Annotation of mandoc/mandocdb.c, Revision 1.207
1.207 ! schwarze 1: /* $Id: mandocdb.c,v 1.206 2015/11/06 16:30:33 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.182 schwarze 4: * Copyright (c) 2011-2015 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>
1.198 schwarze 26: #include <err.h>
1.63 schwarze 27: #include <errno.h>
1.1 kristaps 28: #include <fcntl.h>
1.158 schwarze 29: #if HAVE_FTS
1.50 kristaps 30: #include <fts.h>
1.157 schwarze 31: #else
32: #include "compat_fts.h"
33: #endif
1.1 kristaps 34: #include <getopt.h>
1.58 schwarze 35: #include <limits.h>
1.50 kristaps 36: #include <stddef.h>
1.59 schwarze 37: #include <stdio.h>
1.1 kristaps 38: #include <stdint.h>
39: #include <stdlib.h>
40: #include <string.h>
1.17 schwarze 41: #include <unistd.h>
1.1 kristaps 42:
1.50 kristaps 43: #include <sqlite3.h>
1.1 kristaps 44:
1.188 schwarze 45: #include "mandoc_aux.h"
1.202 schwarze 46: #include "mandoc_ohash.h"
1.188 schwarze 47: #include "mandoc.h"
48: #include "roff.h"
1.50 kristaps 49: #include "mdoc.h"
1.1 kristaps 50: #include "man.h"
1.187 schwarze 51: #include "manconf.h"
1.55 kristaps 52: #include "mansearch.h"
1.1 kristaps 53:
1.111 schwarze 54: extern int mansearch_keymax;
55: extern const char *const mansearch_keynames[];
56:
1.52 kristaps 57: #define SQL_EXEC(_v) \
58: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.134 schwarze 59: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.52 kristaps 60: #define SQL_BIND_TEXT(_s, _i, _v) \
61: if (SQLITE_OK != sqlite3_bind_text \
62: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.134 schwarze 63: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 64: #define SQL_BIND_INT(_s, _i, _v) \
65: if (SQLITE_OK != sqlite3_bind_int \
66: ((_s), (_i)++, (_v))) \
1.134 schwarze 67: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 68: #define SQL_BIND_INT64(_s, _i, _v) \
69: if (SQLITE_OK != sqlite3_bind_int64 \
70: ((_s), (_i)++, (_v))) \
1.134 schwarze 71: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 72: #define SQL_STEP(_s) \
73: if (SQLITE_DONE != sqlite3_step((_s))) \
1.134 schwarze 74: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 75:
1.50 kristaps 76: enum op {
77: OP_DEFAULT = 0, /* new dbs from dir list or default config */
78: OP_CONFFILE, /* new databases from custom config file */
79: OP_UPDATE, /* delete/add entries in existing database */
80: OP_DELETE, /* delete entries from existing database */
81: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 82: };
83:
1.50 kristaps 84: struct str {
1.78 schwarze 85: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 86: uint64_t mask; /* bitmask in sequence */
1.176 schwarze 87: char key[]; /* rendered text */
1.38 schwarze 88: };
89:
1.79 schwarze 90: struct inodev {
91: ino_t st_ino;
92: dev_t st_dev;
1.50 kristaps 93: };
1.5 kristaps 94:
1.78 schwarze 95: struct mpage {
1.79 schwarze 96: struct inodev inodev; /* used for hashing routine */
1.137 schwarze 97: int64_t pageid; /* pageid in mpages SQL table */
1.82 schwarze 98: char *sec; /* section from file content */
99: char *arch; /* architecture from file content */
100: char *title; /* title from file content */
101: char *desc; /* description from file content */
102: struct mlink *mlinks; /* singly linked list */
1.161 schwarze 103: int form; /* format from file content */
1.174 schwarze 104: int name_head_done;
1.82 schwarze 105: };
106:
107: struct mlink {
1.58 schwarze 108: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 109: char *dsec; /* section from directory */
110: char *arch; /* architecture from directory */
111: char *name; /* name from file name (not empty) */
112: char *fsec; /* section from file name suffix */
1.85 schwarze 113: struct mlink *next; /* singly linked list */
1.118 schwarze 114: struct mpage *mpage; /* parent */
1.161 schwarze 115: int dform; /* format from directory */
116: int fform; /* format from file name suffix */
1.124 schwarze 117: int gzip; /* filename has a .gz suffix */
1.3 kristaps 118: };
119:
1.50 kristaps 120: enum stmt {
1.81 schwarze 121: STMT_DELETE_PAGE = 0, /* delete mpage */
122: STMT_INSERT_PAGE, /* insert mpage */
123: STMT_INSERT_LINK, /* insert mlink */
1.133 schwarze 124: STMT_INSERT_NAME, /* insert name */
1.175 schwarze 125: STMT_SELECT_NAME, /* retrieve existing name flags */
1.81 schwarze 126: STMT_INSERT_KEY, /* insert parsed key */
1.50 kristaps 127: STMT__MAX
1.1 kristaps 128: };
129:
1.190 schwarze 130: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.189 schwarze 131: const struct roff_node *);
1.1 kristaps 132:
1.50 kristaps 133: struct mdoc_handler {
134: mdoc_fp fp; /* optional handler */
135: uint64_t mask; /* set unless handler returns 0 */
1.1 kristaps 136: };
137:
1.59 schwarze 138: static void dbclose(int);
1.176 schwarze 139: static void dbadd(struct mpage *);
1.118 schwarze 140: static void dbadd_mlink(const struct mlink *mlink);
1.169 schwarze 141: static void dbadd_mlink_name(const struct mlink *mlink);
1.59 schwarze 142: static int dbopen(int);
143: static void dbprune(void);
144: static void filescan(const char *);
1.84 schwarze 145: static void mlink_add(struct mlink *, const struct stat *);
1.131 schwarze 146: static void mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 147: static void mlink_free(struct mlink *);
1.89 schwarze 148: static void mlinks_undupe(struct mpage *);
1.78 schwarze 149: static void mpages_free(void);
1.176 schwarze 150: static void mpages_merge(struct mparse *);
1.130 schwarze 151: static void names_check(void);
1.124 schwarze 152: static void parse_cat(struct mpage *, int);
1.190 schwarze 153: static void parse_man(struct mpage *, const struct roff_meta *,
1.189 schwarze 154: const struct roff_node *);
1.190 schwarze 155: static void parse_mdoc(struct mpage *, const struct roff_meta *,
1.189 schwarze 156: const struct roff_node *);
1.190 schwarze 157: static int parse_mdoc_body(struct mpage *, const struct roff_meta *,
1.189 schwarze 158: const struct roff_node *);
1.190 schwarze 159: static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
1.189 schwarze 160: const struct roff_node *);
1.190 schwarze 161: static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
1.189 schwarze 162: const struct roff_node *);
163: static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
1.190 schwarze 164: static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
1.189 schwarze 165: const struct roff_node *);
1.190 schwarze 166: static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
1.189 schwarze 167: const struct roff_node *);
1.190 schwarze 168: static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
1.189 schwarze 169: const struct roff_node *);
1.190 schwarze 170: static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
1.189 schwarze 171: const struct roff_node *);
1.190 schwarze 172: static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
1.189 schwarze 173: const struct roff_node *);
1.190 schwarze 174: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.189 schwarze 175: const struct roff_node *);
1.112 schwarze 176: static void putkey(const struct mpage *, char *, uint64_t);
1.176 schwarze 177: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.78 schwarze 178: static void putmdockey(const struct mpage *,
1.189 schwarze 179: const struct roff_node *, uint64_t);
1.176 schwarze 180: static int render_string(char **, size_t *);
1.59 schwarze 181: static void say(const char *, const char *, ...);
1.165 schwarze 182: static int set_basedir(const char *, int);
1.59 schwarze 183: static int treescan(void);
1.50 kristaps 184: static size_t utf8(unsigned int, char [7]);
185:
1.115 schwarze 186: static char tempfilename[32];
1.102 schwarze 187: static int nodb; /* no database changes */
1.116 schwarze 188: static int mparse_options; /* abort the parse early */
1.141 schwarze 189: static int use_all; /* use all found files */
190: static int debug; /* print what we're doing */
191: static int warnings; /* warn about crap */
1.95 schwarze 192: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 193: static int exitcode; /* to be returned by main */
1.141 schwarze 194: static enum op op; /* operational mode */
1.59 schwarze 195: static char basedir[PATH_MAX]; /* current base directory */
1.78 schwarze 196: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 197: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 198: static struct ohash names; /* table of all names */
1.50 kristaps 199: static struct ohash strings; /* table of all strings */
200: static sqlite3 *db = NULL; /* current database */
201: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.133 schwarze 202: static uint64_t name_mask;
1.25 schwarze 203:
204: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 205: { NULL, 0 }, /* Ap */
206: { NULL, 0 }, /* Dd */
207: { NULL, 0 }, /* Dt */
208: { NULL, 0 }, /* Os */
209: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
210: { parse_mdoc_head, TYPE_Ss }, /* Ss */
211: { NULL, 0 }, /* Pp */
212: { NULL, 0 }, /* D1 */
213: { NULL, 0 }, /* Dl */
214: { NULL, 0 }, /* Bd */
215: { NULL, 0 }, /* Ed */
216: { NULL, 0 }, /* Bl */
217: { NULL, 0 }, /* El */
218: { NULL, 0 }, /* It */
219: { NULL, 0 }, /* Ad */
220: { NULL, TYPE_An }, /* An */
221: { NULL, TYPE_Ar }, /* Ar */
222: { NULL, TYPE_Cd }, /* Cd */
223: { NULL, TYPE_Cm }, /* Cm */
224: { NULL, TYPE_Dv }, /* Dv */
225: { NULL, TYPE_Er }, /* Er */
226: { NULL, TYPE_Ev }, /* Ev */
227: { NULL, 0 }, /* Ex */
228: { NULL, TYPE_Fa }, /* Fa */
229: { parse_mdoc_Fd, 0 }, /* Fd */
230: { NULL, TYPE_Fl }, /* Fl */
231: { parse_mdoc_Fn, 0 }, /* Fn */
232: { NULL, TYPE_Ft }, /* Ft */
233: { NULL, TYPE_Ic }, /* Ic */
1.92 schwarze 234: { NULL, TYPE_In }, /* In */
1.68 schwarze 235: { NULL, TYPE_Li }, /* Li */
1.132 schwarze 236: { parse_mdoc_Nd, 0 }, /* Nd */
1.129 schwarze 237: { parse_mdoc_Nm, 0 }, /* Nm */
1.68 schwarze 238: { NULL, 0 }, /* Op */
239: { NULL, 0 }, /* Ot */
240: { NULL, TYPE_Pa }, /* Pa */
241: { NULL, 0 }, /* Rv */
1.92 schwarze 242: { NULL, TYPE_St }, /* St */
1.68 schwarze 243: { NULL, TYPE_Va }, /* Va */
244: { parse_mdoc_body, TYPE_Va }, /* Vt */
245: { parse_mdoc_Xr, 0 }, /* Xr */
246: { NULL, 0 }, /* %A */
247: { NULL, 0 }, /* %B */
248: { NULL, 0 }, /* %D */
249: { NULL, 0 }, /* %I */
250: { NULL, 0 }, /* %J */
251: { NULL, 0 }, /* %N */
252: { NULL, 0 }, /* %O */
253: { NULL, 0 }, /* %P */
254: { NULL, 0 }, /* %R */
255: { NULL, 0 }, /* %T */
256: { NULL, 0 }, /* %V */
257: { NULL, 0 }, /* Ac */
258: { NULL, 0 }, /* Ao */
259: { NULL, 0 }, /* Aq */
260: { NULL, TYPE_At }, /* At */
261: { NULL, 0 }, /* Bc */
262: { NULL, 0 }, /* Bf */
263: { NULL, 0 }, /* Bo */
264: { NULL, 0 }, /* Bq */
265: { NULL, TYPE_Bsx }, /* Bsx */
266: { NULL, TYPE_Bx }, /* Bx */
267: { NULL, 0 }, /* Db */
268: { NULL, 0 }, /* Dc */
269: { NULL, 0 }, /* Do */
270: { NULL, 0 }, /* Dq */
271: { NULL, 0 }, /* Ec */
272: { NULL, 0 }, /* Ef */
273: { NULL, TYPE_Em }, /* Em */
274: { NULL, 0 }, /* Eo */
275: { NULL, TYPE_Fx }, /* Fx */
276: { NULL, TYPE_Ms }, /* Ms */
277: { NULL, 0 }, /* No */
278: { NULL, 0 }, /* Ns */
279: { NULL, TYPE_Nx }, /* Nx */
280: { NULL, TYPE_Ox }, /* Ox */
281: { NULL, 0 }, /* Pc */
282: { NULL, 0 }, /* Pf */
283: { NULL, 0 }, /* Po */
284: { NULL, 0 }, /* Pq */
285: { NULL, 0 }, /* Qc */
286: { NULL, 0 }, /* Ql */
287: { NULL, 0 }, /* Qo */
288: { NULL, 0 }, /* Qq */
289: { NULL, 0 }, /* Re */
290: { NULL, 0 }, /* Rs */
291: { NULL, 0 }, /* Sc */
292: { NULL, 0 }, /* So */
293: { NULL, 0 }, /* Sq */
294: { NULL, 0 }, /* Sm */
295: { NULL, 0 }, /* Sx */
296: { NULL, TYPE_Sy }, /* Sy */
297: { NULL, TYPE_Tn }, /* Tn */
298: { NULL, 0 }, /* Ux */
299: { NULL, 0 }, /* Xc */
300: { NULL, 0 }, /* Xo */
1.173 schwarze 301: { parse_mdoc_Fo, 0 }, /* Fo */
1.68 schwarze 302: { NULL, 0 }, /* Fc */
303: { NULL, 0 }, /* Oo */
304: { NULL, 0 }, /* Oc */
305: { NULL, 0 }, /* Bk */
306: { NULL, 0 }, /* Ek */
307: { NULL, 0 }, /* Bt */
308: { NULL, 0 }, /* Hf */
309: { NULL, 0 }, /* Fr */
310: { NULL, 0 }, /* Ud */
311: { NULL, TYPE_Lb }, /* Lb */
312: { NULL, 0 }, /* Lp */
313: { NULL, TYPE_Lk }, /* Lk */
314: { NULL, TYPE_Mt }, /* Mt */
315: { NULL, 0 }, /* Brq */
316: { NULL, 0 }, /* Bro */
317: { NULL, 0 }, /* Brc */
318: { NULL, 0 }, /* %C */
319: { NULL, 0 }, /* Es */
320: { NULL, 0 }, /* En */
321: { NULL, TYPE_Dx }, /* Dx */
322: { NULL, 0 }, /* %Q */
323: { NULL, 0 }, /* br */
324: { NULL, 0 }, /* sp */
325: { NULL, 0 }, /* %U */
326: { NULL, 0 }, /* Ta */
1.166 schwarze 327: { NULL, 0 }, /* ll */
1.1 kristaps 328: };
329:
1.141 schwarze 330:
1.1 kristaps 331: int
1.179 schwarze 332: mandocdb(int argc, char *argv[])
1.1 kristaps 333: {
1.187 schwarze 334: struct manconf conf;
335: struct mparse *mp;
1.206 schwarze 336: const char *path_arg, *progname;
1.187 schwarze 337: size_t j, sz;
1.59 schwarze 338: int ch, i;
1.50 kristaps 339:
1.207 ! schwarze 340: #if HAVE_PLEDGE
! 341: if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
! 342: perror("pledge");
! 343: return (int)MANDOCLEVEL_SYSERR;
! 344: }
! 345: #endif
! 346:
1.187 schwarze 347: memset(&conf, 0, sizeof(conf));
1.50 kristaps 348: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
349:
350: /*
1.141 schwarze 351: * We accept a few different invocations.
1.50 kristaps 352: * The CHECKOP macro makes sure that invocation styles don't
353: * clobber each other.
354: */
355: #define CHECKOP(_op, _ch) do \
356: if (OP_DEFAULT != (_op)) { \
1.198 schwarze 357: warnx("-%c: Conflicting option", (_ch)); \
1.50 kristaps 358: goto usage; \
359: } while (/*CONSTCOND*/0)
1.10 kristaps 360:
1.59 schwarze 361: path_arg = NULL;
1.38 schwarze 362: op = OP_DEFAULT;
1.1 kristaps 363:
1.126 schwarze 364: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 365: switch (ch) {
1.141 schwarze 366: case 'a':
1.12 schwarze 367: use_all = 1;
368: break;
1.141 schwarze 369: case 'C':
1.50 kristaps 370: CHECKOP(op, ch);
1.59 schwarze 371: path_arg = optarg;
1.38 schwarze 372: op = OP_CONFFILE;
1.34 schwarze 373: break;
1.141 schwarze 374: case 'D':
1.125 schwarze 375: debug++;
376: break;
1.141 schwarze 377: case 'd':
1.50 kristaps 378: CHECKOP(op, ch);
1.59 schwarze 379: path_arg = optarg;
1.5 kristaps 380: op = OP_UPDATE;
381: break;
1.141 schwarze 382: case 'n':
1.50 kristaps 383: nodb = 1;
384: break;
1.141 schwarze 385: case 'p':
1.126 schwarze 386: warnings = 1;
387: break;
1.141 schwarze 388: case 'Q':
1.116 schwarze 389: mparse_options |= MPARSE_QUICK;
1.102 schwarze 390: break;
1.141 schwarze 391: case 'T':
1.95 schwarze 392: if (strcmp(optarg, "utf8")) {
1.198 schwarze 393: warnx("-T%s: Unsupported output format",
394: optarg);
1.95 schwarze 395: goto usage;
396: }
397: write_utf8 = 1;
398: break;
1.141 schwarze 399: case 't':
1.50 kristaps 400: CHECKOP(op, ch);
1.38 schwarze 401: dup2(STDOUT_FILENO, STDERR_FILENO);
402: op = OP_TEST;
1.50 kristaps 403: nodb = warnings = 1;
1.38 schwarze 404: break;
1.141 schwarze 405: case 'u':
1.50 kristaps 406: CHECKOP(op, ch);
1.59 schwarze 407: path_arg = optarg;
1.5 kristaps 408: op = OP_DELETE;
409: break;
1.141 schwarze 410: case 'v':
1.126 schwarze 411: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 412: break;
1.1 kristaps 413: default:
1.38 schwarze 414: goto usage;
1.1 kristaps 415: }
416:
417: argc -= optind;
418: argv += optind;
419:
1.207 ! schwarze 420: #if HAVE_PLEDGE
! 421: if (nodb && pledge("stdio rpath", NULL) == -1) {
! 422: perror("pledge");
! 423: return (int)MANDOCLEVEL_SYSERR;
! 424: }
! 425: #endif
! 426:
1.38 schwarze 427: if (OP_CONFFILE == op && argc > 0) {
1.198 schwarze 428: warnx("-C: Too many arguments");
1.38 schwarze 429: goto usage;
430: }
431:
1.59 schwarze 432: exitcode = (int)MANDOCLEVEL_OK;
1.203 schwarze 433: mchars_alloc();
434: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
1.202 schwarze 435: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
436: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.50 kristaps 437:
438: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 439:
1.50 kristaps 440: /*
1.150 schwarze 441: * Most of these deal with a specific directory.
1.140 schwarze 442: * Jump into that directory first.
1.50 kristaps 443: */
1.165 schwarze 444: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.50 kristaps 445: goto out;
1.140 schwarze 446:
1.138 schwarze 447: if (dbopen(1)) {
1.140 schwarze 448: /*
449: * The existing database is usable. Process
450: * all files specified on the command-line.
451: */
1.207 ! schwarze 452: #if HAVE_PLEDGE
! 453: if (!nodb && pledge("stdio rpath wpath cpath fattr flock",
! 454: NULL) == -1) {
! 455: perror("pledge");
! 456: exitcode = (int)MANDOCLEVEL_SYSERR;
! 457: goto out;
! 458: }
! 459: #endif
1.140 schwarze 460: use_all = 1;
1.138 schwarze 461: for (i = 0; i < argc; i++)
462: filescan(argv[i]);
463: if (OP_TEST != op)
464: dbprune();
465: } else {
466: /*
467: * Database missing or corrupt.
468: * Recreate from scratch.
469: */
1.140 schwarze 470: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 471: op = OP_DEFAULT;
472: if (0 == treescan())
473: goto out;
474: if (0 == dbopen(0))
475: goto out;
476: }
1.50 kristaps 477: if (OP_DELETE != op)
1.176 schwarze 478: mpages_merge(mp);
1.138 schwarze 479: dbclose(OP_DEFAULT == op ? 0 : 1);
1.50 kristaps 480: } else {
481: /*
482: * If we have arguments, use them as our manpaths.
483: * If we don't, grok from manpath(1) or however else
1.187 schwarze 484: * manconf_parse() wants to do it.
1.50 kristaps 485: */
486: if (argc > 0) {
1.187 schwarze 487: conf.manpath.paths = mandoc_reallocarray(NULL,
1.144 schwarze 488: argc, sizeof(char *));
1.187 schwarze 489: conf.manpath.sz = (size_t)argc;
1.50 kristaps 490: for (i = 0; i < argc; i++)
1.187 schwarze 491: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.50 kristaps 492: } else
1.187 schwarze 493: manconf_parse(&conf, path_arg, NULL, NULL);
1.127 schwarze 494:
1.187 schwarze 495: if (conf.manpath.sz == 0) {
1.127 schwarze 496: exitcode = (int)MANDOCLEVEL_BADARG;
497: say("", "Empty manpath");
498: }
1.50 kristaps 499:
500: /*
1.69 schwarze 501: * First scan the tree rooted at a base directory, then
502: * build a new database and finally move it into place.
1.50 kristaps 503: * Ignore zero-length directories and strip trailing
504: * slashes.
505: */
1.187 schwarze 506: for (j = 0; j < conf.manpath.sz; j++) {
507: sz = strlen(conf.manpath.paths[j]);
508: if (sz && conf.manpath.paths[j][sz - 1] == '/')
509: conf.manpath.paths[j][--sz] = '\0';
1.50 kristaps 510: if (0 == sz)
511: continue;
1.66 schwarze 512:
513: if (j) {
1.202 schwarze 514: mandoc_ohash_init(&mpages, 6,
515: offsetof(struct mpage, inodev));
516: mandoc_ohash_init(&mlinks, 6,
517: offsetof(struct mlink, file));
1.66 schwarze 518: }
519:
1.187 schwarze 520: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.165 schwarze 521: continue;
1.59 schwarze 522: if (0 == treescan())
1.165 schwarze 523: continue;
1.59 schwarze 524: if (0 == dbopen(0))
1.165 schwarze 525: continue;
1.52 kristaps 526:
1.176 schwarze 527: mpages_merge(mp);
1.147 schwarze 528: if (warnings && !nodb &&
1.130 schwarze 529: ! (MPARSE_QUICK & mparse_options))
530: names_check();
1.59 schwarze 531: dbclose(0);
1.66 schwarze 532:
1.187 schwarze 533: if (j + 1 < conf.manpath.sz) {
1.80 schwarze 534: mpages_free();
1.78 schwarze 535: ohash_delete(&mpages);
1.83 schwarze 536: ohash_delete(&mlinks);
1.66 schwarze 537: }
1.50 kristaps 538: }
539: }
540: out:
1.187 schwarze 541: manconf_free(&conf);
1.168 schwarze 542: mparse_free(mp);
1.203 schwarze 543: mchars_free();
1.80 schwarze 544: mpages_free();
1.78 schwarze 545: ohash_delete(&mpages);
1.83 schwarze 546: ohash_delete(&mlinks);
1.197 schwarze 547: return exitcode;
1.50 kristaps 548: usage:
1.206 schwarze 549: progname = getprogname();
1.126 schwarze 550: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
551: " %s [-aDnpQ] [-Tutf8] dir ...\n"
552: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
553: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 554: " %s [-Q] -t file ...\n",
1.206 schwarze 555: progname, progname, progname, progname, progname);
1.50 kristaps 556:
1.197 schwarze 557: return (int)MANDOCLEVEL_BADARG;
1.50 kristaps 558: }
559:
560: /*
1.59 schwarze 561: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 562: * We use fts(), scanning directory parts along the way for clues to our
563: * section and architecture.
564: *
565: * If use_all has been specified, grok all files.
566: * If not, sanitise paths to the following:
567: *
1.141 schwarze 568: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 569: * or
570: * [./]cat<section>[/<arch>]/<name>.0
571: *
572: * TODO: accomodate for multi-language directories.
573: */
574: static int
1.59 schwarze 575: treescan(void)
1.50 kristaps 576: {
1.139 schwarze 577: char buf[PATH_MAX];
1.50 kristaps 578: FTS *f;
579: FTSENT *ff;
1.84 schwarze 580: struct mlink *mlink;
1.124 schwarze 581: int dform, gzip;
1.94 schwarze 582: char *dsec, *arch, *fsec, *cp;
583: const char *path;
1.50 kristaps 584: const char *argv[2];
585:
586: argv[0] = ".";
587: argv[1] = (char *)NULL;
588:
1.139 schwarze 589: f = fts_open((char * const *)argv,
590: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.200 schwarze 591: if (f == NULL) {
1.59 schwarze 592: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 593: say("", "&fts_open");
1.197 schwarze 594: return 0;
1.50 kristaps 595: }
596:
597: dsec = arch = NULL;
598: dform = FORM_NONE;
599:
1.200 schwarze 600: while ((ff = fts_read(f)) != NULL) {
1.50 kristaps 601: path = ff->fts_path + 2;
1.139 schwarze 602: switch (ff->fts_info) {
603:
604: /*
605: * Symbolic links require various sanity checks,
606: * then get handled just like regular files.
607: */
1.141 schwarze 608: case FTS_SL:
1.200 schwarze 609: if (realpath(path, buf) == NULL) {
1.139 schwarze 610: if (warnings)
611: say(path, "&realpath");
612: continue;
613: }
1.184 schwarze 614: if (strstr(buf, basedir) != buf
615: #ifdef HOMEBREWDIR
616: && strstr(buf, HOMEBREWDIR) != buf
617: #endif
618: ) {
1.139 schwarze 619: if (warnings) say("",
620: "%s: outside base directory", buf);
621: continue;
622: }
623: /* Use logical inode to avoid mpages dupe. */
1.200 schwarze 624: if (stat(path, ff->fts_statp) == -1) {
1.139 schwarze 625: if (warnings)
626: say(path, "&stat");
627: continue;
628: }
629: /* FALLTHROUGH */
630:
1.50 kristaps 631: /*
1.83 schwarze 632: * If we're a regular file, add an mlink by using the
1.50 kristaps 633: * stored directory data and handling the filename.
634: */
1.141 schwarze 635: case FTS_F:
1.200 schwarze 636: if ( ! strcmp(path, MANDOC_DB))
1.60 schwarze 637: continue;
1.50 kristaps 638: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 639: if (warnings)
1.59 schwarze 640: say(path, "Extraneous file");
1.50 kristaps 641: continue;
1.124 schwarze 642: }
643: gzip = 0;
644: fsec = NULL;
1.200 schwarze 645: while (fsec == NULL) {
1.124 schwarze 646: fsec = strrchr(ff->fts_name, '.');
1.200 schwarze 647: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.124 schwarze 648: break;
649: gzip = 1;
650: *fsec = '\0';
651: fsec = NULL;
652: }
1.200 schwarze 653: if (fsec == NULL) {
1.60 schwarze 654: if ( ! use_all) {
1.56 schwarze 655: if (warnings)
1.60 schwarze 656: say(path,
657: "No filename suffix");
1.50 kristaps 658: continue;
659: }
1.200 schwarze 660: } else if ( ! strcmp(++fsec, "html")) {
1.60 schwarze 661: if (warnings)
662: say(path, "Skip html");
663: continue;
1.200 schwarze 664: } else if ( ! strcmp(fsec, "ps")) {
1.60 schwarze 665: if (warnings)
666: say(path, "Skip ps");
667: continue;
1.200 schwarze 668: } else if ( ! strcmp(fsec, "pdf")) {
1.60 schwarze 669: if (warnings)
670: say(path, "Skip pdf");
671: continue;
672: } else if ( ! use_all &&
1.200 schwarze 673: ((dform == FORM_SRC &&
1.185 schwarze 674: strncmp(fsec, dsec, strlen(dsec))) ||
1.200 schwarze 675: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.60 schwarze 676: if (warnings)
677: say(path, "Wrong filename suffix");
678: continue;
1.66 schwarze 679: } else
1.84 schwarze 680: fsec[-1] = '\0';
1.94 schwarze 681:
1.84 schwarze 682: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 683: if (strlcpy(mlink->file, path,
684: sizeof(mlink->file)) >=
685: sizeof(mlink->file)) {
686: say(path, "Filename too long");
687: free(mlink);
688: continue;
689: }
1.84 schwarze 690: mlink->dform = dform;
1.94 schwarze 691: mlink->dsec = dsec;
692: mlink->arch = arch;
693: mlink->name = ff->fts_name;
694: mlink->fsec = fsec;
1.124 schwarze 695: mlink->gzip = gzip;
1.84 schwarze 696: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 697: continue;
1.139 schwarze 698:
1.141 schwarze 699: case FTS_D:
700: case FTS_DP:
1.139 schwarze 701: break;
702:
703: default:
1.60 schwarze 704: if (warnings)
705: say(path, "Not a regular file");
1.50 kristaps 706: continue;
1.60 schwarze 707: }
1.1 kristaps 708:
1.50 kristaps 709: switch (ff->fts_level) {
1.141 schwarze 710: case 0:
1.50 kristaps 711: /* Ignore the root directory. */
712: break;
1.141 schwarze 713: case 1:
1.50 kristaps 714: /*
715: * This might contain manX/ or catX/.
716: * Try to infer this from the name.
717: * If we're not in use_all, enforce it.
718: */
719: cp = ff->fts_name;
1.200 schwarze 720: if (ff->fts_info == FTS_DP) {
721: dform = FORM_NONE;
722: dsec = NULL;
1.50 kristaps 723: break;
1.200 schwarze 724: }
1.1 kristaps 725:
1.200 schwarze 726: if ( ! strncmp(cp, "man", 3)) {
1.50 kristaps 727: dform = FORM_SRC;
1.66 schwarze 728: dsec = cp + 3;
1.200 schwarze 729: } else if ( ! strncmp(cp, "cat", 3)) {
1.50 kristaps 730: dform = FORM_CAT;
1.66 schwarze 731: dsec = cp + 3;
1.94 schwarze 732: } else {
733: dform = FORM_NONE;
734: dsec = NULL;
1.50 kristaps 735: }
1.1 kristaps 736:
1.200 schwarze 737: if (dsec != NULL || use_all)
1.50 kristaps 738: break;
1.1 kristaps 739:
1.56 schwarze 740: if (warnings)
1.59 schwarze 741: say(path, "Unknown directory part");
1.50 kristaps 742: fts_set(f, ff, FTS_SKIP);
743: break;
1.141 schwarze 744: case 2:
1.50 kristaps 745: /*
746: * Possibly our architecture.
747: * If we're descending, keep tabs on it.
748: */
1.200 schwarze 749: if (ff->fts_info != FTS_DP && dsec != NULL)
1.66 schwarze 750: arch = ff->fts_name;
1.94 schwarze 751: else
752: arch = NULL;
1.50 kristaps 753: break;
754: default:
1.200 schwarze 755: if (ff->fts_info == FTS_DP || use_all)
1.50 kristaps 756: break;
1.56 schwarze 757: if (warnings)
1.59 schwarze 758: say(path, "Extraneous directory part");
1.50 kristaps 759: fts_set(f, ff, FTS_SKIP);
760: break;
1.5 kristaps 761: }
1.50 kristaps 762: }
763:
764: fts_close(f);
1.197 schwarze 765: return 1;
1.50 kristaps 766: }
1.5 kristaps 767:
1.50 kristaps 768: /*
1.89 schwarze 769: * Add a file to the mlinks table.
1.50 kristaps 770: * Do not verify that it's a "valid" looking manpage (we'll do that
771: * later).
772: *
773: * Try to infer the manual section, architecture, and page name from the
774: * path, assuming it looks like
775: *
1.141 schwarze 776: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 777: * or
778: * [./]cat<section>[/<arch>]/<name>.0
779: *
780: * See treescan() for the fts(3) version of this.
781: */
782: static void
1.59 schwarze 783: filescan(const char *file)
1.50 kristaps 784: {
1.59 schwarze 785: char buf[PATH_MAX];
1.84 schwarze 786: struct stat st;
787: struct mlink *mlink;
1.59 schwarze 788: char *p, *start;
1.5 kristaps 789:
1.50 kristaps 790: assert(use_all);
1.5 kristaps 791:
1.50 kristaps 792: if (0 == strncmp(file, "./", 2))
793: file += 2;
1.5 kristaps 794:
1.139 schwarze 795: /*
796: * We have to do lstat(2) before realpath(3) loses
797: * the information whether this is a symbolic link.
798: * We need to know that because for symbolic links,
799: * we want to use the orginal file name, while for
800: * regular files, we want to use the real path.
801: */
802: if (-1 == lstat(file, &st)) {
803: exitcode = (int)MANDOCLEVEL_BADARG;
804: say(file, "&lstat");
805: return;
806: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
807: exitcode = (int)MANDOCLEVEL_BADARG;
808: say(file, "Not a regular file");
809: return;
810: }
811:
812: /*
813: * We have to resolve the file name to the real path
814: * in any case for the base directory check.
815: */
1.59 schwarze 816: if (NULL == realpath(file, buf)) {
817: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 818: say(file, "&realpath");
1.59 schwarze 819: return;
1.106 schwarze 820: }
821:
1.150 schwarze 822: if (OP_TEST == op)
1.106 schwarze 823: start = buf;
1.150 schwarze 824: else if (strstr(buf, basedir) == buf)
825: start = buf + strlen(basedir);
1.184 schwarze 826: #ifdef HOMEBREWDIR
827: else if (strstr(buf, HOMEBREWDIR) == buf)
828: start = buf;
829: #endif
1.106 schwarze 830: else {
1.59 schwarze 831: exitcode = (int)MANDOCLEVEL_BADARG;
832: say("", "%s: outside base directory", buf);
833: return;
1.106 schwarze 834: }
835:
1.139 schwarze 836: /*
837: * Now we are sure the file is inside our tree.
838: * If it is a symbolic link, ignore the real path
839: * and use the original name.
840: * This implies passing stuff like "cat1/../man1/foo.1"
841: * on the command line won't work. So don't do that.
842: * Note the stat(2) can still fail if the link target
843: * doesn't exist.
844: */
845: if (S_IFLNK & st.st_mode) {
846: if (-1 == stat(buf, &st)) {
847: exitcode = (int)MANDOCLEVEL_BADARG;
848: say(file, "&stat");
849: return;
850: }
1.143 schwarze 851: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
852: say(file, "Filename too long");
853: return;
854: }
1.150 schwarze 855: start = buf;
856: if (OP_TEST != op && strstr(buf, basedir) == buf)
857: start += strlen(basedir);
1.50 kristaps 858: }
1.106 schwarze 859:
1.84 schwarze 860: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 861: mlink->dform = FORM_NONE;
1.143 schwarze 862: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
863: sizeof(mlink->file)) {
864: say(start, "Filename too long");
1.180 schwarze 865: free(mlink);
1.143 schwarze 866: return;
867: }
1.17 schwarze 868:
1.50 kristaps 869: /*
870: * First try to guess our directory structure.
871: * If we find a separator, try to look for man* or cat*.
872: * If we find one of these and what's underneath is a directory,
873: * assume it's an architecture.
874: */
875: if (NULL != (p = strchr(start, '/'))) {
876: *p++ = '\0';
877: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 878: mlink->dform = FORM_SRC;
1.94 schwarze 879: mlink->dsec = start + 3;
1.50 kristaps 880: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 881: mlink->dform = FORM_CAT;
1.94 schwarze 882: mlink->dsec = start + 3;
1.17 schwarze 883: }
1.5 kristaps 884:
1.50 kristaps 885: start = p;
1.84 schwarze 886: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 887: *p++ = '\0';
1.94 schwarze 888: mlink->arch = start;
1.50 kristaps 889: start = p;
1.84 schwarze 890: }
1.50 kristaps 891: }
892:
893: /*
894: * Now check the file suffix.
895: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
896: */
897: p = strrchr(start, '\0');
898: while (p-- > start && '/' != *p && '.' != *p)
899: /* Loop. */ ;
900:
901: if ('.' == *p) {
902: *p++ = '\0';
1.94 schwarze 903: mlink->fsec = p;
1.5 kristaps 904: }
905:
1.10 kristaps 906: /*
1.50 kristaps 907: * Now try to parse the name.
908: * Use the filename portion of the path.
1.10 kristaps 909: */
1.84 schwarze 910: mlink->name = start;
1.50 kristaps 911: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 912: mlink->name = p + 1;
1.50 kristaps 913: *p = '\0';
1.84 schwarze 914: }
915: mlink_add(mlink, &st);
1.50 kristaps 916: }
1.5 kristaps 917:
1.50 kristaps 918: static void
1.84 schwarze 919: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 920: {
1.83 schwarze 921: struct inodev inodev;
1.78 schwarze 922: struct mpage *mpage;
1.83 schwarze 923: unsigned int slot;
1.50 kristaps 924:
1.84 schwarze 925: assert(NULL != mlink->file);
1.50 kristaps 926:
1.94 schwarze 927: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
928: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
929: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
930: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 931:
932: if ('0' == *mlink->fsec) {
933: free(mlink->fsec);
934: mlink->fsec = mandoc_strdup(mlink->dsec);
935: mlink->fform = FORM_CAT;
936: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
937: mlink->fform = FORM_SRC;
1.74 schwarze 938: else
1.84 schwarze 939: mlink->fform = FORM_NONE;
1.82 schwarze 940:
1.83 schwarze 941: slot = ohash_qlookup(&mlinks, mlink->file);
942: assert(NULL == ohash_find(&mlinks, slot));
943: ohash_insert(&mlinks, slot, mlink);
944:
1.186 schwarze 945: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 946: inodev.st_ino = st->st_ino;
947: inodev.st_dev = st->st_dev;
948: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
949: sizeof(struct inodev), inodev.st_ino);
950: mpage = ohash_find(&mpages, slot);
951: if (NULL == mpage) {
952: mpage = mandoc_calloc(1, sizeof(struct mpage));
953: mpage->inodev.st_ino = inodev.st_ino;
954: mpage->inodev.st_dev = inodev.st_dev;
955: ohash_insert(&mpages, slot, mpage);
956: } else
1.85 schwarze 957: mlink->next = mpage->mlinks;
1.82 schwarze 958: mpage->mlinks = mlink;
1.118 schwarze 959: mlink->mpage = mpage;
1.82 schwarze 960: }
961:
962: static void
963: mlink_free(struct mlink *mlink)
964: {
965:
966: free(mlink->dsec);
967: free(mlink->arch);
968: free(mlink->name);
969: free(mlink->fsec);
970: free(mlink);
1.50 kristaps 971: }
1.3 kristaps 972:
1.50 kristaps 973: static void
1.78 schwarze 974: mpages_free(void)
1.50 kristaps 975: {
1.78 schwarze 976: struct mpage *mpage;
1.82 schwarze 977: struct mlink *mlink;
1.80 schwarze 978: unsigned int slot;
1.3 kristaps 979:
1.80 schwarze 980: mpage = ohash_first(&mpages, &slot);
981: while (NULL != mpage) {
1.82 schwarze 982: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 983: mpage->mlinks = mlink->next;
1.82 schwarze 984: mlink_free(mlink);
985: }
1.78 schwarze 986: free(mpage->sec);
987: free(mpage->arch);
1.82 schwarze 988: free(mpage->title);
989: free(mpage->desc);
1.78 schwarze 990: free(mpage);
1.80 schwarze 991: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 992: }
993: }
1.38 schwarze 994:
1.50 kristaps 995: /*
1.89 schwarze 996: * For each mlink to the mpage, check whether the path looks like
997: * it is formatted, and if it does, check whether a source manual
998: * exists by the same name, ignoring the suffix.
999: * If both conditions hold, drop the mlink.
1000: */
1001: static void
1002: mlinks_undupe(struct mpage *mpage)
1003: {
1004: char buf[PATH_MAX];
1005: struct mlink **prev;
1006: struct mlink *mlink;
1007: char *bufp;
1008:
1009: mpage->form = FORM_CAT;
1.90 schwarze 1010: prev = &mpage->mlinks;
1011: while (NULL != (mlink = *prev)) {
1.89 schwarze 1012: if (FORM_CAT != mlink->dform) {
1013: mpage->form = FORM_NONE;
1.90 schwarze 1014: goto nextlink;
1.89 schwarze 1015: }
1.143 schwarze 1016: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1017: bufp = strstr(buf, "cat");
1018: assert(NULL != bufp);
1019: memcpy(bufp, "man", 3);
1020: if (NULL != (bufp = strrchr(buf, '.')))
1021: *++bufp = '\0';
1.143 schwarze 1022: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1023: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1024: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1025: goto nextlink;
1.89 schwarze 1026: if (warnings)
1027: say(mlink->file, "Man source exists: %s", buf);
1028: if (use_all)
1.90 schwarze 1029: goto nextlink;
1.89 schwarze 1030: *prev = mlink->next;
1031: mlink_free(mlink);
1.90 schwarze 1032: continue;
1033: nextlink:
1034: prev = &(*prev)->next;
1.89 schwarze 1035: }
1036: }
1037:
1.131 schwarze 1038: static void
1.93 schwarze 1039: mlink_check(struct mpage *mpage, struct mlink *mlink)
1040: {
1.131 schwarze 1041: struct str *str;
1042: unsigned int slot;
1.93 schwarze 1043:
1044: /*
1045: * Check whether the manual section given in a file
1046: * agrees with the directory where the file is located.
1047: * Some manuals have suffixes like (3p) on their
1048: * section number either inside the file or in the
1049: * directory name, some are linked into more than one
1050: * section, like encrypt(1) = makekey(8).
1051: */
1052:
1053: if (FORM_SRC == mpage->form &&
1.131 schwarze 1054: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1055: say(mlink->file, "Section \"%s\" manual in %s directory",
1056: mpage->sec, mlink->dsec);
1057:
1058: /*
1059: * Manual page directories exist for each kernel
1060: * architecture as returned by machine(1).
1061: * However, many manuals only depend on the
1062: * application architecture as returned by arch(1).
1063: * For example, some (2/ARM) manuals are shared
1064: * across the "armish" and "zaurus" kernel
1065: * architectures.
1066: * A few manuals are even shared across completely
1067: * different architectures, for example fdformat(1)
1068: * on amd64, i386, sparc, and sparc64.
1069: */
1070:
1.131 schwarze 1071: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1072: say(mlink->file, "Architecture \"%s\" manual in "
1073: "\"%s\" directory", mpage->arch, mlink->arch);
1074:
1.131 schwarze 1075: /*
1076: * XXX
1.133 schwarze 1077: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1078: */
1079:
1080: if (FORM_CAT == mpage->form)
1081: return;
1082:
1083: /*
1084: * Check whether this mlink
1085: * appears as a name in the NAME section.
1086: */
1.93 schwarze 1087:
1.133 schwarze 1088: slot = ohash_qlookup(&names, mlink->name);
1089: str = ohash_find(&names, slot);
1.131 schwarze 1090: assert(NULL != str);
1.133 schwarze 1091: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1092: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1093: }
1094:
1.89 schwarze 1095: /*
1.80 schwarze 1096: * Run through the files in the global vector "mpages"
1097: * and add them to the database specified in "basedir".
1.50 kristaps 1098: *
1099: * This handles the parsing scheme itself, using the cues of directory
1100: * and filename to determine whether the file is parsable or not.
1101: */
1.59 schwarze 1102: static void
1.176 schwarze 1103: mpages_merge(struct mparse *mp)
1.50 kristaps 1104: {
1.113 schwarze 1105: char any[] = "any";
1.118 schwarze 1106: struct mpage *mpage, *mpage_dest;
1107: struct mlink *mlink, *mlink_dest;
1.191 schwarze 1108: struct roff_man *man;
1.118 schwarze 1109: char *sodest;
1.112 schwarze 1110: char *cp;
1.162 schwarze 1111: int fd;
1.101 schwarze 1112: unsigned int pslot;
1.69 schwarze 1113:
1.171 schwarze 1114: if ( ! nodb)
1.107 schwarze 1115: SQL_EXEC("BEGIN TRANSACTION");
1116:
1.80 schwarze 1117: mpage = ohash_first(&mpages, &pslot);
1.171 schwarze 1118: while (mpage != NULL) {
1.89 schwarze 1119: mlinks_undupe(mpage);
1.182 schwarze 1120: if ((mlink = mpage->mlinks) == NULL) {
1.89 schwarze 1121: mpage = ohash_next(&mpages, &pslot);
1122: continue;
1.50 kristaps 1123: }
1124:
1.133 schwarze 1125: name_mask = NAME_MASK;
1.202 schwarze 1126: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1127: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.50 kristaps 1128: mparse_reset(mp);
1129: man = NULL;
1.124 schwarze 1130: sodest = NULL;
1131:
1.182 schwarze 1132: mparse_open(mp, &fd, mlink->file);
1.162 schwarze 1133: if (fd == -1) {
1.182 schwarze 1134: say(mlink->file, "&open");
1.162 schwarze 1135: goto nextpage;
1.124 schwarze 1136: }
1.14 schwarze 1137:
1138: /*
1.183 schwarze 1139: * Interpret the file as mdoc(7) or man(7) source
1140: * code, unless it is known to be formatted.
1.14 schwarze 1141: */
1.182 schwarze 1142: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1143: mparse_readfd(mp, fd, mlink->file);
1.192 schwarze 1144: mparse_result(mp, &man, &sodest);
1.76 schwarze 1145: }
1.14 schwarze 1146:
1.171 schwarze 1147: if (sodest != NULL) {
1.118 schwarze 1148: mlink_dest = ohash_find(&mlinks,
1149: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1150: if (mlink_dest == NULL) {
1151: mandoc_asprintf(&cp, "%s.gz", sodest);
1152: mlink_dest = ohash_find(&mlinks,
1153: ohash_qlookup(&mlinks, cp));
1154: free(cp);
1155: }
1156: if (mlink_dest != NULL) {
1.118 schwarze 1157:
1158: /* The .so target exists. */
1159:
1160: mpage_dest = mlink_dest->mpage;
1161: while (1) {
1162: mlink->mpage = mpage_dest;
1163:
1164: /*
1165: * If the target was already
1166: * processed, add the links
1167: * to the database now.
1168: * Otherwise, this will
1169: * happen when we come
1170: * to the target.
1171: */
1172:
1.137 schwarze 1173: if (mpage_dest->pageid)
1.169 schwarze 1174: dbadd_mlink_name(mlink);
1.118 schwarze 1175:
1.171 schwarze 1176: if (mlink->next == NULL)
1.118 schwarze 1177: break;
1178: mlink = mlink->next;
1179: }
1180:
1181: /* Move all links to the target. */
1182:
1183: mlink->next = mlink_dest->next;
1184: mlink_dest->next = mpage->mlinks;
1185: mpage->mlinks = NULL;
1186: }
1.124 schwarze 1187: goto nextpage;
1.192 schwarze 1188: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.204 schwarze 1189: mdoc_validate(man);
1.82 schwarze 1190: mpage->form = FORM_SRC;
1.193 schwarze 1191: mpage->sec = man->meta.msec;
1.155 schwarze 1192: mpage->sec = mandoc_strdup(
1.171 schwarze 1193: mpage->sec == NULL ? "" : mpage->sec);
1.193 schwarze 1194: mpage->arch = man->meta.arch;
1.84 schwarze 1195: mpage->arch = mandoc_strdup(
1.171 schwarze 1196: mpage->arch == NULL ? "" : mpage->arch);
1.193 schwarze 1197: mpage->title = mandoc_strdup(man->meta.title);
1.192 schwarze 1198: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.205 schwarze 1199: man_validate(man);
1.82 schwarze 1200: mpage->form = FORM_SRC;
1.193 schwarze 1201: mpage->sec = mandoc_strdup(man->meta.msec);
1.182 schwarze 1202: mpage->arch = mandoc_strdup(mlink->arch);
1.193 schwarze 1203: mpage->title = mandoc_strdup(man->meta.title);
1.76 schwarze 1204: } else {
1.82 schwarze 1205: mpage->form = FORM_CAT;
1.182 schwarze 1206: mpage->sec = mandoc_strdup(mlink->dsec);
1207: mpage->arch = mandoc_strdup(mlink->arch);
1208: mpage->title = mandoc_strdup(mlink->name);
1.76 schwarze 1209: }
1.97 schwarze 1210: putkey(mpage, mpage->sec, TYPE_sec);
1.164 schwarze 1211: if (*mpage->arch != '\0')
1212: putkey(mpage, mpage->arch, TYPE_arch);
1.50 kristaps 1213:
1.182 schwarze 1214: for ( ; mlink != NULL; mlink = mlink->next) {
1.97 schwarze 1215: if ('\0' != *mlink->dsec)
1216: putkey(mpage, mlink->dsec, TYPE_sec);
1217: if ('\0' != *mlink->fsec)
1218: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1219: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1220: any : mlink->arch, TYPE_arch);
1.133 schwarze 1221: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1222: }
1.12 schwarze 1223:
1.174 schwarze 1224: assert(mpage->desc == NULL);
1.192 schwarze 1225: if (man != NULL && man->macroset == MACROSET_MDOC)
1.193 schwarze 1226: parse_mdoc(mpage, &man->meta, man->first);
1.174 schwarze 1227: else if (man != NULL)
1.193 schwarze 1228: parse_man(mpage, &man->meta, man->first);
1.50 kristaps 1229: else
1.162 schwarze 1230: parse_cat(mpage, fd);
1.174 schwarze 1231: if (mpage->desc == NULL)
1.135 schwarze 1232: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1233:
1234: if (warnings && !use_all)
1235: for (mlink = mpage->mlinks; mlink;
1236: mlink = mlink->next)
1237: mlink_check(mpage, mlink);
1.44 kristaps 1238:
1.176 schwarze 1239: dbadd(mpage);
1.182 schwarze 1240: mlink = mpage->mlinks;
1.124 schwarze 1241:
1242: nextpage:
1.66 schwarze 1243: ohash_delete(&strings);
1.133 schwarze 1244: ohash_delete(&names);
1.80 schwarze 1245: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1246: }
1.107 schwarze 1247:
1248: if (0 == nodb)
1249: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1250: }
1251:
1252: static void
1253: names_check(void)
1254: {
1255: sqlite3_stmt *stmt;
1256: const char *name, *sec, *arch, *key;
1257:
1258: sqlite3_prepare_v2(db,
1259: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1260: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1261: "WHERE bits & ? AND NOT EXISTS ("
1262: "SELECT pageid FROM mlinks "
1.133 schwarze 1263: "WHERE mlinks.pageid == names.pageid "
1264: "AND mlinks.name == names.name"
1.130 schwarze 1265: ")"
1266: ") JOIN ("
1.142 schwarze 1267: "SELECT sec, arch, name, pageid FROM mlinks "
1268: "GROUP BY pageid"
1.130 schwarze 1269: ") USING (pageid);",
1270: -1, &stmt, NULL);
1271:
1.201 schwarze 1272: if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK)
1.134 schwarze 1273: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1274:
1.201 schwarze 1275: while (sqlite3_step(stmt) == SQLITE_ROW) {
1.154 schwarze 1276: name = (const char *)sqlite3_column_text(stmt, 0);
1277: sec = (const char *)sqlite3_column_text(stmt, 1);
1278: arch = (const char *)sqlite3_column_text(stmt, 2);
1279: key = (const char *)sqlite3_column_text(stmt, 3);
1.130 schwarze 1280: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1281: '\0' == *arch ? "" : "/",
1282: '\0' == *arch ? "" : arch, key);
1283: }
1284: sqlite3_finalize(stmt);
1.50 kristaps 1285: }
1.12 schwarze 1286:
1.50 kristaps 1287: static void
1.124 schwarze 1288: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1289: {
1290: FILE *stream;
1291: char *line, *p, *title;
1292: size_t len, plen, titlesz;
1.12 schwarze 1293:
1.124 schwarze 1294: stream = (-1 == fd) ?
1295: fopen(mpage->mlinks->file, "r") :
1296: fdopen(fd, "r");
1297: if (NULL == stream) {
1.159 schwarze 1298: if (-1 != fd)
1299: close(fd);
1.56 schwarze 1300: if (warnings)
1.123 schwarze 1301: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1302: return;
1303: }
1.1 kristaps 1304:
1.50 kristaps 1305: /* Skip to first blank line. */
1.1 kristaps 1306:
1.50 kristaps 1307: while (NULL != (line = fgetln(stream, &len)))
1308: if ('\n' == *line)
1309: break;
1.1 kristaps 1310:
1.50 kristaps 1311: /*
1312: * Assume the first line that is not indented
1313: * is the first section header. Skip to it.
1314: */
1.1 kristaps 1315:
1.50 kristaps 1316: while (NULL != (line = fgetln(stream, &len)))
1317: if ('\n' != *line && ' ' != *line)
1318: break;
1.141 schwarze 1319:
1.50 kristaps 1320: /*
1321: * Read up until the next section into a buffer.
1322: * Strip the leading and trailing newline from each read line,
1323: * appending a trailing space.
1324: * Ignore empty (whitespace-only) lines.
1325: */
1.1 kristaps 1326:
1.50 kristaps 1327: titlesz = 0;
1328: title = NULL;
1.38 schwarze 1329:
1.50 kristaps 1330: while (NULL != (line = fgetln(stream, &len))) {
1331: if (' ' != *line || '\n' != line[len - 1])
1332: break;
1333: while (len > 0 && isspace((unsigned char)*line)) {
1334: line++;
1335: len--;
1336: }
1337: if (1 == len)
1.38 schwarze 1338: continue;
1.50 kristaps 1339: title = mandoc_realloc(title, titlesz + len);
1340: memcpy(title + titlesz, line, len);
1341: titlesz += len;
1342: title[titlesz - 1] = ' ';
1343: }
1.38 schwarze 1344:
1.50 kristaps 1345: /*
1346: * If no page content can be found, or the input line
1347: * is already the next section header, or there is no
1348: * trailing newline, reuse the page title as the page
1349: * description.
1350: */
1.44 kristaps 1351:
1.50 kristaps 1352: if (NULL == title || '\0' == *title) {
1.56 schwarze 1353: if (warnings)
1.82 schwarze 1354: say(mpage->mlinks->file,
1355: "Cannot find NAME section");
1.50 kristaps 1356: fclose(stream);
1357: free(title);
1358: return;
1359: }
1.1 kristaps 1360:
1.50 kristaps 1361: title = mandoc_realloc(title, titlesz + 1);
1362: title[titlesz] = '\0';
1.33 schwarze 1363:
1.50 kristaps 1364: /*
1365: * Skip to the first dash.
1366: * Use the remaining line as the description (no more than 70
1367: * bytes).
1368: */
1.33 schwarze 1369:
1.50 kristaps 1370: if (NULL != (p = strstr(title, "- "))) {
1371: for (p += 2; ' ' == *p || '\b' == *p; p++)
1372: /* Skip to next word. */ ;
1373: } else {
1.56 schwarze 1374: if (warnings)
1.82 schwarze 1375: say(mpage->mlinks->file,
1376: "No dash in title line");
1.50 kristaps 1377: p = title;
1378: }
1.38 schwarze 1379:
1.50 kristaps 1380: plen = strlen(p);
1.1 kristaps 1381:
1.50 kristaps 1382: /* Strip backspace-encoding from line. */
1.1 kristaps 1383:
1.50 kristaps 1384: while (NULL != (line = memchr(p, '\b', plen))) {
1385: len = line - p;
1386: if (0 == len) {
1387: memmove(line, line + 1, plen--);
1388: continue;
1.141 schwarze 1389: }
1.50 kristaps 1390: memmove(line - 1, line + 1, plen - len);
1391: plen -= 2;
1392: }
1.1 kristaps 1393:
1.78 schwarze 1394: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1395: fclose(stream);
1396: free(title);
1397: }
1.1 kristaps 1398:
1.50 kristaps 1399: /*
1400: * Put a type/word pair into the word database for this particular file.
1401: */
1402: static void
1.112 schwarze 1403: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1404: {
1.112 schwarze 1405: char *cp;
1.18 kristaps 1406:
1.50 kristaps 1407: assert(NULL != value);
1.112 schwarze 1408: if (TYPE_arch == type)
1409: for (cp = value; *cp; cp++)
1410: if (isupper((unsigned char)*cp))
1411: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1412: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1413: }
1414:
1415: /*
1.50 kristaps 1416: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1417: */
1418: static void
1.78 schwarze 1419: putmdockey(const struct mpage *mpage,
1.189 schwarze 1420: const struct roff_node *n, uint64_t m)
1.3 kristaps 1421: {
1.18 kristaps 1422:
1.50 kristaps 1423: for ( ; NULL != n; n = n->next) {
1424: if (NULL != n->child)
1.78 schwarze 1425: putmdockey(mpage, n->child, m);
1.188 schwarze 1426: if (n->type == ROFFT_TEXT)
1.78 schwarze 1427: putkey(mpage, n->string, m);
1.50 kristaps 1428: }
1429: }
1.18 kristaps 1430:
1.61 schwarze 1431: static void
1.190 schwarze 1432: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1433: const struct roff_node *n)
1.50 kristaps 1434: {
1.189 schwarze 1435: const struct roff_node *head, *body;
1.121 schwarze 1436: char *start, *title;
1.50 kristaps 1437: char byte;
1.121 schwarze 1438: size_t sz;
1.18 kristaps 1439:
1.50 kristaps 1440: if (NULL == n)
1.61 schwarze 1441: return;
1.18 kristaps 1442:
1.50 kristaps 1443: /*
1444: * We're only searching for one thing: the first text child in
1445: * the BODY of a NAME section. Since we don't keep track of
1446: * sections in -man, run some hoops to find out whether we're in
1447: * the correct section or not.
1448: */
1.18 kristaps 1449:
1.188 schwarze 1450: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.50 kristaps 1451: body = n;
1452: assert(body->parent);
1453: if (NULL != (head = body->parent->head) &&
1.141 schwarze 1454: 1 == head->nchild &&
1455: NULL != (head = (head->child)) &&
1.188 schwarze 1456: head->type == ROFFT_TEXT &&
1.141 schwarze 1457: 0 == strcmp(head->string, "NAME") &&
1458: NULL != body->child) {
1.3 kristaps 1459:
1.50 kristaps 1460: /*
1461: * Suck the entire NAME section into memory.
1462: * Yes, we might run away.
1463: * But too many manuals have big, spread-out
1464: * NAME sections over many lines.
1465: */
1.3 kristaps 1466:
1.121 schwarze 1467: title = NULL;
1.194 schwarze 1468: deroff(&title, body);
1.50 kristaps 1469: if (NULL == title)
1.61 schwarze 1470: return;
1.18 kristaps 1471:
1.141 schwarze 1472: /*
1.50 kristaps 1473: * Go through a special heuristic dance here.
1474: * Conventionally, one or more manual names are
1475: * comma-specified prior to a whitespace, then a
1476: * dash, then a description. Try to puzzle out
1477: * the name parts here.
1478: */
1.18 kristaps 1479:
1.121 schwarze 1480: start = title;
1.50 kristaps 1481: for ( ;; ) {
1482: sz = strcspn(start, " ,");
1483: if ('\0' == start[sz])
1484: break;
1.1 kristaps 1485:
1.50 kristaps 1486: byte = start[sz];
1487: start[sz] = '\0';
1.110 schwarze 1488:
1489: /*
1490: * Assume a stray trailing comma in the
1491: * name list if a name begins with a dash.
1492: */
1493:
1494: if ('-' == start[0] ||
1495: ('\\' == start[0] && '-' == start[1]))
1496: break;
1.1 kristaps 1497:
1.133 schwarze 1498: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1499: if ( ! (mpage->name_head_done ||
1500: strcasecmp(start, meta->title))) {
1501: putkey(mpage, start, NAME_HEAD);
1502: mpage->name_head_done = 1;
1503: }
1.1 kristaps 1504:
1.50 kristaps 1505: if (' ' == byte) {
1506: start += sz + 1;
1507: break;
1508: }
1.1 kristaps 1509:
1.50 kristaps 1510: assert(',' == byte);
1511: start += sz + 1;
1512: while (' ' == *start)
1513: start++;
1514: }
1.1 kristaps 1515:
1.121 schwarze 1516: if (start == title) {
1.133 schwarze 1517: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1518: if ( ! (mpage->name_head_done ||
1519: strcasecmp(start, meta->title))) {
1520: putkey(mpage, start, NAME_HEAD);
1521: mpage->name_head_done = 1;
1522: }
1.50 kristaps 1523: free(title);
1.61 schwarze 1524: return;
1.50 kristaps 1525: }
1.1 kristaps 1526:
1.50 kristaps 1527: while (isspace((unsigned char)*start))
1528: start++;
1.1 kristaps 1529:
1.50 kristaps 1530: if (0 == strncmp(start, "-", 1))
1531: start += 1;
1532: else if (0 == strncmp(start, "\\-\\-", 4))
1533: start += 4;
1534: else if (0 == strncmp(start, "\\-", 2))
1535: start += 2;
1536: else if (0 == strncmp(start, "\\(en", 4))
1537: start += 4;
1538: else if (0 == strncmp(start, "\\(em", 4))
1539: start += 4;
1.1 kristaps 1540:
1.50 kristaps 1541: while (' ' == *start)
1542: start++;
1.1 kristaps 1543:
1.78 schwarze 1544: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1545: free(title);
1.61 schwarze 1546: return;
1.50 kristaps 1547: }
1548: }
1.1 kristaps 1549:
1.77 schwarze 1550: for (n = n->child; n; n = n->next) {
1.78 schwarze 1551: if (NULL != mpage->desc)
1.77 schwarze 1552: break;
1.174 schwarze 1553: parse_man(mpage, meta, n);
1.77 schwarze 1554: }
1.1 kristaps 1555: }
1556:
1557: static void
1.190 schwarze 1558: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1559: const struct roff_node *n)
1.1 kristaps 1560: {
1561:
1.50 kristaps 1562: assert(NULL != n);
1563: for (n = n->child; NULL != n; n = n->next) {
1564: switch (n->type) {
1.188 schwarze 1565: case ROFFT_ELEM:
1566: case ROFFT_BLOCK:
1567: case ROFFT_HEAD:
1568: case ROFFT_BODY:
1569: case ROFFT_TAIL:
1.50 kristaps 1570: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1571: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1572: break;
1.68 schwarze 1573: if (mdocs[n->tok].mask)
1.78 schwarze 1574: putmdockey(mpage, n->child,
1575: mdocs[n->tok].mask);
1.50 kristaps 1576: break;
1577: default:
1.188 schwarze 1578: assert(n->type != ROFFT_ROOT);
1.50 kristaps 1579: continue;
1580: }
1581: if (NULL != n->child)
1.172 schwarze 1582: parse_mdoc(mpage, meta, n);
1.1 kristaps 1583: }
1584: }
1585:
1.25 schwarze 1586: static int
1.190 schwarze 1587: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1588: const struct roff_node *n)
1.1 kristaps 1589: {
1.176 schwarze 1590: char *start, *end;
1.1 kristaps 1591: size_t sz;
1.25 schwarze 1592:
1.50 kristaps 1593: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1594: NULL == (n = n->child) ||
1.188 schwarze 1595: n->type != ROFFT_TEXT)
1.197 schwarze 1596: return 0;
1.1 kristaps 1597:
1598: /*
1599: * Only consider those `Fd' macro fields that begin with an
1600: * "inclusion" token (versus, e.g., #define).
1601: */
1.50 kristaps 1602:
1.1 kristaps 1603: if (strcmp("#include", n->string))
1.197 schwarze 1604: return 0;
1.1 kristaps 1605:
1.188 schwarze 1606: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1607: return 0;
1.1 kristaps 1608:
1609: /*
1610: * Strip away the enclosing angle brackets and make sure we're
1611: * not zero-length.
1612: */
1613:
1614: start = n->string;
1615: if ('<' == *start || '"' == *start)
1616: start++;
1617:
1618: if (0 == (sz = strlen(start)))
1.197 schwarze 1619: return 0;
1.1 kristaps 1620:
1621: end = &start[(int)sz - 1];
1622: if ('>' == *end || '"' == *end)
1623: end--;
1624:
1.50 kristaps 1625: if (end > start)
1.78 schwarze 1626: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1627: return 0;
1.1 kristaps 1628: }
1629:
1.178 schwarze 1630: static void
1.189 schwarze 1631: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1632: {
1.112 schwarze 1633: char *cp;
1.178 schwarze 1634: size_t sz;
1.1 kristaps 1635:
1.188 schwarze 1636: if (n->type != ROFFT_TEXT)
1.178 schwarze 1637: return;
1.25 schwarze 1638:
1.178 schwarze 1639: /* Skip function pointer punctuation. */
1.1 kristaps 1640:
1.178 schwarze 1641: cp = n->string;
1642: while (*cp == '(' || *cp == '*')
1.1 kristaps 1643: cp++;
1.178 schwarze 1644: sz = strcspn(cp, "()");
1.1 kristaps 1645:
1.178 schwarze 1646: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1647: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1648: putkeys(mpage, cp, sz, NAME_SYN);
1649: }
1.25 schwarze 1650:
1.178 schwarze 1651: static int
1.190 schwarze 1652: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1653: const struct roff_node *n)
1.178 schwarze 1654: {
1655:
1656: if (n->child == NULL)
1.197 schwarze 1657: return 0;
1.178 schwarze 1658:
1659: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1660:
1.178 schwarze 1661: for (n = n->child->next; n != NULL; n = n->next)
1.188 schwarze 1662: if (n->type == ROFFT_TEXT)
1.78 schwarze 1663: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1664:
1.197 schwarze 1665: return 0;
1.173 schwarze 1666: }
1667:
1668: static int
1.190 schwarze 1669: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1670: const struct roff_node *n)
1.173 schwarze 1671: {
1.177 schwarze 1672:
1.188 schwarze 1673: if (n->type != ROFFT_HEAD)
1.197 schwarze 1674: return 1;
1.173 schwarze 1675:
1.178 schwarze 1676: if (n->child != NULL)
1677: parse_mdoc_fname(mpage, n->child);
1678:
1.197 schwarze 1679: return 0;
1.1 kristaps 1680: }
1681:
1.25 schwarze 1682: static int
1.190 schwarze 1683: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1684: const struct roff_node *n)
1.1 kristaps 1685: {
1.67 schwarze 1686: char *cp;
1.1 kristaps 1687:
1688: if (NULL == (n = n->child))
1.197 schwarze 1689: return 0;
1.1 kristaps 1690:
1.67 schwarze 1691: if (NULL == n->next) {
1.78 schwarze 1692: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1693: return 0;
1.67 schwarze 1694: }
1695:
1.120 schwarze 1696: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1697: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1698: free(cp);
1.197 schwarze 1699: return 0;
1.1 kristaps 1700: }
1701:
1.25 schwarze 1702: static int
1.190 schwarze 1703: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1704: const struct roff_node *n)
1.1 kristaps 1705: {
1706:
1.188 schwarze 1707: if (n->type == ROFFT_BODY)
1.194 schwarze 1708: deroff(&mpage->desc, n);
1.197 schwarze 1709: return 0;
1.1 kristaps 1710: }
1711:
1.25 schwarze 1712: static int
1.190 schwarze 1713: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1714: const struct roff_node *n)
1.1 kristaps 1715: {
1716:
1.129 schwarze 1717: if (SEC_NAME == n->sec)
1.133 schwarze 1718: putmdockey(mpage, n->child, NAME_TITLE);
1.188 schwarze 1719: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1720: if (n->child == NULL)
1721: putkey(mpage, meta->name, NAME_SYN);
1722: else
1723: putmdockey(mpage, n->child, NAME_SYN);
1.174 schwarze 1724: }
1725: if ( ! (mpage->name_head_done ||
1726: n->child == NULL || n->child->string == NULL ||
1727: strcasecmp(n->child->string, meta->title))) {
1.188 schwarze 1728: putkey(mpage, n->child->string, ROFFT_HEAD);
1.174 schwarze 1729: mpage->name_head_done = 1;
1.172 schwarze 1730: }
1.197 schwarze 1731: return 0;
1.1 kristaps 1732: }
1733:
1.25 schwarze 1734: static int
1.190 schwarze 1735: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1736: const struct roff_node *n)
1.1 kristaps 1737: {
1738:
1.197 schwarze 1739: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1740: }
1741:
1.50 kristaps 1742: static int
1.190 schwarze 1743: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1744: const struct roff_node *n)
1.1 kristaps 1745: {
1746:
1.197 schwarze 1747: return n->type == ROFFT_HEAD;
1.50 kristaps 1748: }
1.1 kristaps 1749:
1.50 kristaps 1750: static int
1.190 schwarze 1751: parse_mdoc_body(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1752: const struct roff_node *n)
1.50 kristaps 1753: {
1.1 kristaps 1754:
1.197 schwarze 1755: return n->type == ROFFT_BODY;
1.1 kristaps 1756: }
1757:
1.50 kristaps 1758: /*
1.66 schwarze 1759: * Add a string to the hash table for the current manual.
1760: * Each string has a bitmask telling which macros it belongs to.
1761: * When we finish the manual, we'll dump the table.
1.50 kristaps 1762: */
1763: static void
1.176 schwarze 1764: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1765: {
1.133 schwarze 1766: struct ohash *htab;
1.50 kristaps 1767: struct str *s;
1.111 schwarze 1768: const char *end;
1.65 schwarze 1769: unsigned int slot;
1.176 schwarze 1770: int i, mustfree;
1.25 schwarze 1771:
1.50 kristaps 1772: if (0 == sz)
1773: return;
1.111 schwarze 1774:
1.176 schwarze 1775: mustfree = render_string(&cp, &sz);
1776:
1.133 schwarze 1777: if (TYPE_Nm & v) {
1778: htab = &names;
1779: v &= name_mask;
1.169 schwarze 1780: if (v & NAME_FIRST)
1781: name_mask &= ~NAME_FIRST;
1.133 schwarze 1782: if (debug > 1)
1783: say(mpage->mlinks->file,
1.175 schwarze 1784: "Adding name %*s, bits=%d", sz, cp, v);
1.133 schwarze 1785: } else {
1786: htab = &strings;
1787: if (debug > 1)
1788: for (i = 0; i < mansearch_keymax; i++)
1.163 schwarze 1789: if ((uint64_t)1 << i & v)
1.133 schwarze 1790: say(mpage->mlinks->file,
1791: "Adding key %s=%*s",
1792: mansearch_keynames[i], sz, cp);
1.111 schwarze 1793: }
1.25 schwarze 1794:
1.65 schwarze 1795: end = cp + sz;
1.133 schwarze 1796: slot = ohash_qlookupi(htab, cp, &end);
1797: s = ohash_find(htab, slot);
1.25 schwarze 1798:
1.78 schwarze 1799: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1800: s->mask |= v;
1801: return;
1802: } else if (NULL == s) {
1.144 schwarze 1803: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1804: memcpy(s->key, cp, sz);
1.133 schwarze 1805: ohash_insert(htab, slot, s);
1.1 kristaps 1806: }
1.78 schwarze 1807: s->mpage = mpage;
1.50 kristaps 1808: s->mask = v;
1.176 schwarze 1809:
1810: if (mustfree)
1811: free(cp);
1.1 kristaps 1812: }
1813:
1.50 kristaps 1814: /*
1815: * Take a Unicode codepoint and produce its UTF-8 encoding.
1816: * This isn't the best way to do this, but it works.
1817: * The magic numbers are from the UTF-8 packaging.
1818: * They're not as scary as they seem: read the UTF-8 spec for details.
1819: */
1820: static size_t
1821: utf8(unsigned int cp, char out[7])
1.1 kristaps 1822: {
1.50 kristaps 1823: size_t rc;
1.1 kristaps 1824:
1.50 kristaps 1825: rc = 0;
1826: if (cp <= 0x0000007F) {
1827: rc = 1;
1828: out[0] = (char)cp;
1829: } else if (cp <= 0x000007FF) {
1830: rc = 2;
1831: out[0] = (cp >> 6 & 31) | 192;
1832: out[1] = (cp & 63) | 128;
1833: } else if (cp <= 0x0000FFFF) {
1834: rc = 3;
1835: out[0] = (cp >> 12 & 15) | 224;
1836: out[1] = (cp >> 6 & 63) | 128;
1837: out[2] = (cp & 63) | 128;
1838: } else if (cp <= 0x001FFFFF) {
1839: rc = 4;
1840: out[0] = (cp >> 18 & 7) | 240;
1841: out[1] = (cp >> 12 & 63) | 128;
1842: out[2] = (cp >> 6 & 63) | 128;
1843: out[3] = (cp & 63) | 128;
1844: } else if (cp <= 0x03FFFFFF) {
1845: rc = 5;
1846: out[0] = (cp >> 24 & 3) | 248;
1847: out[1] = (cp >> 18 & 63) | 128;
1848: out[2] = (cp >> 12 & 63) | 128;
1849: out[3] = (cp >> 6 & 63) | 128;
1850: out[4] = (cp & 63) | 128;
1851: } else if (cp <= 0x7FFFFFFF) {
1852: rc = 6;
1853: out[0] = (cp >> 30 & 1) | 252;
1854: out[1] = (cp >> 24 & 63) | 128;
1855: out[2] = (cp >> 18 & 63) | 128;
1856: out[3] = (cp >> 12 & 63) | 128;
1857: out[4] = (cp >> 6 & 63) | 128;
1858: out[5] = (cp & 63) | 128;
1859: } else
1.197 schwarze 1860: return 0;
1.1 kristaps 1861:
1.50 kristaps 1862: out[rc] = '\0';
1.197 schwarze 1863: return rc;
1.50 kristaps 1864: }
1.1 kristaps 1865:
1.50 kristaps 1866: /*
1.176 schwarze 1867: * If the string contains escape sequences,
1868: * replace it with an allocated rendering and return 1,
1869: * such that the caller can free it after use.
1870: * Otherwise, do nothing and return 0.
1.50 kristaps 1871: */
1.176 schwarze 1872: static int
1873: render_string(char **public, size_t *psz)
1.50 kristaps 1874: {
1.176 schwarze 1875: const char *src, *scp, *addcp, *seq;
1876: char *dst;
1877: size_t ssz, dsz, addsz;
1.114 schwarze 1878: char utfbuf[7], res[6];
1.176 schwarze 1879: int seqlen, unicode;
1.50 kristaps 1880:
1881: res[0] = '\\';
1882: res[1] = '\t';
1883: res[2] = ASCII_NBRSP;
1884: res[3] = ASCII_HYPH;
1.114 schwarze 1885: res[4] = ASCII_BREAK;
1886: res[5] = '\0';
1.1 kristaps 1887:
1.176 schwarze 1888: src = scp = *public;
1889: ssz = *psz;
1890: dst = NULL;
1891: dsz = 0;
1892:
1893: while (scp < src + *psz) {
1894:
1895: /* Leave normal characters unchanged. */
1896:
1897: if (strchr(res, *scp) == NULL) {
1898: if (dst != NULL)
1899: dst[dsz++] = *scp;
1900: scp++;
1901: continue;
1902: }
1.46 kristaps 1903:
1.50 kristaps 1904: /*
1.176 schwarze 1905: * Found something that requires replacing,
1906: * make sure we have a destination buffer.
1.50 kristaps 1907: */
1.176 schwarze 1908:
1909: if (dst == NULL) {
1910: dst = mandoc_malloc(ssz + 1);
1911: dsz = scp - src;
1912: memcpy(dst, src, dsz);
1.50 kristaps 1913: }
1.46 kristaps 1914:
1.176 schwarze 1915: /* Handle single-char special characters. */
1916:
1917: switch (*scp) {
1918: case '\\':
1919: break;
1.141 schwarze 1920: case '\t':
1921: case ASCII_NBRSP:
1.176 schwarze 1922: dst[dsz++] = ' ';
1923: scp++;
1924: continue;
1925: case ASCII_HYPH:
1926: dst[dsz++] = '-';
1.114 schwarze 1927: /* FALLTHROUGH */
1.141 schwarze 1928: case ASCII_BREAK:
1.176 schwarze 1929: scp++;
1.50 kristaps 1930: continue;
1.114 schwarze 1931: default:
1.176 schwarze 1932: abort();
1.114 schwarze 1933: }
1.46 kristaps 1934:
1.50 kristaps 1935: /*
1.176 schwarze 1936: * Found an escape sequence.
1937: * Read past the slash, then parse it.
1938: * Ignore everything except characters.
1.50 kristaps 1939: */
1.95 schwarze 1940:
1.176 schwarze 1941: scp++;
1942: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1943: continue;
1.1 kristaps 1944:
1.50 kristaps 1945: /*
1.95 schwarze 1946: * Render the special character
1947: * as either UTF-8 or ASCII.
1.50 kristaps 1948: */
1.95 schwarze 1949:
1950: if (write_utf8) {
1.203 schwarze 1951: unicode = mchars_spec2cp(seq, seqlen);
1.176 schwarze 1952: if (unicode <= 0)
1.95 schwarze 1953: continue;
1.176 schwarze 1954: addsz = utf8(unicode, utfbuf);
1955: if (addsz == 0)
1.95 schwarze 1956: continue;
1.176 schwarze 1957: addcp = utfbuf;
1.95 schwarze 1958: } else {
1.203 schwarze 1959: addcp = mchars_spec2str(seq, seqlen, &addsz);
1.176 schwarze 1960: if (addcp == NULL)
1.95 schwarze 1961: continue;
1.176 schwarze 1962: if (*addcp == ASCII_NBRSP) {
1963: addcp = " ";
1964: addsz = 1;
1.95 schwarze 1965: }
1966: }
1.1 kristaps 1967:
1.50 kristaps 1968: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1969:
1.176 schwarze 1970: ssz += addsz;
1971: dst = mandoc_realloc(dst, ssz + 1);
1972: memcpy(dst + dsz, addcp, addsz);
1973: dsz += addsz;
1974: }
1975: if (dst != NULL) {
1976: *public = dst;
1977: *psz = dsz;
1978: }
1979:
1980: /* Trim trailing whitespace and NUL-terminate. */
1981:
1982: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1983: --*psz;
1984: if (dst != NULL) {
1985: (*public)[*psz] = '\0';
1.197 schwarze 1986: return 1;
1.176 schwarze 1987: } else
1.197 schwarze 1988: return 0;
1.1 kristaps 1989: }
1990:
1.118 schwarze 1991: static void
1992: dbadd_mlink(const struct mlink *mlink)
1993: {
1994: size_t i;
1995:
1996: i = 1;
1997: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1998: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1999: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 2000: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 2001: SQL_STEP(stmts[STMT_INSERT_LINK]);
2002: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.169 schwarze 2003: }
2004:
2005: static void
2006: dbadd_mlink_name(const struct mlink *mlink)
2007: {
1.175 schwarze 2008: uint64_t bits;
1.169 schwarze 2009: size_t i;
2010:
2011: dbadd_mlink(mlink);
1.160 schwarze 2012:
2013: i = 1;
1.175 schwarze 2014: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2015: bits = NAME_FILE & NAME_MASK;
2016: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2017: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2018: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2019: }
2020:
2021: i = 1;
2022: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.160 schwarze 2023: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2024: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2025: SQL_STEP(stmts[STMT_INSERT_NAME]);
2026: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.118 schwarze 2027: }
2028:
1.14 schwarze 2029: /*
1.50 kristaps 2030: * Flush the current page's terms (and their bits) into the database.
2031: * Wrap the entire set of additions in a transaction to make sqlite be a
2032: * little faster.
1.96 schwarze 2033: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2034: */
2035: static void
1.176 schwarze 2036: dbadd(struct mpage *mpage)
1.14 schwarze 2037: {
1.87 schwarze 2038: struct mlink *mlink;
1.50 kristaps 2039: struct str *key;
1.176 schwarze 2040: char *cp;
1.52 kristaps 2041: size_t i;
1.66 schwarze 2042: unsigned int slot;
1.176 schwarze 2043: int mustfree;
1.14 schwarze 2044:
1.128 schwarze 2045: mlink = mpage->mlinks;
1.43 kristaps 2046:
1.128 schwarze 2047: if (nodb) {
1.147 schwarze 2048: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2049: key = ohash_next(&names, &slot))
1.147 schwarze 2050: free(key);
2051: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2052: key = ohash_next(&strings, &slot))
1.147 schwarze 2053: free(key);
1.145 schwarze 2054: if (0 == debug)
2055: return;
1.128 schwarze 2056: while (NULL != mlink) {
2057: fputs(mlink->name, stdout);
2058: if (NULL == mlink->next ||
2059: strcmp(mlink->dsec, mlink->next->dsec) ||
2060: strcmp(mlink->fsec, mlink->next->fsec) ||
2061: strcmp(mlink->arch, mlink->next->arch)) {
2062: putchar('(');
2063: if ('\0' == *mlink->dsec)
2064: fputs(mlink->fsec, stdout);
2065: else
2066: fputs(mlink->dsec, stdout);
2067: if ('\0' != *mlink->arch)
2068: printf("/%s", mlink->arch);
2069: putchar(')');
2070: }
2071: mlink = mlink->next;
2072: if (NULL != mlink)
2073: fputs(", ", stdout);
2074: }
1.132 schwarze 2075: printf(" - %s\n", mpage->desc);
1.14 schwarze 2076: return;
1.128 schwarze 2077: }
2078:
2079: if (debug)
2080: say(mlink->file, "Adding to database");
1.28 kristaps 2081:
1.176 schwarze 2082: cp = mpage->desc;
2083: i = strlen(cp);
2084: mustfree = render_string(&cp, &i);
1.52 kristaps 2085: i = 1;
1.176 schwarze 2086: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.161 schwarze 2087: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.81 schwarze 2088: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2089: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2090: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.176 schwarze 2091: if (mustfree)
2092: free(cp);
1.81 schwarze 2093:
1.128 schwarze 2094: while (NULL != mlink) {
1.118 schwarze 2095: dbadd_mlink(mlink);
1.128 schwarze 2096: mlink = mlink->next;
2097: }
1.134 schwarze 2098: mlink = mpage->mlinks;
1.50 kristaps 2099:
1.133 schwarze 2100: for (key = ohash_first(&names, &slot); NULL != key;
2101: key = ohash_next(&names, &slot)) {
2102: assert(key->mpage == mpage);
2103: i = 1;
2104: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.176 schwarze 2105: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.137 schwarze 2106: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2107: SQL_STEP(stmts[STMT_INSERT_NAME]);
2108: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2109: free(key);
2110: }
1.66 schwarze 2111: for (key = ohash_first(&strings, &slot); NULL != key;
2112: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2113: assert(key->mpage == mpage);
1.52 kristaps 2114: i = 1;
2115: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.176 schwarze 2116: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.137 schwarze 2117: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2118: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2119: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 2120: free(key);
1.38 schwarze 2121: }
1.14 schwarze 2122: }
2123:
1.5 kristaps 2124: static void
1.59 schwarze 2125: dbprune(void)
1.5 kristaps 2126: {
1.78 schwarze 2127: struct mpage *mpage;
1.82 schwarze 2128: struct mlink *mlink;
1.52 kristaps 2129: size_t i;
1.80 schwarze 2130: unsigned int slot;
1.5 kristaps 2131:
1.106 schwarze 2132: if (0 == nodb)
2133: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2134:
1.106 schwarze 2135: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2136: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2137: mlink = mpage->mlinks;
1.125 schwarze 2138: if (debug)
1.106 schwarze 2139: say(mlink->file, "Deleting from database");
2140: if (nodb)
2141: continue;
2142: for ( ; NULL != mlink; mlink = mlink->next) {
2143: i = 1;
2144: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2145: i, mlink->dsec);
2146: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2147: i, mlink->arch);
2148: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2149: i, mlink->name);
2150: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2151: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2152: }
1.5 kristaps 2153: }
1.106 schwarze 2154:
2155: if (0 == nodb)
2156: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2157: }
2158:
1.4 kristaps 2159: /*
1.50 kristaps 2160: * Close an existing database and its prepared statements.
2161: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2162: */
1.35 kristaps 2163: static void
1.59 schwarze 2164: dbclose(int real)
1.4 kristaps 2165: {
1.50 kristaps 2166: size_t i;
1.115 schwarze 2167: int status;
2168: pid_t child;
1.4 kristaps 2169:
1.50 kristaps 2170: if (nodb)
1.38 schwarze 2171: return;
1.50 kristaps 2172:
2173: for (i = 0; i < STMT__MAX; i++) {
2174: sqlite3_finalize(stmts[i]);
2175: stmts[i] = NULL;
1.4 kristaps 2176: }
2177:
1.50 kristaps 2178: sqlite3_close(db);
2179: db = NULL;
1.12 schwarze 2180:
1.50 kristaps 2181: if (real)
2182: return;
1.12 schwarze 2183:
1.115 schwarze 2184: if ('\0' == *tempfilename) {
2185: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2186: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2187: say(MANDOC_DB, "&rename");
1.115 schwarze 2188: }
2189: return;
2190: }
2191:
2192: switch (child = fork()) {
1.141 schwarze 2193: case -1:
1.115 schwarze 2194: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2195: say("", "&fork cmp");
1.115 schwarze 2196: return;
1.141 schwarze 2197: case 0:
1.115 schwarze 2198: execlp("cmp", "cmp", "-s",
1.196 schwarze 2199: tempfilename, MANDOC_DB, (char *)NULL);
1.123 schwarze 2200: say("", "&exec cmp");
1.115 schwarze 2201: exit(0);
2202: default:
2203: break;
2204: }
2205: if (-1 == waitpid(child, &status, 0)) {
2206: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2207: say("", "&wait cmp");
1.115 schwarze 2208: } else if (WIFSIGNALED(status)) {
2209: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2210: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2211: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2212: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2213: say(MANDOC_DB,
2214: "Data changed, but cannot replace database");
2215: }
2216:
2217: *strrchr(tempfilename, '/') = '\0';
2218: switch (child = fork()) {
1.141 schwarze 2219: case -1:
1.115 schwarze 2220: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2221: say("", "&fork rm");
1.115 schwarze 2222: return;
1.141 schwarze 2223: case 0:
1.196 schwarze 2224: execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
1.123 schwarze 2225: say("", "&exec rm");
1.115 schwarze 2226: exit((int)MANDOCLEVEL_SYSERR);
2227: default:
2228: break;
2229: }
2230: if (-1 == waitpid(child, &status, 0)) {
2231: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2232: say("", "&wait rm");
1.115 schwarze 2233: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2234: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2235: say("", "%s: Cannot remove temporary directory",
2236: tempfilename);
1.59 schwarze 2237: }
1.50 kristaps 2238: }
1.14 schwarze 2239:
1.50 kristaps 2240: /*
2241: * This is straightforward stuff.
2242: * Open a database connection to a "temporary" database, then open a set
2243: * of prepared statements we'll use over and over again.
2244: * If "real" is set, we use the existing database; if not, we truncate a
2245: * temporary one.
2246: * Must be matched by dbclose().
2247: */
2248: static int
1.59 schwarze 2249: dbopen(int real)
1.50 kristaps 2250: {
1.115 schwarze 2251: const char *sql;
1.50 kristaps 2252: int rc, ofl;
1.12 schwarze 2253:
1.141 schwarze 2254: if (nodb)
1.197 schwarze 2255: return 1;
1.12 schwarze 2256:
1.115 schwarze 2257: *tempfilename = '\0';
1.63 schwarze 2258: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2259:
2260: if (real) {
2261: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2262: if (SQLITE_OK != rc) {
1.63 schwarze 2263: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2264: if (SQLITE_CANTOPEN != rc)
2265: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.197 schwarze 2266: return 0;
1.63 schwarze 2267: }
1.115 schwarze 2268: goto prepare_statements;
2269: }
2270:
2271: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2272:
1.115 schwarze 2273: remove(MANDOC_DB "~");
2274: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2275: if (SQLITE_OK == rc)
1.115 schwarze 2276: goto create_tables;
1.116 schwarze 2277: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2278: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2279: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.197 schwarze 2280: return 0;
1.50 kristaps 2281: }
1.12 schwarze 2282:
1.143 schwarze 2283: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2284: sizeof(tempfilename));
1.115 schwarze 2285: if (NULL == mkdtemp(tempfilename)) {
2286: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2287: say("", "&%s", tempfilename);
1.197 schwarze 2288: return 0;
1.115 schwarze 2289: }
1.143 schwarze 2290: (void)strlcat(tempfilename, "/" MANDOC_DB,
2291: sizeof(tempfilename));
1.115 schwarze 2292: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2293: if (SQLITE_OK != rc) {
1.59 schwarze 2294: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2295: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.197 schwarze 2296: return 0;
1.50 kristaps 2297: }
1.12 schwarze 2298:
1.115 schwarze 2299: create_tables:
1.81 schwarze 2300: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2301: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2302: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2303: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2304: ");\n"
2305: "\n"
1.81 schwarze 2306: "CREATE TABLE \"mlinks\" (\n"
2307: " \"sec\" TEXT NOT NULL,\n"
2308: " \"arch\" TEXT NOT NULL,\n"
2309: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2310: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2311: "ON DELETE CASCADE\n"
1.81 schwarze 2312: ");\n"
1.136 schwarze 2313: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2314: "\n"
1.133 schwarze 2315: "CREATE TABLE \"names\" (\n"
2316: " \"bits\" INTEGER NOT NULL,\n"
2317: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2318: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.175 schwarze 2319: "ON DELETE CASCADE,\n"
2320: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.133 schwarze 2321: ");\n"
2322: "\n"
1.50 kristaps 2323: "CREATE TABLE \"keys\" (\n"
2324: " \"bits\" INTEGER NOT NULL,\n"
2325: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2326: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2327: "ON DELETE CASCADE\n"
1.136 schwarze 2328: ");\n"
2329: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2330:
1.50 kristaps 2331: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2332: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2333: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2334: sqlite3_close(db);
1.197 schwarze 2335: return 0;
1.50 kristaps 2336: }
1.4 kristaps 2337:
1.57 schwarze 2338: prepare_statements:
1.146 schwarze 2339: if (SQLITE_OK != sqlite3_exec(db,
2340: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2341: exitcode = (int)MANDOCLEVEL_SYSERR;
2342: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2343: sqlite3_errmsg(db));
2344: sqlite3_close(db);
1.197 schwarze 2345: return 0;
1.146 schwarze 2346: }
2347:
1.137 schwarze 2348: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2349: "(SELECT pageid FROM mlinks WHERE "
2350: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2351: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2352: sql = "INSERT INTO mpages "
1.132 schwarze 2353: "(desc,form) VALUES (?,?)";
1.81 schwarze 2354: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2355: sql = "INSERT INTO mlinks "
1.104 schwarze 2356: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2357: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.175 schwarze 2358: sql = "SELECT bits FROM names where pageid = ?";
2359: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.133 schwarze 2360: sql = "INSERT INTO names "
2361: "(bits,name,pageid) VALUES (?,?,?)";
2362: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2363: sql = "INSERT INTO keys "
1.81 schwarze 2364: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2365: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2366:
2367: #ifndef __APPLE__
2368: /*
2369: * When opening a new database, we can turn off
2370: * synchronous mode for much better performance.
2371: */
2372:
1.146 schwarze 2373: if (real && SQLITE_OK != sqlite3_exec(db,
2374: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2375: exitcode = (int)MANDOCLEVEL_SYSERR;
2376: say(MANDOC_DB, "PRAGMA synchronous: %s",
1.162 schwarze 2377: sqlite3_errmsg(db));
1.146 schwarze 2378: sqlite3_close(db);
1.197 schwarze 2379: return 0;
1.146 schwarze 2380: }
1.70 schwarze 2381: #endif
2382:
1.197 schwarze 2383: return 1;
1.4 kristaps 2384: }
2385:
1.50 kristaps 2386: static int
1.165 schwarze 2387: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2388: {
1.59 schwarze 2389: static char startdir[PATH_MAX];
1.151 schwarze 2390: static int getcwd_status; /* 1 = ok, 2 = failure */
2391: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2392: char *cp;
1.4 kristaps 2393:
1.59 schwarze 2394: /*
1.151 schwarze 2395: * Remember the original working directory, if possible.
2396: * This will be needed if the second or a later directory
2397: * on the command line is given as a relative path.
2398: * Do not error out if the current directory is not
2399: * searchable: Maybe it won't be needed after all.
2400: */
2401: if (0 == getcwd_status) {
2402: if (NULL == getcwd(startdir, sizeof(startdir))) {
2403: getcwd_status = 2;
2404: (void)strlcpy(startdir, strerror(errno),
2405: sizeof(startdir));
2406: } else
2407: getcwd_status = 1;
2408: }
2409:
2410: /*
2411: * We are leaving the old base directory.
2412: * Do not use it any longer, not even for messages.
2413: */
2414: *basedir = '\0';
2415:
2416: /*
2417: * If and only if the directory was changed earlier and
2418: * the next directory to process is given as a relative path,
2419: * first go back, or bail out if that is impossible.
1.59 schwarze 2420: */
1.151 schwarze 2421: if (chdir_status && '/' != *targetdir) {
2422: if (2 == getcwd_status) {
1.59 schwarze 2423: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2424: say("", "getcwd: %s", startdir);
1.197 schwarze 2425: return 0;
1.59 schwarze 2426: }
1.151 schwarze 2427: if (-1 == chdir(startdir)) {
1.59 schwarze 2428: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2429: say("", "&chdir %s", startdir);
1.197 schwarze 2430: return 0;
1.59 schwarze 2431: }
2432: }
1.151 schwarze 2433:
2434: /*
2435: * Always resolve basedir to the canonicalized absolute
2436: * pathname and append a trailing slash, such that
2437: * we can reliably check whether files are inside.
2438: */
1.59 schwarze 2439: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2440: if (report_baddir || errno != ENOENT) {
2441: exitcode = (int)MANDOCLEVEL_BADARG;
2442: say("", "&%s: realpath", targetdir);
2443: }
1.197 schwarze 2444: return 0;
1.59 schwarze 2445: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2446: if (report_baddir || errno != ENOENT) {
2447: exitcode = (int)MANDOCLEVEL_BADARG;
2448: say("", "&chdir");
2449: }
1.197 schwarze 2450: return 0;
1.150 schwarze 2451: }
1.151 schwarze 2452: chdir_status = 1;
1.150 schwarze 2453: cp = strchr(basedir, '\0');
2454: if ('/' != cp[-1]) {
2455: if (cp - basedir >= PATH_MAX - 1) {
2456: exitcode = (int)MANDOCLEVEL_SYSERR;
2457: say("", "Filename too long");
1.197 schwarze 2458: return 0;
1.150 schwarze 2459: }
2460: *cp++ = '/';
2461: *cp = '\0';
1.4 kristaps 2462: }
1.197 schwarze 2463: return 1;
1.56 schwarze 2464: }
2465:
2466: static void
1.59 schwarze 2467: say(const char *file, const char *format, ...)
1.56 schwarze 2468: {
2469: va_list ap;
1.123 schwarze 2470: int use_errno;
1.56 schwarze 2471:
1.59 schwarze 2472: if ('\0' != *basedir)
2473: fprintf(stderr, "%s", basedir);
2474: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2475: fputc('/', stderr);
1.56 schwarze 2476: if ('\0' != *file)
1.59 schwarze 2477: fprintf(stderr, "%s", file);
2478:
1.123 schwarze 2479: use_errno = 1;
2480: if (NULL != format) {
2481: switch (*format) {
1.141 schwarze 2482: case '&':
1.123 schwarze 2483: format++;
2484: break;
1.141 schwarze 2485: case '\0':
1.123 schwarze 2486: format = NULL;
2487: break;
2488: default:
2489: use_errno = 0;
2490: break;
2491: }
2492: }
2493: if (NULL != format) {
2494: if ('\0' != *basedir || '\0' != *file)
2495: fputs(": ", stderr);
2496: va_start(ap, format);
2497: vfprintf(stderr, format, ap);
2498: va_end(ap);
2499: }
2500: if (use_errno) {
2501: if ('\0' != *basedir || '\0' != *file || NULL != format)
2502: fputs(": ", stderr);
1.59 schwarze 2503: perror(NULL);
1.123 schwarze 2504: } else
2505: fputc('\n', stderr);
1.1 kristaps 2506: }
CVSweb