Annotation of mandoc/mandocdb.c, Revision 1.216
1.216 ! schwarze 1: /* $Id: mandocdb.c,v 1.215 2016/01/08 17:48:09 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
1.1 kristaps 36: #include <getopt.h>
1.58 schwarze 37: #include <limits.h>
1.50 kristaps 38: #include <stddef.h>
1.59 schwarze 39: #include <stdio.h>
1.1 kristaps 40: #include <stdint.h>
41: #include <stdlib.h>
42: #include <string.h>
1.17 schwarze 43: #include <unistd.h>
1.1 kristaps 44:
1.50 kristaps 45: #include <sqlite3.h>
1.1 kristaps 46:
1.188 schwarze 47: #include "mandoc_aux.h"
1.202 schwarze 48: #include "mandoc_ohash.h"
1.188 schwarze 49: #include "mandoc.h"
50: #include "roff.h"
1.50 kristaps 51: #include "mdoc.h"
1.1 kristaps 52: #include "man.h"
1.187 schwarze 53: #include "manconf.h"
1.55 kristaps 54: #include "mansearch.h"
1.1 kristaps 55:
1.111 schwarze 56: extern int mansearch_keymax;
57: extern const char *const mansearch_keynames[];
58:
1.52 kristaps 59: #define SQL_EXEC(_v) \
60: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.134 schwarze 61: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.52 kristaps 62: #define SQL_BIND_TEXT(_s, _i, _v) \
63: if (SQLITE_OK != sqlite3_bind_text \
64: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.134 schwarze 65: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 66: #define SQL_BIND_INT(_s, _i, _v) \
67: if (SQLITE_OK != sqlite3_bind_int \
68: ((_s), (_i)++, (_v))) \
1.134 schwarze 69: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 70: #define SQL_BIND_INT64(_s, _i, _v) \
71: if (SQLITE_OK != sqlite3_bind_int64 \
72: ((_s), (_i)++, (_v))) \
1.134 schwarze 73: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 74: #define SQL_STEP(_s) \
75: if (SQLITE_DONE != sqlite3_step((_s))) \
1.134 schwarze 76: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 77:
1.50 kristaps 78: enum op {
79: OP_DEFAULT = 0, /* new dbs from dir list or default config */
80: OP_CONFFILE, /* new databases from custom config file */
81: OP_UPDATE, /* delete/add entries in existing database */
82: OP_DELETE, /* delete entries from existing database */
83: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 84: };
85:
1.50 kristaps 86: struct str {
1.78 schwarze 87: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 88: uint64_t mask; /* bitmask in sequence */
1.176 schwarze 89: char key[]; /* rendered text */
1.38 schwarze 90: };
91:
1.79 schwarze 92: struct inodev {
93: ino_t st_ino;
94: dev_t st_dev;
1.50 kristaps 95: };
1.5 kristaps 96:
1.78 schwarze 97: struct mpage {
1.79 schwarze 98: struct inodev inodev; /* used for hashing routine */
1.137 schwarze 99: int64_t pageid; /* pageid in mpages SQL table */
1.82 schwarze 100: char *sec; /* section from file content */
101: char *arch; /* architecture from file content */
102: char *title; /* title from file content */
103: char *desc; /* description from file content */
104: struct mlink *mlinks; /* singly linked list */
1.161 schwarze 105: int form; /* format from file content */
1.174 schwarze 106: int name_head_done;
1.82 schwarze 107: };
108:
109: struct mlink {
1.58 schwarze 110: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 111: char *dsec; /* section from directory */
112: char *arch; /* architecture from directory */
113: char *name; /* name from file name (not empty) */
114: char *fsec; /* section from file name suffix */
1.85 schwarze 115: struct mlink *next; /* singly linked list */
1.118 schwarze 116: struct mpage *mpage; /* parent */
1.161 schwarze 117: int dform; /* format from directory */
118: int fform; /* format from file name suffix */
1.124 schwarze 119: int gzip; /* filename has a .gz suffix */
1.3 kristaps 120: };
121:
1.50 kristaps 122: enum stmt {
1.81 schwarze 123: STMT_DELETE_PAGE = 0, /* delete mpage */
124: STMT_INSERT_PAGE, /* insert mpage */
125: STMT_INSERT_LINK, /* insert mlink */
1.133 schwarze 126: STMT_INSERT_NAME, /* insert name */
1.175 schwarze 127: STMT_SELECT_NAME, /* retrieve existing name flags */
1.81 schwarze 128: STMT_INSERT_KEY, /* insert parsed key */
1.50 kristaps 129: STMT__MAX
1.1 kristaps 130: };
131:
1.190 schwarze 132: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.189 schwarze 133: const struct roff_node *);
1.1 kristaps 134:
1.50 kristaps 135: struct mdoc_handler {
136: mdoc_fp fp; /* optional handler */
137: uint64_t mask; /* set unless handler returns 0 */
1.1 kristaps 138: };
139:
1.59 schwarze 140: static void dbclose(int);
1.176 schwarze 141: static void dbadd(struct mpage *);
1.118 schwarze 142: static void dbadd_mlink(const struct mlink *mlink);
1.169 schwarze 143: static void dbadd_mlink_name(const struct mlink *mlink);
1.59 schwarze 144: static int dbopen(int);
145: static void dbprune(void);
146: static void filescan(const char *);
1.84 schwarze 147: static void mlink_add(struct mlink *, const struct stat *);
1.131 schwarze 148: static void mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 149: static void mlink_free(struct mlink *);
1.89 schwarze 150: static void mlinks_undupe(struct mpage *);
1.78 schwarze 151: static void mpages_free(void);
1.176 schwarze 152: static void mpages_merge(struct mparse *);
1.130 schwarze 153: static void names_check(void);
1.124 schwarze 154: static void parse_cat(struct mpage *, int);
1.190 schwarze 155: static void parse_man(struct mpage *, const struct roff_meta *,
1.189 schwarze 156: const struct roff_node *);
1.190 schwarze 157: static void parse_mdoc(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 *);
1.211 schwarze 174: static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
175: const struct roff_node *);
1.190 schwarze 176: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.189 schwarze 177: const struct roff_node *);
1.112 schwarze 178: static void putkey(const struct mpage *, char *, uint64_t);
1.176 schwarze 179: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.78 schwarze 180: static void putmdockey(const struct mpage *,
1.189 schwarze 181: const struct roff_node *, uint64_t);
1.176 schwarze 182: static int render_string(char **, size_t *);
1.59 schwarze 183: static void say(const char *, const char *, ...);
1.165 schwarze 184: static int set_basedir(const char *, int);
1.59 schwarze 185: static int treescan(void);
1.50 kristaps 186: static size_t utf8(unsigned int, char [7]);
187:
1.115 schwarze 188: static char tempfilename[32];
1.102 schwarze 189: static int nodb; /* no database changes */
1.116 schwarze 190: static int mparse_options; /* abort the parse early */
1.141 schwarze 191: static int use_all; /* use all found files */
192: static int debug; /* print what we're doing */
193: static int warnings; /* warn about crap */
1.95 schwarze 194: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 195: static int exitcode; /* to be returned by main */
1.141 schwarze 196: static enum op op; /* operational mode */
1.59 schwarze 197: static char basedir[PATH_MAX]; /* current base directory */
1.78 schwarze 198: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 199: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 200: static struct ohash names; /* table of all names */
1.50 kristaps 201: static struct ohash strings; /* table of all strings */
202: static sqlite3 *db = NULL; /* current database */
203: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.133 schwarze 204: static uint64_t name_mask;
1.25 schwarze 205:
206: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 207: { NULL, 0 }, /* Ap */
208: { NULL, 0 }, /* Dd */
209: { NULL, 0 }, /* Dt */
210: { NULL, 0 }, /* Os */
211: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
212: { parse_mdoc_head, TYPE_Ss }, /* Ss */
213: { NULL, 0 }, /* Pp */
214: { NULL, 0 }, /* D1 */
215: { NULL, 0 }, /* Dl */
216: { NULL, 0 }, /* Bd */
217: { NULL, 0 }, /* Ed */
218: { NULL, 0 }, /* Bl */
219: { NULL, 0 }, /* El */
220: { NULL, 0 }, /* It */
221: { NULL, 0 }, /* Ad */
222: { NULL, TYPE_An }, /* An */
223: { NULL, TYPE_Ar }, /* Ar */
224: { NULL, TYPE_Cd }, /* Cd */
225: { NULL, TYPE_Cm }, /* Cm */
226: { NULL, TYPE_Dv }, /* Dv */
227: { NULL, TYPE_Er }, /* Er */
228: { NULL, TYPE_Ev }, /* Ev */
229: { NULL, 0 }, /* Ex */
230: { NULL, TYPE_Fa }, /* Fa */
231: { parse_mdoc_Fd, 0 }, /* Fd */
232: { NULL, TYPE_Fl }, /* Fl */
233: { parse_mdoc_Fn, 0 }, /* Fn */
234: { NULL, TYPE_Ft }, /* Ft */
235: { NULL, TYPE_Ic }, /* Ic */
1.92 schwarze 236: { NULL, TYPE_In }, /* In */
1.68 schwarze 237: { NULL, TYPE_Li }, /* Li */
1.132 schwarze 238: { parse_mdoc_Nd, 0 }, /* Nd */
1.129 schwarze 239: { parse_mdoc_Nm, 0 }, /* Nm */
1.68 schwarze 240: { NULL, 0 }, /* Op */
241: { NULL, 0 }, /* Ot */
242: { NULL, TYPE_Pa }, /* Pa */
243: { NULL, 0 }, /* Rv */
1.92 schwarze 244: { NULL, TYPE_St }, /* St */
1.211 schwarze 245: { parse_mdoc_Va, TYPE_Va }, /* Va */
246: { parse_mdoc_Va, TYPE_Vt }, /* Vt */
1.68 schwarze 247: { parse_mdoc_Xr, 0 }, /* Xr */
248: { NULL, 0 }, /* %A */
249: { NULL, 0 }, /* %B */
250: { NULL, 0 }, /* %D */
251: { NULL, 0 }, /* %I */
252: { NULL, 0 }, /* %J */
253: { NULL, 0 }, /* %N */
254: { NULL, 0 }, /* %O */
255: { NULL, 0 }, /* %P */
256: { NULL, 0 }, /* %R */
257: { NULL, 0 }, /* %T */
258: { NULL, 0 }, /* %V */
259: { NULL, 0 }, /* Ac */
260: { NULL, 0 }, /* Ao */
261: { NULL, 0 }, /* Aq */
262: { NULL, TYPE_At }, /* At */
263: { NULL, 0 }, /* Bc */
264: { NULL, 0 }, /* Bf */
265: { NULL, 0 }, /* Bo */
266: { NULL, 0 }, /* Bq */
267: { NULL, TYPE_Bsx }, /* Bsx */
268: { NULL, TYPE_Bx }, /* Bx */
269: { NULL, 0 }, /* Db */
270: { NULL, 0 }, /* Dc */
271: { NULL, 0 }, /* Do */
272: { NULL, 0 }, /* Dq */
273: { NULL, 0 }, /* Ec */
274: { NULL, 0 }, /* Ef */
275: { NULL, TYPE_Em }, /* Em */
276: { NULL, 0 }, /* Eo */
277: { NULL, TYPE_Fx }, /* Fx */
278: { NULL, TYPE_Ms }, /* Ms */
279: { NULL, 0 }, /* No */
280: { NULL, 0 }, /* Ns */
281: { NULL, TYPE_Nx }, /* Nx */
282: { NULL, TYPE_Ox }, /* Ox */
283: { NULL, 0 }, /* Pc */
284: { NULL, 0 }, /* Pf */
285: { NULL, 0 }, /* Po */
286: { NULL, 0 }, /* Pq */
287: { NULL, 0 }, /* Qc */
288: { NULL, 0 }, /* Ql */
289: { NULL, 0 }, /* Qo */
290: { NULL, 0 }, /* Qq */
291: { NULL, 0 }, /* Re */
292: { NULL, 0 }, /* Rs */
293: { NULL, 0 }, /* Sc */
294: { NULL, 0 }, /* So */
295: { NULL, 0 }, /* Sq */
296: { NULL, 0 }, /* Sm */
297: { NULL, 0 }, /* Sx */
298: { NULL, TYPE_Sy }, /* Sy */
299: { NULL, TYPE_Tn }, /* Tn */
300: { NULL, 0 }, /* Ux */
301: { NULL, 0 }, /* Xc */
302: { NULL, 0 }, /* Xo */
1.173 schwarze 303: { parse_mdoc_Fo, 0 }, /* Fo */
1.68 schwarze 304: { NULL, 0 }, /* Fc */
305: { NULL, 0 }, /* Oo */
306: { NULL, 0 }, /* Oc */
307: { NULL, 0 }, /* Bk */
308: { NULL, 0 }, /* Ek */
309: { NULL, 0 }, /* Bt */
310: { NULL, 0 }, /* Hf */
311: { NULL, 0 }, /* Fr */
312: { NULL, 0 }, /* Ud */
313: { NULL, TYPE_Lb }, /* Lb */
314: { NULL, 0 }, /* Lp */
315: { NULL, TYPE_Lk }, /* Lk */
316: { NULL, TYPE_Mt }, /* Mt */
317: { NULL, 0 }, /* Brq */
318: { NULL, 0 }, /* Bro */
319: { NULL, 0 }, /* Brc */
320: { NULL, 0 }, /* %C */
321: { NULL, 0 }, /* Es */
322: { NULL, 0 }, /* En */
323: { NULL, TYPE_Dx }, /* Dx */
324: { NULL, 0 }, /* %Q */
325: { NULL, 0 }, /* br */
326: { NULL, 0 }, /* sp */
327: { NULL, 0 }, /* %U */
328: { NULL, 0 }, /* Ta */
1.166 schwarze 329: { NULL, 0 }, /* ll */
1.1 kristaps 330: };
331:
1.141 schwarze 332:
1.1 kristaps 333: int
1.179 schwarze 334: mandocdb(int argc, char *argv[])
1.1 kristaps 335: {
1.187 schwarze 336: struct manconf conf;
337: struct mparse *mp;
1.206 schwarze 338: const char *path_arg, *progname;
1.187 schwarze 339: size_t j, sz;
1.59 schwarze 340: int ch, i;
1.50 kristaps 341:
1.207 schwarze 342: #if HAVE_PLEDGE
343: if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
1.214 schwarze 344: warn("pledge");
1.207 schwarze 345: return (int)MANDOCLEVEL_SYSERR;
346: }
347: #endif
348:
1.187 schwarze 349: memset(&conf, 0, sizeof(conf));
1.50 kristaps 350: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
351:
352: /*
1.141 schwarze 353: * We accept a few different invocations.
1.50 kristaps 354: * The CHECKOP macro makes sure that invocation styles don't
355: * clobber each other.
356: */
357: #define CHECKOP(_op, _ch) do \
358: if (OP_DEFAULT != (_op)) { \
1.198 schwarze 359: warnx("-%c: Conflicting option", (_ch)); \
1.50 kristaps 360: goto usage; \
361: } while (/*CONSTCOND*/0)
1.10 kristaps 362:
1.59 schwarze 363: path_arg = NULL;
1.38 schwarze 364: op = OP_DEFAULT;
1.1 kristaps 365:
1.126 schwarze 366: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 367: switch (ch) {
1.141 schwarze 368: case 'a':
1.12 schwarze 369: use_all = 1;
370: break;
1.141 schwarze 371: case 'C':
1.50 kristaps 372: CHECKOP(op, ch);
1.59 schwarze 373: path_arg = optarg;
1.38 schwarze 374: op = OP_CONFFILE;
1.34 schwarze 375: break;
1.141 schwarze 376: case 'D':
1.125 schwarze 377: debug++;
378: break;
1.141 schwarze 379: case 'd':
1.50 kristaps 380: CHECKOP(op, ch);
1.59 schwarze 381: path_arg = optarg;
1.5 kristaps 382: op = OP_UPDATE;
383: break;
1.141 schwarze 384: case 'n':
1.50 kristaps 385: nodb = 1;
386: break;
1.141 schwarze 387: case 'p':
1.126 schwarze 388: warnings = 1;
389: break;
1.141 schwarze 390: case 'Q':
1.116 schwarze 391: mparse_options |= MPARSE_QUICK;
1.102 schwarze 392: break;
1.141 schwarze 393: case 'T':
1.95 schwarze 394: if (strcmp(optarg, "utf8")) {
1.198 schwarze 395: warnx("-T%s: Unsupported output format",
396: optarg);
1.95 schwarze 397: goto usage;
398: }
399: write_utf8 = 1;
400: break;
1.141 schwarze 401: case 't':
1.50 kristaps 402: CHECKOP(op, ch);
1.38 schwarze 403: dup2(STDOUT_FILENO, STDERR_FILENO);
404: op = OP_TEST;
1.50 kristaps 405: nodb = warnings = 1;
1.38 schwarze 406: break;
1.141 schwarze 407: case 'u':
1.50 kristaps 408: CHECKOP(op, ch);
1.59 schwarze 409: path_arg = optarg;
1.5 kristaps 410: op = OP_DELETE;
411: break;
1.141 schwarze 412: case 'v':
1.126 schwarze 413: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 414: break;
1.1 kristaps 415: default:
1.38 schwarze 416: goto usage;
1.1 kristaps 417: }
418:
419: argc -= optind;
420: argv += optind;
421:
1.207 schwarze 422: #if HAVE_PLEDGE
1.210 schwarze 423: if (nodb) {
424: if (pledge("stdio rpath", NULL) == -1) {
1.214 schwarze 425: warn("pledge");
1.210 schwarze 426: return (int)MANDOCLEVEL_SYSERR;
427: }
1.207 schwarze 428: }
429: #endif
430:
1.38 schwarze 431: if (OP_CONFFILE == op && argc > 0) {
1.198 schwarze 432: warnx("-C: Too many arguments");
1.38 schwarze 433: goto usage;
434: }
435:
1.59 schwarze 436: exitcode = (int)MANDOCLEVEL_OK;
1.203 schwarze 437: mchars_alloc();
438: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
1.202 schwarze 439: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
440: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.50 kristaps 441:
442: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 443:
1.50 kristaps 444: /*
1.150 schwarze 445: * Most of these deal with a specific directory.
1.140 schwarze 446: * Jump into that directory first.
1.50 kristaps 447: */
1.165 schwarze 448: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.50 kristaps 449: goto out;
1.140 schwarze 450:
1.138 schwarze 451: if (dbopen(1)) {
1.140 schwarze 452: /*
453: * The existing database is usable. Process
454: * all files specified on the command-line.
455: */
1.207 schwarze 456: #if HAVE_PLEDGE
1.210 schwarze 457: if (!nodb) {
458: if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
1.214 schwarze 459: warn("pledge");
1.210 schwarze 460: exitcode = (int)MANDOCLEVEL_SYSERR;
461: goto out;
462: }
1.207 schwarze 463: }
464: #endif
1.140 schwarze 465: use_all = 1;
1.138 schwarze 466: for (i = 0; i < argc; i++)
467: filescan(argv[i]);
468: if (OP_TEST != op)
469: dbprune();
470: } else {
471: /*
472: * Database missing or corrupt.
473: * Recreate from scratch.
474: */
1.140 schwarze 475: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 476: op = OP_DEFAULT;
477: if (0 == treescan())
478: goto out;
479: if (0 == dbopen(0))
480: goto out;
481: }
1.50 kristaps 482: if (OP_DELETE != op)
1.176 schwarze 483: mpages_merge(mp);
1.138 schwarze 484: dbclose(OP_DEFAULT == op ? 0 : 1);
1.50 kristaps 485: } else {
486: /*
487: * If we have arguments, use them as our manpaths.
488: * If we don't, grok from manpath(1) or however else
1.187 schwarze 489: * manconf_parse() wants to do it.
1.50 kristaps 490: */
491: if (argc > 0) {
1.187 schwarze 492: conf.manpath.paths = mandoc_reallocarray(NULL,
1.144 schwarze 493: argc, sizeof(char *));
1.187 schwarze 494: conf.manpath.sz = (size_t)argc;
1.50 kristaps 495: for (i = 0; i < argc; i++)
1.187 schwarze 496: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.50 kristaps 497: } else
1.187 schwarze 498: manconf_parse(&conf, path_arg, NULL, NULL);
1.127 schwarze 499:
1.187 schwarze 500: if (conf.manpath.sz == 0) {
1.127 schwarze 501: exitcode = (int)MANDOCLEVEL_BADARG;
502: say("", "Empty manpath");
503: }
1.50 kristaps 504:
505: /*
1.69 schwarze 506: * First scan the tree rooted at a base directory, then
507: * build a new database and finally move it into place.
1.50 kristaps 508: * Ignore zero-length directories and strip trailing
509: * slashes.
510: */
1.187 schwarze 511: for (j = 0; j < conf.manpath.sz; j++) {
512: sz = strlen(conf.manpath.paths[j]);
513: if (sz && conf.manpath.paths[j][sz - 1] == '/')
514: conf.manpath.paths[j][--sz] = '\0';
1.50 kristaps 515: if (0 == sz)
516: continue;
1.66 schwarze 517:
518: if (j) {
1.202 schwarze 519: mandoc_ohash_init(&mpages, 6,
520: offsetof(struct mpage, inodev));
521: mandoc_ohash_init(&mlinks, 6,
522: offsetof(struct mlink, file));
1.66 schwarze 523: }
524:
1.187 schwarze 525: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.165 schwarze 526: continue;
1.59 schwarze 527: if (0 == treescan())
1.165 schwarze 528: continue;
1.59 schwarze 529: if (0 == dbopen(0))
1.165 schwarze 530: continue;
1.52 kristaps 531:
1.176 schwarze 532: mpages_merge(mp);
1.147 schwarze 533: if (warnings && !nodb &&
1.130 schwarze 534: ! (MPARSE_QUICK & mparse_options))
535: names_check();
1.59 schwarze 536: dbclose(0);
1.66 schwarze 537:
1.187 schwarze 538: if (j + 1 < conf.manpath.sz) {
1.80 schwarze 539: mpages_free();
1.78 schwarze 540: ohash_delete(&mpages);
1.83 schwarze 541: ohash_delete(&mlinks);
1.66 schwarze 542: }
1.50 kristaps 543: }
544: }
545: out:
1.187 schwarze 546: manconf_free(&conf);
1.168 schwarze 547: mparse_free(mp);
1.203 schwarze 548: mchars_free();
1.80 schwarze 549: mpages_free();
1.78 schwarze 550: ohash_delete(&mpages);
1.83 schwarze 551: ohash_delete(&mlinks);
1.197 schwarze 552: return exitcode;
1.50 kristaps 553: usage:
1.206 schwarze 554: progname = getprogname();
1.126 schwarze 555: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
556: " %s [-aDnpQ] [-Tutf8] dir ...\n"
557: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
558: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 559: " %s [-Q] -t file ...\n",
1.206 schwarze 560: progname, progname, progname, progname, progname);
1.50 kristaps 561:
1.197 schwarze 562: return (int)MANDOCLEVEL_BADARG;
1.50 kristaps 563: }
564:
565: /*
1.59 schwarze 566: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 567: * We use fts(), scanning directory parts along the way for clues to our
568: * section and architecture.
569: *
570: * If use_all has been specified, grok all files.
571: * If not, sanitise paths to the following:
572: *
1.141 schwarze 573: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 574: * or
575: * [./]cat<section>[/<arch>]/<name>.0
576: *
1.216 ! schwarze 577: * TODO: accommodate for multi-language directories.
1.50 kristaps 578: */
579: static int
1.59 schwarze 580: treescan(void)
1.50 kristaps 581: {
1.139 schwarze 582: char buf[PATH_MAX];
1.50 kristaps 583: FTS *f;
584: FTSENT *ff;
1.84 schwarze 585: struct mlink *mlink;
1.124 schwarze 586: int dform, gzip;
1.94 schwarze 587: char *dsec, *arch, *fsec, *cp;
588: const char *path;
1.50 kristaps 589: const char *argv[2];
590:
591: argv[0] = ".";
592: argv[1] = (char *)NULL;
593:
1.139 schwarze 594: f = fts_open((char * const *)argv,
595: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.200 schwarze 596: if (f == NULL) {
1.59 schwarze 597: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 598: say("", "&fts_open");
1.197 schwarze 599: return 0;
1.50 kristaps 600: }
601:
602: dsec = arch = NULL;
603: dform = FORM_NONE;
604:
1.200 schwarze 605: while ((ff = fts_read(f)) != NULL) {
1.50 kristaps 606: path = ff->fts_path + 2;
1.139 schwarze 607: switch (ff->fts_info) {
608:
609: /*
610: * Symbolic links require various sanity checks,
611: * then get handled just like regular files.
612: */
1.141 schwarze 613: case FTS_SL:
1.200 schwarze 614: if (realpath(path, buf) == NULL) {
1.139 schwarze 615: if (warnings)
616: say(path, "&realpath");
617: continue;
618: }
1.184 schwarze 619: if (strstr(buf, basedir) != buf
620: #ifdef HOMEBREWDIR
621: && strstr(buf, HOMEBREWDIR) != buf
622: #endif
623: ) {
1.139 schwarze 624: if (warnings) say("",
625: "%s: outside base directory", buf);
626: continue;
627: }
628: /* Use logical inode to avoid mpages dupe. */
1.200 schwarze 629: if (stat(path, ff->fts_statp) == -1) {
1.139 schwarze 630: if (warnings)
631: say(path, "&stat");
632: continue;
633: }
634: /* FALLTHROUGH */
635:
1.50 kristaps 636: /*
1.83 schwarze 637: * If we're a regular file, add an mlink by using the
1.50 kristaps 638: * stored directory data and handling the filename.
639: */
1.141 schwarze 640: case FTS_F:
1.200 schwarze 641: if ( ! strcmp(path, MANDOC_DB))
1.60 schwarze 642: continue;
1.50 kristaps 643: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 644: if (warnings)
1.59 schwarze 645: say(path, "Extraneous file");
1.50 kristaps 646: continue;
1.124 schwarze 647: }
648: gzip = 0;
649: fsec = NULL;
1.200 schwarze 650: while (fsec == NULL) {
1.124 schwarze 651: fsec = strrchr(ff->fts_name, '.');
1.200 schwarze 652: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.124 schwarze 653: break;
654: gzip = 1;
655: *fsec = '\0';
656: fsec = NULL;
657: }
1.200 schwarze 658: if (fsec == NULL) {
1.60 schwarze 659: if ( ! use_all) {
1.56 schwarze 660: if (warnings)
1.60 schwarze 661: say(path,
662: "No filename suffix");
1.50 kristaps 663: continue;
664: }
1.200 schwarze 665: } else if ( ! strcmp(++fsec, "html")) {
1.60 schwarze 666: if (warnings)
667: say(path, "Skip html");
668: continue;
1.200 schwarze 669: } else if ( ! strcmp(fsec, "ps")) {
1.60 schwarze 670: if (warnings)
671: say(path, "Skip ps");
672: continue;
1.200 schwarze 673: } else if ( ! strcmp(fsec, "pdf")) {
1.60 schwarze 674: if (warnings)
675: say(path, "Skip pdf");
676: continue;
677: } else if ( ! use_all &&
1.200 schwarze 678: ((dform == FORM_SRC &&
1.185 schwarze 679: strncmp(fsec, dsec, strlen(dsec))) ||
1.200 schwarze 680: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.60 schwarze 681: if (warnings)
682: say(path, "Wrong filename suffix");
683: continue;
1.66 schwarze 684: } else
1.84 schwarze 685: fsec[-1] = '\0';
1.94 schwarze 686:
1.84 schwarze 687: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 688: if (strlcpy(mlink->file, path,
689: sizeof(mlink->file)) >=
690: sizeof(mlink->file)) {
691: say(path, "Filename too long");
692: free(mlink);
693: continue;
694: }
1.84 schwarze 695: mlink->dform = dform;
1.94 schwarze 696: mlink->dsec = dsec;
697: mlink->arch = arch;
698: mlink->name = ff->fts_name;
699: mlink->fsec = fsec;
1.124 schwarze 700: mlink->gzip = gzip;
1.84 schwarze 701: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 702: continue;
1.139 schwarze 703:
1.141 schwarze 704: case FTS_D:
705: case FTS_DP:
1.139 schwarze 706: break;
707:
708: default:
1.60 schwarze 709: if (warnings)
710: say(path, "Not a regular file");
1.50 kristaps 711: continue;
1.60 schwarze 712: }
1.1 kristaps 713:
1.50 kristaps 714: switch (ff->fts_level) {
1.141 schwarze 715: case 0:
1.50 kristaps 716: /* Ignore the root directory. */
717: break;
1.141 schwarze 718: case 1:
1.50 kristaps 719: /*
720: * This might contain manX/ or catX/.
721: * Try to infer this from the name.
722: * If we're not in use_all, enforce it.
723: */
724: cp = ff->fts_name;
1.200 schwarze 725: if (ff->fts_info == FTS_DP) {
726: dform = FORM_NONE;
727: dsec = NULL;
1.50 kristaps 728: break;
1.200 schwarze 729: }
1.1 kristaps 730:
1.200 schwarze 731: if ( ! strncmp(cp, "man", 3)) {
1.50 kristaps 732: dform = FORM_SRC;
1.66 schwarze 733: dsec = cp + 3;
1.200 schwarze 734: } else if ( ! strncmp(cp, "cat", 3)) {
1.50 kristaps 735: dform = FORM_CAT;
1.66 schwarze 736: dsec = cp + 3;
1.94 schwarze 737: } else {
738: dform = FORM_NONE;
739: dsec = NULL;
1.50 kristaps 740: }
1.1 kristaps 741:
1.200 schwarze 742: if (dsec != NULL || use_all)
1.50 kristaps 743: break;
1.1 kristaps 744:
1.56 schwarze 745: if (warnings)
1.59 schwarze 746: say(path, "Unknown directory part");
1.50 kristaps 747: fts_set(f, ff, FTS_SKIP);
748: break;
1.141 schwarze 749: case 2:
1.50 kristaps 750: /*
751: * Possibly our architecture.
752: * If we're descending, keep tabs on it.
753: */
1.200 schwarze 754: if (ff->fts_info != FTS_DP && dsec != NULL)
1.66 schwarze 755: arch = ff->fts_name;
1.94 schwarze 756: else
757: arch = NULL;
1.50 kristaps 758: break;
759: default:
1.200 schwarze 760: if (ff->fts_info == FTS_DP || use_all)
1.50 kristaps 761: break;
1.56 schwarze 762: if (warnings)
1.59 schwarze 763: say(path, "Extraneous directory part");
1.50 kristaps 764: fts_set(f, ff, FTS_SKIP);
765: break;
1.5 kristaps 766: }
1.50 kristaps 767: }
768:
769: fts_close(f);
1.197 schwarze 770: return 1;
1.50 kristaps 771: }
1.5 kristaps 772:
1.50 kristaps 773: /*
1.89 schwarze 774: * Add a file to the mlinks table.
1.50 kristaps 775: * Do not verify that it's a "valid" looking manpage (we'll do that
776: * later).
777: *
778: * Try to infer the manual section, architecture, and page name from the
779: * path, assuming it looks like
780: *
1.141 schwarze 781: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 782: * or
783: * [./]cat<section>[/<arch>]/<name>.0
784: *
785: * See treescan() for the fts(3) version of this.
786: */
787: static void
1.59 schwarze 788: filescan(const char *file)
1.50 kristaps 789: {
1.59 schwarze 790: char buf[PATH_MAX];
1.84 schwarze 791: struct stat st;
792: struct mlink *mlink;
1.59 schwarze 793: char *p, *start;
1.5 kristaps 794:
1.50 kristaps 795: assert(use_all);
1.5 kristaps 796:
1.50 kristaps 797: if (0 == strncmp(file, "./", 2))
798: file += 2;
1.5 kristaps 799:
1.139 schwarze 800: /*
801: * We have to do lstat(2) before realpath(3) loses
802: * the information whether this is a symbolic link.
803: * We need to know that because for symbolic links,
804: * we want to use the orginal file name, while for
805: * regular files, we want to use the real path.
806: */
807: if (-1 == lstat(file, &st)) {
808: exitcode = (int)MANDOCLEVEL_BADARG;
809: say(file, "&lstat");
810: return;
811: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
812: exitcode = (int)MANDOCLEVEL_BADARG;
813: say(file, "Not a regular file");
814: return;
815: }
816:
817: /*
818: * We have to resolve the file name to the real path
819: * in any case for the base directory check.
820: */
1.59 schwarze 821: if (NULL == realpath(file, buf)) {
822: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 823: say(file, "&realpath");
1.59 schwarze 824: return;
1.106 schwarze 825: }
826:
1.150 schwarze 827: if (OP_TEST == op)
1.106 schwarze 828: start = buf;
1.150 schwarze 829: else if (strstr(buf, basedir) == buf)
830: start = buf + strlen(basedir);
1.184 schwarze 831: #ifdef HOMEBREWDIR
832: else if (strstr(buf, HOMEBREWDIR) == buf)
833: start = buf;
834: #endif
1.106 schwarze 835: else {
1.59 schwarze 836: exitcode = (int)MANDOCLEVEL_BADARG;
837: say("", "%s: outside base directory", buf);
838: return;
1.106 schwarze 839: }
840:
1.139 schwarze 841: /*
842: * Now we are sure the file is inside our tree.
843: * If it is a symbolic link, ignore the real path
844: * and use the original name.
845: * This implies passing stuff like "cat1/../man1/foo.1"
846: * on the command line won't work. So don't do that.
847: * Note the stat(2) can still fail if the link target
848: * doesn't exist.
849: */
850: if (S_IFLNK & st.st_mode) {
851: if (-1 == stat(buf, &st)) {
852: exitcode = (int)MANDOCLEVEL_BADARG;
853: say(file, "&stat");
854: return;
855: }
1.143 schwarze 856: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
857: say(file, "Filename too long");
858: return;
859: }
1.150 schwarze 860: start = buf;
861: if (OP_TEST != op && strstr(buf, basedir) == buf)
862: start += strlen(basedir);
1.50 kristaps 863: }
1.106 schwarze 864:
1.84 schwarze 865: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 866: mlink->dform = FORM_NONE;
1.143 schwarze 867: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
868: sizeof(mlink->file)) {
869: say(start, "Filename too long");
1.180 schwarze 870: free(mlink);
1.143 schwarze 871: return;
872: }
1.17 schwarze 873:
1.50 kristaps 874: /*
875: * First try to guess our directory structure.
876: * If we find a separator, try to look for man* or cat*.
877: * If we find one of these and what's underneath is a directory,
878: * assume it's an architecture.
879: */
880: if (NULL != (p = strchr(start, '/'))) {
881: *p++ = '\0';
882: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 883: mlink->dform = FORM_SRC;
1.94 schwarze 884: mlink->dsec = start + 3;
1.50 kristaps 885: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 886: mlink->dform = FORM_CAT;
1.94 schwarze 887: mlink->dsec = start + 3;
1.17 schwarze 888: }
1.5 kristaps 889:
1.50 kristaps 890: start = p;
1.84 schwarze 891: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 892: *p++ = '\0';
1.94 schwarze 893: mlink->arch = start;
1.50 kristaps 894: start = p;
1.84 schwarze 895: }
1.50 kristaps 896: }
897:
898: /*
899: * Now check the file suffix.
900: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
901: */
902: p = strrchr(start, '\0');
903: while (p-- > start && '/' != *p && '.' != *p)
904: /* Loop. */ ;
905:
906: if ('.' == *p) {
907: *p++ = '\0';
1.94 schwarze 908: mlink->fsec = p;
1.5 kristaps 909: }
910:
1.10 kristaps 911: /*
1.50 kristaps 912: * Now try to parse the name.
913: * Use the filename portion of the path.
1.10 kristaps 914: */
1.84 schwarze 915: mlink->name = start;
1.50 kristaps 916: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 917: mlink->name = p + 1;
1.50 kristaps 918: *p = '\0';
1.84 schwarze 919: }
920: mlink_add(mlink, &st);
1.50 kristaps 921: }
1.5 kristaps 922:
1.50 kristaps 923: static void
1.84 schwarze 924: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 925: {
1.83 schwarze 926: struct inodev inodev;
1.78 schwarze 927: struct mpage *mpage;
1.83 schwarze 928: unsigned int slot;
1.50 kristaps 929:
1.84 schwarze 930: assert(NULL != mlink->file);
1.50 kristaps 931:
1.94 schwarze 932: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
933: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
934: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
935: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 936:
937: if ('0' == *mlink->fsec) {
938: free(mlink->fsec);
939: mlink->fsec = mandoc_strdup(mlink->dsec);
940: mlink->fform = FORM_CAT;
941: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
942: mlink->fform = FORM_SRC;
1.74 schwarze 943: else
1.84 schwarze 944: mlink->fform = FORM_NONE;
1.82 schwarze 945:
1.83 schwarze 946: slot = ohash_qlookup(&mlinks, mlink->file);
947: assert(NULL == ohash_find(&mlinks, slot));
948: ohash_insert(&mlinks, slot, mlink);
949:
1.186 schwarze 950: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 951: inodev.st_ino = st->st_ino;
952: inodev.st_dev = st->st_dev;
953: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
954: sizeof(struct inodev), inodev.st_ino);
955: mpage = ohash_find(&mpages, slot);
956: if (NULL == mpage) {
957: mpage = mandoc_calloc(1, sizeof(struct mpage));
958: mpage->inodev.st_ino = inodev.st_ino;
959: mpage->inodev.st_dev = inodev.st_dev;
960: ohash_insert(&mpages, slot, mpage);
961: } else
1.85 schwarze 962: mlink->next = mpage->mlinks;
1.82 schwarze 963: mpage->mlinks = mlink;
1.118 schwarze 964: mlink->mpage = mpage;
1.82 schwarze 965: }
966:
967: static void
968: mlink_free(struct mlink *mlink)
969: {
970:
971: free(mlink->dsec);
972: free(mlink->arch);
973: free(mlink->name);
974: free(mlink->fsec);
975: free(mlink);
1.50 kristaps 976: }
1.3 kristaps 977:
1.50 kristaps 978: static void
1.78 schwarze 979: mpages_free(void)
1.50 kristaps 980: {
1.78 schwarze 981: struct mpage *mpage;
1.82 schwarze 982: struct mlink *mlink;
1.80 schwarze 983: unsigned int slot;
1.3 kristaps 984:
1.80 schwarze 985: mpage = ohash_first(&mpages, &slot);
986: while (NULL != mpage) {
1.82 schwarze 987: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 988: mpage->mlinks = mlink->next;
1.82 schwarze 989: mlink_free(mlink);
990: }
1.78 schwarze 991: free(mpage->sec);
992: free(mpage->arch);
1.82 schwarze 993: free(mpage->title);
994: free(mpage->desc);
1.78 schwarze 995: free(mpage);
1.80 schwarze 996: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 997: }
998: }
1.38 schwarze 999:
1.50 kristaps 1000: /*
1.89 schwarze 1001: * For each mlink to the mpage, check whether the path looks like
1002: * it is formatted, and if it does, check whether a source manual
1003: * exists by the same name, ignoring the suffix.
1004: * If both conditions hold, drop the mlink.
1005: */
1006: static void
1007: mlinks_undupe(struct mpage *mpage)
1008: {
1009: char buf[PATH_MAX];
1010: struct mlink **prev;
1011: struct mlink *mlink;
1012: char *bufp;
1013:
1014: mpage->form = FORM_CAT;
1.90 schwarze 1015: prev = &mpage->mlinks;
1016: while (NULL != (mlink = *prev)) {
1.89 schwarze 1017: if (FORM_CAT != mlink->dform) {
1018: mpage->form = FORM_NONE;
1.90 schwarze 1019: goto nextlink;
1.89 schwarze 1020: }
1.143 schwarze 1021: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1022: bufp = strstr(buf, "cat");
1023: assert(NULL != bufp);
1024: memcpy(bufp, "man", 3);
1025: if (NULL != (bufp = strrchr(buf, '.')))
1026: *++bufp = '\0';
1.143 schwarze 1027: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1028: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1029: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1030: goto nextlink;
1.89 schwarze 1031: if (warnings)
1032: say(mlink->file, "Man source exists: %s", buf);
1033: if (use_all)
1.90 schwarze 1034: goto nextlink;
1.89 schwarze 1035: *prev = mlink->next;
1036: mlink_free(mlink);
1.90 schwarze 1037: continue;
1038: nextlink:
1039: prev = &(*prev)->next;
1.89 schwarze 1040: }
1041: }
1042:
1.131 schwarze 1043: static void
1.93 schwarze 1044: mlink_check(struct mpage *mpage, struct mlink *mlink)
1045: {
1.131 schwarze 1046: struct str *str;
1047: unsigned int slot;
1.93 schwarze 1048:
1049: /*
1050: * Check whether the manual section given in a file
1051: * agrees with the directory where the file is located.
1052: * Some manuals have suffixes like (3p) on their
1053: * section number either inside the file or in the
1054: * directory name, some are linked into more than one
1055: * section, like encrypt(1) = makekey(8).
1056: */
1057:
1058: if (FORM_SRC == mpage->form &&
1.131 schwarze 1059: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1060: say(mlink->file, "Section \"%s\" manual in %s directory",
1061: mpage->sec, mlink->dsec);
1062:
1063: /*
1064: * Manual page directories exist for each kernel
1065: * architecture as returned by machine(1).
1066: * However, many manuals only depend on the
1067: * application architecture as returned by arch(1).
1068: * For example, some (2/ARM) manuals are shared
1069: * across the "armish" and "zaurus" kernel
1070: * architectures.
1071: * A few manuals are even shared across completely
1072: * different architectures, for example fdformat(1)
1073: * on amd64, i386, sparc, and sparc64.
1074: */
1075:
1.131 schwarze 1076: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1077: say(mlink->file, "Architecture \"%s\" manual in "
1078: "\"%s\" directory", mpage->arch, mlink->arch);
1079:
1.131 schwarze 1080: /*
1081: * XXX
1.133 schwarze 1082: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1083: */
1084:
1085: if (FORM_CAT == mpage->form)
1086: return;
1087:
1088: /*
1089: * Check whether this mlink
1090: * appears as a name in the NAME section.
1091: */
1.93 schwarze 1092:
1.133 schwarze 1093: slot = ohash_qlookup(&names, mlink->name);
1094: str = ohash_find(&names, slot);
1.131 schwarze 1095: assert(NULL != str);
1.133 schwarze 1096: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1097: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1098: }
1099:
1.89 schwarze 1100: /*
1.80 schwarze 1101: * Run through the files in the global vector "mpages"
1102: * and add them to the database specified in "basedir".
1.50 kristaps 1103: *
1104: * This handles the parsing scheme itself, using the cues of directory
1105: * and filename to determine whether the file is parsable or not.
1106: */
1.59 schwarze 1107: static void
1.176 schwarze 1108: mpages_merge(struct mparse *mp)
1.50 kristaps 1109: {
1.113 schwarze 1110: char any[] = "any";
1.118 schwarze 1111: struct mpage *mpage, *mpage_dest;
1112: struct mlink *mlink, *mlink_dest;
1.191 schwarze 1113: struct roff_man *man;
1.118 schwarze 1114: char *sodest;
1.112 schwarze 1115: char *cp;
1.162 schwarze 1116: int fd;
1.101 schwarze 1117: unsigned int pslot;
1.69 schwarze 1118:
1.171 schwarze 1119: if ( ! nodb)
1.107 schwarze 1120: SQL_EXEC("BEGIN TRANSACTION");
1121:
1.80 schwarze 1122: mpage = ohash_first(&mpages, &pslot);
1.171 schwarze 1123: while (mpage != NULL) {
1.89 schwarze 1124: mlinks_undupe(mpage);
1.182 schwarze 1125: if ((mlink = mpage->mlinks) == NULL) {
1.89 schwarze 1126: mpage = ohash_next(&mpages, &pslot);
1127: continue;
1.50 kristaps 1128: }
1129:
1.133 schwarze 1130: name_mask = NAME_MASK;
1.202 schwarze 1131: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1132: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.50 kristaps 1133: mparse_reset(mp);
1134: man = NULL;
1.124 schwarze 1135: sodest = NULL;
1136:
1.213 schwarze 1137: if ((fd = mparse_open(mp, mlink->file)) == -1) {
1.182 schwarze 1138: say(mlink->file, "&open");
1.162 schwarze 1139: goto nextpage;
1.124 schwarze 1140: }
1.14 schwarze 1141:
1142: /*
1.183 schwarze 1143: * Interpret the file as mdoc(7) or man(7) source
1144: * code, unless it is known to be formatted.
1.14 schwarze 1145: */
1.182 schwarze 1146: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1147: mparse_readfd(mp, fd, mlink->file);
1.212 schwarze 1148: close(fd);
1.192 schwarze 1149: mparse_result(mp, &man, &sodest);
1.76 schwarze 1150: }
1.14 schwarze 1151:
1.171 schwarze 1152: if (sodest != NULL) {
1.118 schwarze 1153: mlink_dest = ohash_find(&mlinks,
1154: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1155: if (mlink_dest == NULL) {
1156: mandoc_asprintf(&cp, "%s.gz", sodest);
1157: mlink_dest = ohash_find(&mlinks,
1158: ohash_qlookup(&mlinks, cp));
1159: free(cp);
1160: }
1161: if (mlink_dest != NULL) {
1.118 schwarze 1162:
1163: /* The .so target exists. */
1164:
1165: mpage_dest = mlink_dest->mpage;
1166: while (1) {
1167: mlink->mpage = mpage_dest;
1168:
1169: /*
1170: * If the target was already
1171: * processed, add the links
1172: * to the database now.
1173: * Otherwise, this will
1174: * happen when we come
1175: * to the target.
1176: */
1177:
1.137 schwarze 1178: if (mpage_dest->pageid)
1.169 schwarze 1179: dbadd_mlink_name(mlink);
1.118 schwarze 1180:
1.171 schwarze 1181: if (mlink->next == NULL)
1.118 schwarze 1182: break;
1183: mlink = mlink->next;
1184: }
1185:
1186: /* Move all links to the target. */
1187:
1188: mlink->next = mlink_dest->next;
1189: mlink_dest->next = mpage->mlinks;
1190: mpage->mlinks = NULL;
1191: }
1.124 schwarze 1192: goto nextpage;
1.192 schwarze 1193: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.204 schwarze 1194: mdoc_validate(man);
1.82 schwarze 1195: mpage->form = FORM_SRC;
1.193 schwarze 1196: mpage->sec = man->meta.msec;
1.155 schwarze 1197: mpage->sec = mandoc_strdup(
1.171 schwarze 1198: mpage->sec == NULL ? "" : mpage->sec);
1.193 schwarze 1199: mpage->arch = man->meta.arch;
1.84 schwarze 1200: mpage->arch = mandoc_strdup(
1.171 schwarze 1201: mpage->arch == NULL ? "" : mpage->arch);
1.193 schwarze 1202: mpage->title = mandoc_strdup(man->meta.title);
1.192 schwarze 1203: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.205 schwarze 1204: man_validate(man);
1.82 schwarze 1205: mpage->form = FORM_SRC;
1.193 schwarze 1206: mpage->sec = mandoc_strdup(man->meta.msec);
1.182 schwarze 1207: mpage->arch = mandoc_strdup(mlink->arch);
1.193 schwarze 1208: mpage->title = mandoc_strdup(man->meta.title);
1.76 schwarze 1209: } else {
1.82 schwarze 1210: mpage->form = FORM_CAT;
1.182 schwarze 1211: mpage->sec = mandoc_strdup(mlink->dsec);
1212: mpage->arch = mandoc_strdup(mlink->arch);
1213: mpage->title = mandoc_strdup(mlink->name);
1.76 schwarze 1214: }
1.97 schwarze 1215: putkey(mpage, mpage->sec, TYPE_sec);
1.164 schwarze 1216: if (*mpage->arch != '\0')
1217: putkey(mpage, mpage->arch, TYPE_arch);
1.50 kristaps 1218:
1.182 schwarze 1219: for ( ; mlink != NULL; mlink = mlink->next) {
1.97 schwarze 1220: if ('\0' != *mlink->dsec)
1221: putkey(mpage, mlink->dsec, TYPE_sec);
1222: if ('\0' != *mlink->fsec)
1223: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1224: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1225: any : mlink->arch, TYPE_arch);
1.133 schwarze 1226: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1227: }
1.12 schwarze 1228:
1.174 schwarze 1229: assert(mpage->desc == NULL);
1.192 schwarze 1230: if (man != NULL && man->macroset == MACROSET_MDOC)
1.193 schwarze 1231: parse_mdoc(mpage, &man->meta, man->first);
1.174 schwarze 1232: else if (man != NULL)
1.193 schwarze 1233: parse_man(mpage, &man->meta, man->first);
1.50 kristaps 1234: else
1.162 schwarze 1235: parse_cat(mpage, fd);
1.174 schwarze 1236: if (mpage->desc == NULL)
1.135 schwarze 1237: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1238:
1239: if (warnings && !use_all)
1240: for (mlink = mpage->mlinks; mlink;
1241: mlink = mlink->next)
1242: mlink_check(mpage, mlink);
1.44 kristaps 1243:
1.176 schwarze 1244: dbadd(mpage);
1.182 schwarze 1245: mlink = mpage->mlinks;
1.124 schwarze 1246:
1247: nextpage:
1.66 schwarze 1248: ohash_delete(&strings);
1.133 schwarze 1249: ohash_delete(&names);
1.80 schwarze 1250: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1251: }
1.107 schwarze 1252:
1253: if (0 == nodb)
1254: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1255: }
1256:
1257: static void
1258: names_check(void)
1259: {
1260: sqlite3_stmt *stmt;
1261: const char *name, *sec, *arch, *key;
1262:
1263: sqlite3_prepare_v2(db,
1264: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1265: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1266: "WHERE bits & ? AND NOT EXISTS ("
1267: "SELECT pageid FROM mlinks "
1.133 schwarze 1268: "WHERE mlinks.pageid == names.pageid "
1269: "AND mlinks.name == names.name"
1.130 schwarze 1270: ")"
1271: ") JOIN ("
1.142 schwarze 1272: "SELECT sec, arch, name, pageid FROM mlinks "
1273: "GROUP BY pageid"
1.130 schwarze 1274: ") USING (pageid);",
1275: -1, &stmt, NULL);
1276:
1.201 schwarze 1277: if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK)
1.134 schwarze 1278: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1279:
1.201 schwarze 1280: while (sqlite3_step(stmt) == SQLITE_ROW) {
1.154 schwarze 1281: name = (const char *)sqlite3_column_text(stmt, 0);
1282: sec = (const char *)sqlite3_column_text(stmt, 1);
1283: arch = (const char *)sqlite3_column_text(stmt, 2);
1284: key = (const char *)sqlite3_column_text(stmt, 3);
1.130 schwarze 1285: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1286: '\0' == *arch ? "" : "/",
1287: '\0' == *arch ? "" : arch, key);
1288: }
1289: sqlite3_finalize(stmt);
1.50 kristaps 1290: }
1.12 schwarze 1291:
1.50 kristaps 1292: static void
1.124 schwarze 1293: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1294: {
1295: FILE *stream;
1296: char *line, *p, *title;
1.209 schwarze 1297: size_t linesz, plen, titlesz;
1298: ssize_t len;
1299: int offs;
1.12 schwarze 1300:
1.124 schwarze 1301: stream = (-1 == fd) ?
1302: fopen(mpage->mlinks->file, "r") :
1303: fdopen(fd, "r");
1304: if (NULL == stream) {
1.159 schwarze 1305: if (-1 != fd)
1306: close(fd);
1.56 schwarze 1307: if (warnings)
1.123 schwarze 1308: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1309: return;
1310: }
1.1 kristaps 1311:
1.209 schwarze 1312: line = NULL;
1313: linesz = 0;
1314:
1.50 kristaps 1315: /* Skip to first blank line. */
1.1 kristaps 1316:
1.209 schwarze 1317: while (getline(&line, &linesz, stream) != -1)
1318: if (*line == '\n')
1.50 kristaps 1319: break;
1.1 kristaps 1320:
1.50 kristaps 1321: /*
1322: * Assume the first line that is not indented
1323: * is the first section header. Skip to it.
1324: */
1.1 kristaps 1325:
1.209 schwarze 1326: while (getline(&line, &linesz, stream) != -1)
1327: if (*line != '\n' && *line != ' ')
1.50 kristaps 1328: break;
1.141 schwarze 1329:
1.50 kristaps 1330: /*
1331: * Read up until the next section into a buffer.
1332: * Strip the leading and trailing newline from each read line,
1333: * appending a trailing space.
1334: * Ignore empty (whitespace-only) lines.
1335: */
1.1 kristaps 1336:
1.50 kristaps 1337: titlesz = 0;
1338: title = NULL;
1.38 schwarze 1339:
1.209 schwarze 1340: while ((len = getline(&line, &linesz, stream)) != -1) {
1341: if (*line != ' ')
1.50 kristaps 1342: break;
1.209 schwarze 1343: offs = 0;
1344: while (isspace((unsigned char)line[offs]))
1345: offs++;
1346: if (line[offs] == '\0')
1.38 schwarze 1347: continue;
1.209 schwarze 1348: title = mandoc_realloc(title, titlesz + len - offs);
1349: memcpy(title + titlesz, line + offs, len - offs);
1350: titlesz += len - offs;
1.50 kristaps 1351: title[titlesz - 1] = ' ';
1352: }
1.209 schwarze 1353: free(line);
1.38 schwarze 1354:
1.50 kristaps 1355: /*
1356: * If no page content can be found, or the input line
1357: * is already the next section header, or there is no
1358: * trailing newline, reuse the page title as the page
1359: * description.
1360: */
1.44 kristaps 1361:
1.50 kristaps 1362: if (NULL == title || '\0' == *title) {
1.56 schwarze 1363: if (warnings)
1.82 schwarze 1364: say(mpage->mlinks->file,
1365: "Cannot find NAME section");
1.50 kristaps 1366: fclose(stream);
1367: free(title);
1368: return;
1369: }
1.1 kristaps 1370:
1.209 schwarze 1371: title[titlesz - 1] = '\0';
1.33 schwarze 1372:
1.50 kristaps 1373: /*
1374: * Skip to the first dash.
1375: * Use the remaining line as the description (no more than 70
1376: * bytes).
1377: */
1.33 schwarze 1378:
1.50 kristaps 1379: if (NULL != (p = strstr(title, "- "))) {
1380: for (p += 2; ' ' == *p || '\b' == *p; p++)
1381: /* Skip to next word. */ ;
1382: } else {
1.56 schwarze 1383: if (warnings)
1.82 schwarze 1384: say(mpage->mlinks->file,
1385: "No dash in title line");
1.50 kristaps 1386: p = title;
1387: }
1.38 schwarze 1388:
1.50 kristaps 1389: plen = strlen(p);
1.1 kristaps 1390:
1.50 kristaps 1391: /* Strip backspace-encoding from line. */
1.1 kristaps 1392:
1.50 kristaps 1393: while (NULL != (line = memchr(p, '\b', plen))) {
1394: len = line - p;
1395: if (0 == len) {
1396: memmove(line, line + 1, plen--);
1397: continue;
1.141 schwarze 1398: }
1.50 kristaps 1399: memmove(line - 1, line + 1, plen - len);
1400: plen -= 2;
1401: }
1.1 kristaps 1402:
1.78 schwarze 1403: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1404: fclose(stream);
1405: free(title);
1406: }
1.1 kristaps 1407:
1.50 kristaps 1408: /*
1409: * Put a type/word pair into the word database for this particular file.
1410: */
1411: static void
1.112 schwarze 1412: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1413: {
1.112 schwarze 1414: char *cp;
1.18 kristaps 1415:
1.50 kristaps 1416: assert(NULL != value);
1.112 schwarze 1417: if (TYPE_arch == type)
1418: for (cp = value; *cp; cp++)
1419: if (isupper((unsigned char)*cp))
1420: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1421: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1422: }
1423:
1424: /*
1.50 kristaps 1425: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1426: */
1427: static void
1.78 schwarze 1428: putmdockey(const struct mpage *mpage,
1.189 schwarze 1429: const struct roff_node *n, uint64_t m)
1.3 kristaps 1430: {
1.18 kristaps 1431:
1.50 kristaps 1432: for ( ; NULL != n; n = n->next) {
1433: if (NULL != n->child)
1.78 schwarze 1434: putmdockey(mpage, n->child, m);
1.188 schwarze 1435: if (n->type == ROFFT_TEXT)
1.78 schwarze 1436: putkey(mpage, n->string, m);
1.50 kristaps 1437: }
1438: }
1.18 kristaps 1439:
1.61 schwarze 1440: static void
1.190 schwarze 1441: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1442: const struct roff_node *n)
1.50 kristaps 1443: {
1.189 schwarze 1444: const struct roff_node *head, *body;
1.121 schwarze 1445: char *start, *title;
1.50 kristaps 1446: char byte;
1.121 schwarze 1447: size_t sz;
1.18 kristaps 1448:
1.215 schwarze 1449: if (n == NULL)
1.61 schwarze 1450: return;
1.18 kristaps 1451:
1.50 kristaps 1452: /*
1453: * We're only searching for one thing: the first text child in
1454: * the BODY of a NAME section. Since we don't keep track of
1455: * sections in -man, run some hoops to find out whether we're in
1456: * the correct section or not.
1457: */
1.18 kristaps 1458:
1.188 schwarze 1459: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.50 kristaps 1460: body = n;
1.215 schwarze 1461: if ((head = body->parent->head) != NULL &&
1462: (head = head->child) != NULL &&
1463: head->next == NULL &&
1.188 schwarze 1464: head->type == ROFFT_TEXT &&
1.215 schwarze 1465: strcmp(head->string, "NAME") == 0 &&
1466: body->child != NULL) {
1.3 kristaps 1467:
1.50 kristaps 1468: /*
1469: * Suck the entire NAME section into memory.
1470: * Yes, we might run away.
1471: * But too many manuals have big, spread-out
1472: * NAME sections over many lines.
1473: */
1.3 kristaps 1474:
1.121 schwarze 1475: title = NULL;
1.194 schwarze 1476: deroff(&title, body);
1.50 kristaps 1477: if (NULL == title)
1.61 schwarze 1478: return;
1.18 kristaps 1479:
1.141 schwarze 1480: /*
1.50 kristaps 1481: * Go through a special heuristic dance here.
1482: * Conventionally, one or more manual names are
1483: * comma-specified prior to a whitespace, then a
1484: * dash, then a description. Try to puzzle out
1485: * the name parts here.
1486: */
1.18 kristaps 1487:
1.121 schwarze 1488: start = title;
1.50 kristaps 1489: for ( ;; ) {
1490: sz = strcspn(start, " ,");
1491: if ('\0' == start[sz])
1492: break;
1.1 kristaps 1493:
1.50 kristaps 1494: byte = start[sz];
1495: start[sz] = '\0';
1.110 schwarze 1496:
1497: /*
1498: * Assume a stray trailing comma in the
1499: * name list if a name begins with a dash.
1500: */
1501:
1502: if ('-' == start[0] ||
1503: ('\\' == start[0] && '-' == start[1]))
1504: break;
1.1 kristaps 1505:
1.133 schwarze 1506: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1507: if ( ! (mpage->name_head_done ||
1508: strcasecmp(start, meta->title))) {
1509: putkey(mpage, start, NAME_HEAD);
1510: mpage->name_head_done = 1;
1511: }
1.1 kristaps 1512:
1.50 kristaps 1513: if (' ' == byte) {
1514: start += sz + 1;
1515: break;
1516: }
1.1 kristaps 1517:
1.50 kristaps 1518: assert(',' == byte);
1519: start += sz + 1;
1520: while (' ' == *start)
1521: start++;
1522: }
1.1 kristaps 1523:
1.121 schwarze 1524: if (start == title) {
1.133 schwarze 1525: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1526: if ( ! (mpage->name_head_done ||
1527: strcasecmp(start, meta->title))) {
1528: putkey(mpage, start, NAME_HEAD);
1529: mpage->name_head_done = 1;
1530: }
1.50 kristaps 1531: free(title);
1.61 schwarze 1532: return;
1.50 kristaps 1533: }
1.1 kristaps 1534:
1.50 kristaps 1535: while (isspace((unsigned char)*start))
1536: start++;
1.1 kristaps 1537:
1.50 kristaps 1538: if (0 == strncmp(start, "-", 1))
1539: start += 1;
1540: else if (0 == strncmp(start, "\\-\\-", 4))
1541: start += 4;
1542: else if (0 == strncmp(start, "\\-", 2))
1543: start += 2;
1544: else if (0 == strncmp(start, "\\(en", 4))
1545: start += 4;
1546: else if (0 == strncmp(start, "\\(em", 4))
1547: start += 4;
1.1 kristaps 1548:
1.50 kristaps 1549: while (' ' == *start)
1550: start++;
1.1 kristaps 1551:
1.78 schwarze 1552: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1553: free(title);
1.61 schwarze 1554: return;
1.50 kristaps 1555: }
1556: }
1.1 kristaps 1557:
1.77 schwarze 1558: for (n = n->child; n; n = n->next) {
1.78 schwarze 1559: if (NULL != mpage->desc)
1.77 schwarze 1560: break;
1.174 schwarze 1561: parse_man(mpage, meta, n);
1.77 schwarze 1562: }
1.1 kristaps 1563: }
1564:
1565: static void
1.190 schwarze 1566: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1567: const struct roff_node *n)
1.1 kristaps 1568: {
1569:
1.50 kristaps 1570: assert(NULL != n);
1571: for (n = n->child; NULL != n; n = n->next) {
1572: switch (n->type) {
1.188 schwarze 1573: case ROFFT_ELEM:
1574: case ROFFT_BLOCK:
1575: case ROFFT_HEAD:
1576: case ROFFT_BODY:
1577: case ROFFT_TAIL:
1.50 kristaps 1578: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1579: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1580: break;
1.68 schwarze 1581: if (mdocs[n->tok].mask)
1.78 schwarze 1582: putmdockey(mpage, n->child,
1583: mdocs[n->tok].mask);
1.50 kristaps 1584: break;
1585: default:
1.188 schwarze 1586: assert(n->type != ROFFT_ROOT);
1.50 kristaps 1587: continue;
1588: }
1589: if (NULL != n->child)
1.172 schwarze 1590: parse_mdoc(mpage, meta, n);
1.1 kristaps 1591: }
1592: }
1593:
1.25 schwarze 1594: static int
1.190 schwarze 1595: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1596: const struct roff_node *n)
1.1 kristaps 1597: {
1.176 schwarze 1598: char *start, *end;
1.1 kristaps 1599: size_t sz;
1.25 schwarze 1600:
1.50 kristaps 1601: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1602: NULL == (n = n->child) ||
1.188 schwarze 1603: n->type != ROFFT_TEXT)
1.197 schwarze 1604: return 0;
1.1 kristaps 1605:
1606: /*
1607: * Only consider those `Fd' macro fields that begin with an
1608: * "inclusion" token (versus, e.g., #define).
1609: */
1.50 kristaps 1610:
1.1 kristaps 1611: if (strcmp("#include", n->string))
1.197 schwarze 1612: return 0;
1.1 kristaps 1613:
1.188 schwarze 1614: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1615: return 0;
1.1 kristaps 1616:
1617: /*
1618: * Strip away the enclosing angle brackets and make sure we're
1619: * not zero-length.
1620: */
1621:
1622: start = n->string;
1623: if ('<' == *start || '"' == *start)
1624: start++;
1625:
1626: if (0 == (sz = strlen(start)))
1.197 schwarze 1627: return 0;
1.1 kristaps 1628:
1629: end = &start[(int)sz - 1];
1630: if ('>' == *end || '"' == *end)
1631: end--;
1632:
1.50 kristaps 1633: if (end > start)
1.78 schwarze 1634: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1635: return 0;
1.1 kristaps 1636: }
1637:
1.178 schwarze 1638: static void
1.189 schwarze 1639: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1640: {
1.112 schwarze 1641: char *cp;
1.178 schwarze 1642: size_t sz;
1.1 kristaps 1643:
1.188 schwarze 1644: if (n->type != ROFFT_TEXT)
1.178 schwarze 1645: return;
1.25 schwarze 1646:
1.178 schwarze 1647: /* Skip function pointer punctuation. */
1.1 kristaps 1648:
1.178 schwarze 1649: cp = n->string;
1650: while (*cp == '(' || *cp == '*')
1.1 kristaps 1651: cp++;
1.178 schwarze 1652: sz = strcspn(cp, "()");
1.1 kristaps 1653:
1.178 schwarze 1654: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1655: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1656: putkeys(mpage, cp, sz, NAME_SYN);
1657: }
1.25 schwarze 1658:
1.178 schwarze 1659: static int
1.190 schwarze 1660: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1661: const struct roff_node *n)
1.178 schwarze 1662: {
1663:
1664: if (n->child == NULL)
1.197 schwarze 1665: return 0;
1.178 schwarze 1666:
1667: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1668:
1.178 schwarze 1669: for (n = n->child->next; n != NULL; n = n->next)
1.188 schwarze 1670: if (n->type == ROFFT_TEXT)
1.78 schwarze 1671: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1672:
1.197 schwarze 1673: return 0;
1.173 schwarze 1674: }
1675:
1676: static int
1.190 schwarze 1677: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1678: const struct roff_node *n)
1.173 schwarze 1679: {
1.177 schwarze 1680:
1.188 schwarze 1681: if (n->type != ROFFT_HEAD)
1.197 schwarze 1682: return 1;
1.173 schwarze 1683:
1.178 schwarze 1684: if (n->child != NULL)
1685: parse_mdoc_fname(mpage, n->child);
1686:
1.197 schwarze 1687: return 0;
1.1 kristaps 1688: }
1689:
1.25 schwarze 1690: static int
1.211 schwarze 1691: parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1692: const struct roff_node *n)
1693: {
1694: char *cp;
1695:
1696: if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1697: return 0;
1698:
1.215 schwarze 1699: if (n->child != NULL &&
1700: n->child->next == NULL &&
1701: n->child->type == ROFFT_TEXT)
1.211 schwarze 1702: return 1;
1703:
1704: cp = NULL;
1705: deroff(&cp, n);
1706: if (cp != NULL) {
1707: putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1708: n->type == ROFFT_BODY ? TYPE_Va : 0));
1709: free(cp);
1710: }
1711:
1712: return 0;
1713: }
1714:
1715: static int
1.190 schwarze 1716: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1717: const struct roff_node *n)
1.1 kristaps 1718: {
1.67 schwarze 1719: char *cp;
1.1 kristaps 1720:
1721: if (NULL == (n = n->child))
1.197 schwarze 1722: return 0;
1.1 kristaps 1723:
1.67 schwarze 1724: if (NULL == n->next) {
1.78 schwarze 1725: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1726: return 0;
1.67 schwarze 1727: }
1728:
1.120 schwarze 1729: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1730: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1731: free(cp);
1.197 schwarze 1732: return 0;
1.1 kristaps 1733: }
1734:
1.25 schwarze 1735: static int
1.190 schwarze 1736: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1737: const struct roff_node *n)
1.1 kristaps 1738: {
1739:
1.188 schwarze 1740: if (n->type == ROFFT_BODY)
1.194 schwarze 1741: deroff(&mpage->desc, n);
1.197 schwarze 1742: return 0;
1.1 kristaps 1743: }
1744:
1.25 schwarze 1745: static int
1.190 schwarze 1746: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1747: const struct roff_node *n)
1.1 kristaps 1748: {
1749:
1.129 schwarze 1750: if (SEC_NAME == n->sec)
1.133 schwarze 1751: putmdockey(mpage, n->child, NAME_TITLE);
1.188 schwarze 1752: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1753: if (n->child == NULL)
1754: putkey(mpage, meta->name, NAME_SYN);
1755: else
1756: putmdockey(mpage, n->child, NAME_SYN);
1.174 schwarze 1757: }
1758: if ( ! (mpage->name_head_done ||
1759: n->child == NULL || n->child->string == NULL ||
1760: strcasecmp(n->child->string, meta->title))) {
1.188 schwarze 1761: putkey(mpage, n->child->string, ROFFT_HEAD);
1.174 schwarze 1762: mpage->name_head_done = 1;
1.172 schwarze 1763: }
1.197 schwarze 1764: return 0;
1.1 kristaps 1765: }
1766:
1.25 schwarze 1767: static int
1.190 schwarze 1768: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1769: const struct roff_node *n)
1.1 kristaps 1770: {
1771:
1.197 schwarze 1772: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1773: }
1774:
1.50 kristaps 1775: static int
1.190 schwarze 1776: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1777: const struct roff_node *n)
1.1 kristaps 1778: {
1779:
1.197 schwarze 1780: return n->type == ROFFT_HEAD;
1.1 kristaps 1781: }
1782:
1.50 kristaps 1783: /*
1.66 schwarze 1784: * Add a string to the hash table for the current manual.
1785: * Each string has a bitmask telling which macros it belongs to.
1786: * When we finish the manual, we'll dump the table.
1.50 kristaps 1787: */
1788: static void
1.176 schwarze 1789: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1790: {
1.133 schwarze 1791: struct ohash *htab;
1.50 kristaps 1792: struct str *s;
1.111 schwarze 1793: const char *end;
1.65 schwarze 1794: unsigned int slot;
1.176 schwarze 1795: int i, mustfree;
1.25 schwarze 1796:
1.50 kristaps 1797: if (0 == sz)
1798: return;
1.111 schwarze 1799:
1.176 schwarze 1800: mustfree = render_string(&cp, &sz);
1801:
1.133 schwarze 1802: if (TYPE_Nm & v) {
1803: htab = &names;
1804: v &= name_mask;
1.169 schwarze 1805: if (v & NAME_FIRST)
1806: name_mask &= ~NAME_FIRST;
1.133 schwarze 1807: if (debug > 1)
1808: say(mpage->mlinks->file,
1.175 schwarze 1809: "Adding name %*s, bits=%d", sz, cp, v);
1.133 schwarze 1810: } else {
1811: htab = &strings;
1812: if (debug > 1)
1813: for (i = 0; i < mansearch_keymax; i++)
1.163 schwarze 1814: if ((uint64_t)1 << i & v)
1.133 schwarze 1815: say(mpage->mlinks->file,
1816: "Adding key %s=%*s",
1817: mansearch_keynames[i], sz, cp);
1.111 schwarze 1818: }
1.25 schwarze 1819:
1.65 schwarze 1820: end = cp + sz;
1.133 schwarze 1821: slot = ohash_qlookupi(htab, cp, &end);
1822: s = ohash_find(htab, slot);
1.25 schwarze 1823:
1.78 schwarze 1824: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1825: s->mask |= v;
1826: return;
1827: } else if (NULL == s) {
1.144 schwarze 1828: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1829: memcpy(s->key, cp, sz);
1.133 schwarze 1830: ohash_insert(htab, slot, s);
1.1 kristaps 1831: }
1.78 schwarze 1832: s->mpage = mpage;
1.50 kristaps 1833: s->mask = v;
1.176 schwarze 1834:
1835: if (mustfree)
1836: free(cp);
1.1 kristaps 1837: }
1838:
1.50 kristaps 1839: /*
1840: * Take a Unicode codepoint and produce its UTF-8 encoding.
1841: * This isn't the best way to do this, but it works.
1842: * The magic numbers are from the UTF-8 packaging.
1843: * They're not as scary as they seem: read the UTF-8 spec for details.
1844: */
1845: static size_t
1846: utf8(unsigned int cp, char out[7])
1.1 kristaps 1847: {
1.50 kristaps 1848: size_t rc;
1.1 kristaps 1849:
1.50 kristaps 1850: rc = 0;
1851: if (cp <= 0x0000007F) {
1852: rc = 1;
1853: out[0] = (char)cp;
1854: } else if (cp <= 0x000007FF) {
1855: rc = 2;
1856: out[0] = (cp >> 6 & 31) | 192;
1857: out[1] = (cp & 63) | 128;
1858: } else if (cp <= 0x0000FFFF) {
1859: rc = 3;
1860: out[0] = (cp >> 12 & 15) | 224;
1861: out[1] = (cp >> 6 & 63) | 128;
1862: out[2] = (cp & 63) | 128;
1863: } else if (cp <= 0x001FFFFF) {
1864: rc = 4;
1865: out[0] = (cp >> 18 & 7) | 240;
1866: out[1] = (cp >> 12 & 63) | 128;
1867: out[2] = (cp >> 6 & 63) | 128;
1868: out[3] = (cp & 63) | 128;
1869: } else if (cp <= 0x03FFFFFF) {
1870: rc = 5;
1871: out[0] = (cp >> 24 & 3) | 248;
1872: out[1] = (cp >> 18 & 63) | 128;
1873: out[2] = (cp >> 12 & 63) | 128;
1874: out[3] = (cp >> 6 & 63) | 128;
1875: out[4] = (cp & 63) | 128;
1876: } else if (cp <= 0x7FFFFFFF) {
1877: rc = 6;
1878: out[0] = (cp >> 30 & 1) | 252;
1879: out[1] = (cp >> 24 & 63) | 128;
1880: out[2] = (cp >> 18 & 63) | 128;
1881: out[3] = (cp >> 12 & 63) | 128;
1882: out[4] = (cp >> 6 & 63) | 128;
1883: out[5] = (cp & 63) | 128;
1884: } else
1.197 schwarze 1885: return 0;
1.1 kristaps 1886:
1.50 kristaps 1887: out[rc] = '\0';
1.197 schwarze 1888: return rc;
1.50 kristaps 1889: }
1.1 kristaps 1890:
1.50 kristaps 1891: /*
1.176 schwarze 1892: * If the string contains escape sequences,
1893: * replace it with an allocated rendering and return 1,
1894: * such that the caller can free it after use.
1895: * Otherwise, do nothing and return 0.
1.50 kristaps 1896: */
1.176 schwarze 1897: static int
1898: render_string(char **public, size_t *psz)
1.50 kristaps 1899: {
1.176 schwarze 1900: const char *src, *scp, *addcp, *seq;
1901: char *dst;
1902: size_t ssz, dsz, addsz;
1.114 schwarze 1903: char utfbuf[7], res[6];
1.176 schwarze 1904: int seqlen, unicode;
1.50 kristaps 1905:
1906: res[0] = '\\';
1907: res[1] = '\t';
1908: res[2] = ASCII_NBRSP;
1909: res[3] = ASCII_HYPH;
1.114 schwarze 1910: res[4] = ASCII_BREAK;
1911: res[5] = '\0';
1.1 kristaps 1912:
1.176 schwarze 1913: src = scp = *public;
1914: ssz = *psz;
1915: dst = NULL;
1916: dsz = 0;
1917:
1918: while (scp < src + *psz) {
1919:
1920: /* Leave normal characters unchanged. */
1921:
1922: if (strchr(res, *scp) == NULL) {
1923: if (dst != NULL)
1924: dst[dsz++] = *scp;
1925: scp++;
1926: continue;
1927: }
1.46 kristaps 1928:
1.50 kristaps 1929: /*
1.176 schwarze 1930: * Found something that requires replacing,
1931: * make sure we have a destination buffer.
1.50 kristaps 1932: */
1.176 schwarze 1933:
1934: if (dst == NULL) {
1935: dst = mandoc_malloc(ssz + 1);
1936: dsz = scp - src;
1937: memcpy(dst, src, dsz);
1.50 kristaps 1938: }
1.46 kristaps 1939:
1.176 schwarze 1940: /* Handle single-char special characters. */
1941:
1942: switch (*scp) {
1943: case '\\':
1944: break;
1.141 schwarze 1945: case '\t':
1946: case ASCII_NBRSP:
1.176 schwarze 1947: dst[dsz++] = ' ';
1948: scp++;
1949: continue;
1950: case ASCII_HYPH:
1951: dst[dsz++] = '-';
1.114 schwarze 1952: /* FALLTHROUGH */
1.141 schwarze 1953: case ASCII_BREAK:
1.176 schwarze 1954: scp++;
1.50 kristaps 1955: continue;
1.114 schwarze 1956: default:
1.176 schwarze 1957: abort();
1.114 schwarze 1958: }
1.46 kristaps 1959:
1.50 kristaps 1960: /*
1.176 schwarze 1961: * Found an escape sequence.
1962: * Read past the slash, then parse it.
1963: * Ignore everything except characters.
1.50 kristaps 1964: */
1.95 schwarze 1965:
1.176 schwarze 1966: scp++;
1967: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1968: continue;
1.1 kristaps 1969:
1.50 kristaps 1970: /*
1.95 schwarze 1971: * Render the special character
1972: * as either UTF-8 or ASCII.
1.50 kristaps 1973: */
1.95 schwarze 1974:
1975: if (write_utf8) {
1.203 schwarze 1976: unicode = mchars_spec2cp(seq, seqlen);
1.176 schwarze 1977: if (unicode <= 0)
1.95 schwarze 1978: continue;
1.176 schwarze 1979: addsz = utf8(unicode, utfbuf);
1980: if (addsz == 0)
1.95 schwarze 1981: continue;
1.176 schwarze 1982: addcp = utfbuf;
1.95 schwarze 1983: } else {
1.203 schwarze 1984: addcp = mchars_spec2str(seq, seqlen, &addsz);
1.176 schwarze 1985: if (addcp == NULL)
1.95 schwarze 1986: continue;
1.176 schwarze 1987: if (*addcp == ASCII_NBRSP) {
1988: addcp = " ";
1989: addsz = 1;
1.95 schwarze 1990: }
1991: }
1.1 kristaps 1992:
1.50 kristaps 1993: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1994:
1.176 schwarze 1995: ssz += addsz;
1996: dst = mandoc_realloc(dst, ssz + 1);
1997: memcpy(dst + dsz, addcp, addsz);
1998: dsz += addsz;
1999: }
2000: if (dst != NULL) {
2001: *public = dst;
2002: *psz = dsz;
2003: }
2004:
2005: /* Trim trailing whitespace and NUL-terminate. */
2006:
2007: while (*psz > 0 && (*public)[*psz - 1] == ' ')
2008: --*psz;
2009: if (dst != NULL) {
2010: (*public)[*psz] = '\0';
1.197 schwarze 2011: return 1;
1.176 schwarze 2012: } else
1.197 schwarze 2013: return 0;
1.1 kristaps 2014: }
2015:
1.118 schwarze 2016: static void
2017: dbadd_mlink(const struct mlink *mlink)
2018: {
2019: size_t i;
2020:
2021: i = 1;
2022: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2023: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2024: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 2025: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 2026: SQL_STEP(stmts[STMT_INSERT_LINK]);
2027: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.169 schwarze 2028: }
2029:
2030: static void
2031: dbadd_mlink_name(const struct mlink *mlink)
2032: {
1.175 schwarze 2033: uint64_t bits;
1.169 schwarze 2034: size_t i;
2035:
2036: dbadd_mlink(mlink);
1.160 schwarze 2037:
2038: i = 1;
1.175 schwarze 2039: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2040: bits = NAME_FILE & NAME_MASK;
2041: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2042: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2043: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2044: }
2045:
2046: i = 1;
2047: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.160 schwarze 2048: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2049: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2050: SQL_STEP(stmts[STMT_INSERT_NAME]);
2051: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.118 schwarze 2052: }
2053:
1.14 schwarze 2054: /*
1.50 kristaps 2055: * Flush the current page's terms (and their bits) into the database.
2056: * Wrap the entire set of additions in a transaction to make sqlite be a
2057: * little faster.
1.96 schwarze 2058: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2059: */
2060: static void
1.176 schwarze 2061: dbadd(struct mpage *mpage)
1.14 schwarze 2062: {
1.87 schwarze 2063: struct mlink *mlink;
1.50 kristaps 2064: struct str *key;
1.176 schwarze 2065: char *cp;
1.52 kristaps 2066: size_t i;
1.66 schwarze 2067: unsigned int slot;
1.176 schwarze 2068: int mustfree;
1.14 schwarze 2069:
1.128 schwarze 2070: mlink = mpage->mlinks;
1.43 kristaps 2071:
1.128 schwarze 2072: if (nodb) {
1.147 schwarze 2073: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2074: key = ohash_next(&names, &slot))
1.147 schwarze 2075: free(key);
2076: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2077: key = ohash_next(&strings, &slot))
1.147 schwarze 2078: free(key);
1.145 schwarze 2079: if (0 == debug)
2080: return;
1.128 schwarze 2081: while (NULL != mlink) {
2082: fputs(mlink->name, stdout);
2083: if (NULL == mlink->next ||
2084: strcmp(mlink->dsec, mlink->next->dsec) ||
2085: strcmp(mlink->fsec, mlink->next->fsec) ||
2086: strcmp(mlink->arch, mlink->next->arch)) {
2087: putchar('(');
2088: if ('\0' == *mlink->dsec)
2089: fputs(mlink->fsec, stdout);
2090: else
2091: fputs(mlink->dsec, stdout);
2092: if ('\0' != *mlink->arch)
2093: printf("/%s", mlink->arch);
2094: putchar(')');
2095: }
2096: mlink = mlink->next;
2097: if (NULL != mlink)
2098: fputs(", ", stdout);
2099: }
1.132 schwarze 2100: printf(" - %s\n", mpage->desc);
1.14 schwarze 2101: return;
1.128 schwarze 2102: }
2103:
2104: if (debug)
2105: say(mlink->file, "Adding to database");
1.28 kristaps 2106:
1.176 schwarze 2107: cp = mpage->desc;
2108: i = strlen(cp);
2109: mustfree = render_string(&cp, &i);
1.52 kristaps 2110: i = 1;
1.176 schwarze 2111: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.161 schwarze 2112: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.81 schwarze 2113: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2114: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2115: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.176 schwarze 2116: if (mustfree)
2117: free(cp);
1.81 schwarze 2118:
1.128 schwarze 2119: while (NULL != mlink) {
1.118 schwarze 2120: dbadd_mlink(mlink);
1.128 schwarze 2121: mlink = mlink->next;
2122: }
1.134 schwarze 2123: mlink = mpage->mlinks;
1.50 kristaps 2124:
1.133 schwarze 2125: for (key = ohash_first(&names, &slot); NULL != key;
2126: key = ohash_next(&names, &slot)) {
2127: assert(key->mpage == mpage);
2128: i = 1;
2129: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.176 schwarze 2130: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.137 schwarze 2131: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2132: SQL_STEP(stmts[STMT_INSERT_NAME]);
2133: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2134: free(key);
2135: }
1.66 schwarze 2136: for (key = ohash_first(&strings, &slot); NULL != key;
2137: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2138: assert(key->mpage == mpage);
1.52 kristaps 2139: i = 1;
2140: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.176 schwarze 2141: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.137 schwarze 2142: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2143: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2144: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 2145: free(key);
1.38 schwarze 2146: }
1.14 schwarze 2147: }
2148:
1.5 kristaps 2149: static void
1.59 schwarze 2150: dbprune(void)
1.5 kristaps 2151: {
1.78 schwarze 2152: struct mpage *mpage;
1.82 schwarze 2153: struct mlink *mlink;
1.52 kristaps 2154: size_t i;
1.80 schwarze 2155: unsigned int slot;
1.5 kristaps 2156:
1.106 schwarze 2157: if (0 == nodb)
2158: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2159:
1.106 schwarze 2160: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2161: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2162: mlink = mpage->mlinks;
1.125 schwarze 2163: if (debug)
1.106 schwarze 2164: say(mlink->file, "Deleting from database");
2165: if (nodb)
2166: continue;
2167: for ( ; NULL != mlink; mlink = mlink->next) {
2168: i = 1;
2169: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2170: i, mlink->dsec);
2171: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2172: i, mlink->arch);
2173: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2174: i, mlink->name);
2175: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2176: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2177: }
1.5 kristaps 2178: }
1.106 schwarze 2179:
2180: if (0 == nodb)
2181: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2182: }
2183:
1.4 kristaps 2184: /*
1.50 kristaps 2185: * Close an existing database and its prepared statements.
2186: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2187: */
1.35 kristaps 2188: static void
1.59 schwarze 2189: dbclose(int real)
1.4 kristaps 2190: {
1.50 kristaps 2191: size_t i;
1.115 schwarze 2192: int status;
2193: pid_t child;
1.4 kristaps 2194:
1.50 kristaps 2195: if (nodb)
1.38 schwarze 2196: return;
1.50 kristaps 2197:
2198: for (i = 0; i < STMT__MAX; i++) {
2199: sqlite3_finalize(stmts[i]);
2200: stmts[i] = NULL;
1.4 kristaps 2201: }
2202:
1.50 kristaps 2203: sqlite3_close(db);
2204: db = NULL;
1.12 schwarze 2205:
1.50 kristaps 2206: if (real)
2207: return;
1.12 schwarze 2208:
1.115 schwarze 2209: if ('\0' == *tempfilename) {
2210: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2211: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2212: say(MANDOC_DB, "&rename");
1.115 schwarze 2213: }
2214: return;
2215: }
2216:
2217: switch (child = fork()) {
1.141 schwarze 2218: case -1:
1.115 schwarze 2219: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2220: say("", "&fork cmp");
1.115 schwarze 2221: return;
1.141 schwarze 2222: case 0:
1.115 schwarze 2223: execlp("cmp", "cmp", "-s",
1.196 schwarze 2224: tempfilename, MANDOC_DB, (char *)NULL);
1.123 schwarze 2225: say("", "&exec cmp");
1.115 schwarze 2226: exit(0);
2227: default:
2228: break;
2229: }
2230: if (-1 == waitpid(child, &status, 0)) {
2231: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2232: say("", "&wait cmp");
1.115 schwarze 2233: } else if (WIFSIGNALED(status)) {
2234: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2235: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2236: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2237: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2238: say(MANDOC_DB,
2239: "Data changed, but cannot replace database");
2240: }
2241:
2242: *strrchr(tempfilename, '/') = '\0';
2243: switch (child = fork()) {
1.141 schwarze 2244: case -1:
1.115 schwarze 2245: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2246: say("", "&fork rm");
1.115 schwarze 2247: return;
1.141 schwarze 2248: case 0:
1.196 schwarze 2249: execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
1.123 schwarze 2250: say("", "&exec rm");
1.115 schwarze 2251: exit((int)MANDOCLEVEL_SYSERR);
2252: default:
2253: break;
2254: }
2255: if (-1 == waitpid(child, &status, 0)) {
2256: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2257: say("", "&wait rm");
1.115 schwarze 2258: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2259: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2260: say("", "%s: Cannot remove temporary directory",
2261: tempfilename);
1.59 schwarze 2262: }
1.50 kristaps 2263: }
1.14 schwarze 2264:
1.50 kristaps 2265: /*
2266: * This is straightforward stuff.
2267: * Open a database connection to a "temporary" database, then open a set
2268: * of prepared statements we'll use over and over again.
2269: * If "real" is set, we use the existing database; if not, we truncate a
2270: * temporary one.
2271: * Must be matched by dbclose().
2272: */
2273: static int
1.59 schwarze 2274: dbopen(int real)
1.50 kristaps 2275: {
1.115 schwarze 2276: const char *sql;
1.50 kristaps 2277: int rc, ofl;
1.12 schwarze 2278:
1.141 schwarze 2279: if (nodb)
1.197 schwarze 2280: return 1;
1.12 schwarze 2281:
1.115 schwarze 2282: *tempfilename = '\0';
1.63 schwarze 2283: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2284:
2285: if (real) {
2286: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2287: if (SQLITE_OK != rc) {
1.63 schwarze 2288: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2289: if (SQLITE_CANTOPEN != rc)
2290: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.197 schwarze 2291: return 0;
1.63 schwarze 2292: }
1.115 schwarze 2293: goto prepare_statements;
2294: }
2295:
2296: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2297:
1.115 schwarze 2298: remove(MANDOC_DB "~");
2299: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2300: if (SQLITE_OK == rc)
1.115 schwarze 2301: goto create_tables;
1.116 schwarze 2302: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2303: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2304: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.197 schwarze 2305: return 0;
1.50 kristaps 2306: }
1.12 schwarze 2307:
1.143 schwarze 2308: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2309: sizeof(tempfilename));
1.115 schwarze 2310: if (NULL == mkdtemp(tempfilename)) {
2311: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2312: say("", "&%s", tempfilename);
1.197 schwarze 2313: return 0;
1.115 schwarze 2314: }
1.143 schwarze 2315: (void)strlcat(tempfilename, "/" MANDOC_DB,
2316: sizeof(tempfilename));
1.115 schwarze 2317: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2318: if (SQLITE_OK != rc) {
1.59 schwarze 2319: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2320: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.197 schwarze 2321: return 0;
1.50 kristaps 2322: }
1.12 schwarze 2323:
1.115 schwarze 2324: create_tables:
1.81 schwarze 2325: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2326: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2327: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2328: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2329: ");\n"
2330: "\n"
1.81 schwarze 2331: "CREATE TABLE \"mlinks\" (\n"
2332: " \"sec\" TEXT NOT NULL,\n"
2333: " \"arch\" TEXT NOT NULL,\n"
2334: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2335: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2336: "ON DELETE CASCADE\n"
1.81 schwarze 2337: ");\n"
1.136 schwarze 2338: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2339: "\n"
1.133 schwarze 2340: "CREATE TABLE \"names\" (\n"
2341: " \"bits\" INTEGER NOT NULL,\n"
2342: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2343: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.175 schwarze 2344: "ON DELETE CASCADE,\n"
2345: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.133 schwarze 2346: ");\n"
2347: "\n"
1.50 kristaps 2348: "CREATE TABLE \"keys\" (\n"
2349: " \"bits\" INTEGER NOT NULL,\n"
2350: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2351: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2352: "ON DELETE CASCADE\n"
1.136 schwarze 2353: ");\n"
2354: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2355:
1.50 kristaps 2356: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2357: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2358: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2359: sqlite3_close(db);
1.197 schwarze 2360: return 0;
1.50 kristaps 2361: }
1.4 kristaps 2362:
1.57 schwarze 2363: prepare_statements:
1.146 schwarze 2364: if (SQLITE_OK != sqlite3_exec(db,
2365: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2366: exitcode = (int)MANDOCLEVEL_SYSERR;
2367: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2368: sqlite3_errmsg(db));
2369: sqlite3_close(db);
1.197 schwarze 2370: return 0;
1.146 schwarze 2371: }
2372:
1.137 schwarze 2373: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2374: "(SELECT pageid FROM mlinks WHERE "
2375: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2376: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2377: sql = "INSERT INTO mpages "
1.132 schwarze 2378: "(desc,form) VALUES (?,?)";
1.81 schwarze 2379: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2380: sql = "INSERT INTO mlinks "
1.104 schwarze 2381: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2382: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.175 schwarze 2383: sql = "SELECT bits FROM names where pageid = ?";
2384: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.133 schwarze 2385: sql = "INSERT INTO names "
2386: "(bits,name,pageid) VALUES (?,?,?)";
2387: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2388: sql = "INSERT INTO keys "
1.81 schwarze 2389: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2390: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2391:
2392: #ifndef __APPLE__
2393: /*
2394: * When opening a new database, we can turn off
2395: * synchronous mode for much better performance.
2396: */
2397:
1.146 schwarze 2398: if (real && SQLITE_OK != sqlite3_exec(db,
2399: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2400: exitcode = (int)MANDOCLEVEL_SYSERR;
2401: say(MANDOC_DB, "PRAGMA synchronous: %s",
1.162 schwarze 2402: sqlite3_errmsg(db));
1.146 schwarze 2403: sqlite3_close(db);
1.197 schwarze 2404: return 0;
1.146 schwarze 2405: }
1.70 schwarze 2406: #endif
2407:
1.197 schwarze 2408: return 1;
1.4 kristaps 2409: }
2410:
1.50 kristaps 2411: static int
1.165 schwarze 2412: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2413: {
1.59 schwarze 2414: static char startdir[PATH_MAX];
1.151 schwarze 2415: static int getcwd_status; /* 1 = ok, 2 = failure */
2416: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2417: char *cp;
1.4 kristaps 2418:
1.59 schwarze 2419: /*
1.151 schwarze 2420: * Remember the original working directory, if possible.
2421: * This will be needed if the second or a later directory
2422: * on the command line is given as a relative path.
2423: * Do not error out if the current directory is not
2424: * searchable: Maybe it won't be needed after all.
2425: */
2426: if (0 == getcwd_status) {
2427: if (NULL == getcwd(startdir, sizeof(startdir))) {
2428: getcwd_status = 2;
2429: (void)strlcpy(startdir, strerror(errno),
2430: sizeof(startdir));
2431: } else
2432: getcwd_status = 1;
2433: }
2434:
2435: /*
2436: * We are leaving the old base directory.
2437: * Do not use it any longer, not even for messages.
2438: */
2439: *basedir = '\0';
2440:
2441: /*
2442: * If and only if the directory was changed earlier and
2443: * the next directory to process is given as a relative path,
2444: * first go back, or bail out if that is impossible.
1.59 schwarze 2445: */
1.151 schwarze 2446: if (chdir_status && '/' != *targetdir) {
2447: if (2 == getcwd_status) {
1.59 schwarze 2448: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2449: say("", "getcwd: %s", startdir);
1.197 schwarze 2450: return 0;
1.59 schwarze 2451: }
1.151 schwarze 2452: if (-1 == chdir(startdir)) {
1.59 schwarze 2453: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2454: say("", "&chdir %s", startdir);
1.197 schwarze 2455: return 0;
1.59 schwarze 2456: }
2457: }
1.151 schwarze 2458:
2459: /*
2460: * Always resolve basedir to the canonicalized absolute
2461: * pathname and append a trailing slash, such that
2462: * we can reliably check whether files are inside.
2463: */
1.59 schwarze 2464: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2465: if (report_baddir || errno != ENOENT) {
2466: exitcode = (int)MANDOCLEVEL_BADARG;
2467: say("", "&%s: realpath", targetdir);
2468: }
1.197 schwarze 2469: return 0;
1.59 schwarze 2470: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2471: if (report_baddir || errno != ENOENT) {
2472: exitcode = (int)MANDOCLEVEL_BADARG;
2473: say("", "&chdir");
2474: }
1.197 schwarze 2475: return 0;
1.150 schwarze 2476: }
1.151 schwarze 2477: chdir_status = 1;
1.150 schwarze 2478: cp = strchr(basedir, '\0');
2479: if ('/' != cp[-1]) {
2480: if (cp - basedir >= PATH_MAX - 1) {
2481: exitcode = (int)MANDOCLEVEL_SYSERR;
2482: say("", "Filename too long");
1.197 schwarze 2483: return 0;
1.150 schwarze 2484: }
2485: *cp++ = '/';
2486: *cp = '\0';
1.4 kristaps 2487: }
1.197 schwarze 2488: return 1;
1.56 schwarze 2489: }
2490:
2491: static void
1.59 schwarze 2492: say(const char *file, const char *format, ...)
1.56 schwarze 2493: {
2494: va_list ap;
1.123 schwarze 2495: int use_errno;
1.56 schwarze 2496:
1.59 schwarze 2497: if ('\0' != *basedir)
2498: fprintf(stderr, "%s", basedir);
2499: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2500: fputc('/', stderr);
1.56 schwarze 2501: if ('\0' != *file)
1.59 schwarze 2502: fprintf(stderr, "%s", file);
2503:
1.123 schwarze 2504: use_errno = 1;
2505: if (NULL != format) {
2506: switch (*format) {
1.141 schwarze 2507: case '&':
1.123 schwarze 2508: format++;
2509: break;
1.141 schwarze 2510: case '\0':
1.123 schwarze 2511: format = NULL;
2512: break;
2513: default:
2514: use_errno = 0;
2515: break;
2516: }
2517: }
2518: if (NULL != format) {
2519: if ('\0' != *basedir || '\0' != *file)
2520: fputs(": ", stderr);
2521: va_start(ap, format);
2522: vfprintf(stderr, format, ap);
2523: va_end(ap);
2524: }
2525: if (use_errno) {
2526: if ('\0' != *basedir || '\0' != *file || NULL != format)
2527: fputs(": ", stderr);
1.59 schwarze 2528: perror(NULL);
1.123 schwarze 2529: } else
2530: fputc('\n', stderr);
1.1 kristaps 2531: }
CVSweb