int rc, ofl;
1.12 schwarze 2277:
1.141 schwarze 2278: if (nodb)
1.197 schwarze 2279: return 1;
1.12 schwarze 2280:
1.115 schwarze 2281: *tempfilename = '\0';
1.63 schwarze 2282: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2283:
2284: if (real) {
2285: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2286: if (SQLITE_OK != rc) {
1.63 schwarze 2287: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2288: if (SQLITE_CANTOPEN != rc)
2289: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.197 schwarze 2290: return 0;
1.63 schwarze 2291: }
1.115 schwarze 2292: goto prepare_statements;
2293: }
2294:
2295: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2296:
1.115 schwarze 2297: remove(MANDOC_DB "~");
2298: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2299: if (SQLITE_OK == rc)
1.115 schwarze 2300: goto create_tables;
1.116 schwarze 2301: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2302: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2303: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.197 schwarze 2304: return 0;
1.50 kristaps 2305: }
1.12 schwarze 2306:
1.143 schwarze 2307: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2308: sizeof(tempfilename));
1.115 schwarze 2309: if (NULL == mkdtemp(tempfilename)) {
2310: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2311: say("", "&%s", tempfilename);
1.197 schwarze 2312: return 0;
1.115 schwarze 2313: }
1.143 schwarze 2314: (void)strlcat(tempfilename, "/" MANDOC_DB,
2315: sizeof(tempfilename));
1.115 schwarze 2316: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2317: if (SQLITE_OK != rc) {
1.59 schwarze 2318: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2319: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.197 schwarze 2320: return 0;
1.50 kristaps 2321: }
1.12 schwarze 2322:
1.115 schwarze 2323: create_tables:
1.81 schwarze 2324: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2325: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2326: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2327: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2328: ");\n"
2329: "\n"
1.81 schwarze 2330: "CREATE TABLE \"mlinks\" (\n"
2331: " \"sec\" TEXT NOT NULL,\n"
2332: " \"arch\" TEXT NOT NULL,\n"
2333: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2334: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2335: "ON DELETE CASCADE\n"
1.81 schwarze 2336: ");\n"
1.136 schwarze 2337: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2338: "\n"
1.133 schwarze 2339: "CREATE TABLE \"names\" (\n"
2340: " \"bits\" INTEGER NOT NULL,\n"
2341: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2342: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.175 schwarze 2343: "ON DELETE CASCADE,\n"
2344: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.133 schwarze 2345: ");\n"
2346: "\n"
1.50 kristaps 2347: "CREATE TABLE \"keys\" (\n"
2348: " \"bits\" INTEGER NOT NULL,\n"
2349: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2350: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2351: "ON DELETE CASCADE\n"
1.136 schwarze 2352: ");\n"
2353: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2354:
1.50 kristaps 2355: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2356: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2357: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2358: sqlite3_close(db);
1.197 schwarze 2359: return 0;
1.50 kristaps 2360: }
1.4 kristaps 2361:
1.57 schwarze 2362: prepare_statements:
1.146 schwarze 2363: if (SQLITE_OK != sqlite3_exec(db,
2364: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2365: exitcode = (int)MANDOCLEVEL_SYSERR;
2366: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2367: sqlite3_errmsg(db));
2368: sqlite3_close(db);
1.197 schwarze 2369: return 0;
1.146 schwarze 2370: }
2371:
1.137 schwarze 2372: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2373: "(SELECT pageid FROM mlinks WHERE "
2374: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2375: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2376: sql = "INSERT INTO mpages "
1.132 schwarze 2377: "(desc,form) VALUES (?,?)";
1.81 schwarze 2378: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2379: sql = "INSERT INTO mlinks "
1.104 schwarze 2380: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2381: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.175 schwarze 2382: sql = "SELECT bits FROM names where pageid = ?";
2383: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.133 schwarze 2384: sql = "INSERT INTO names "
2385: "(bits,name,pageid) VALUES (?,?,?)";
2386: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2387: sql = "INSERT INTO keys "
1.81 schwarze 2388: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2389: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2390:
2391: #ifndef __APPLE__
2392: /*
2393: * When opening a new database, we can turn off
2394: * synchronous mode for much better performance.
2395: */
2396:
1.146 schwarze 2397: if (real && SQLITE_OK != sqlite3_exec(db,
2398: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2399: exitcode = (int)MANDOCLEVEL_SYSERR;
2400: say(MANDOC_DB, "PRAGMA synchronous: %s",
1.162 schwarze 2401: sqlite3_errmsg(db));
1.146 schwarze 2402: sqlite3_close(db);
1.197 schwarze 2403: return 0;
1.146 schwarze 2404: }
1.70 schwarze 2405: #endif
2406:
1.197 schwarze 2407: return 1;
1.4 kristaps 2408: }
2409:
1.50 kristaps 2410: static int
1.165 schwarze 2411: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2412: {
1.59 schwarze 2413: static char startdir[PATH_MAX];
1.151 schwarze 2414: static int getcwd_status; /* 1 = ok, 2 = failure */
2415: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2416: char *cp;
1.4 kristaps 2417:
1.59 schwarze 2418: /*
1.151 schwarze 2419: * Remember the original working directory, if possible.
2420: * This will be needed if the second or a later directory
2421: * on the command line is given as a relative path.
2422: * Do not error out if the current directory is not
2423: * searchable: Maybe it won't be needed after all.
2424: */
2425: if (0 == getcwd_status) {
2426: if (NULL == getcwd(startdir, sizeof(startdir))) {
2427: getcwd_status = 2;
2428: (void)strlcpy(startdir, strerror(errno),
2429: sizeof(startdir));
2430: } else
2431: getcwd_status = 1;
2432: }
2433:
2434: /*
2435: * We are leaving the old base directory.
2436: * Do not use it any longer, not even for messages.
2437: */
2438: *basedir = '\0';
2439:
2440: /*
2441: * If and only if the directory was changed earlier and
2442: * the next directory to process is given as a relative path,
2443: * first go back, or bail out if that is impossible.
1.59 schwarze 2444: */
1.151 schwarze 2445: if (chdir_status && '/' != *targetdir) {
2446: if (2 == getcwd_status) {
1.59 schwarze 2447: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2448: say("", "getcwd: %s", startdir);
1.197 schwarze 2449: return 0;
1.59 schwarze 2450: }
1.151 schwarze 2451: if (-1 == chdir(startdir)) {
1.59 schwarze 2452: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2453: say("", "&chdir %s", startdir);
1.197 schwarze 2454: return 0;
1.59 schwarze 2455: }
2456: }
1.151 schwarze 2457:
2458: /*
2459: * Always resolve basedir to the canonicalized absolute
2460: * pathname and append a trailing slash, such that
2461: * we can reliably check whether files are inside.
2462: */
1.59 schwarze 2463: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2464: if (report_baddir || errno != ENOENT) {
2465: exitcode = (int)MANDOCLEVEL_BADARG;
2466: say("", "&%s: realpath", targetdir);
2467: }
1.197 schwarze 2468: return 0;
1.59 schwarze 2469: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2470: if (report_baddir || errno != ENOENT) {
2471: exitcode = (int)MANDOCLEVEL_BADARG;
2472: say("", "&chdir");
2473: }
1.197 schwarze 2474: return 0;
1.150 schwarze 2475: }
1.151 schwarze 2476: chdir_status = 1;
1.150 schwarze 2477: cp = strchr(basedir, '\0');
2478: if ('/' != cp[-1]) {
2479: if (cp - basedir >= PATH_MAX - 1) {
2480: exitcode = (int)MANDOCLEVEL_SYSERR;
2481: say("", "Filename too long");
1.197 schwarze 2482: return 0;
1.150 schwarze 2483: }
2484: *cp++ = '/';
2485: *cp = '\0';
1.4 kristaps 2486: }
1.197 schwarze 2487: return 1;
1.56 schwarze 2488: }
2489:
2490: static void
1.59 schwarze 2491: say(const char *file, const char *format, ...)
1.56 schwarze 2492: {
2493: va_list ap;
1.123 schwarze 2494: int use_errno;
1.56 schwarze 2495:
1.59 schwarze 2496: if ('\0' != *basedir)
2497: fprintf(stderr, "%s", basedir);
2498: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2499: fputc('/', stderr);
1.56 schwarze 2500: if ('\0' != *file)
1.59 schwarze 2501: fprintf(stderr, "%s", file);
2502:
1.123 schwarze 2503: use_errno = 1;
2504: if (NULL != format) {
2505: switch (*format) {
1.141 schwarze 2506: case '&':
1.123 schwarze 2507: format++;
2508: break;
1.141 schwarze 2509: case '\0':
1.123 schwarze 2510: format = NULL;
2511: break;
2512: default:
2513: use_errno = 0;
2514: break;
2515: }
2516: }
2517: if (NULL != format) {
2518: if ('\0' != *basedir || '\0' != *file)
2519: fputs(": ", stderr);
2520: va_start(ap, format);
2521: vfprintf(stderr, format, ap);
2522: va_end(ap);
2523: }
2524: if (use_errno) {
2525: if ('\0' != *basedir || '\0' != *file || NULL != format)
2526: fputs(": ", stderr);
1.59 schwarze 2527: perror(NULL);
1.123 schwarze 2528: } else
2529: fputc('\n', stderr);
1.1 kristaps 2530: }
CVSweb