Annotation of mandoc/mandocdb.c, Revision 1.220
1.220 ! schwarze 1: /* $Id: mandocdb.c,v 1.219 2016/07/15 18:03:45 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>
1.218 kristaps 37: #if HAVE_SANDBOX_INIT
38: #include <sandbox.h>
39: #endif
1.50 kristaps 40: #include <stddef.h>
1.59 schwarze 41: #include <stdio.h>
1.1 kristaps 42: #include <stdint.h>
43: #include <stdlib.h>
44: #include <string.h>
1.17 schwarze 45: #include <unistd.h>
1.1 kristaps 46:
1.50 kristaps 47: #include <sqlite3.h>
1.1 kristaps 48:
1.188 schwarze 49: #include "mandoc_aux.h"
1.202 schwarze 50: #include "mandoc_ohash.h"
1.188 schwarze 51: #include "mandoc.h"
52: #include "roff.h"
1.50 kristaps 53: #include "mdoc.h"
1.1 kristaps 54: #include "man.h"
1.187 schwarze 55: #include "manconf.h"
1.55 kristaps 56: #include "mansearch.h"
1.1 kristaps 57:
1.111 schwarze 58: extern int mansearch_keymax;
59: extern const char *const mansearch_keynames[];
60:
1.52 kristaps 61: #define SQL_EXEC(_v) \
62: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.134 schwarze 63: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.52 kristaps 64: #define SQL_BIND_TEXT(_s, _i, _v) \
65: if (SQLITE_OK != sqlite3_bind_text \
66: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.134 schwarze 67: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 68: #define SQL_BIND_INT(_s, _i, _v) \
69: if (SQLITE_OK != sqlite3_bind_int \
70: ((_s), (_i)++, (_v))) \
1.134 schwarze 71: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 72: #define SQL_BIND_INT64(_s, _i, _v) \
73: if (SQLITE_OK != sqlite3_bind_int64 \
74: ((_s), (_i)++, (_v))) \
1.134 schwarze 75: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 76: #define SQL_STEP(_s) \
77: if (SQLITE_DONE != sqlite3_step((_s))) \
1.134 schwarze 78: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 79:
1.50 kristaps 80: enum op {
81: OP_DEFAULT = 0, /* new dbs from dir list or default config */
82: OP_CONFFILE, /* new databases from custom config file */
83: OP_UPDATE, /* delete/add entries in existing database */
84: OP_DELETE, /* delete entries from existing database */
85: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 86: };
87:
1.50 kristaps 88: struct str {
1.78 schwarze 89: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 90: uint64_t mask; /* bitmask in sequence */
1.176 schwarze 91: char key[]; /* rendered text */
1.38 schwarze 92: };
93:
1.79 schwarze 94: struct inodev {
95: ino_t st_ino;
96: dev_t st_dev;
1.50 kristaps 97: };
1.5 kristaps 98:
1.78 schwarze 99: struct mpage {
1.79 schwarze 100: struct inodev inodev; /* used for hashing routine */
1.137 schwarze 101: int64_t pageid; /* pageid in mpages SQL table */
1.82 schwarze 102: char *sec; /* section from file content */
103: char *arch; /* architecture from file content */
104: char *title; /* title from file content */
105: char *desc; /* description from file content */
106: struct mlink *mlinks; /* singly linked list */
1.161 schwarze 107: int form; /* format from file content */
1.174 schwarze 108: int name_head_done;
1.82 schwarze 109: };
110:
111: struct mlink {
1.58 schwarze 112: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 113: char *dsec; /* section from directory */
114: char *arch; /* architecture from directory */
115: char *name; /* name from file name (not empty) */
116: char *fsec; /* section from file name suffix */
1.85 schwarze 117: struct mlink *next; /* singly linked list */
1.118 schwarze 118: struct mpage *mpage; /* parent */
1.161 schwarze 119: int dform; /* format from directory */
120: int fform; /* format from file name suffix */
1.124 schwarze 121: int gzip; /* filename has a .gz suffix */
1.3 kristaps 122: };
123:
1.50 kristaps 124: enum stmt {
1.81 schwarze 125: STMT_DELETE_PAGE = 0, /* delete mpage */
126: STMT_INSERT_PAGE, /* insert mpage */
127: STMT_INSERT_LINK, /* insert mlink */
1.133 schwarze 128: STMT_INSERT_NAME, /* insert name */
1.175 schwarze 129: STMT_SELECT_NAME, /* retrieve existing name flags */
1.81 schwarze 130: STMT_INSERT_KEY, /* insert parsed key */
1.50 kristaps 131: STMT__MAX
1.1 kristaps 132: };
133:
1.190 schwarze 134: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.189 schwarze 135: const struct roff_node *);
1.1 kristaps 136:
1.50 kristaps 137: struct mdoc_handler {
138: mdoc_fp fp; /* optional handler */
139: uint64_t mask; /* set unless handler returns 0 */
1.1 kristaps 140: };
1.219 schwarze 141:
142:
143: int mandocdb(int, char *[]);
1.1 kristaps 144:
1.59 schwarze 145: static void dbclose(int);
1.176 schwarze 146: static void dbadd(struct mpage *);
1.118 schwarze 147: static void dbadd_mlink(const struct mlink *mlink);
1.169 schwarze 148: static void dbadd_mlink_name(const struct mlink *mlink);
1.59 schwarze 149: static int dbopen(int);
150: static void dbprune(void);
151: static void filescan(const char *);
1.84 schwarze 152: static void mlink_add(struct mlink *, const struct stat *);
1.131 schwarze 153: static void mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 154: static void mlink_free(struct mlink *);
1.89 schwarze 155: static void mlinks_undupe(struct mpage *);
1.78 schwarze 156: static void mpages_free(void);
1.176 schwarze 157: static void mpages_merge(struct mparse *);
1.130 schwarze 158: static void names_check(void);
1.124 schwarze 159: static void parse_cat(struct mpage *, int);
1.190 schwarze 160: static void parse_man(struct mpage *, const struct roff_meta *,
1.189 schwarze 161: const struct roff_node *);
1.190 schwarze 162: static void parse_mdoc(struct mpage *, const struct roff_meta *,
1.189 schwarze 163: const struct roff_node *);
1.190 schwarze 164: static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
1.189 schwarze 165: const struct roff_node *);
1.190 schwarze 166: static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
1.189 schwarze 167: const struct roff_node *);
168: static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
1.190 schwarze 169: static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
1.189 schwarze 170: const struct roff_node *);
1.190 schwarze 171: static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
1.189 schwarze 172: const struct roff_node *);
1.190 schwarze 173: static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
1.189 schwarze 174: const struct roff_node *);
1.190 schwarze 175: static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
1.189 schwarze 176: const struct roff_node *);
1.190 schwarze 177: static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
1.189 schwarze 178: const struct roff_node *);
1.211 schwarze 179: static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
180: const struct roff_node *);
1.190 schwarze 181: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.189 schwarze 182: const struct roff_node *);
1.112 schwarze 183: static void putkey(const struct mpage *, char *, uint64_t);
1.176 schwarze 184: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.78 schwarze 185: static void putmdockey(const struct mpage *,
1.189 schwarze 186: const struct roff_node *, uint64_t);
1.176 schwarze 187: static int render_string(char **, size_t *);
1.220 ! schwarze 188: static void say(const char *, const char *, ...)
! 189: __attribute__((__format__ (printf, 2, 3)));
1.165 schwarze 190: static int set_basedir(const char *, int);
1.59 schwarze 191: static int treescan(void);
1.50 kristaps 192: static size_t utf8(unsigned int, char [7]);
193:
1.115 schwarze 194: static char tempfilename[32];
1.102 schwarze 195: static int nodb; /* no database changes */
1.116 schwarze 196: static int mparse_options; /* abort the parse early */
1.141 schwarze 197: static int use_all; /* use all found files */
198: static int debug; /* print what we're doing */
199: static int warnings; /* warn about crap */
1.95 schwarze 200: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 201: static int exitcode; /* to be returned by main */
1.141 schwarze 202: static enum op op; /* operational mode */
1.59 schwarze 203: static char basedir[PATH_MAX]; /* current base directory */
1.78 schwarze 204: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 205: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 206: static struct ohash names; /* table of all names */
1.50 kristaps 207: static struct ohash strings; /* table of all strings */
208: static sqlite3 *db = NULL; /* current database */
209: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.133 schwarze 210: static uint64_t name_mask;
1.25 schwarze 211:
212: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 213: { NULL, 0 }, /* Ap */
214: { NULL, 0 }, /* Dd */
215: { NULL, 0 }, /* Dt */
216: { NULL, 0 }, /* Os */
217: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
218: { parse_mdoc_head, TYPE_Ss }, /* Ss */
219: { NULL, 0 }, /* Pp */
220: { NULL, 0 }, /* D1 */
221: { NULL, 0 }, /* Dl */
222: { NULL, 0 }, /* Bd */
223: { NULL, 0 }, /* Ed */
224: { NULL, 0 }, /* Bl */
225: { NULL, 0 }, /* El */
226: { NULL, 0 }, /* It */
227: { NULL, 0 }, /* Ad */
228: { NULL, TYPE_An }, /* An */
229: { NULL, TYPE_Ar }, /* Ar */
230: { NULL, TYPE_Cd }, /* Cd */
231: { NULL, TYPE_Cm }, /* Cm */
232: { NULL, TYPE_Dv }, /* Dv */
233: { NULL, TYPE_Er }, /* Er */
234: { NULL, TYPE_Ev }, /* Ev */
235: { NULL, 0 }, /* Ex */
236: { NULL, TYPE_Fa }, /* Fa */
237: { parse_mdoc_Fd, 0 }, /* Fd */
238: { NULL, TYPE_Fl }, /* Fl */
239: { parse_mdoc_Fn, 0 }, /* Fn */
240: { NULL, TYPE_Ft }, /* Ft */
241: { NULL, TYPE_Ic }, /* Ic */
1.92 schwarze 242: { NULL, TYPE_In }, /* In */
1.68 schwarze 243: { NULL, TYPE_Li }, /* Li */
1.132 schwarze 244: { parse_mdoc_Nd, 0 }, /* Nd */
1.129 schwarze 245: { parse_mdoc_Nm, 0 }, /* Nm */
1.68 schwarze 246: { NULL, 0 }, /* Op */
247: { NULL, 0 }, /* Ot */
248: { NULL, TYPE_Pa }, /* Pa */
249: { NULL, 0 }, /* Rv */
1.92 schwarze 250: { NULL, TYPE_St }, /* St */
1.211 schwarze 251: { parse_mdoc_Va, TYPE_Va }, /* Va */
252: { parse_mdoc_Va, TYPE_Vt }, /* Vt */
1.68 schwarze 253: { parse_mdoc_Xr, 0 }, /* Xr */
254: { NULL, 0 }, /* %A */
255: { NULL, 0 }, /* %B */
256: { NULL, 0 }, /* %D */
257: { NULL, 0 }, /* %I */
258: { NULL, 0 }, /* %J */
259: { NULL, 0 }, /* %N */
260: { NULL, 0 }, /* %O */
261: { NULL, 0 }, /* %P */
262: { NULL, 0 }, /* %R */
263: { NULL, 0 }, /* %T */
264: { NULL, 0 }, /* %V */
265: { NULL, 0 }, /* Ac */
266: { NULL, 0 }, /* Ao */
267: { NULL, 0 }, /* Aq */
268: { NULL, TYPE_At }, /* At */
269: { NULL, 0 }, /* Bc */
270: { NULL, 0 }, /* Bf */
271: { NULL, 0 }, /* Bo */
272: { NULL, 0 }, /* Bq */
273: { NULL, TYPE_Bsx }, /* Bsx */
274: { NULL, TYPE_Bx }, /* Bx */
275: { NULL, 0 }, /* Db */
276: { NULL, 0 }, /* Dc */
277: { NULL, 0 }, /* Do */
278: { NULL, 0 }, /* Dq */
279: { NULL, 0 }, /* Ec */
280: { NULL, 0 }, /* Ef */
281: { NULL, TYPE_Em }, /* Em */
282: { NULL, 0 }, /* Eo */
283: { NULL, TYPE_Fx }, /* Fx */
284: { NULL, TYPE_Ms }, /* Ms */
285: { NULL, 0 }, /* No */
286: { NULL, 0 }, /* Ns */
287: { NULL, TYPE_Nx }, /* Nx */
288: { NULL, TYPE_Ox }, /* Ox */
289: { NULL, 0 }, /* Pc */
290: { NULL, 0 }, /* Pf */
291: { NULL, 0 }, /* Po */
292: { NULL, 0 }, /* Pq */
293: { NULL, 0 }, /* Qc */
294: { NULL, 0 }, /* Ql */
295: { NULL, 0 }, /* Qo */
296: { NULL, 0 }, /* Qq */
297: { NULL, 0 }, /* Re */
298: { NULL, 0 }, /* Rs */
299: { NULL, 0 }, /* Sc */
300: { NULL, 0 }, /* So */
301: { NULL, 0 }, /* Sq */
302: { NULL, 0 }, /* Sm */
303: { NULL, 0 }, /* Sx */
304: { NULL, TYPE_Sy }, /* Sy */
305: { NULL, TYPE_Tn }, /* Tn */
306: { NULL, 0 }, /* Ux */
307: { NULL, 0 }, /* Xc */
308: { NULL, 0 }, /* Xo */
1.173 schwarze 309: { parse_mdoc_Fo, 0 }, /* Fo */
1.68 schwarze 310: { NULL, 0 }, /* Fc */
311: { NULL, 0 }, /* Oo */
312: { NULL, 0 }, /* Oc */
313: { NULL, 0 }, /* Bk */
314: { NULL, 0 }, /* Ek */
315: { NULL, 0 }, /* Bt */
316: { NULL, 0 }, /* Hf */
317: { NULL, 0 }, /* Fr */
318: { NULL, 0 }, /* Ud */
319: { NULL, TYPE_Lb }, /* Lb */
320: { NULL, 0 }, /* Lp */
321: { NULL, TYPE_Lk }, /* Lk */
322: { NULL, TYPE_Mt }, /* Mt */
323: { NULL, 0 }, /* Brq */
324: { NULL, 0 }, /* Bro */
325: { NULL, 0 }, /* Brc */
326: { NULL, 0 }, /* %C */
327: { NULL, 0 }, /* Es */
328: { NULL, 0 }, /* En */
329: { NULL, TYPE_Dx }, /* Dx */
330: { NULL, 0 }, /* %Q */
331: { NULL, 0 }, /* br */
332: { NULL, 0 }, /* sp */
333: { NULL, 0 }, /* %U */
334: { NULL, 0 }, /* Ta */
1.166 schwarze 335: { NULL, 0 }, /* ll */
1.1 kristaps 336: };
337:
1.141 schwarze 338:
1.1 kristaps 339: int
1.179 schwarze 340: mandocdb(int argc, char *argv[])
1.1 kristaps 341: {
1.187 schwarze 342: struct manconf conf;
343: struct mparse *mp;
1.206 schwarze 344: const char *path_arg, *progname;
1.187 schwarze 345: size_t j, sz;
1.59 schwarze 346: int ch, i;
1.50 kristaps 347:
1.207 schwarze 348: #if HAVE_PLEDGE
349: if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
1.214 schwarze 350: warn("pledge");
1.218 kristaps 351: return (int)MANDOCLEVEL_SYSERR;
352: }
353: #endif
354:
355: #if HAVE_SANDBOX_INIT
356: if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) {
357: warnx("sandbox_init");
1.207 schwarze 358: return (int)MANDOCLEVEL_SYSERR;
359: }
360: #endif
361:
1.187 schwarze 362: memset(&conf, 0, sizeof(conf));
1.50 kristaps 363: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
364:
365: /*
1.141 schwarze 366: * We accept a few different invocations.
1.50 kristaps 367: * The CHECKOP macro makes sure that invocation styles don't
368: * clobber each other.
369: */
370: #define CHECKOP(_op, _ch) do \
371: if (OP_DEFAULT != (_op)) { \
1.198 schwarze 372: warnx("-%c: Conflicting option", (_ch)); \
1.50 kristaps 373: goto usage; \
374: } while (/*CONSTCOND*/0)
1.10 kristaps 375:
1.59 schwarze 376: path_arg = NULL;
1.38 schwarze 377: op = OP_DEFAULT;
1.1 kristaps 378:
1.126 schwarze 379: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 380: switch (ch) {
1.141 schwarze 381: case 'a':
1.12 schwarze 382: use_all = 1;
383: break;
1.141 schwarze 384: case 'C':
1.50 kristaps 385: CHECKOP(op, ch);
1.59 schwarze 386: path_arg = optarg;
1.38 schwarze 387: op = OP_CONFFILE;
1.34 schwarze 388: break;
1.141 schwarze 389: case 'D':
1.125 schwarze 390: debug++;
391: break;
1.141 schwarze 392: case 'd':
1.50 kristaps 393: CHECKOP(op, ch);
1.59 schwarze 394: path_arg = optarg;
1.5 kristaps 395: op = OP_UPDATE;
396: break;
1.141 schwarze 397: case 'n':
1.50 kristaps 398: nodb = 1;
399: break;
1.141 schwarze 400: case 'p':
1.126 schwarze 401: warnings = 1;
402: break;
1.141 schwarze 403: case 'Q':
1.116 schwarze 404: mparse_options |= MPARSE_QUICK;
1.102 schwarze 405: break;
1.141 schwarze 406: case 'T':
1.95 schwarze 407: if (strcmp(optarg, "utf8")) {
1.198 schwarze 408: warnx("-T%s: Unsupported output format",
409: optarg);
1.95 schwarze 410: goto usage;
411: }
412: write_utf8 = 1;
413: break;
1.141 schwarze 414: case 't':
1.50 kristaps 415: CHECKOP(op, ch);
1.38 schwarze 416: dup2(STDOUT_FILENO, STDERR_FILENO);
417: op = OP_TEST;
1.50 kristaps 418: nodb = warnings = 1;
1.38 schwarze 419: break;
1.141 schwarze 420: case 'u':
1.50 kristaps 421: CHECKOP(op, ch);
1.59 schwarze 422: path_arg = optarg;
1.5 kristaps 423: op = OP_DELETE;
424: break;
1.141 schwarze 425: case 'v':
1.126 schwarze 426: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 427: break;
1.1 kristaps 428: default:
1.38 schwarze 429: goto usage;
1.1 kristaps 430: }
431:
432: argc -= optind;
433: argv += optind;
434:
1.207 schwarze 435: #if HAVE_PLEDGE
1.210 schwarze 436: if (nodb) {
437: if (pledge("stdio rpath", NULL) == -1) {
1.214 schwarze 438: warn("pledge");
1.210 schwarze 439: return (int)MANDOCLEVEL_SYSERR;
440: }
1.207 schwarze 441: }
442: #endif
443:
1.38 schwarze 444: if (OP_CONFFILE == op && argc > 0) {
1.198 schwarze 445: warnx("-C: Too many arguments");
1.38 schwarze 446: goto usage;
447: }
448:
1.59 schwarze 449: exitcode = (int)MANDOCLEVEL_OK;
1.203 schwarze 450: mchars_alloc();
451: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
1.202 schwarze 452: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
453: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.50 kristaps 454:
455: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 456:
1.50 kristaps 457: /*
1.150 schwarze 458: * Most of these deal with a specific directory.
1.140 schwarze 459: * Jump into that directory first.
1.50 kristaps 460: */
1.165 schwarze 461: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.50 kristaps 462: goto out;
1.140 schwarze 463:
1.138 schwarze 464: if (dbopen(1)) {
1.140 schwarze 465: /*
466: * The existing database is usable. Process
467: * all files specified on the command-line.
468: */
1.207 schwarze 469: #if HAVE_PLEDGE
1.210 schwarze 470: if (!nodb) {
471: if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
1.214 schwarze 472: warn("pledge");
1.210 schwarze 473: exitcode = (int)MANDOCLEVEL_SYSERR;
474: goto out;
475: }
1.207 schwarze 476: }
477: #endif
1.140 schwarze 478: use_all = 1;
1.138 schwarze 479: for (i = 0; i < argc; i++)
480: filescan(argv[i]);
481: if (OP_TEST != op)
482: dbprune();
483: } else {
484: /*
485: * Database missing or corrupt.
486: * Recreate from scratch.
487: */
1.140 schwarze 488: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 489: op = OP_DEFAULT;
490: if (0 == treescan())
491: goto out;
492: if (0 == dbopen(0))
493: goto out;
494: }
1.50 kristaps 495: if (OP_DELETE != op)
1.176 schwarze 496: mpages_merge(mp);
1.138 schwarze 497: dbclose(OP_DEFAULT == op ? 0 : 1);
1.50 kristaps 498: } else {
499: /*
500: * If we have arguments, use them as our manpaths.
501: * If we don't, grok from manpath(1) or however else
1.187 schwarze 502: * manconf_parse() wants to do it.
1.50 kristaps 503: */
504: if (argc > 0) {
1.187 schwarze 505: conf.manpath.paths = mandoc_reallocarray(NULL,
1.144 schwarze 506: argc, sizeof(char *));
1.187 schwarze 507: conf.manpath.sz = (size_t)argc;
1.50 kristaps 508: for (i = 0; i < argc; i++)
1.187 schwarze 509: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.50 kristaps 510: } else
1.187 schwarze 511: manconf_parse(&conf, path_arg, NULL, NULL);
1.127 schwarze 512:
1.187 schwarze 513: if (conf.manpath.sz == 0) {
1.127 schwarze 514: exitcode = (int)MANDOCLEVEL_BADARG;
515: say("", "Empty manpath");
516: }
1.50 kristaps 517:
518: /*
1.69 schwarze 519: * First scan the tree rooted at a base directory, then
520: * build a new database and finally move it into place.
1.50 kristaps 521: * Ignore zero-length directories and strip trailing
522: * slashes.
523: */
1.187 schwarze 524: for (j = 0; j < conf.manpath.sz; j++) {
525: sz = strlen(conf.manpath.paths[j]);
526: if (sz && conf.manpath.paths[j][sz - 1] == '/')
527: conf.manpath.paths[j][--sz] = '\0';
1.50 kristaps 528: if (0 == sz)
529: continue;
1.66 schwarze 530:
531: if (j) {
1.202 schwarze 532: mandoc_ohash_init(&mpages, 6,
533: offsetof(struct mpage, inodev));
534: mandoc_ohash_init(&mlinks, 6,
535: offsetof(struct mlink, file));
1.66 schwarze 536: }
537:
1.187 schwarze 538: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.165 schwarze 539: continue;
1.59 schwarze 540: if (0 == treescan())
1.165 schwarze 541: continue;
1.59 schwarze 542: if (0 == dbopen(0))
1.165 schwarze 543: continue;
1.52 kristaps 544:
1.176 schwarze 545: mpages_merge(mp);
1.147 schwarze 546: if (warnings && !nodb &&
1.130 schwarze 547: ! (MPARSE_QUICK & mparse_options))
548: names_check();
1.59 schwarze 549: dbclose(0);
1.66 schwarze 550:
1.187 schwarze 551: if (j + 1 < conf.manpath.sz) {
1.80 schwarze 552: mpages_free();
1.78 schwarze 553: ohash_delete(&mpages);
1.83 schwarze 554: ohash_delete(&mlinks);
1.66 schwarze 555: }
1.50 kristaps 556: }
557: }
558: out:
1.187 schwarze 559: manconf_free(&conf);
1.168 schwarze 560: mparse_free(mp);
1.203 schwarze 561: mchars_free();
1.80 schwarze 562: mpages_free();
1.78 schwarze 563: ohash_delete(&mpages);
1.83 schwarze 564: ohash_delete(&mlinks);
1.197 schwarze 565: return exitcode;
1.50 kristaps 566: usage:
1.206 schwarze 567: progname = getprogname();
1.126 schwarze 568: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
569: " %s [-aDnpQ] [-Tutf8] dir ...\n"
570: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
571: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 572: " %s [-Q] -t file ...\n",
1.206 schwarze 573: progname, progname, progname, progname, progname);
1.50 kristaps 574:
1.197 schwarze 575: return (int)MANDOCLEVEL_BADARG;
1.50 kristaps 576: }
577:
578: /*
1.59 schwarze 579: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 580: * We use fts(), scanning directory parts along the way for clues to our
581: * section and architecture.
582: *
583: * If use_all has been specified, grok all files.
584: * If not, sanitise paths to the following:
585: *
1.141 schwarze 586: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 587: * or
588: * [./]cat<section>[/<arch>]/<name>.0
589: *
1.216 schwarze 590: * TODO: accommodate for multi-language directories.
1.50 kristaps 591: */
592: static int
1.59 schwarze 593: treescan(void)
1.50 kristaps 594: {
1.139 schwarze 595: char buf[PATH_MAX];
1.50 kristaps 596: FTS *f;
597: FTSENT *ff;
1.84 schwarze 598: struct mlink *mlink;
1.124 schwarze 599: int dform, gzip;
1.94 schwarze 600: char *dsec, *arch, *fsec, *cp;
601: const char *path;
1.50 kristaps 602: const char *argv[2];
603:
604: argv[0] = ".";
605: argv[1] = (char *)NULL;
606:
1.139 schwarze 607: f = fts_open((char * const *)argv,
608: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.200 schwarze 609: if (f == NULL) {
1.59 schwarze 610: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 611: say("", "&fts_open");
1.197 schwarze 612: return 0;
1.50 kristaps 613: }
614:
615: dsec = arch = NULL;
616: dform = FORM_NONE;
617:
1.200 schwarze 618: while ((ff = fts_read(f)) != NULL) {
1.50 kristaps 619: path = ff->fts_path + 2;
1.139 schwarze 620: switch (ff->fts_info) {
621:
622: /*
623: * Symbolic links require various sanity checks,
624: * then get handled just like regular files.
625: */
1.141 schwarze 626: case FTS_SL:
1.200 schwarze 627: if (realpath(path, buf) == NULL) {
1.139 schwarze 628: if (warnings)
629: say(path, "&realpath");
630: continue;
631: }
1.184 schwarze 632: if (strstr(buf, basedir) != buf
633: #ifdef HOMEBREWDIR
634: && strstr(buf, HOMEBREWDIR) != buf
635: #endif
636: ) {
1.139 schwarze 637: if (warnings) say("",
638: "%s: outside base directory", buf);
639: continue;
640: }
641: /* Use logical inode to avoid mpages dupe. */
1.200 schwarze 642: if (stat(path, ff->fts_statp) == -1) {
1.139 schwarze 643: if (warnings)
644: say(path, "&stat");
645: continue;
646: }
647: /* FALLTHROUGH */
648:
1.50 kristaps 649: /*
1.83 schwarze 650: * If we're a regular file, add an mlink by using the
1.50 kristaps 651: * stored directory data and handling the filename.
652: */
1.141 schwarze 653: case FTS_F:
1.200 schwarze 654: if ( ! strcmp(path, MANDOC_DB))
1.60 schwarze 655: continue;
1.50 kristaps 656: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 657: if (warnings)
1.59 schwarze 658: say(path, "Extraneous file");
1.50 kristaps 659: continue;
1.124 schwarze 660: }
661: gzip = 0;
662: fsec = NULL;
1.200 schwarze 663: while (fsec == NULL) {
1.124 schwarze 664: fsec = strrchr(ff->fts_name, '.');
1.200 schwarze 665: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.124 schwarze 666: break;
667: gzip = 1;
668: *fsec = '\0';
669: fsec = NULL;
670: }
1.200 schwarze 671: if (fsec == NULL) {
1.60 schwarze 672: if ( ! use_all) {
1.56 schwarze 673: if (warnings)
1.60 schwarze 674: say(path,
675: "No filename suffix");
1.50 kristaps 676: continue;
677: }
1.200 schwarze 678: } else if ( ! strcmp(++fsec, "html")) {
1.60 schwarze 679: if (warnings)
680: say(path, "Skip html");
681: continue;
1.200 schwarze 682: } else if ( ! strcmp(fsec, "ps")) {
1.60 schwarze 683: if (warnings)
684: say(path, "Skip ps");
685: continue;
1.200 schwarze 686: } else if ( ! strcmp(fsec, "pdf")) {
1.60 schwarze 687: if (warnings)
688: say(path, "Skip pdf");
689: continue;
690: } else if ( ! use_all &&
1.200 schwarze 691: ((dform == FORM_SRC &&
1.185 schwarze 692: strncmp(fsec, dsec, strlen(dsec))) ||
1.200 schwarze 693: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.60 schwarze 694: if (warnings)
695: say(path, "Wrong filename suffix");
696: continue;
1.66 schwarze 697: } else
1.84 schwarze 698: fsec[-1] = '\0';
1.94 schwarze 699:
1.84 schwarze 700: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 701: if (strlcpy(mlink->file, path,
702: sizeof(mlink->file)) >=
703: sizeof(mlink->file)) {
704: say(path, "Filename too long");
705: free(mlink);
706: continue;
707: }
1.84 schwarze 708: mlink->dform = dform;
1.94 schwarze 709: mlink->dsec = dsec;
710: mlink->arch = arch;
711: mlink->name = ff->fts_name;
712: mlink->fsec = fsec;
1.124 schwarze 713: mlink->gzip = gzip;
1.84 schwarze 714: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 715: continue;
1.139 schwarze 716:
1.141 schwarze 717: case FTS_D:
718: case FTS_DP:
1.139 schwarze 719: break;
720:
721: default:
1.60 schwarze 722: if (warnings)
723: say(path, "Not a regular file");
1.50 kristaps 724: continue;
1.60 schwarze 725: }
1.1 kristaps 726:
1.50 kristaps 727: switch (ff->fts_level) {
1.141 schwarze 728: case 0:
1.50 kristaps 729: /* Ignore the root directory. */
730: break;
1.141 schwarze 731: case 1:
1.50 kristaps 732: /*
733: * This might contain manX/ or catX/.
734: * Try to infer this from the name.
735: * If we're not in use_all, enforce it.
736: */
737: cp = ff->fts_name;
1.200 schwarze 738: if (ff->fts_info == FTS_DP) {
739: dform = FORM_NONE;
740: dsec = NULL;
1.50 kristaps 741: break;
1.200 schwarze 742: }
1.1 kristaps 743:
1.200 schwarze 744: if ( ! strncmp(cp, "man", 3)) {
1.50 kristaps 745: dform = FORM_SRC;
1.66 schwarze 746: dsec = cp + 3;
1.200 schwarze 747: } else if ( ! strncmp(cp, "cat", 3)) {
1.50 kristaps 748: dform = FORM_CAT;
1.66 schwarze 749: dsec = cp + 3;
1.94 schwarze 750: } else {
751: dform = FORM_NONE;
752: dsec = NULL;
1.50 kristaps 753: }
1.1 kristaps 754:
1.200 schwarze 755: if (dsec != NULL || use_all)
1.50 kristaps 756: break;
1.1 kristaps 757:
1.56 schwarze 758: if (warnings)
1.59 schwarze 759: say(path, "Unknown directory part");
1.50 kristaps 760: fts_set(f, ff, FTS_SKIP);
761: break;
1.141 schwarze 762: case 2:
1.50 kristaps 763: /*
764: * Possibly our architecture.
765: * If we're descending, keep tabs on it.
766: */
1.200 schwarze 767: if (ff->fts_info != FTS_DP && dsec != NULL)
1.66 schwarze 768: arch = ff->fts_name;
1.94 schwarze 769: else
770: arch = NULL;
1.50 kristaps 771: break;
772: default:
1.200 schwarze 773: if (ff->fts_info == FTS_DP || use_all)
1.50 kristaps 774: break;
1.56 schwarze 775: if (warnings)
1.59 schwarze 776: say(path, "Extraneous directory part");
1.50 kristaps 777: fts_set(f, ff, FTS_SKIP);
778: break;
1.5 kristaps 779: }
1.50 kristaps 780: }
781:
782: fts_close(f);
1.197 schwarze 783: return 1;
1.50 kristaps 784: }
1.5 kristaps 785:
1.50 kristaps 786: /*
1.89 schwarze 787: * Add a file to the mlinks table.
1.50 kristaps 788: * Do not verify that it's a "valid" looking manpage (we'll do that
789: * later).
790: *
791: * Try to infer the manual section, architecture, and page name from the
792: * path, assuming it looks like
793: *
1.141 schwarze 794: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 795: * or
796: * [./]cat<section>[/<arch>]/<name>.0
797: *
798: * See treescan() for the fts(3) version of this.
799: */
800: static void
1.59 schwarze 801: filescan(const char *file)
1.50 kristaps 802: {
1.59 schwarze 803: char buf[PATH_MAX];
1.84 schwarze 804: struct stat st;
805: struct mlink *mlink;
1.59 schwarze 806: char *p, *start;
1.5 kristaps 807:
1.50 kristaps 808: assert(use_all);
1.5 kristaps 809:
1.50 kristaps 810: if (0 == strncmp(file, "./", 2))
811: file += 2;
1.5 kristaps 812:
1.139 schwarze 813: /*
814: * We have to do lstat(2) before realpath(3) loses
815: * the information whether this is a symbolic link.
816: * We need to know that because for symbolic links,
817: * we want to use the orginal file name, while for
818: * regular files, we want to use the real path.
819: */
820: if (-1 == lstat(file, &st)) {
821: exitcode = (int)MANDOCLEVEL_BADARG;
822: say(file, "&lstat");
823: return;
824: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
825: exitcode = (int)MANDOCLEVEL_BADARG;
826: say(file, "Not a regular file");
827: return;
828: }
829:
830: /*
831: * We have to resolve the file name to the real path
832: * in any case for the base directory check.
833: */
1.59 schwarze 834: if (NULL == realpath(file, buf)) {
835: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 836: say(file, "&realpath");
1.59 schwarze 837: return;
1.106 schwarze 838: }
839:
1.150 schwarze 840: if (OP_TEST == op)
1.106 schwarze 841: start = buf;
1.150 schwarze 842: else if (strstr(buf, basedir) == buf)
843: start = buf + strlen(basedir);
1.184 schwarze 844: #ifdef HOMEBREWDIR
845: else if (strstr(buf, HOMEBREWDIR) == buf)
846: start = buf;
847: #endif
1.106 schwarze 848: else {
1.59 schwarze 849: exitcode = (int)MANDOCLEVEL_BADARG;
850: say("", "%s: outside base directory", buf);
851: return;
1.106 schwarze 852: }
853:
1.139 schwarze 854: /*
855: * Now we are sure the file is inside our tree.
856: * If it is a symbolic link, ignore the real path
857: * and use the original name.
858: * This implies passing stuff like "cat1/../man1/foo.1"
859: * on the command line won't work. So don't do that.
860: * Note the stat(2) can still fail if the link target
861: * doesn't exist.
862: */
863: if (S_IFLNK & st.st_mode) {
864: if (-1 == stat(buf, &st)) {
865: exitcode = (int)MANDOCLEVEL_BADARG;
866: say(file, "&stat");
867: return;
868: }
1.143 schwarze 869: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
870: say(file, "Filename too long");
871: return;
872: }
1.150 schwarze 873: start = buf;
874: if (OP_TEST != op && strstr(buf, basedir) == buf)
875: start += strlen(basedir);
1.50 kristaps 876: }
1.106 schwarze 877:
1.84 schwarze 878: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 879: mlink->dform = FORM_NONE;
1.143 schwarze 880: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
881: sizeof(mlink->file)) {
882: say(start, "Filename too long");
1.180 schwarze 883: free(mlink);
1.143 schwarze 884: return;
885: }
1.17 schwarze 886:
1.50 kristaps 887: /*
888: * First try to guess our directory structure.
889: * If we find a separator, try to look for man* or cat*.
890: * If we find one of these and what's underneath is a directory,
891: * assume it's an architecture.
892: */
893: if (NULL != (p = strchr(start, '/'))) {
894: *p++ = '\0';
895: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 896: mlink->dform = FORM_SRC;
1.94 schwarze 897: mlink->dsec = start + 3;
1.50 kristaps 898: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 899: mlink->dform = FORM_CAT;
1.94 schwarze 900: mlink->dsec = start + 3;
1.17 schwarze 901: }
1.5 kristaps 902:
1.50 kristaps 903: start = p;
1.84 schwarze 904: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 905: *p++ = '\0';
1.94 schwarze 906: mlink->arch = start;
1.50 kristaps 907: start = p;
1.84 schwarze 908: }
1.50 kristaps 909: }
910:
911: /*
912: * Now check the file suffix.
913: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
914: */
915: p = strrchr(start, '\0');
916: while (p-- > start && '/' != *p && '.' != *p)
917: /* Loop. */ ;
918:
919: if ('.' == *p) {
920: *p++ = '\0';
1.94 schwarze 921: mlink->fsec = p;
1.5 kristaps 922: }
923:
1.10 kristaps 924: /*
1.50 kristaps 925: * Now try to parse the name.
926: * Use the filename portion of the path.
1.10 kristaps 927: */
1.84 schwarze 928: mlink->name = start;
1.50 kristaps 929: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 930: mlink->name = p + 1;
1.50 kristaps 931: *p = '\0';
1.84 schwarze 932: }
933: mlink_add(mlink, &st);
1.50 kristaps 934: }
1.5 kristaps 935:
1.50 kristaps 936: static void
1.84 schwarze 937: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 938: {
1.83 schwarze 939: struct inodev inodev;
1.78 schwarze 940: struct mpage *mpage;
1.83 schwarze 941: unsigned int slot;
1.50 kristaps 942:
1.84 schwarze 943: assert(NULL != mlink->file);
1.50 kristaps 944:
1.94 schwarze 945: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
946: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
947: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
948: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 949:
950: if ('0' == *mlink->fsec) {
951: free(mlink->fsec);
952: mlink->fsec = mandoc_strdup(mlink->dsec);
953: mlink->fform = FORM_CAT;
954: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
955: mlink->fform = FORM_SRC;
1.74 schwarze 956: else
1.84 schwarze 957: mlink->fform = FORM_NONE;
1.82 schwarze 958:
1.83 schwarze 959: slot = ohash_qlookup(&mlinks, mlink->file);
960: assert(NULL == ohash_find(&mlinks, slot));
961: ohash_insert(&mlinks, slot, mlink);
962:
1.186 schwarze 963: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 964: inodev.st_ino = st->st_ino;
965: inodev.st_dev = st->st_dev;
966: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
967: sizeof(struct inodev), inodev.st_ino);
968: mpage = ohash_find(&mpages, slot);
969: if (NULL == mpage) {
970: mpage = mandoc_calloc(1, sizeof(struct mpage));
971: mpage->inodev.st_ino = inodev.st_ino;
972: mpage->inodev.st_dev = inodev.st_dev;
973: ohash_insert(&mpages, slot, mpage);
974: } else
1.85 schwarze 975: mlink->next = mpage->mlinks;
1.82 schwarze 976: mpage->mlinks = mlink;
1.118 schwarze 977: mlink->mpage = mpage;
1.82 schwarze 978: }
979:
980: static void
981: mlink_free(struct mlink *mlink)
982: {
983:
984: free(mlink->dsec);
985: free(mlink->arch);
986: free(mlink->name);
987: free(mlink->fsec);
988: free(mlink);
1.50 kristaps 989: }
1.3 kristaps 990:
1.50 kristaps 991: static void
1.78 schwarze 992: mpages_free(void)
1.50 kristaps 993: {
1.78 schwarze 994: struct mpage *mpage;
1.82 schwarze 995: struct mlink *mlink;
1.80 schwarze 996: unsigned int slot;
1.3 kristaps 997:
1.80 schwarze 998: mpage = ohash_first(&mpages, &slot);
999: while (NULL != mpage) {
1.82 schwarze 1000: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 1001: mpage->mlinks = mlink->next;
1.82 schwarze 1002: mlink_free(mlink);
1003: }
1.78 schwarze 1004: free(mpage->sec);
1005: free(mpage->arch);
1.82 schwarze 1006: free(mpage->title);
1007: free(mpage->desc);
1.78 schwarze 1008: free(mpage);
1.80 schwarze 1009: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 1010: }
1011: }
1.38 schwarze 1012:
1.50 kristaps 1013: /*
1.89 schwarze 1014: * For each mlink to the mpage, check whether the path looks like
1015: * it is formatted, and if it does, check whether a source manual
1016: * exists by the same name, ignoring the suffix.
1017: * If both conditions hold, drop the mlink.
1018: */
1019: static void
1020: mlinks_undupe(struct mpage *mpage)
1021: {
1022: char buf[PATH_MAX];
1023: struct mlink **prev;
1024: struct mlink *mlink;
1025: char *bufp;
1026:
1027: mpage->form = FORM_CAT;
1.90 schwarze 1028: prev = &mpage->mlinks;
1029: while (NULL != (mlink = *prev)) {
1.89 schwarze 1030: if (FORM_CAT != mlink->dform) {
1031: mpage->form = FORM_NONE;
1.90 schwarze 1032: goto nextlink;
1.89 schwarze 1033: }
1.143 schwarze 1034: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1035: bufp = strstr(buf, "cat");
1036: assert(NULL != bufp);
1037: memcpy(bufp, "man", 3);
1038: if (NULL != (bufp = strrchr(buf, '.')))
1039: *++bufp = '\0';
1.143 schwarze 1040: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1041: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1042: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1043: goto nextlink;
1.89 schwarze 1044: if (warnings)
1045: say(mlink->file, "Man source exists: %s", buf);
1046: if (use_all)
1.90 schwarze 1047: goto nextlink;
1.89 schwarze 1048: *prev = mlink->next;
1049: mlink_free(mlink);
1.90 schwarze 1050: continue;
1051: nextlink:
1052: prev = &(*prev)->next;
1.89 schwarze 1053: }
1054: }
1055:
1.131 schwarze 1056: static void
1.93 schwarze 1057: mlink_check(struct mpage *mpage, struct mlink *mlink)
1058: {
1.131 schwarze 1059: struct str *str;
1060: unsigned int slot;
1.93 schwarze 1061:
1062: /*
1063: * Check whether the manual section given in a file
1064: * agrees with the directory where the file is located.
1065: * Some manuals have suffixes like (3p) on their
1066: * section number either inside the file or in the
1067: * directory name, some are linked into more than one
1068: * section, like encrypt(1) = makekey(8).
1069: */
1070:
1071: if (FORM_SRC == mpage->form &&
1.131 schwarze 1072: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1073: say(mlink->file, "Section \"%s\" manual in %s directory",
1074: mpage->sec, mlink->dsec);
1075:
1076: /*
1077: * Manual page directories exist for each kernel
1078: * architecture as returned by machine(1).
1079: * However, many manuals only depend on the
1080: * application architecture as returned by arch(1).
1081: * For example, some (2/ARM) manuals are shared
1082: * across the "armish" and "zaurus" kernel
1083: * architectures.
1084: * A few manuals are even shared across completely
1085: * different architectures, for example fdformat(1)
1086: * on amd64, i386, sparc, and sparc64.
1087: */
1088:
1.131 schwarze 1089: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1090: say(mlink->file, "Architecture \"%s\" manual in "
1091: "\"%s\" directory", mpage->arch, mlink->arch);
1092:
1.131 schwarze 1093: /*
1094: * XXX
1.133 schwarze 1095: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1096: */
1097:
1098: if (FORM_CAT == mpage->form)
1099: return;
1100:
1101: /*
1102: * Check whether this mlink
1103: * appears as a name in the NAME section.
1104: */
1.93 schwarze 1105:
1.133 schwarze 1106: slot = ohash_qlookup(&names, mlink->name);
1107: str = ohash_find(&names, slot);
1.131 schwarze 1108: assert(NULL != str);
1.133 schwarze 1109: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1110: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1111: }
1112:
1.89 schwarze 1113: /*
1.80 schwarze 1114: * Run through the files in the global vector "mpages"
1115: * and add them to the database specified in "basedir".
1.50 kristaps 1116: *
1117: * This handles the parsing scheme itself, using the cues of directory
1118: * and filename to determine whether the file is parsable or not.
1119: */
1.59 schwarze 1120: static void
1.176 schwarze 1121: mpages_merge(struct mparse *mp)
1.50 kristaps 1122: {
1.113 schwarze 1123: char any[] = "any";
1.118 schwarze 1124: struct mpage *mpage, *mpage_dest;
1125: struct mlink *mlink, *mlink_dest;
1.191 schwarze 1126: struct roff_man *man;
1.118 schwarze 1127: char *sodest;
1.112 schwarze 1128: char *cp;
1.162 schwarze 1129: int fd;
1.101 schwarze 1130: unsigned int pslot;
1.69 schwarze 1131:
1.171 schwarze 1132: if ( ! nodb)
1.107 schwarze 1133: SQL_EXEC("BEGIN TRANSACTION");
1134:
1.80 schwarze 1135: mpage = ohash_first(&mpages, &pslot);
1.171 schwarze 1136: while (mpage != NULL) {
1.89 schwarze 1137: mlinks_undupe(mpage);
1.182 schwarze 1138: if ((mlink = mpage->mlinks) == NULL) {
1.89 schwarze 1139: mpage = ohash_next(&mpages, &pslot);
1140: continue;
1.50 kristaps 1141: }
1142:
1.133 schwarze 1143: name_mask = NAME_MASK;
1.202 schwarze 1144: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1145: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.50 kristaps 1146: mparse_reset(mp);
1147: man = NULL;
1.124 schwarze 1148: sodest = NULL;
1149:
1.213 schwarze 1150: if ((fd = mparse_open(mp, mlink->file)) == -1) {
1.182 schwarze 1151: say(mlink->file, "&open");
1.162 schwarze 1152: goto nextpage;
1.124 schwarze 1153: }
1.14 schwarze 1154:
1155: /*
1.183 schwarze 1156: * Interpret the file as mdoc(7) or man(7) source
1157: * code, unless it is known to be formatted.
1.14 schwarze 1158: */
1.182 schwarze 1159: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1160: mparse_readfd(mp, fd, mlink->file);
1.212 schwarze 1161: close(fd);
1.192 schwarze 1162: mparse_result(mp, &man, &sodest);
1.76 schwarze 1163: }
1.14 schwarze 1164:
1.171 schwarze 1165: if (sodest != NULL) {
1.118 schwarze 1166: mlink_dest = ohash_find(&mlinks,
1167: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1168: if (mlink_dest == NULL) {
1169: mandoc_asprintf(&cp, "%s.gz", sodest);
1170: mlink_dest = ohash_find(&mlinks,
1171: ohash_qlookup(&mlinks, cp));
1172: free(cp);
1173: }
1174: if (mlink_dest != NULL) {
1.118 schwarze 1175:
1176: /* The .so target exists. */
1177:
1178: mpage_dest = mlink_dest->mpage;
1179: while (1) {
1180: mlink->mpage = mpage_dest;
1181:
1182: /*
1183: * If the target was already
1184: * processed, add the links
1185: * to the database now.
1186: * Otherwise, this will
1187: * happen when we come
1188: * to the target.
1189: */
1190:
1.137 schwarze 1191: if (mpage_dest->pageid)
1.169 schwarze 1192: dbadd_mlink_name(mlink);
1.118 schwarze 1193:
1.171 schwarze 1194: if (mlink->next == NULL)
1.118 schwarze 1195: break;
1196: mlink = mlink->next;
1197: }
1198:
1199: /* Move all links to the target. */
1200:
1201: mlink->next = mlink_dest->next;
1202: mlink_dest->next = mpage->mlinks;
1203: mpage->mlinks = NULL;
1204: }
1.124 schwarze 1205: goto nextpage;
1.192 schwarze 1206: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.204 schwarze 1207: mdoc_validate(man);
1.82 schwarze 1208: mpage->form = FORM_SRC;
1.193 schwarze 1209: mpage->sec = man->meta.msec;
1.155 schwarze 1210: mpage->sec = mandoc_strdup(
1.171 schwarze 1211: mpage->sec == NULL ? "" : mpage->sec);
1.193 schwarze 1212: mpage->arch = man->meta.arch;
1.84 schwarze 1213: mpage->arch = mandoc_strdup(
1.171 schwarze 1214: mpage->arch == NULL ? "" : mpage->arch);
1.193 schwarze 1215: mpage->title = mandoc_strdup(man->meta.title);
1.192 schwarze 1216: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.205 schwarze 1217: man_validate(man);
1.82 schwarze 1218: mpage->form = FORM_SRC;
1.193 schwarze 1219: mpage->sec = mandoc_strdup(man->meta.msec);
1.182 schwarze 1220: mpage->arch = mandoc_strdup(mlink->arch);
1.193 schwarze 1221: mpage->title = mandoc_strdup(man->meta.title);
1.76 schwarze 1222: } else {
1.82 schwarze 1223: mpage->form = FORM_CAT;
1.182 schwarze 1224: mpage->sec = mandoc_strdup(mlink->dsec);
1225: mpage->arch = mandoc_strdup(mlink->arch);
1226: mpage->title = mandoc_strdup(mlink->name);
1.76 schwarze 1227: }
1.97 schwarze 1228: putkey(mpage, mpage->sec, TYPE_sec);
1.164 schwarze 1229: if (*mpage->arch != '\0')
1230: putkey(mpage, mpage->arch, TYPE_arch);
1.50 kristaps 1231:
1.182 schwarze 1232: for ( ; mlink != NULL; mlink = mlink->next) {
1.97 schwarze 1233: if ('\0' != *mlink->dsec)
1234: putkey(mpage, mlink->dsec, TYPE_sec);
1235: if ('\0' != *mlink->fsec)
1236: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1237: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1238: any : mlink->arch, TYPE_arch);
1.133 schwarze 1239: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1240: }
1.12 schwarze 1241:
1.174 schwarze 1242: assert(mpage->desc == NULL);
1.192 schwarze 1243: if (man != NULL && man->macroset == MACROSET_MDOC)
1.193 schwarze 1244: parse_mdoc(mpage, &man->meta, man->first);
1.174 schwarze 1245: else if (man != NULL)
1.193 schwarze 1246: parse_man(mpage, &man->meta, man->first);
1.50 kristaps 1247: else
1.162 schwarze 1248: parse_cat(mpage, fd);
1.174 schwarze 1249: if (mpage->desc == NULL)
1.135 schwarze 1250: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1251:
1252: if (warnings && !use_all)
1253: for (mlink = mpage->mlinks; mlink;
1254: mlink = mlink->next)
1255: mlink_check(mpage, mlink);
1.44 kristaps 1256:
1.176 schwarze 1257: dbadd(mpage);
1.182 schwarze 1258: mlink = mpage->mlinks;
1.124 schwarze 1259:
1260: nextpage:
1.66 schwarze 1261: ohash_delete(&strings);
1.133 schwarze 1262: ohash_delete(&names);
1.80 schwarze 1263: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1264: }
1.107 schwarze 1265:
1266: if (0 == nodb)
1267: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1268: }
1269:
1270: static void
1271: names_check(void)
1272: {
1273: sqlite3_stmt *stmt;
1274: const char *name, *sec, *arch, *key;
1275:
1276: sqlite3_prepare_v2(db,
1277: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1278: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1279: "WHERE bits & ? AND NOT EXISTS ("
1280: "SELECT pageid FROM mlinks "
1.133 schwarze 1281: "WHERE mlinks.pageid == names.pageid "
1282: "AND mlinks.name == names.name"
1.130 schwarze 1283: ")"
1284: ") JOIN ("
1.142 schwarze 1285: "SELECT sec, arch, name, pageid FROM mlinks "
1286: "GROUP BY pageid"
1.130 schwarze 1287: ") USING (pageid);",
1288: -1, &stmt, NULL);
1289:
1.201 schwarze 1290: if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK)
1.134 schwarze 1291: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1292:
1.201 schwarze 1293: while (sqlite3_step(stmt) == SQLITE_ROW) {
1.154 schwarze 1294: name = (const char *)sqlite3_column_text(stmt, 0);
1295: sec = (const char *)sqlite3_column_text(stmt, 1);
1296: arch = (const char *)sqlite3_column_text(stmt, 2);
1297: key = (const char *)sqlite3_column_text(stmt, 3);
1.130 schwarze 1298: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1299: '\0' == *arch ? "" : "/",
1300: '\0' == *arch ? "" : arch, key);
1301: }
1302: sqlite3_finalize(stmt);
1.50 kristaps 1303: }
1.12 schwarze 1304:
1.50 kristaps 1305: static void
1.124 schwarze 1306: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1307: {
1308: FILE *stream;
1309: char *line, *p, *title;
1.209 schwarze 1310: size_t linesz, plen, titlesz;
1311: ssize_t len;
1312: int offs;
1.12 schwarze 1313:
1.124 schwarze 1314: stream = (-1 == fd) ?
1315: fopen(mpage->mlinks->file, "r") :
1316: fdopen(fd, "r");
1317: if (NULL == stream) {
1.159 schwarze 1318: if (-1 != fd)
1319: close(fd);
1.56 schwarze 1320: if (warnings)
1.123 schwarze 1321: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1322: return;
1323: }
1.1 kristaps 1324:
1.209 schwarze 1325: line = NULL;
1326: linesz = 0;
1327:
1.50 kristaps 1328: /* Skip to first blank line. */
1.1 kristaps 1329:
1.209 schwarze 1330: while (getline(&line, &linesz, stream) != -1)
1331: if (*line == '\n')
1.50 kristaps 1332: break;
1.1 kristaps 1333:
1.50 kristaps 1334: /*
1335: * Assume the first line that is not indented
1336: * is the first section header. Skip to it.
1337: */
1.1 kristaps 1338:
1.209 schwarze 1339: while (getline(&line, &linesz, stream) != -1)
1340: if (*line != '\n' && *line != ' ')
1.50 kristaps 1341: break;
1.141 schwarze 1342:
1.50 kristaps 1343: /*
1344: * Read up until the next section into a buffer.
1345: * Strip the leading and trailing newline from each read line,
1346: * appending a trailing space.
1347: * Ignore empty (whitespace-only) lines.
1348: */
1.1 kristaps 1349:
1.50 kristaps 1350: titlesz = 0;
1351: title = NULL;
1.38 schwarze 1352:
1.209 schwarze 1353: while ((len = getline(&line, &linesz, stream)) != -1) {
1354: if (*line != ' ')
1.50 kristaps 1355: break;
1.209 schwarze 1356: offs = 0;
1357: while (isspace((unsigned char)line[offs]))
1358: offs++;
1359: if (line[offs] == '\0')
1.38 schwarze 1360: continue;
1.209 schwarze 1361: title = mandoc_realloc(title, titlesz + len - offs);
1362: memcpy(title + titlesz, line + offs, len - offs);
1363: titlesz += len - offs;
1.50 kristaps 1364: title[titlesz - 1] = ' ';
1365: }
1.209 schwarze 1366: free(line);
1.38 schwarze 1367:
1.50 kristaps 1368: /*
1369: * If no page content can be found, or the input line
1370: * is already the next section header, or there is no
1371: * trailing newline, reuse the page title as the page
1372: * description.
1373: */
1.44 kristaps 1374:
1.50 kristaps 1375: if (NULL == title || '\0' == *title) {
1.56 schwarze 1376: if (warnings)
1.82 schwarze 1377: say(mpage->mlinks->file,
1378: "Cannot find NAME section");
1.50 kristaps 1379: fclose(stream);
1380: free(title);
1381: return;
1382: }
1.1 kristaps 1383:
1.209 schwarze 1384: title[titlesz - 1] = '\0';
1.33 schwarze 1385:
1.50 kristaps 1386: /*
1387: * Skip to the first dash.
1388: * Use the remaining line as the description (no more than 70
1389: * bytes).
1390: */
1.33 schwarze 1391:
1.50 kristaps 1392: if (NULL != (p = strstr(title, "- "))) {
1393: for (p += 2; ' ' == *p || '\b' == *p; p++)
1394: /* Skip to next word. */ ;
1395: } else {
1.56 schwarze 1396: if (warnings)
1.82 schwarze 1397: say(mpage->mlinks->file,
1398: "No dash in title line");
1.50 kristaps 1399: p = title;
1400: }
1.38 schwarze 1401:
1.50 kristaps 1402: plen = strlen(p);
1.1 kristaps 1403:
1.50 kristaps 1404: /* Strip backspace-encoding from line. */
1.1 kristaps 1405:
1.50 kristaps 1406: while (NULL != (line = memchr(p, '\b', plen))) {
1407: len = line - p;
1408: if (0 == len) {
1409: memmove(line, line + 1, plen--);
1410: continue;
1.141 schwarze 1411: }
1.50 kristaps 1412: memmove(line - 1, line + 1, plen - len);
1413: plen -= 2;
1414: }
1.1 kristaps 1415:
1.78 schwarze 1416: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1417: fclose(stream);
1418: free(title);
1419: }
1.1 kristaps 1420:
1.50 kristaps 1421: /*
1422: * Put a type/word pair into the word database for this particular file.
1423: */
1424: static void
1.112 schwarze 1425: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1426: {
1.112 schwarze 1427: char *cp;
1.18 kristaps 1428:
1.50 kristaps 1429: assert(NULL != value);
1.112 schwarze 1430: if (TYPE_arch == type)
1431: for (cp = value; *cp; cp++)
1432: if (isupper((unsigned char)*cp))
1433: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1434: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1435: }
1436:
1437: /*
1.50 kristaps 1438: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1439: */
1440: static void
1.78 schwarze 1441: putmdockey(const struct mpage *mpage,
1.189 schwarze 1442: const struct roff_node *n, uint64_t m)
1.3 kristaps 1443: {
1.18 kristaps 1444:
1.50 kristaps 1445: for ( ; NULL != n; n = n->next) {
1446: if (NULL != n->child)
1.78 schwarze 1447: putmdockey(mpage, n->child, m);
1.188 schwarze 1448: if (n->type == ROFFT_TEXT)
1.78 schwarze 1449: putkey(mpage, n->string, m);
1.50 kristaps 1450: }
1451: }
1.18 kristaps 1452:
1.61 schwarze 1453: static void
1.190 schwarze 1454: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1455: const struct roff_node *n)
1.50 kristaps 1456: {
1.189 schwarze 1457: const struct roff_node *head, *body;
1.121 schwarze 1458: char *start, *title;
1.50 kristaps 1459: char byte;
1.121 schwarze 1460: size_t sz;
1.18 kristaps 1461:
1.215 schwarze 1462: if (n == NULL)
1.61 schwarze 1463: return;
1.18 kristaps 1464:
1.50 kristaps 1465: /*
1466: * We're only searching for one thing: the first text child in
1467: * the BODY of a NAME section. Since we don't keep track of
1468: * sections in -man, run some hoops to find out whether we're in
1469: * the correct section or not.
1470: */
1.18 kristaps 1471:
1.188 schwarze 1472: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.50 kristaps 1473: body = n;
1.215 schwarze 1474: if ((head = body->parent->head) != NULL &&
1475: (head = head->child) != NULL &&
1476: head->next == NULL &&
1.188 schwarze 1477: head->type == ROFFT_TEXT &&
1.215 schwarze 1478: strcmp(head->string, "NAME") == 0 &&
1479: body->child != NULL) {
1.3 kristaps 1480:
1.50 kristaps 1481: /*
1482: * Suck the entire NAME section into memory.
1483: * Yes, we might run away.
1484: * But too many manuals have big, spread-out
1485: * NAME sections over many lines.
1486: */
1.3 kristaps 1487:
1.121 schwarze 1488: title = NULL;
1.194 schwarze 1489: deroff(&title, body);
1.50 kristaps 1490: if (NULL == title)
1.61 schwarze 1491: return;
1.18 kristaps 1492:
1.141 schwarze 1493: /*
1.50 kristaps 1494: * Go through a special heuristic dance here.
1495: * Conventionally, one or more manual names are
1496: * comma-specified prior to a whitespace, then a
1497: * dash, then a description. Try to puzzle out
1498: * the name parts here.
1499: */
1.18 kristaps 1500:
1.121 schwarze 1501: start = title;
1.50 kristaps 1502: for ( ;; ) {
1503: sz = strcspn(start, " ,");
1504: if ('\0' == start[sz])
1505: break;
1.1 kristaps 1506:
1.50 kristaps 1507: byte = start[sz];
1508: start[sz] = '\0';
1.110 schwarze 1509:
1510: /*
1511: * Assume a stray trailing comma in the
1512: * name list if a name begins with a dash.
1513: */
1514:
1515: if ('-' == start[0] ||
1516: ('\\' == start[0] && '-' == start[1]))
1517: break;
1.1 kristaps 1518:
1.133 schwarze 1519: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1520: if ( ! (mpage->name_head_done ||
1521: strcasecmp(start, meta->title))) {
1522: putkey(mpage, start, NAME_HEAD);
1523: mpage->name_head_done = 1;
1524: }
1.1 kristaps 1525:
1.50 kristaps 1526: if (' ' == byte) {
1527: start += sz + 1;
1528: break;
1529: }
1.1 kristaps 1530:
1.50 kristaps 1531: assert(',' == byte);
1532: start += sz + 1;
1533: while (' ' == *start)
1534: start++;
1535: }
1.1 kristaps 1536:
1.121 schwarze 1537: if (start == title) {
1.133 schwarze 1538: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1539: if ( ! (mpage->name_head_done ||
1540: strcasecmp(start, meta->title))) {
1541: putkey(mpage, start, NAME_HEAD);
1542: mpage->name_head_done = 1;
1543: }
1.50 kristaps 1544: free(title);
1.61 schwarze 1545: return;
1.50 kristaps 1546: }
1.1 kristaps 1547:
1.50 kristaps 1548: while (isspace((unsigned char)*start))
1549: start++;
1.1 kristaps 1550:
1.50 kristaps 1551: if (0 == strncmp(start, "-", 1))
1552: start += 1;
1553: else if (0 == strncmp(start, "\\-\\-", 4))
1554: start += 4;
1555: else if (0 == strncmp(start, "\\-", 2))
1556: start += 2;
1557: else if (0 == strncmp(start, "\\(en", 4))
1558: start += 4;
1559: else if (0 == strncmp(start, "\\(em", 4))
1560: start += 4;
1.1 kristaps 1561:
1.50 kristaps 1562: while (' ' == *start)
1563: start++;
1.1 kristaps 1564:
1.78 schwarze 1565: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1566: free(title);
1.61 schwarze 1567: return;
1.50 kristaps 1568: }
1569: }
1.1 kristaps 1570:
1.77 schwarze 1571: for (n = n->child; n; n = n->next) {
1.78 schwarze 1572: if (NULL != mpage->desc)
1.77 schwarze 1573: break;
1.174 schwarze 1574: parse_man(mpage, meta, n);
1.77 schwarze 1575: }
1.1 kristaps 1576: }
1577:
1578: static void
1.190 schwarze 1579: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1580: const struct roff_node *n)
1.1 kristaps 1581: {
1582:
1.50 kristaps 1583: assert(NULL != n);
1584: for (n = n->child; NULL != n; n = n->next) {
1585: switch (n->type) {
1.188 schwarze 1586: case ROFFT_ELEM:
1587: case ROFFT_BLOCK:
1588: case ROFFT_HEAD:
1589: case ROFFT_BODY:
1590: case ROFFT_TAIL:
1.50 kristaps 1591: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1592: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1593: break;
1.68 schwarze 1594: if (mdocs[n->tok].mask)
1.78 schwarze 1595: putmdockey(mpage, n->child,
1596: mdocs[n->tok].mask);
1.50 kristaps 1597: break;
1598: default:
1.188 schwarze 1599: assert(n->type != ROFFT_ROOT);
1.50 kristaps 1600: continue;
1601: }
1602: if (NULL != n->child)
1.172 schwarze 1603: parse_mdoc(mpage, meta, n);
1.1 kristaps 1604: }
1605: }
1606:
1.25 schwarze 1607: static int
1.190 schwarze 1608: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1609: const struct roff_node *n)
1.1 kristaps 1610: {
1.176 schwarze 1611: char *start, *end;
1.1 kristaps 1612: size_t sz;
1.25 schwarze 1613:
1.50 kristaps 1614: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1615: NULL == (n = n->child) ||
1.188 schwarze 1616: n->type != ROFFT_TEXT)
1.197 schwarze 1617: return 0;
1.1 kristaps 1618:
1619: /*
1620: * Only consider those `Fd' macro fields that begin with an
1621: * "inclusion" token (versus, e.g., #define).
1622: */
1.50 kristaps 1623:
1.1 kristaps 1624: if (strcmp("#include", n->string))
1.197 schwarze 1625: return 0;
1.1 kristaps 1626:
1.188 schwarze 1627: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1628: return 0;
1.1 kristaps 1629:
1630: /*
1631: * Strip away the enclosing angle brackets and make sure we're
1632: * not zero-length.
1633: */
1634:
1635: start = n->string;
1636: if ('<' == *start || '"' == *start)
1637: start++;
1638:
1639: if (0 == (sz = strlen(start)))
1.197 schwarze 1640: return 0;
1.1 kristaps 1641:
1642: end = &start[(int)sz - 1];
1643: if ('>' == *end || '"' == *end)
1644: end--;
1645:
1.50 kristaps 1646: if (end > start)
1.78 schwarze 1647: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1648: return 0;
1.1 kristaps 1649: }
1650:
1.178 schwarze 1651: static void
1.189 schwarze 1652: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1653: {
1.112 schwarze 1654: char *cp;
1.178 schwarze 1655: size_t sz;
1.1 kristaps 1656:
1.188 schwarze 1657: if (n->type != ROFFT_TEXT)
1.178 schwarze 1658: return;
1.25 schwarze 1659:
1.178 schwarze 1660: /* Skip function pointer punctuation. */
1.1 kristaps 1661:
1.178 schwarze 1662: cp = n->string;
1663: while (*cp == '(' || *cp == '*')
1.1 kristaps 1664: cp++;
1.178 schwarze 1665: sz = strcspn(cp, "()");
1.1 kristaps 1666:
1.178 schwarze 1667: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1668: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1669: putkeys(mpage, cp, sz, NAME_SYN);
1670: }
1.25 schwarze 1671:
1.178 schwarze 1672: static int
1.190 schwarze 1673: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1674: const struct roff_node *n)
1.178 schwarze 1675: {
1676:
1677: if (n->child == NULL)
1.197 schwarze 1678: return 0;
1.178 schwarze 1679:
1680: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1681:
1.178 schwarze 1682: for (n = n->child->next; n != NULL; n = n->next)
1.188 schwarze 1683: if (n->type == ROFFT_TEXT)
1.78 schwarze 1684: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1685:
1.197 schwarze 1686: return 0;
1.173 schwarze 1687: }
1688:
1689: static int
1.190 schwarze 1690: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1691: const struct roff_node *n)
1.173 schwarze 1692: {
1.177 schwarze 1693:
1.188 schwarze 1694: if (n->type != ROFFT_HEAD)
1.197 schwarze 1695: return 1;
1.173 schwarze 1696:
1.178 schwarze 1697: if (n->child != NULL)
1698: parse_mdoc_fname(mpage, n->child);
1699:
1.197 schwarze 1700: return 0;
1.1 kristaps 1701: }
1702:
1.25 schwarze 1703: static int
1.211 schwarze 1704: parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1705: const struct roff_node *n)
1706: {
1707: char *cp;
1708:
1709: if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1710: return 0;
1711:
1.215 schwarze 1712: if (n->child != NULL &&
1713: n->child->next == NULL &&
1714: n->child->type == ROFFT_TEXT)
1.211 schwarze 1715: return 1;
1716:
1717: cp = NULL;
1718: deroff(&cp, n);
1719: if (cp != NULL) {
1720: putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1721: n->type == ROFFT_BODY ? TYPE_Va : 0));
1722: free(cp);
1723: }
1724:
1725: return 0;
1726: }
1727:
1728: static int
1.190 schwarze 1729: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1730: const struct roff_node *n)
1.1 kristaps 1731: {
1.67 schwarze 1732: char *cp;
1.1 kristaps 1733:
1734: if (NULL == (n = n->child))
1.197 schwarze 1735: return 0;
1.1 kristaps 1736:
1.67 schwarze 1737: if (NULL == n->next) {
1.78 schwarze 1738: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1739: return 0;
1.67 schwarze 1740: }
1741:
1.120 schwarze 1742: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1743: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1744: free(cp);
1.197 schwarze 1745: return 0;
1.1 kristaps 1746: }
1747:
1.25 schwarze 1748: static int
1.190 schwarze 1749: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1750: const struct roff_node *n)
1.1 kristaps 1751: {
1752:
1.188 schwarze 1753: if (n->type == ROFFT_BODY)
1.194 schwarze 1754: deroff(&mpage->desc, n);
1.197 schwarze 1755: return 0;
1.1 kristaps 1756: }
1757:
1.25 schwarze 1758: static int
1.190 schwarze 1759: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1760: const struct roff_node *n)
1.1 kristaps 1761: {
1762:
1.129 schwarze 1763: if (SEC_NAME == n->sec)
1.133 schwarze 1764: putmdockey(mpage, n->child, NAME_TITLE);
1.188 schwarze 1765: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1766: if (n->child == NULL)
1767: putkey(mpage, meta->name, NAME_SYN);
1768: else
1769: putmdockey(mpage, n->child, NAME_SYN);
1.174 schwarze 1770: }
1771: if ( ! (mpage->name_head_done ||
1772: n->child == NULL || n->child->string == NULL ||
1773: strcasecmp(n->child->string, meta->title))) {
1.188 schwarze 1774: putkey(mpage, n->child->string, ROFFT_HEAD);
1.174 schwarze 1775: mpage->name_head_done = 1;
1.172 schwarze 1776: }
1.197 schwarze 1777: return 0;
1.1 kristaps 1778: }
1779:
1.25 schwarze 1780: static int
1.190 schwarze 1781: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1782: const struct roff_node *n)
1.1 kristaps 1783: {
1784:
1.197 schwarze 1785: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1786: }
1787:
1.50 kristaps 1788: static int
1.190 schwarze 1789: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1790: const struct roff_node *n)
1.1 kristaps 1791: {
1792:
1.197 schwarze 1793: return n->type == ROFFT_HEAD;
1.1 kristaps 1794: }
1795:
1.50 kristaps 1796: /*
1.66 schwarze 1797: * Add a string to the hash table for the current manual.
1798: * Each string has a bitmask telling which macros it belongs to.
1799: * When we finish the manual, we'll dump the table.
1.50 kristaps 1800: */
1801: static void
1.176 schwarze 1802: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1803: {
1.133 schwarze 1804: struct ohash *htab;
1.50 kristaps 1805: struct str *s;
1.111 schwarze 1806: const char *end;
1.65 schwarze 1807: unsigned int slot;
1.176 schwarze 1808: int i, mustfree;
1.25 schwarze 1809:
1.50 kristaps 1810: if (0 == sz)
1811: return;
1.111 schwarze 1812:
1.176 schwarze 1813: mustfree = render_string(&cp, &sz);
1814:
1.133 schwarze 1815: if (TYPE_Nm & v) {
1816: htab = &names;
1817: v &= name_mask;
1.169 schwarze 1818: if (v & NAME_FIRST)
1819: name_mask &= ~NAME_FIRST;
1.133 schwarze 1820: if (debug > 1)
1821: say(mpage->mlinks->file,
1.220 ! schwarze 1822: "Adding name %*s, bits=0x%llu", (int)sz, cp, v);
1.133 schwarze 1823: } else {
1824: htab = &strings;
1825: if (debug > 1)
1826: for (i = 0; i < mansearch_keymax; i++)
1.163 schwarze 1827: if ((uint64_t)1 << i & v)
1.133 schwarze 1828: say(mpage->mlinks->file,
1829: "Adding key %s=%*s",
1.220 ! schwarze 1830: mansearch_keynames[i], (int)sz, cp);
1.111 schwarze 1831: }
1.25 schwarze 1832:
1.65 schwarze 1833: end = cp + sz;
1.133 schwarze 1834: slot = ohash_qlookupi(htab, cp, &end);
1835: s = ohash_find(htab, slot);
1.25 schwarze 1836:
1.78 schwarze 1837: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1838: s->mask |= v;
1839: return;
1840: } else if (NULL == s) {
1.144 schwarze 1841: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1842: memcpy(s->key, cp, sz);
1.133 schwarze 1843: ohash_insert(htab, slot, s);
1.1 kristaps 1844: }
1.78 schwarze 1845: s->mpage = mpage;
1.50 kristaps 1846: s->mask = v;
1.176 schwarze 1847:
1848: if (mustfree)
1849: free(cp);
1.1 kristaps 1850: }
1851:
1.50 kristaps 1852: /*
1853: * Take a Unicode codepoint and produce its UTF-8 encoding.
1854: * This isn't the best way to do this, but it works.
1855: * The magic numbers are from the UTF-8 packaging.
1856: * They're not as scary as they seem: read the UTF-8 spec for details.
1857: */
1858: static size_t
1859: utf8(unsigned int cp, char out[7])
1.1 kristaps 1860: {
1.50 kristaps 1861: size_t rc;
1.1 kristaps 1862:
1.50 kristaps 1863: rc = 0;
1864: if (cp <= 0x0000007F) {
1865: rc = 1;
1866: out[0] = (char)cp;
1867: } else if (cp <= 0x000007FF) {
1868: rc = 2;
1869: out[0] = (cp >> 6 & 31) | 192;
1870: out[1] = (cp & 63) | 128;
1871: } else if (cp <= 0x0000FFFF) {
1872: rc = 3;
1873: out[0] = (cp >> 12 & 15) | 224;
1874: out[1] = (cp >> 6 & 63) | 128;
1875: out[2] = (cp & 63) | 128;
1876: } else if (cp <= 0x001FFFFF) {
1877: rc = 4;
1878: out[0] = (cp >> 18 & 7) | 240;
1879: out[1] = (cp >> 12 & 63) | 128;
1880: out[2] = (cp >> 6 & 63) | 128;
1881: out[3] = (cp & 63) | 128;
1882: } else if (cp <= 0x03FFFFFF) {
1883: rc = 5;
1884: out[0] = (cp >> 24 & 3) | 248;
1885: out[1] = (cp >> 18 & 63) | 128;
1886: out[2] = (cp >> 12 & 63) | 128;
1887: out[3] = (cp >> 6 & 63) | 128;
1888: out[4] = (cp & 63) | 128;
1889: } else if (cp <= 0x7FFFFFFF) {
1890: rc = 6;
1891: out[0] = (cp >> 30 & 1) | 252;
1892: out[1] = (cp >> 24 & 63) | 128;
1893: out[2] = (cp >> 18 & 63) | 128;
1894: out[3] = (cp >> 12 & 63) | 128;
1895: out[4] = (cp >> 6 & 63) | 128;
1896: out[5] = (cp & 63) | 128;
1897: } else
1.197 schwarze 1898: return 0;
1.1 kristaps 1899:
1.50 kristaps 1900: out[rc] = '\0';
1.197 schwarze 1901: return rc;
1.50 kristaps 1902: }
1.1 kristaps 1903:
1.50 kristaps 1904: /*
1.176 schwarze 1905: * If the string contains escape sequences,
1906: * replace it with an allocated rendering and return 1,
1907: * such that the caller can free it after use.
1908: * Otherwise, do nothing and return 0.
1.50 kristaps 1909: */
1.176 schwarze 1910: static int
1911: render_string(char **public, size_t *psz)
1.50 kristaps 1912: {
1.176 schwarze 1913: const char *src, *scp, *addcp, *seq;
1914: char *dst;
1915: size_t ssz, dsz, addsz;
1.114 schwarze 1916: char utfbuf[7], res[6];
1.176 schwarze 1917: int seqlen, unicode;
1.50 kristaps 1918:
1919: res[0] = '\\';
1920: res[1] = '\t';
1921: res[2] = ASCII_NBRSP;
1922: res[3] = ASCII_HYPH;
1.114 schwarze 1923: res[4] = ASCII_BREAK;
1924: res[5] = '\0';
1.1 kristaps 1925:
1.176 schwarze 1926: src = scp = *public;
1927: ssz = *psz;
1928: dst = NULL;
1929: dsz = 0;
1930:
1931: while (scp < src + *psz) {
1932:
1933: /* Leave normal characters unchanged. */
1934:
1935: if (strchr(res, *scp) == NULL) {
1936: if (dst != NULL)
1937: dst[dsz++] = *scp;
1938: scp++;
1939: continue;
1940: }
1.46 kristaps 1941:
1.50 kristaps 1942: /*
1.176 schwarze 1943: * Found something that requires replacing,
1944: * make sure we have a destination buffer.
1.50 kristaps 1945: */
1.176 schwarze 1946:
1947: if (dst == NULL) {
1948: dst = mandoc_malloc(ssz + 1);
1949: dsz = scp - src;
1950: memcpy(dst, src, dsz);
1.50 kristaps 1951: }
1.46 kristaps 1952:
1.176 schwarze 1953: /* Handle single-char special characters. */
1954:
1955: switch (*scp) {
1956: case '\\':
1957: break;
1.141 schwarze 1958: case '\t':
1959: case ASCII_NBRSP:
1.176 schwarze 1960: dst[dsz++] = ' ';
1961: scp++;
1962: continue;
1963: case ASCII_HYPH:
1964: dst[dsz++] = '-';
1.114 schwarze 1965: /* FALLTHROUGH */
1.141 schwarze 1966: case ASCII_BREAK:
1.176 schwarze 1967: scp++;
1.50 kristaps 1968: continue;
1.114 schwarze 1969: default:
1.176 schwarze 1970: abort();
1.114 schwarze 1971: }
1.46 kristaps 1972:
1.50 kristaps 1973: /*
1.176 schwarze 1974: * Found an escape sequence.
1975: * Read past the slash, then parse it.
1976: * Ignore everything except characters.
1.50 kristaps 1977: */
1.95 schwarze 1978:
1.176 schwarze 1979: scp++;
1980: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1981: continue;
1.1 kristaps 1982:
1.50 kristaps 1983: /*
1.95 schwarze 1984: * Render the special character
1985: * as either UTF-8 or ASCII.
1.50 kristaps 1986: */
1.95 schwarze 1987:
1988: if (write_utf8) {
1.203 schwarze 1989: unicode = mchars_spec2cp(seq, seqlen);
1.176 schwarze 1990: if (unicode <= 0)
1.95 schwarze 1991: continue;
1.176 schwarze 1992: addsz = utf8(unicode, utfbuf);
1993: if (addsz == 0)
1.95 schwarze 1994: continue;
1.176 schwarze 1995: addcp = utfbuf;
1.95 schwarze 1996: } else {
1.203 schwarze 1997: addcp = mchars_spec2str(seq, seqlen, &addsz);
1.176 schwarze 1998: if (addcp == NULL)
1.95 schwarze 1999: continue;
1.176 schwarze 2000: if (*addcp == ASCII_NBRSP) {
2001: addcp = " ";
2002: addsz = 1;
1.95 schwarze 2003: }
2004: }
1.1 kristaps 2005:
1.50 kristaps 2006: /* Copy the rendered glyph into the stream. */
1.1 kristaps 2007:
1.176 schwarze 2008: ssz += addsz;
2009: dst = mandoc_realloc(dst, ssz + 1);
2010: memcpy(dst + dsz, addcp, addsz);
2011: dsz += addsz;
2012: }
2013: if (dst != NULL) {
2014: *public = dst;
2015: *psz = dsz;
2016: }
2017:
2018: /* Trim trailing whitespace and NUL-terminate. */
2019:
2020: while (*psz > 0 && (*public)[*psz - 1] == ' ')
2021: --*psz;
2022: if (dst != NULL) {
2023: (*public)[*psz] = '\0';
1.197 schwarze 2024: return 1;
1.176 schwarze 2025: } else
1.197 schwarze 2026: return 0;
1.1 kristaps 2027: }
2028:
1.118 schwarze 2029: static void
2030: dbadd_mlink(const struct mlink *mlink)
2031: {
2032: size_t i;
2033:
2034: i = 1;
2035: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2036: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2037: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 2038: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 2039: SQL_STEP(stmts[STMT_INSERT_LINK]);
2040: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.169 schwarze 2041: }
2042:
2043: static void
2044: dbadd_mlink_name(const struct mlink *mlink)
2045: {
1.175 schwarze 2046: uint64_t bits;
1.169 schwarze 2047: size_t i;
2048:
2049: dbadd_mlink(mlink);
1.160 schwarze 2050:
2051: i = 1;
1.175 schwarze 2052: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2053: bits = NAME_FILE & NAME_MASK;
2054: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2055: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2056: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2057: }
2058:
2059: i = 1;
2060: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.160 schwarze 2061: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2062: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2063: SQL_STEP(stmts[STMT_INSERT_NAME]);
2064: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.118 schwarze 2065: }
2066:
1.14 schwarze 2067: /*
1.50 kristaps 2068: * Flush the current page's terms (and their bits) into the database.
2069: * Wrap the entire set of additions in a transaction to make sqlite be a
2070: * little faster.
1.96 schwarze 2071: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2072: */
2073: static void
1.176 schwarze 2074: dbadd(struct mpage *mpage)
1.14 schwarze 2075: {
1.87 schwarze 2076: struct mlink *mlink;
1.50 kristaps 2077: struct str *key;
1.176 schwarze 2078: char *cp;
1.52 kristaps 2079: size_t i;
1.66 schwarze 2080: unsigned int slot;
1.176 schwarze 2081: int mustfree;
1.14 schwarze 2082:
1.128 schwarze 2083: mlink = mpage->mlinks;
1.43 kristaps 2084:
1.128 schwarze 2085: if (nodb) {
1.147 schwarze 2086: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2087: key = ohash_next(&names, &slot))
1.147 schwarze 2088: free(key);
2089: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2090: key = ohash_next(&strings, &slot))
1.147 schwarze 2091: free(key);
1.145 schwarze 2092: if (0 == debug)
2093: return;
1.128 schwarze 2094: while (NULL != mlink) {
2095: fputs(mlink->name, stdout);
2096: if (NULL == mlink->next ||
2097: strcmp(mlink->dsec, mlink->next->dsec) ||
2098: strcmp(mlink->fsec, mlink->next->fsec) ||
2099: strcmp(mlink->arch, mlink->next->arch)) {
2100: putchar('(');
2101: if ('\0' == *mlink->dsec)
2102: fputs(mlink->fsec, stdout);
2103: else
2104: fputs(mlink->dsec, stdout);
2105: if ('\0' != *mlink->arch)
2106: printf("/%s", mlink->arch);
2107: putchar(')');
2108: }
2109: mlink = mlink->next;
2110: if (NULL != mlink)
2111: fputs(", ", stdout);
2112: }
1.132 schwarze 2113: printf(" - %s\n", mpage->desc);
1.14 schwarze 2114: return;
1.128 schwarze 2115: }
2116:
2117: if (debug)
2118: say(mlink->file, "Adding to database");
1.28 kristaps 2119:
1.176 schwarze 2120: cp = mpage->desc;
2121: i = strlen(cp);
2122: mustfree = render_string(&cp, &i);
1.52 kristaps 2123: i = 1;
1.176 schwarze 2124: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.161 schwarze 2125: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.81 schwarze 2126: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2127: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2128: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.176 schwarze 2129: if (mustfree)
2130: free(cp);
1.81 schwarze 2131:
1.128 schwarze 2132: while (NULL != mlink) {
1.118 schwarze 2133: dbadd_mlink(mlink);
1.128 schwarze 2134: mlink = mlink->next;
2135: }
1.134 schwarze 2136: mlink = mpage->mlinks;
1.50 kristaps 2137:
1.133 schwarze 2138: for (key = ohash_first(&names, &slot); NULL != key;
2139: key = ohash_next(&names, &slot)) {
2140: assert(key->mpage == mpage);
2141: i = 1;
2142: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.176 schwarze 2143: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.137 schwarze 2144: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2145: SQL_STEP(stmts[STMT_INSERT_NAME]);
2146: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2147: free(key);
2148: }
1.66 schwarze 2149: for (key = ohash_first(&strings, &slot); NULL != key;
2150: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2151: assert(key->mpage == mpage);
1.52 kristaps 2152: i = 1;
2153: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.176 schwarze 2154: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.137 schwarze 2155: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2156: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2157: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 2158: free(key);
1.38 schwarze 2159: }
1.14 schwarze 2160: }
2161:
1.5 kristaps 2162: static void
1.59 schwarze 2163: dbprune(void)
1.5 kristaps 2164: {
1.78 schwarze 2165: struct mpage *mpage;
1.82 schwarze 2166: struct mlink *mlink;
1.52 kristaps 2167: size_t i;
1.80 schwarze 2168: unsigned int slot;
1.5 kristaps 2169:
1.106 schwarze 2170: if (0 == nodb)
2171: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2172:
1.106 schwarze 2173: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2174: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2175: mlink = mpage->mlinks;
1.125 schwarze 2176: if (debug)
1.106 schwarze 2177: say(mlink->file, "Deleting from database");
2178: if (nodb)
2179: continue;
2180: for ( ; NULL != mlink; mlink = mlink->next) {
2181: i = 1;
2182: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2183: i, mlink->dsec);
2184: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2185: i, mlink->arch);
2186: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2187: i, mlink->name);
2188: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2189: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2190: }
1.5 kristaps 2191: }
1.106 schwarze 2192:
2193: if (0 == nodb)
2194: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2195: }
2196:
1.4 kristaps 2197: /*
1.50 kristaps 2198: * Close an existing database and its prepared statements.
2199: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2200: */
1.35 kristaps 2201: static void
1.59 schwarze 2202: dbclose(int real)
1.4 kristaps 2203: {
1.50 kristaps 2204: size_t i;
1.115 schwarze 2205: int status;
2206: pid_t child;
1.4 kristaps 2207:
1.50 kristaps 2208: if (nodb)
1.38 schwarze 2209: return;
1.50 kristaps 2210:
2211: for (i = 0; i < STMT__MAX; i++) {
2212: sqlite3_finalize(stmts[i]);
2213: stmts[i] = NULL;
1.4 kristaps 2214: }
2215:
1.50 kristaps 2216: sqlite3_close(db);
2217: db = NULL;
1.12 schwarze 2218:
1.50 kristaps 2219: if (real)
2220: return;
1.12 schwarze 2221:
1.115 schwarze 2222: if ('\0' == *tempfilename) {
2223: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2224: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2225: say(MANDOC_DB, "&rename");
1.115 schwarze 2226: }
2227: return;
2228: }
2229:
2230: switch (child = fork()) {
1.141 schwarze 2231: case -1:
1.115 schwarze 2232: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2233: say("", "&fork cmp");
1.115 schwarze 2234: return;
1.141 schwarze 2235: case 0:
1.115 schwarze 2236: execlp("cmp", "cmp", "-s",
1.196 schwarze 2237: tempfilename, MANDOC_DB, (char *)NULL);
1.123 schwarze 2238: say("", "&exec cmp");
1.115 schwarze 2239: exit(0);
2240: default:
2241: break;
2242: }
2243: if (-1 == waitpid(child, &status, 0)) {
2244: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2245: say("", "&wait cmp");
1.115 schwarze 2246: } else if (WIFSIGNALED(status)) {
2247: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2248: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2249: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2250: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2251: say(MANDOC_DB,
2252: "Data changed, but cannot replace database");
2253: }
2254:
2255: *strrchr(tempfilename, '/') = '\0';
2256: switch (child = fork()) {
1.141 schwarze 2257: case -1:
1.115 schwarze 2258: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2259: say("", "&fork rm");
1.115 schwarze 2260: return;
1.141 schwarze 2261: case 0:
1.196 schwarze 2262: execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
1.123 schwarze 2263: say("", "&exec rm");
1.115 schwarze 2264: exit((int)MANDOCLEVEL_SYSERR);
2265: default:
2266: break;
2267: }
2268: if (-1 == waitpid(child, &status, 0)) {
2269: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2270: say("", "&wait rm");
1.115 schwarze 2271: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2272: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2273: say("", "%s: Cannot remove temporary directory",
2274: tempfilename);
1.59 schwarze 2275: }
1.50 kristaps 2276: }
1.14 schwarze 2277:
1.50 kristaps 2278: /*
2279: * This is straightforward stuff.
2280: * Open a database connection to a "temporary" database, then open a set
2281: * of prepared statements we'll use over and over again.
2282: * If "real" is set, we use the existing database; if not, we truncate a
2283: * temporary one.
2284: * Must be matched by dbclose().
2285: */
2286: static int
1.59 schwarze 2287: dbopen(int real)
1.50 kristaps 2288: {
1.115 schwarze 2289: const char *sql;
1.50 kristaps 2290: int rc, ofl;
1.12 schwarze 2291:
1.141 schwarze 2292: if (nodb)
1.197 schwarze 2293: return 1;
1.12 schwarze 2294:
1.115 schwarze 2295: *tempfilename = '\0';
1.63 schwarze 2296: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2297:
2298: if (real) {
2299: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2300: if (SQLITE_OK != rc) {
1.63 schwarze 2301: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2302: if (SQLITE_CANTOPEN != rc)
2303: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.197 schwarze 2304: return 0;
1.63 schwarze 2305: }
1.115 schwarze 2306: goto prepare_statements;
2307: }
2308:
2309: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2310:
1.115 schwarze 2311: remove(MANDOC_DB "~");
2312: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2313: if (SQLITE_OK == rc)
1.115 schwarze 2314: goto create_tables;
1.116 schwarze 2315: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2316: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2317: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.197 schwarze 2318: return 0;
1.50 kristaps 2319: }
1.12 schwarze 2320:
1.143 schwarze 2321: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2322: sizeof(tempfilename));
1.115 schwarze 2323: if (NULL == mkdtemp(tempfilename)) {
2324: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2325: say("", "&%s", tempfilename);
1.197 schwarze 2326: return 0;
1.115 schwarze 2327: }
1.143 schwarze 2328: (void)strlcat(tempfilename, "/" MANDOC_DB,
2329: sizeof(tempfilename));
1.115 schwarze 2330: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2331: if (SQLITE_OK != rc) {
1.59 schwarze 2332: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2333: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.197 schwarze 2334: return 0;
1.50 kristaps 2335: }
1.12 schwarze 2336:
1.115 schwarze 2337: create_tables:
1.81 schwarze 2338: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2339: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2340: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2341: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2342: ");\n"
2343: "\n"
1.81 schwarze 2344: "CREATE TABLE \"mlinks\" (\n"
2345: " \"sec\" TEXT NOT NULL,\n"
2346: " \"arch\" TEXT NOT NULL,\n"
2347: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2348: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2349: "ON DELETE CASCADE\n"
1.81 schwarze 2350: ");\n"
1.136 schwarze 2351: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2352: "\n"
1.133 schwarze 2353: "CREATE TABLE \"names\" (\n"
2354: " \"bits\" INTEGER NOT NULL,\n"
2355: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2356: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.175 schwarze 2357: "ON DELETE CASCADE,\n"
2358: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.133 schwarze 2359: ");\n"
2360: "\n"
1.50 kristaps 2361: "CREATE TABLE \"keys\" (\n"
2362: " \"bits\" INTEGER NOT NULL,\n"
2363: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2364: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2365: "ON DELETE CASCADE\n"
1.136 schwarze 2366: ");\n"
2367: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2368:
1.50 kristaps 2369: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2370: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2371: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2372: sqlite3_close(db);
1.197 schwarze 2373: return 0;
1.50 kristaps 2374: }
1.4 kristaps 2375:
1.57 schwarze 2376: prepare_statements:
1.146 schwarze 2377: if (SQLITE_OK != sqlite3_exec(db,
2378: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2379: exitcode = (int)MANDOCLEVEL_SYSERR;
2380: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2381: sqlite3_errmsg(db));
2382: sqlite3_close(db);
1.197 schwarze 2383: return 0;
1.146 schwarze 2384: }
2385:
1.137 schwarze 2386: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2387: "(SELECT pageid FROM mlinks WHERE "
2388: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2389: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2390: sql = "INSERT INTO mpages "
1.132 schwarze 2391: "(desc,form) VALUES (?,?)";
1.81 schwarze 2392: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2393: sql = "INSERT INTO mlinks "
1.104 schwarze 2394: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2395: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.175 schwarze 2396: sql = "SELECT bits FROM names where pageid = ?";
2397: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.133 schwarze 2398: sql = "INSERT INTO names "
2399: "(bits,name,pageid) VALUES (?,?,?)";
2400: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2401: sql = "INSERT INTO keys "
1.81 schwarze 2402: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2403: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2404:
2405: #ifndef __APPLE__
2406: /*
2407: * When opening a new database, we can turn off
2408: * synchronous mode for much better performance.
2409: */
2410:
1.146 schwarze 2411: if (real && SQLITE_OK != sqlite3_exec(db,
2412: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2413: exitcode = (int)MANDOCLEVEL_SYSERR;
2414: say(MANDOC_DB, "PRAGMA synchronous: %s",
1.162 schwarze 2415: sqlite3_errmsg(db));
1.146 schwarze 2416: sqlite3_close(db);
1.197 schwarze 2417: return 0;
1.146 schwarze 2418: }
1.70 schwarze 2419: #endif
2420:
1.197 schwarze 2421: return 1;
1.4 kristaps 2422: }
2423:
1.50 kristaps 2424: static int
1.165 schwarze 2425: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2426: {
1.59 schwarze 2427: static char startdir[PATH_MAX];
1.151 schwarze 2428: static int getcwd_status; /* 1 = ok, 2 = failure */
2429: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2430: char *cp;
1.4 kristaps 2431:
1.59 schwarze 2432: /*
1.151 schwarze 2433: * Remember the original working directory, if possible.
2434: * This will be needed if the second or a later directory
2435: * on the command line is given as a relative path.
2436: * Do not error out if the current directory is not
2437: * searchable: Maybe it won't be needed after all.
2438: */
2439: if (0 == getcwd_status) {
2440: if (NULL == getcwd(startdir, sizeof(startdir))) {
2441: getcwd_status = 2;
2442: (void)strlcpy(startdir, strerror(errno),
2443: sizeof(startdir));
2444: } else
2445: getcwd_status = 1;
2446: }
2447:
2448: /*
2449: * We are leaving the old base directory.
2450: * Do not use it any longer, not even for messages.
2451: */
2452: *basedir = '\0';
2453:
2454: /*
2455: * If and only if the directory was changed earlier and
2456: * the next directory to process is given as a relative path,
2457: * first go back, or bail out if that is impossible.
1.59 schwarze 2458: */
1.151 schwarze 2459: if (chdir_status && '/' != *targetdir) {
2460: if (2 == getcwd_status) {
1.59 schwarze 2461: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2462: say("", "getcwd: %s", startdir);
1.197 schwarze 2463: return 0;
1.59 schwarze 2464: }
1.151 schwarze 2465: if (-1 == chdir(startdir)) {
1.59 schwarze 2466: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2467: say("", "&chdir %s", startdir);
1.197 schwarze 2468: return 0;
1.59 schwarze 2469: }
2470: }
1.151 schwarze 2471:
2472: /*
2473: * Always resolve basedir to the canonicalized absolute
2474: * pathname and append a trailing slash, such that
2475: * we can reliably check whether files are inside.
2476: */
1.59 schwarze 2477: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2478: if (report_baddir || errno != ENOENT) {
2479: exitcode = (int)MANDOCLEVEL_BADARG;
2480: say("", "&%s: realpath", targetdir);
2481: }
1.197 schwarze 2482: return 0;
1.59 schwarze 2483: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2484: if (report_baddir || errno != ENOENT) {
2485: exitcode = (int)MANDOCLEVEL_BADARG;
2486: say("", "&chdir");
2487: }
1.197 schwarze 2488: return 0;
1.150 schwarze 2489: }
1.151 schwarze 2490: chdir_status = 1;
1.150 schwarze 2491: cp = strchr(basedir, '\0');
2492: if ('/' != cp[-1]) {
2493: if (cp - basedir >= PATH_MAX - 1) {
2494: exitcode = (int)MANDOCLEVEL_SYSERR;
2495: say("", "Filename too long");
1.197 schwarze 2496: return 0;
1.150 schwarze 2497: }
2498: *cp++ = '/';
2499: *cp = '\0';
1.4 kristaps 2500: }
1.197 schwarze 2501: return 1;
1.56 schwarze 2502: }
2503:
2504: static void
1.59 schwarze 2505: say(const char *file, const char *format, ...)
1.56 schwarze 2506: {
2507: va_list ap;
1.123 schwarze 2508: int use_errno;
1.56 schwarze 2509:
1.59 schwarze 2510: if ('\0' != *basedir)
2511: fprintf(stderr, "%s", basedir);
2512: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2513: fputc('/', stderr);
1.56 schwarze 2514: if ('\0' != *file)
1.59 schwarze 2515: fprintf(stderr, "%s", file);
2516:
1.123 schwarze 2517: use_errno = 1;
2518: if (NULL != format) {
2519: switch (*format) {
1.141 schwarze 2520: case '&':
1.123 schwarze 2521: format++;
2522: break;
1.141 schwarze 2523: case '\0':
1.123 schwarze 2524: format = NULL;
2525: break;
2526: default:
2527: use_errno = 0;
2528: break;
2529: }
2530: }
2531: if (NULL != format) {
2532: if ('\0' != *basedir || '\0' != *file)
2533: fputs(": ", stderr);
2534: va_start(ap, format);
2535: vfprintf(stderr, format, ap);
2536: va_end(ap);
2537: }
2538: if (use_errno) {
2539: if ('\0' != *basedir || '\0' != *file || NULL != format)
2540: fputs(": ", stderr);
1.59 schwarze 2541: perror(NULL);
1.123 schwarze 2542: } else
2543: fputc('\n', stderr);
1.1 kristaps 2544: }
CVSweb