Annotation of mandoc/mandocdb.c, Revision 1.202
1.202 ! schwarze 1: /* $Id: mandocdb.c,v 1.201 2015/10/12 21:17:15 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:
1.198 schwarze 186: extern char *__progname;
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 */
1.176 schwarze 198: static struct mchars *mchars; /* table of named characters */
1.78 schwarze 199: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 200: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 201: static struct ohash names; /* table of all names */
1.50 kristaps 202: static struct ohash strings; /* table of all strings */
203: static sqlite3 *db = NULL; /* current database */
204: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.133 schwarze 205: static uint64_t name_mask;
1.25 schwarze 206:
207: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 208: { NULL, 0 }, /* Ap */
209: { NULL, 0 }, /* Dd */
210: { NULL, 0 }, /* Dt */
211: { NULL, 0 }, /* Os */
212: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
213: { parse_mdoc_head, TYPE_Ss }, /* Ss */
214: { NULL, 0 }, /* Pp */
215: { NULL, 0 }, /* D1 */
216: { NULL, 0 }, /* Dl */
217: { NULL, 0 }, /* Bd */
218: { NULL, 0 }, /* Ed */
219: { NULL, 0 }, /* Bl */
220: { NULL, 0 }, /* El */
221: { NULL, 0 }, /* It */
222: { NULL, 0 }, /* Ad */
223: { NULL, TYPE_An }, /* An */
224: { NULL, TYPE_Ar }, /* Ar */
225: { NULL, TYPE_Cd }, /* Cd */
226: { NULL, TYPE_Cm }, /* Cm */
227: { NULL, TYPE_Dv }, /* Dv */
228: { NULL, TYPE_Er }, /* Er */
229: { NULL, TYPE_Ev }, /* Ev */
230: { NULL, 0 }, /* Ex */
231: { NULL, TYPE_Fa }, /* Fa */
232: { parse_mdoc_Fd, 0 }, /* Fd */
233: { NULL, TYPE_Fl }, /* Fl */
234: { parse_mdoc_Fn, 0 }, /* Fn */
235: { NULL, TYPE_Ft }, /* Ft */
236: { NULL, TYPE_Ic }, /* Ic */
1.92 schwarze 237: { NULL, TYPE_In }, /* In */
1.68 schwarze 238: { NULL, TYPE_Li }, /* Li */
1.132 schwarze 239: { parse_mdoc_Nd, 0 }, /* Nd */
1.129 schwarze 240: { parse_mdoc_Nm, 0 }, /* Nm */
1.68 schwarze 241: { NULL, 0 }, /* Op */
242: { NULL, 0 }, /* Ot */
243: { NULL, TYPE_Pa }, /* Pa */
244: { NULL, 0 }, /* Rv */
1.92 schwarze 245: { NULL, TYPE_St }, /* St */
1.68 schwarze 246: { NULL, TYPE_Va }, /* Va */
247: { parse_mdoc_body, TYPE_Va }, /* Vt */
248: { parse_mdoc_Xr, 0 }, /* Xr */
249: { NULL, 0 }, /* %A */
250: { NULL, 0 }, /* %B */
251: { NULL, 0 }, /* %D */
252: { NULL, 0 }, /* %I */
253: { NULL, 0 }, /* %J */
254: { NULL, 0 }, /* %N */
255: { NULL, 0 }, /* %O */
256: { NULL, 0 }, /* %P */
257: { NULL, 0 }, /* %R */
258: { NULL, 0 }, /* %T */
259: { NULL, 0 }, /* %V */
260: { NULL, 0 }, /* Ac */
261: { NULL, 0 }, /* Ao */
262: { NULL, 0 }, /* Aq */
263: { NULL, TYPE_At }, /* At */
264: { NULL, 0 }, /* Bc */
265: { NULL, 0 }, /* Bf */
266: { NULL, 0 }, /* Bo */
267: { NULL, 0 }, /* Bq */
268: { NULL, TYPE_Bsx }, /* Bsx */
269: { NULL, TYPE_Bx }, /* Bx */
270: { NULL, 0 }, /* Db */
271: { NULL, 0 }, /* Dc */
272: { NULL, 0 }, /* Do */
273: { NULL, 0 }, /* Dq */
274: { NULL, 0 }, /* Ec */
275: { NULL, 0 }, /* Ef */
276: { NULL, TYPE_Em }, /* Em */
277: { NULL, 0 }, /* Eo */
278: { NULL, TYPE_Fx }, /* Fx */
279: { NULL, TYPE_Ms }, /* Ms */
280: { NULL, 0 }, /* No */
281: { NULL, 0 }, /* Ns */
282: { NULL, TYPE_Nx }, /* Nx */
283: { NULL, TYPE_Ox }, /* Ox */
284: { NULL, 0 }, /* Pc */
285: { NULL, 0 }, /* Pf */
286: { NULL, 0 }, /* Po */
287: { NULL, 0 }, /* Pq */
288: { NULL, 0 }, /* Qc */
289: { NULL, 0 }, /* Ql */
290: { NULL, 0 }, /* Qo */
291: { NULL, 0 }, /* Qq */
292: { NULL, 0 }, /* Re */
293: { NULL, 0 }, /* Rs */
294: { NULL, 0 }, /* Sc */
295: { NULL, 0 }, /* So */
296: { NULL, 0 }, /* Sq */
297: { NULL, 0 }, /* Sm */
298: { NULL, 0 }, /* Sx */
299: { NULL, TYPE_Sy }, /* Sy */
300: { NULL, TYPE_Tn }, /* Tn */
301: { NULL, 0 }, /* Ux */
302: { NULL, 0 }, /* Xc */
303: { NULL, 0 }, /* Xo */
1.173 schwarze 304: { parse_mdoc_Fo, 0 }, /* Fo */
1.68 schwarze 305: { NULL, 0 }, /* Fc */
306: { NULL, 0 }, /* Oo */
307: { NULL, 0 }, /* Oc */
308: { NULL, 0 }, /* Bk */
309: { NULL, 0 }, /* Ek */
310: { NULL, 0 }, /* Bt */
311: { NULL, 0 }, /* Hf */
312: { NULL, 0 }, /* Fr */
313: { NULL, 0 }, /* Ud */
314: { NULL, TYPE_Lb }, /* Lb */
315: { NULL, 0 }, /* Lp */
316: { NULL, TYPE_Lk }, /* Lk */
317: { NULL, TYPE_Mt }, /* Mt */
318: { NULL, 0 }, /* Brq */
319: { NULL, 0 }, /* Bro */
320: { NULL, 0 }, /* Brc */
321: { NULL, 0 }, /* %C */
322: { NULL, 0 }, /* Es */
323: { NULL, 0 }, /* En */
324: { NULL, TYPE_Dx }, /* Dx */
325: { NULL, 0 }, /* %Q */
326: { NULL, 0 }, /* br */
327: { NULL, 0 }, /* sp */
328: { NULL, 0 }, /* %U */
329: { NULL, 0 }, /* Ta */
1.166 schwarze 330: { NULL, 0 }, /* ll */
1.1 kristaps 331: };
332:
1.141 schwarze 333:
1.1 kristaps 334: int
1.179 schwarze 335: mandocdb(int argc, char *argv[])
1.1 kristaps 336: {
1.187 schwarze 337: struct manconf conf;
338: struct mparse *mp;
339: const char *path_arg;
340: size_t j, sz;
1.59 schwarze 341: int ch, i;
1.50 kristaps 342:
1.187 schwarze 343: memset(&conf, 0, sizeof(conf));
1.50 kristaps 344: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
345:
346: /*
1.141 schwarze 347: * We accept a few different invocations.
1.50 kristaps 348: * The CHECKOP macro makes sure that invocation styles don't
349: * clobber each other.
350: */
351: #define CHECKOP(_op, _ch) do \
352: if (OP_DEFAULT != (_op)) { \
1.198 schwarze 353: warnx("-%c: Conflicting option", (_ch)); \
1.50 kristaps 354: goto usage; \
355: } while (/*CONSTCOND*/0)
1.10 kristaps 356:
1.59 schwarze 357: path_arg = NULL;
1.38 schwarze 358: op = OP_DEFAULT;
1.1 kristaps 359:
1.126 schwarze 360: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 361: switch (ch) {
1.141 schwarze 362: case 'a':
1.12 schwarze 363: use_all = 1;
364: break;
1.141 schwarze 365: case 'C':
1.50 kristaps 366: CHECKOP(op, ch);
1.59 schwarze 367: path_arg = optarg;
1.38 schwarze 368: op = OP_CONFFILE;
1.34 schwarze 369: break;
1.141 schwarze 370: case 'D':
1.125 schwarze 371: debug++;
372: break;
1.141 schwarze 373: case 'd':
1.50 kristaps 374: CHECKOP(op, ch);
1.59 schwarze 375: path_arg = optarg;
1.5 kristaps 376: op = OP_UPDATE;
377: break;
1.141 schwarze 378: case 'n':
1.50 kristaps 379: nodb = 1;
380: break;
1.141 schwarze 381: case 'p':
1.126 schwarze 382: warnings = 1;
383: break;
1.141 schwarze 384: case 'Q':
1.116 schwarze 385: mparse_options |= MPARSE_QUICK;
1.102 schwarze 386: break;
1.141 schwarze 387: case 'T':
1.95 schwarze 388: if (strcmp(optarg, "utf8")) {
1.198 schwarze 389: warnx("-T%s: Unsupported output format",
390: optarg);
1.95 schwarze 391: goto usage;
392: }
393: write_utf8 = 1;
394: break;
1.141 schwarze 395: case 't':
1.50 kristaps 396: CHECKOP(op, ch);
1.38 schwarze 397: dup2(STDOUT_FILENO, STDERR_FILENO);
398: op = OP_TEST;
1.50 kristaps 399: nodb = warnings = 1;
1.38 schwarze 400: break;
1.141 schwarze 401: case 'u':
1.50 kristaps 402: CHECKOP(op, ch);
1.59 schwarze 403: path_arg = optarg;
1.5 kristaps 404: op = OP_DELETE;
405: break;
1.141 schwarze 406: case 'v':
1.126 schwarze 407: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 408: break;
1.1 kristaps 409: default:
1.38 schwarze 410: goto usage;
1.1 kristaps 411: }
412:
413: argc -= optind;
414: argv += optind;
415:
1.38 schwarze 416: if (OP_CONFFILE == op && argc > 0) {
1.198 schwarze 417: warnx("-C: Too many arguments");
1.38 schwarze 418: goto usage;
419: }
420:
1.59 schwarze 421: exitcode = (int)MANDOCLEVEL_OK;
1.176 schwarze 422: mchars = mchars_alloc();
1.183 schwarze 423: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL,
1.176 schwarze 424: mchars, NULL);
1.202 ! schwarze 425: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
! 426: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.50 kristaps 427:
428: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 429:
1.50 kristaps 430: /*
1.150 schwarze 431: * Most of these deal with a specific directory.
1.140 schwarze 432: * Jump into that directory first.
1.50 kristaps 433: */
1.165 schwarze 434: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.50 kristaps 435: goto out;
1.140 schwarze 436:
1.138 schwarze 437: if (dbopen(1)) {
1.140 schwarze 438: /*
439: * The existing database is usable. Process
440: * all files specified on the command-line.
441: */
442: use_all = 1;
1.138 schwarze 443: for (i = 0; i < argc; i++)
444: filescan(argv[i]);
445: if (OP_TEST != op)
446: dbprune();
447: } else {
448: /*
449: * Database missing or corrupt.
450: * Recreate from scratch.
451: */
1.140 schwarze 452: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 453: op = OP_DEFAULT;
454: if (0 == treescan())
455: goto out;
456: if (0 == dbopen(0))
457: goto out;
458: }
1.50 kristaps 459: if (OP_DELETE != op)
1.176 schwarze 460: mpages_merge(mp);
1.138 schwarze 461: dbclose(OP_DEFAULT == op ? 0 : 1);
1.50 kristaps 462: } else {
463: /*
464: * If we have arguments, use them as our manpaths.
465: * If we don't, grok from manpath(1) or however else
1.187 schwarze 466: * manconf_parse() wants to do it.
1.50 kristaps 467: */
468: if (argc > 0) {
1.187 schwarze 469: conf.manpath.paths = mandoc_reallocarray(NULL,
1.144 schwarze 470: argc, sizeof(char *));
1.187 schwarze 471: conf.manpath.sz = (size_t)argc;
1.50 kristaps 472: for (i = 0; i < argc; i++)
1.187 schwarze 473: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.50 kristaps 474: } else
1.187 schwarze 475: manconf_parse(&conf, path_arg, NULL, NULL);
1.127 schwarze 476:
1.187 schwarze 477: if (conf.manpath.sz == 0) {
1.127 schwarze 478: exitcode = (int)MANDOCLEVEL_BADARG;
479: say("", "Empty manpath");
480: }
1.50 kristaps 481:
482: /*
1.69 schwarze 483: * First scan the tree rooted at a base directory, then
484: * build a new database and finally move it into place.
1.50 kristaps 485: * Ignore zero-length directories and strip trailing
486: * slashes.
487: */
1.187 schwarze 488: for (j = 0; j < conf.manpath.sz; j++) {
489: sz = strlen(conf.manpath.paths[j]);
490: if (sz && conf.manpath.paths[j][sz - 1] == '/')
491: conf.manpath.paths[j][--sz] = '\0';
1.50 kristaps 492: if (0 == sz)
493: continue;
1.66 schwarze 494:
495: if (j) {
1.202 ! schwarze 496: mandoc_ohash_init(&mpages, 6,
! 497: offsetof(struct mpage, inodev));
! 498: mandoc_ohash_init(&mlinks, 6,
! 499: offsetof(struct mlink, file));
1.66 schwarze 500: }
501:
1.187 schwarze 502: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.165 schwarze 503: continue;
1.59 schwarze 504: if (0 == treescan())
1.165 schwarze 505: continue;
1.59 schwarze 506: if (0 == dbopen(0))
1.165 schwarze 507: continue;
1.52 kristaps 508:
1.176 schwarze 509: mpages_merge(mp);
1.147 schwarze 510: if (warnings && !nodb &&
1.130 schwarze 511: ! (MPARSE_QUICK & mparse_options))
512: names_check();
1.59 schwarze 513: dbclose(0);
1.66 schwarze 514:
1.187 schwarze 515: if (j + 1 < conf.manpath.sz) {
1.80 schwarze 516: mpages_free();
1.78 schwarze 517: ohash_delete(&mpages);
1.83 schwarze 518: ohash_delete(&mlinks);
1.66 schwarze 519: }
1.50 kristaps 520: }
521: }
522: out:
1.187 schwarze 523: manconf_free(&conf);
1.168 schwarze 524: mparse_free(mp);
1.176 schwarze 525: mchars_free(mchars);
1.80 schwarze 526: mpages_free();
1.78 schwarze 527: ohash_delete(&mpages);
1.83 schwarze 528: ohash_delete(&mlinks);
1.197 schwarze 529: return exitcode;
1.50 kristaps 530: usage:
1.126 schwarze 531: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
532: " %s [-aDnpQ] [-Tutf8] dir ...\n"
533: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
534: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 535: " %s [-Q] -t file ...\n",
1.198 schwarze 536: __progname, __progname, __progname,
537: __progname, __progname);
1.50 kristaps 538:
1.197 schwarze 539: return (int)MANDOCLEVEL_BADARG;
1.50 kristaps 540: }
541:
542: /*
1.59 schwarze 543: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 544: * We use fts(), scanning directory parts along the way for clues to our
545: * section and architecture.
546: *
547: * If use_all has been specified, grok all files.
548: * If not, sanitise paths to the following:
549: *
1.141 schwarze 550: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 551: * or
552: * [./]cat<section>[/<arch>]/<name>.0
553: *
554: * TODO: accomodate for multi-language directories.
555: */
556: static int
1.59 schwarze 557: treescan(void)
1.50 kristaps 558: {
1.139 schwarze 559: char buf[PATH_MAX];
1.50 kristaps 560: FTS *f;
561: FTSENT *ff;
1.84 schwarze 562: struct mlink *mlink;
1.124 schwarze 563: int dform, gzip;
1.94 schwarze 564: char *dsec, *arch, *fsec, *cp;
565: const char *path;
1.50 kristaps 566: const char *argv[2];
567:
568: argv[0] = ".";
569: argv[1] = (char *)NULL;
570:
1.139 schwarze 571: f = fts_open((char * const *)argv,
572: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.200 schwarze 573: if (f == NULL) {
1.59 schwarze 574: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 575: say("", "&fts_open");
1.197 schwarze 576: return 0;
1.50 kristaps 577: }
578:
579: dsec = arch = NULL;
580: dform = FORM_NONE;
581:
1.200 schwarze 582: while ((ff = fts_read(f)) != NULL) {
1.50 kristaps 583: path = ff->fts_path + 2;
1.139 schwarze 584: switch (ff->fts_info) {
585:
586: /*
587: * Symbolic links require various sanity checks,
588: * then get handled just like regular files.
589: */
1.141 schwarze 590: case FTS_SL:
1.200 schwarze 591: if (realpath(path, buf) == NULL) {
1.139 schwarze 592: if (warnings)
593: say(path, "&realpath");
594: continue;
595: }
1.184 schwarze 596: if (strstr(buf, basedir) != buf
597: #ifdef HOMEBREWDIR
598: && strstr(buf, HOMEBREWDIR) != buf
599: #endif
600: ) {
1.139 schwarze 601: if (warnings) say("",
602: "%s: outside base directory", buf);
603: continue;
604: }
605: /* Use logical inode to avoid mpages dupe. */
1.200 schwarze 606: if (stat(path, ff->fts_statp) == -1) {
1.139 schwarze 607: if (warnings)
608: say(path, "&stat");
609: continue;
610: }
611: /* FALLTHROUGH */
612:
1.50 kristaps 613: /*
1.83 schwarze 614: * If we're a regular file, add an mlink by using the
1.50 kristaps 615: * stored directory data and handling the filename.
616: */
1.141 schwarze 617: case FTS_F:
1.200 schwarze 618: if ( ! strcmp(path, MANDOC_DB))
1.60 schwarze 619: continue;
1.50 kristaps 620: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 621: if (warnings)
1.59 schwarze 622: say(path, "Extraneous file");
1.50 kristaps 623: continue;
1.124 schwarze 624: }
625: gzip = 0;
626: fsec = NULL;
1.200 schwarze 627: while (fsec == NULL) {
1.124 schwarze 628: fsec = strrchr(ff->fts_name, '.');
1.200 schwarze 629: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.124 schwarze 630: break;
631: gzip = 1;
632: *fsec = '\0';
633: fsec = NULL;
634: }
1.200 schwarze 635: if (fsec == NULL) {
1.60 schwarze 636: if ( ! use_all) {
1.56 schwarze 637: if (warnings)
1.60 schwarze 638: say(path,
639: "No filename suffix");
1.50 kristaps 640: continue;
641: }
1.200 schwarze 642: } else if ( ! strcmp(++fsec, "html")) {
1.60 schwarze 643: if (warnings)
644: say(path, "Skip html");
645: continue;
1.200 schwarze 646: } else if ( ! strcmp(fsec, "ps")) {
1.60 schwarze 647: if (warnings)
648: say(path, "Skip ps");
649: continue;
1.200 schwarze 650: } else if ( ! strcmp(fsec, "pdf")) {
1.60 schwarze 651: if (warnings)
652: say(path, "Skip pdf");
653: continue;
654: } else if ( ! use_all &&
1.200 schwarze 655: ((dform == FORM_SRC &&
1.185 schwarze 656: strncmp(fsec, dsec, strlen(dsec))) ||
1.200 schwarze 657: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.60 schwarze 658: if (warnings)
659: say(path, "Wrong filename suffix");
660: continue;
1.66 schwarze 661: } else
1.84 schwarze 662: fsec[-1] = '\0';
1.94 schwarze 663:
1.84 schwarze 664: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 665: if (strlcpy(mlink->file, path,
666: sizeof(mlink->file)) >=
667: sizeof(mlink->file)) {
668: say(path, "Filename too long");
669: free(mlink);
670: continue;
671: }
1.84 schwarze 672: mlink->dform = dform;
1.94 schwarze 673: mlink->dsec = dsec;
674: mlink->arch = arch;
675: mlink->name = ff->fts_name;
676: mlink->fsec = fsec;
1.124 schwarze 677: mlink->gzip = gzip;
1.84 schwarze 678: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 679: continue;
1.139 schwarze 680:
1.141 schwarze 681: case FTS_D:
682: case FTS_DP:
1.139 schwarze 683: break;
684:
685: default:
1.60 schwarze 686: if (warnings)
687: say(path, "Not a regular file");
1.50 kristaps 688: continue;
1.60 schwarze 689: }
1.1 kristaps 690:
1.50 kristaps 691: switch (ff->fts_level) {
1.141 schwarze 692: case 0:
1.50 kristaps 693: /* Ignore the root directory. */
694: break;
1.141 schwarze 695: case 1:
1.50 kristaps 696: /*
697: * This might contain manX/ or catX/.
698: * Try to infer this from the name.
699: * If we're not in use_all, enforce it.
700: */
701: cp = ff->fts_name;
1.200 schwarze 702: if (ff->fts_info == FTS_DP) {
703: dform = FORM_NONE;
704: dsec = NULL;
1.50 kristaps 705: break;
1.200 schwarze 706: }
1.1 kristaps 707:
1.200 schwarze 708: if ( ! strncmp(cp, "man", 3)) {
1.50 kristaps 709: dform = FORM_SRC;
1.66 schwarze 710: dsec = cp + 3;
1.200 schwarze 711: } else if ( ! strncmp(cp, "cat", 3)) {
1.50 kristaps 712: dform = FORM_CAT;
1.66 schwarze 713: dsec = cp + 3;
1.94 schwarze 714: } else {
715: dform = FORM_NONE;
716: dsec = NULL;
1.50 kristaps 717: }
1.1 kristaps 718:
1.200 schwarze 719: if (dsec != NULL || use_all)
1.50 kristaps 720: break;
1.1 kristaps 721:
1.56 schwarze 722: if (warnings)
1.59 schwarze 723: say(path, "Unknown directory part");
1.50 kristaps 724: fts_set(f, ff, FTS_SKIP);
725: break;
1.141 schwarze 726: case 2:
1.50 kristaps 727: /*
728: * Possibly our architecture.
729: * If we're descending, keep tabs on it.
730: */
1.200 schwarze 731: if (ff->fts_info != FTS_DP && dsec != NULL)
1.66 schwarze 732: arch = ff->fts_name;
1.94 schwarze 733: else
734: arch = NULL;
1.50 kristaps 735: break;
736: default:
1.200 schwarze 737: if (ff->fts_info == FTS_DP || use_all)
1.50 kristaps 738: break;
1.56 schwarze 739: if (warnings)
1.59 schwarze 740: say(path, "Extraneous directory part");
1.50 kristaps 741: fts_set(f, ff, FTS_SKIP);
742: break;
1.5 kristaps 743: }
1.50 kristaps 744: }
745:
746: fts_close(f);
1.197 schwarze 747: return 1;
1.50 kristaps 748: }
1.5 kristaps 749:
1.50 kristaps 750: /*
1.89 schwarze 751: * Add a file to the mlinks table.
1.50 kristaps 752: * Do not verify that it's a "valid" looking manpage (we'll do that
753: * later).
754: *
755: * Try to infer the manual section, architecture, and page name from the
756: * path, assuming it looks like
757: *
1.141 schwarze 758: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 759: * or
760: * [./]cat<section>[/<arch>]/<name>.0
761: *
762: * See treescan() for the fts(3) version of this.
763: */
764: static void
1.59 schwarze 765: filescan(const char *file)
1.50 kristaps 766: {
1.59 schwarze 767: char buf[PATH_MAX];
1.84 schwarze 768: struct stat st;
769: struct mlink *mlink;
1.59 schwarze 770: char *p, *start;
1.5 kristaps 771:
1.50 kristaps 772: assert(use_all);
1.5 kristaps 773:
1.50 kristaps 774: if (0 == strncmp(file, "./", 2))
775: file += 2;
1.5 kristaps 776:
1.139 schwarze 777: /*
778: * We have to do lstat(2) before realpath(3) loses
779: * the information whether this is a symbolic link.
780: * We need to know that because for symbolic links,
781: * we want to use the orginal file name, while for
782: * regular files, we want to use the real path.
783: */
784: if (-1 == lstat(file, &st)) {
785: exitcode = (int)MANDOCLEVEL_BADARG;
786: say(file, "&lstat");
787: return;
788: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
789: exitcode = (int)MANDOCLEVEL_BADARG;
790: say(file, "Not a regular file");
791: return;
792: }
793:
794: /*
795: * We have to resolve the file name to the real path
796: * in any case for the base directory check.
797: */
1.59 schwarze 798: if (NULL == realpath(file, buf)) {
799: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 800: say(file, "&realpath");
1.59 schwarze 801: return;
1.106 schwarze 802: }
803:
1.150 schwarze 804: if (OP_TEST == op)
1.106 schwarze 805: start = buf;
1.150 schwarze 806: else if (strstr(buf, basedir) == buf)
807: start = buf + strlen(basedir);
1.184 schwarze 808: #ifdef HOMEBREWDIR
809: else if (strstr(buf, HOMEBREWDIR) == buf)
810: start = buf;
811: #endif
1.106 schwarze 812: else {
1.59 schwarze 813: exitcode = (int)MANDOCLEVEL_BADARG;
814: say("", "%s: outside base directory", buf);
815: return;
1.106 schwarze 816: }
817:
1.139 schwarze 818: /*
819: * Now we are sure the file is inside our tree.
820: * If it is a symbolic link, ignore the real path
821: * and use the original name.
822: * This implies passing stuff like "cat1/../man1/foo.1"
823: * on the command line won't work. So don't do that.
824: * Note the stat(2) can still fail if the link target
825: * doesn't exist.
826: */
827: if (S_IFLNK & st.st_mode) {
828: if (-1 == stat(buf, &st)) {
829: exitcode = (int)MANDOCLEVEL_BADARG;
830: say(file, "&stat");
831: return;
832: }
1.143 schwarze 833: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
834: say(file, "Filename too long");
835: return;
836: }
1.150 schwarze 837: start = buf;
838: if (OP_TEST != op && strstr(buf, basedir) == buf)
839: start += strlen(basedir);
1.50 kristaps 840: }
1.106 schwarze 841:
1.84 schwarze 842: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 843: mlink->dform = FORM_NONE;
1.143 schwarze 844: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
845: sizeof(mlink->file)) {
846: say(start, "Filename too long");
1.180 schwarze 847: free(mlink);
1.143 schwarze 848: return;
849: }
1.17 schwarze 850:
1.50 kristaps 851: /*
852: * First try to guess our directory structure.
853: * If we find a separator, try to look for man* or cat*.
854: * If we find one of these and what's underneath is a directory,
855: * assume it's an architecture.
856: */
857: if (NULL != (p = strchr(start, '/'))) {
858: *p++ = '\0';
859: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 860: mlink->dform = FORM_SRC;
1.94 schwarze 861: mlink->dsec = start + 3;
1.50 kristaps 862: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 863: mlink->dform = FORM_CAT;
1.94 schwarze 864: mlink->dsec = start + 3;
1.17 schwarze 865: }
1.5 kristaps 866:
1.50 kristaps 867: start = p;
1.84 schwarze 868: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 869: *p++ = '\0';
1.94 schwarze 870: mlink->arch = start;
1.50 kristaps 871: start = p;
1.84 schwarze 872: }
1.50 kristaps 873: }
874:
875: /*
876: * Now check the file suffix.
877: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
878: */
879: p = strrchr(start, '\0');
880: while (p-- > start && '/' != *p && '.' != *p)
881: /* Loop. */ ;
882:
883: if ('.' == *p) {
884: *p++ = '\0';
1.94 schwarze 885: mlink->fsec = p;
1.5 kristaps 886: }
887:
1.10 kristaps 888: /*
1.50 kristaps 889: * Now try to parse the name.
890: * Use the filename portion of the path.
1.10 kristaps 891: */
1.84 schwarze 892: mlink->name = start;
1.50 kristaps 893: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 894: mlink->name = p + 1;
1.50 kristaps 895: *p = '\0';
1.84 schwarze 896: }
897: mlink_add(mlink, &st);
1.50 kristaps 898: }
1.5 kristaps 899:
1.50 kristaps 900: static void
1.84 schwarze 901: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 902: {
1.83 schwarze 903: struct inodev inodev;
1.78 schwarze 904: struct mpage *mpage;
1.83 schwarze 905: unsigned int slot;
1.50 kristaps 906:
1.84 schwarze 907: assert(NULL != mlink->file);
1.50 kristaps 908:
1.94 schwarze 909: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
910: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
911: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
912: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 913:
914: if ('0' == *mlink->fsec) {
915: free(mlink->fsec);
916: mlink->fsec = mandoc_strdup(mlink->dsec);
917: mlink->fform = FORM_CAT;
918: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
919: mlink->fform = FORM_SRC;
1.74 schwarze 920: else
1.84 schwarze 921: mlink->fform = FORM_NONE;
1.82 schwarze 922:
1.83 schwarze 923: slot = ohash_qlookup(&mlinks, mlink->file);
924: assert(NULL == ohash_find(&mlinks, slot));
925: ohash_insert(&mlinks, slot, mlink);
926:
1.186 schwarze 927: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 928: inodev.st_ino = st->st_ino;
929: inodev.st_dev = st->st_dev;
930: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
931: sizeof(struct inodev), inodev.st_ino);
932: mpage = ohash_find(&mpages, slot);
933: if (NULL == mpage) {
934: mpage = mandoc_calloc(1, sizeof(struct mpage));
935: mpage->inodev.st_ino = inodev.st_ino;
936: mpage->inodev.st_dev = inodev.st_dev;
937: ohash_insert(&mpages, slot, mpage);
938: } else
1.85 schwarze 939: mlink->next = mpage->mlinks;
1.82 schwarze 940: mpage->mlinks = mlink;
1.118 schwarze 941: mlink->mpage = mpage;
1.82 schwarze 942: }
943:
944: static void
945: mlink_free(struct mlink *mlink)
946: {
947:
948: free(mlink->dsec);
949: free(mlink->arch);
950: free(mlink->name);
951: free(mlink->fsec);
952: free(mlink);
1.50 kristaps 953: }
1.3 kristaps 954:
1.50 kristaps 955: static void
1.78 schwarze 956: mpages_free(void)
1.50 kristaps 957: {
1.78 schwarze 958: struct mpage *mpage;
1.82 schwarze 959: struct mlink *mlink;
1.80 schwarze 960: unsigned int slot;
1.3 kristaps 961:
1.80 schwarze 962: mpage = ohash_first(&mpages, &slot);
963: while (NULL != mpage) {
1.82 schwarze 964: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 965: mpage->mlinks = mlink->next;
1.82 schwarze 966: mlink_free(mlink);
967: }
1.78 schwarze 968: free(mpage->sec);
969: free(mpage->arch);
1.82 schwarze 970: free(mpage->title);
971: free(mpage->desc);
1.78 schwarze 972: free(mpage);
1.80 schwarze 973: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 974: }
975: }
1.38 schwarze 976:
1.50 kristaps 977: /*
1.89 schwarze 978: * For each mlink to the mpage, check whether the path looks like
979: * it is formatted, and if it does, check whether a source manual
980: * exists by the same name, ignoring the suffix.
981: * If both conditions hold, drop the mlink.
982: */
983: static void
984: mlinks_undupe(struct mpage *mpage)
985: {
986: char buf[PATH_MAX];
987: struct mlink **prev;
988: struct mlink *mlink;
989: char *bufp;
990:
991: mpage->form = FORM_CAT;
1.90 schwarze 992: prev = &mpage->mlinks;
993: while (NULL != (mlink = *prev)) {
1.89 schwarze 994: if (FORM_CAT != mlink->dform) {
995: mpage->form = FORM_NONE;
1.90 schwarze 996: goto nextlink;
1.89 schwarze 997: }
1.143 schwarze 998: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 999: bufp = strstr(buf, "cat");
1000: assert(NULL != bufp);
1001: memcpy(bufp, "man", 3);
1002: if (NULL != (bufp = strrchr(buf, '.')))
1003: *++bufp = '\0';
1.143 schwarze 1004: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1005: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1006: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1007: goto nextlink;
1.89 schwarze 1008: if (warnings)
1009: say(mlink->file, "Man source exists: %s", buf);
1010: if (use_all)
1.90 schwarze 1011: goto nextlink;
1.89 schwarze 1012: *prev = mlink->next;
1013: mlink_free(mlink);
1.90 schwarze 1014: continue;
1015: nextlink:
1016: prev = &(*prev)->next;
1.89 schwarze 1017: }
1018: }
1019:
1.131 schwarze 1020: static void
1.93 schwarze 1021: mlink_check(struct mpage *mpage, struct mlink *mlink)
1022: {
1.131 schwarze 1023: struct str *str;
1024: unsigned int slot;
1.93 schwarze 1025:
1026: /*
1027: * Check whether the manual section given in a file
1028: * agrees with the directory where the file is located.
1029: * Some manuals have suffixes like (3p) on their
1030: * section number either inside the file or in the
1031: * directory name, some are linked into more than one
1032: * section, like encrypt(1) = makekey(8).
1033: */
1034:
1035: if (FORM_SRC == mpage->form &&
1.131 schwarze 1036: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1037: say(mlink->file, "Section \"%s\" manual in %s directory",
1038: mpage->sec, mlink->dsec);
1039:
1040: /*
1041: * Manual page directories exist for each kernel
1042: * architecture as returned by machine(1).
1043: * However, many manuals only depend on the
1044: * application architecture as returned by arch(1).
1045: * For example, some (2/ARM) manuals are shared
1046: * across the "armish" and "zaurus" kernel
1047: * architectures.
1048: * A few manuals are even shared across completely
1049: * different architectures, for example fdformat(1)
1050: * on amd64, i386, sparc, and sparc64.
1051: */
1052:
1.131 schwarze 1053: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1054: say(mlink->file, "Architecture \"%s\" manual in "
1055: "\"%s\" directory", mpage->arch, mlink->arch);
1056:
1.131 schwarze 1057: /*
1058: * XXX
1.133 schwarze 1059: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1060: */
1061:
1062: if (FORM_CAT == mpage->form)
1063: return;
1064:
1065: /*
1066: * Check whether this mlink
1067: * appears as a name in the NAME section.
1068: */
1.93 schwarze 1069:
1.133 schwarze 1070: slot = ohash_qlookup(&names, mlink->name);
1071: str = ohash_find(&names, slot);
1.131 schwarze 1072: assert(NULL != str);
1.133 schwarze 1073: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1074: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1075: }
1076:
1.89 schwarze 1077: /*
1.80 schwarze 1078: * Run through the files in the global vector "mpages"
1079: * and add them to the database specified in "basedir".
1.50 kristaps 1080: *
1081: * This handles the parsing scheme itself, using the cues of directory
1082: * and filename to determine whether the file is parsable or not.
1083: */
1.59 schwarze 1084: static void
1.176 schwarze 1085: mpages_merge(struct mparse *mp)
1.50 kristaps 1086: {
1.113 schwarze 1087: char any[] = "any";
1.118 schwarze 1088: struct mpage *mpage, *mpage_dest;
1089: struct mlink *mlink, *mlink_dest;
1.191 schwarze 1090: struct roff_man *man;
1.118 schwarze 1091: char *sodest;
1.112 schwarze 1092: char *cp;
1.162 schwarze 1093: int fd;
1.101 schwarze 1094: unsigned int pslot;
1.69 schwarze 1095:
1.171 schwarze 1096: if ( ! nodb)
1.107 schwarze 1097: SQL_EXEC("BEGIN TRANSACTION");
1098:
1.80 schwarze 1099: mpage = ohash_first(&mpages, &pslot);
1.171 schwarze 1100: while (mpage != NULL) {
1.89 schwarze 1101: mlinks_undupe(mpage);
1.182 schwarze 1102: if ((mlink = mpage->mlinks) == NULL) {
1.89 schwarze 1103: mpage = ohash_next(&mpages, &pslot);
1104: continue;
1.50 kristaps 1105: }
1106:
1.133 schwarze 1107: name_mask = NAME_MASK;
1.202 ! schwarze 1108: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
! 1109: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.50 kristaps 1110: mparse_reset(mp);
1111: man = NULL;
1.124 schwarze 1112: sodest = NULL;
1113:
1.182 schwarze 1114: mparse_open(mp, &fd, mlink->file);
1.162 schwarze 1115: if (fd == -1) {
1.182 schwarze 1116: say(mlink->file, "&open");
1.162 schwarze 1117: goto nextpage;
1.124 schwarze 1118: }
1.14 schwarze 1119:
1120: /*
1.183 schwarze 1121: * Interpret the file as mdoc(7) or man(7) source
1122: * code, unless it is known to be formatted.
1.14 schwarze 1123: */
1.182 schwarze 1124: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1125: mparse_readfd(mp, fd, mlink->file);
1.192 schwarze 1126: mparse_result(mp, &man, &sodest);
1.76 schwarze 1127: }
1.14 schwarze 1128:
1.171 schwarze 1129: if (sodest != NULL) {
1.118 schwarze 1130: mlink_dest = ohash_find(&mlinks,
1131: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1132: if (mlink_dest == NULL) {
1133: mandoc_asprintf(&cp, "%s.gz", sodest);
1134: mlink_dest = ohash_find(&mlinks,
1135: ohash_qlookup(&mlinks, cp));
1136: free(cp);
1137: }
1138: if (mlink_dest != NULL) {
1.118 schwarze 1139:
1140: /* The .so target exists. */
1141:
1142: mpage_dest = mlink_dest->mpage;
1143: while (1) {
1144: mlink->mpage = mpage_dest;
1145:
1146: /*
1147: * If the target was already
1148: * processed, add the links
1149: * to the database now.
1150: * Otherwise, this will
1151: * happen when we come
1152: * to the target.
1153: */
1154:
1.137 schwarze 1155: if (mpage_dest->pageid)
1.169 schwarze 1156: dbadd_mlink_name(mlink);
1.118 schwarze 1157:
1.171 schwarze 1158: if (mlink->next == NULL)
1.118 schwarze 1159: break;
1160: mlink = mlink->next;
1161: }
1162:
1163: /* Move all links to the target. */
1164:
1165: mlink->next = mlink_dest->next;
1166: mlink_dest->next = mpage->mlinks;
1167: mpage->mlinks = NULL;
1168: }
1.124 schwarze 1169: goto nextpage;
1.192 schwarze 1170: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.82 schwarze 1171: mpage->form = FORM_SRC;
1.193 schwarze 1172: mpage->sec = man->meta.msec;
1.155 schwarze 1173: mpage->sec = mandoc_strdup(
1.171 schwarze 1174: mpage->sec == NULL ? "" : mpage->sec);
1.193 schwarze 1175: mpage->arch = man->meta.arch;
1.84 schwarze 1176: mpage->arch = mandoc_strdup(
1.171 schwarze 1177: mpage->arch == NULL ? "" : mpage->arch);
1.193 schwarze 1178: mpage->title = mandoc_strdup(man->meta.title);
1.192 schwarze 1179: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.82 schwarze 1180: mpage->form = FORM_SRC;
1.193 schwarze 1181: mpage->sec = mandoc_strdup(man->meta.msec);
1.182 schwarze 1182: mpage->arch = mandoc_strdup(mlink->arch);
1.193 schwarze 1183: mpage->title = mandoc_strdup(man->meta.title);
1.76 schwarze 1184: } else {
1.82 schwarze 1185: mpage->form = FORM_CAT;
1.182 schwarze 1186: mpage->sec = mandoc_strdup(mlink->dsec);
1187: mpage->arch = mandoc_strdup(mlink->arch);
1188: mpage->title = mandoc_strdup(mlink->name);
1.76 schwarze 1189: }
1.97 schwarze 1190: putkey(mpage, mpage->sec, TYPE_sec);
1.164 schwarze 1191: if (*mpage->arch != '\0')
1192: putkey(mpage, mpage->arch, TYPE_arch);
1.50 kristaps 1193:
1.182 schwarze 1194: for ( ; mlink != NULL; mlink = mlink->next) {
1.97 schwarze 1195: if ('\0' != *mlink->dsec)
1196: putkey(mpage, mlink->dsec, TYPE_sec);
1197: if ('\0' != *mlink->fsec)
1198: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1199: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1200: any : mlink->arch, TYPE_arch);
1.133 schwarze 1201: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1202: }
1.12 schwarze 1203:
1.174 schwarze 1204: assert(mpage->desc == NULL);
1.192 schwarze 1205: if (man != NULL && man->macroset == MACROSET_MDOC)
1.193 schwarze 1206: parse_mdoc(mpage, &man->meta, man->first);
1.174 schwarze 1207: else if (man != NULL)
1.193 schwarze 1208: parse_man(mpage, &man->meta, man->first);
1.50 kristaps 1209: else
1.162 schwarze 1210: parse_cat(mpage, fd);
1.174 schwarze 1211: if (mpage->desc == NULL)
1.135 schwarze 1212: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1213:
1214: if (warnings && !use_all)
1215: for (mlink = mpage->mlinks; mlink;
1216: mlink = mlink->next)
1217: mlink_check(mpage, mlink);
1.44 kristaps 1218:
1.176 schwarze 1219: dbadd(mpage);
1.182 schwarze 1220: mlink = mpage->mlinks;
1.124 schwarze 1221:
1222: nextpage:
1.66 schwarze 1223: ohash_delete(&strings);
1.133 schwarze 1224: ohash_delete(&names);
1.80 schwarze 1225: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1226: }
1.107 schwarze 1227:
1228: if (0 == nodb)
1229: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1230: }
1231:
1232: static void
1233: names_check(void)
1234: {
1235: sqlite3_stmt *stmt;
1236: const char *name, *sec, *arch, *key;
1237:
1238: sqlite3_prepare_v2(db,
1239: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1240: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1241: "WHERE bits & ? AND NOT EXISTS ("
1242: "SELECT pageid FROM mlinks "
1.133 schwarze 1243: "WHERE mlinks.pageid == names.pageid "
1244: "AND mlinks.name == names.name"
1.130 schwarze 1245: ")"
1246: ") JOIN ("
1.142 schwarze 1247: "SELECT sec, arch, name, pageid FROM mlinks "
1248: "GROUP BY pageid"
1.130 schwarze 1249: ") USING (pageid);",
1250: -1, &stmt, NULL);
1251:
1.201 schwarze 1252: if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK)
1.134 schwarze 1253: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1254:
1.201 schwarze 1255: while (sqlite3_step(stmt) == SQLITE_ROW) {
1.154 schwarze 1256: name = (const char *)sqlite3_column_text(stmt, 0);
1257: sec = (const char *)sqlite3_column_text(stmt, 1);
1258: arch = (const char *)sqlite3_column_text(stmt, 2);
1259: key = (const char *)sqlite3_column_text(stmt, 3);
1.130 schwarze 1260: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1261: '\0' == *arch ? "" : "/",
1262: '\0' == *arch ? "" : arch, key);
1263: }
1264: sqlite3_finalize(stmt);
1.50 kristaps 1265: }
1.12 schwarze 1266:
1.50 kristaps 1267: static void
1.124 schwarze 1268: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1269: {
1270: FILE *stream;
1271: char *line, *p, *title;
1272: size_t len, plen, titlesz;
1.12 schwarze 1273:
1.124 schwarze 1274: stream = (-1 == fd) ?
1275: fopen(mpage->mlinks->file, "r") :
1276: fdopen(fd, "r");
1277: if (NULL == stream) {
1.159 schwarze 1278: if (-1 != fd)
1279: close(fd);
1.56 schwarze 1280: if (warnings)
1.123 schwarze 1281: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1282: return;
1283: }
1.1 kristaps 1284:
1.50 kristaps 1285: /* Skip to first blank line. */
1.1 kristaps 1286:
1.50 kristaps 1287: while (NULL != (line = fgetln(stream, &len)))
1288: if ('\n' == *line)
1289: break;
1.1 kristaps 1290:
1.50 kristaps 1291: /*
1292: * Assume the first line that is not indented
1293: * is the first section header. Skip to it.
1294: */
1.1 kristaps 1295:
1.50 kristaps 1296: while (NULL != (line = fgetln(stream, &len)))
1297: if ('\n' != *line && ' ' != *line)
1298: break;
1.141 schwarze 1299:
1.50 kristaps 1300: /*
1301: * Read up until the next section into a buffer.
1302: * Strip the leading and trailing newline from each read line,
1303: * appending a trailing space.
1304: * Ignore empty (whitespace-only) lines.
1305: */
1.1 kristaps 1306:
1.50 kristaps 1307: titlesz = 0;
1308: title = NULL;
1.38 schwarze 1309:
1.50 kristaps 1310: while (NULL != (line = fgetln(stream, &len))) {
1311: if (' ' != *line || '\n' != line[len - 1])
1312: break;
1313: while (len > 0 && isspace((unsigned char)*line)) {
1314: line++;
1315: len--;
1316: }
1317: if (1 == len)
1.38 schwarze 1318: continue;
1.50 kristaps 1319: title = mandoc_realloc(title, titlesz + len);
1320: memcpy(title + titlesz, line, len);
1321: titlesz += len;
1322: title[titlesz - 1] = ' ';
1323: }
1.38 schwarze 1324:
1.50 kristaps 1325: /*
1326: * If no page content can be found, or the input line
1327: * is already the next section header, or there is no
1328: * trailing newline, reuse the page title as the page
1329: * description.
1330: */
1.44 kristaps 1331:
1.50 kristaps 1332: if (NULL == title || '\0' == *title) {
1.56 schwarze 1333: if (warnings)
1.82 schwarze 1334: say(mpage->mlinks->file,
1335: "Cannot find NAME section");
1.50 kristaps 1336: fclose(stream);
1337: free(title);
1338: return;
1339: }
1.1 kristaps 1340:
1.50 kristaps 1341: title = mandoc_realloc(title, titlesz + 1);
1342: title[titlesz] = '\0';
1.33 schwarze 1343:
1.50 kristaps 1344: /*
1345: * Skip to the first dash.
1346: * Use the remaining line as the description (no more than 70
1347: * bytes).
1348: */
1.33 schwarze 1349:
1.50 kristaps 1350: if (NULL != (p = strstr(title, "- "))) {
1351: for (p += 2; ' ' == *p || '\b' == *p; p++)
1352: /* Skip to next word. */ ;
1353: } else {
1.56 schwarze 1354: if (warnings)
1.82 schwarze 1355: say(mpage->mlinks->file,
1356: "No dash in title line");
1.50 kristaps 1357: p = title;
1358: }
1.38 schwarze 1359:
1.50 kristaps 1360: plen = strlen(p);
1.1 kristaps 1361:
1.50 kristaps 1362: /* Strip backspace-encoding from line. */
1.1 kristaps 1363:
1.50 kristaps 1364: while (NULL != (line = memchr(p, '\b', plen))) {
1365: len = line - p;
1366: if (0 == len) {
1367: memmove(line, line + 1, plen--);
1368: continue;
1.141 schwarze 1369: }
1.50 kristaps 1370: memmove(line - 1, line + 1, plen - len);
1371: plen -= 2;
1372: }
1.1 kristaps 1373:
1.78 schwarze 1374: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1375: fclose(stream);
1376: free(title);
1377: }
1.1 kristaps 1378:
1.50 kristaps 1379: /*
1380: * Put a type/word pair into the word database for this particular file.
1381: */
1382: static void
1.112 schwarze 1383: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1384: {
1.112 schwarze 1385: char *cp;
1.18 kristaps 1386:
1.50 kristaps 1387: assert(NULL != value);
1.112 schwarze 1388: if (TYPE_arch == type)
1389: for (cp = value; *cp; cp++)
1390: if (isupper((unsigned char)*cp))
1391: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1392: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1393: }
1394:
1395: /*
1.50 kristaps 1396: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1397: */
1398: static void
1.78 schwarze 1399: putmdockey(const struct mpage *mpage,
1.189 schwarze 1400: const struct roff_node *n, uint64_t m)
1.3 kristaps 1401: {
1.18 kristaps 1402:
1.50 kristaps 1403: for ( ; NULL != n; n = n->next) {
1404: if (NULL != n->child)
1.78 schwarze 1405: putmdockey(mpage, n->child, m);
1.188 schwarze 1406: if (n->type == ROFFT_TEXT)
1.78 schwarze 1407: putkey(mpage, n->string, m);
1.50 kristaps 1408: }
1409: }
1.18 kristaps 1410:
1.61 schwarze 1411: static void
1.190 schwarze 1412: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1413: const struct roff_node *n)
1.50 kristaps 1414: {
1.189 schwarze 1415: const struct roff_node *head, *body;
1.121 schwarze 1416: char *start, *title;
1.50 kristaps 1417: char byte;
1.121 schwarze 1418: size_t sz;
1.18 kristaps 1419:
1.50 kristaps 1420: if (NULL == n)
1.61 schwarze 1421: return;
1.18 kristaps 1422:
1.50 kristaps 1423: /*
1424: * We're only searching for one thing: the first text child in
1425: * the BODY of a NAME section. Since we don't keep track of
1426: * sections in -man, run some hoops to find out whether we're in
1427: * the correct section or not.
1428: */
1.18 kristaps 1429:
1.188 schwarze 1430: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.50 kristaps 1431: body = n;
1432: assert(body->parent);
1433: if (NULL != (head = body->parent->head) &&
1.141 schwarze 1434: 1 == head->nchild &&
1435: NULL != (head = (head->child)) &&
1.188 schwarze 1436: head->type == ROFFT_TEXT &&
1.141 schwarze 1437: 0 == strcmp(head->string, "NAME") &&
1438: NULL != body->child) {
1.3 kristaps 1439:
1.50 kristaps 1440: /*
1441: * Suck the entire NAME section into memory.
1442: * Yes, we might run away.
1443: * But too many manuals have big, spread-out
1444: * NAME sections over many lines.
1445: */
1.3 kristaps 1446:
1.121 schwarze 1447: title = NULL;
1.194 schwarze 1448: deroff(&title, body);
1.50 kristaps 1449: if (NULL == title)
1.61 schwarze 1450: return;
1.18 kristaps 1451:
1.141 schwarze 1452: /*
1.50 kristaps 1453: * Go through a special heuristic dance here.
1454: * Conventionally, one or more manual names are
1455: * comma-specified prior to a whitespace, then a
1456: * dash, then a description. Try to puzzle out
1457: * the name parts here.
1458: */
1.18 kristaps 1459:
1.121 schwarze 1460: start = title;
1.50 kristaps 1461: for ( ;; ) {
1462: sz = strcspn(start, " ,");
1463: if ('\0' == start[sz])
1464: break;
1.1 kristaps 1465:
1.50 kristaps 1466: byte = start[sz];
1467: start[sz] = '\0';
1.110 schwarze 1468:
1469: /*
1470: * Assume a stray trailing comma in the
1471: * name list if a name begins with a dash.
1472: */
1473:
1474: if ('-' == start[0] ||
1475: ('\\' == start[0] && '-' == start[1]))
1476: break;
1.1 kristaps 1477:
1.133 schwarze 1478: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1479: if ( ! (mpage->name_head_done ||
1480: strcasecmp(start, meta->title))) {
1481: putkey(mpage, start, NAME_HEAD);
1482: mpage->name_head_done = 1;
1483: }
1.1 kristaps 1484:
1.50 kristaps 1485: if (' ' == byte) {
1486: start += sz + 1;
1487: break;
1488: }
1.1 kristaps 1489:
1.50 kristaps 1490: assert(',' == byte);
1491: start += sz + 1;
1492: while (' ' == *start)
1493: start++;
1494: }
1.1 kristaps 1495:
1.121 schwarze 1496: if (start == title) {
1.133 schwarze 1497: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1498: if ( ! (mpage->name_head_done ||
1499: strcasecmp(start, meta->title))) {
1500: putkey(mpage, start, NAME_HEAD);
1501: mpage->name_head_done = 1;
1502: }
1.50 kristaps 1503: free(title);
1.61 schwarze 1504: return;
1.50 kristaps 1505: }
1.1 kristaps 1506:
1.50 kristaps 1507: while (isspace((unsigned char)*start))
1508: start++;
1.1 kristaps 1509:
1.50 kristaps 1510: if (0 == strncmp(start, "-", 1))
1511: start += 1;
1512: else if (0 == strncmp(start, "\\-\\-", 4))
1513: start += 4;
1514: else if (0 == strncmp(start, "\\-", 2))
1515: start += 2;
1516: else if (0 == strncmp(start, "\\(en", 4))
1517: start += 4;
1518: else if (0 == strncmp(start, "\\(em", 4))
1519: start += 4;
1.1 kristaps 1520:
1.50 kristaps 1521: while (' ' == *start)
1522: start++;
1.1 kristaps 1523:
1.78 schwarze 1524: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1525: free(title);
1.61 schwarze 1526: return;
1.50 kristaps 1527: }
1528: }
1.1 kristaps 1529:
1.77 schwarze 1530: for (n = n->child; n; n = n->next) {
1.78 schwarze 1531: if (NULL != mpage->desc)
1.77 schwarze 1532: break;
1.174 schwarze 1533: parse_man(mpage, meta, n);
1.77 schwarze 1534: }
1.1 kristaps 1535: }
1536:
1537: static void
1.190 schwarze 1538: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1539: const struct roff_node *n)
1.1 kristaps 1540: {
1541:
1.50 kristaps 1542: assert(NULL != n);
1543: for (n = n->child; NULL != n; n = n->next) {
1544: switch (n->type) {
1.188 schwarze 1545: case ROFFT_ELEM:
1546: case ROFFT_BLOCK:
1547: case ROFFT_HEAD:
1548: case ROFFT_BODY:
1549: case ROFFT_TAIL:
1.50 kristaps 1550: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1551: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1552: break;
1.68 schwarze 1553: if (mdocs[n->tok].mask)
1.78 schwarze 1554: putmdockey(mpage, n->child,
1555: mdocs[n->tok].mask);
1.50 kristaps 1556: break;
1557: default:
1.188 schwarze 1558: assert(n->type != ROFFT_ROOT);
1.50 kristaps 1559: continue;
1560: }
1561: if (NULL != n->child)
1.172 schwarze 1562: parse_mdoc(mpage, meta, n);
1.1 kristaps 1563: }
1564: }
1565:
1.25 schwarze 1566: static int
1.190 schwarze 1567: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1568: const struct roff_node *n)
1.1 kristaps 1569: {
1.176 schwarze 1570: char *start, *end;
1.1 kristaps 1571: size_t sz;
1.25 schwarze 1572:
1.50 kristaps 1573: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1574: NULL == (n = n->child) ||
1.188 schwarze 1575: n->type != ROFFT_TEXT)
1.197 schwarze 1576: return 0;
1.1 kristaps 1577:
1578: /*
1579: * Only consider those `Fd' macro fields that begin with an
1580: * "inclusion" token (versus, e.g., #define).
1581: */
1.50 kristaps 1582:
1.1 kristaps 1583: if (strcmp("#include", n->string))
1.197 schwarze 1584: return 0;
1.1 kristaps 1585:
1.188 schwarze 1586: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1587: return 0;
1.1 kristaps 1588:
1589: /*
1590: * Strip away the enclosing angle brackets and make sure we're
1591: * not zero-length.
1592: */
1593:
1594: start = n->string;
1595: if ('<' == *start || '"' == *start)
1596: start++;
1597:
1598: if (0 == (sz = strlen(start)))
1.197 schwarze 1599: return 0;
1.1 kristaps 1600:
1601: end = &start[(int)sz - 1];
1602: if ('>' == *end || '"' == *end)
1603: end--;
1604:
1.50 kristaps 1605: if (end > start)
1.78 schwarze 1606: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1607: return 0;
1.1 kristaps 1608: }
1609:
1.178 schwarze 1610: static void
1.189 schwarze 1611: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1612: {
1.112 schwarze 1613: char *cp;
1.178 schwarze 1614: size_t sz;
1.1 kristaps 1615:
1.188 schwarze 1616: if (n->type != ROFFT_TEXT)
1.178 schwarze 1617: return;
1.25 schwarze 1618:
1.178 schwarze 1619: /* Skip function pointer punctuation. */
1.1 kristaps 1620:
1.178 schwarze 1621: cp = n->string;
1622: while (*cp == '(' || *cp == '*')
1.1 kristaps 1623: cp++;
1.178 schwarze 1624: sz = strcspn(cp, "()");
1.1 kristaps 1625:
1.178 schwarze 1626: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1627: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1628: putkeys(mpage, cp, sz, NAME_SYN);
1629: }
1.25 schwarze 1630:
1.178 schwarze 1631: static int
1.190 schwarze 1632: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1633: const struct roff_node *n)
1.178 schwarze 1634: {
1635:
1636: if (n->child == NULL)
1.197 schwarze 1637: return 0;
1.178 schwarze 1638:
1639: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1640:
1.178 schwarze 1641: for (n = n->child->next; n != NULL; n = n->next)
1.188 schwarze 1642: if (n->type == ROFFT_TEXT)
1.78 schwarze 1643: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1644:
1.197 schwarze 1645: return 0;
1.173 schwarze 1646: }
1647:
1648: static int
1.190 schwarze 1649: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1650: const struct roff_node *n)
1.173 schwarze 1651: {
1.177 schwarze 1652:
1.188 schwarze 1653: if (n->type != ROFFT_HEAD)
1.197 schwarze 1654: return 1;
1.173 schwarze 1655:
1.178 schwarze 1656: if (n->child != NULL)
1657: parse_mdoc_fname(mpage, n->child);
1658:
1.197 schwarze 1659: return 0;
1.1 kristaps 1660: }
1661:
1.25 schwarze 1662: static int
1.190 schwarze 1663: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1664: const struct roff_node *n)
1.1 kristaps 1665: {
1.67 schwarze 1666: char *cp;
1.1 kristaps 1667:
1668: if (NULL == (n = n->child))
1.197 schwarze 1669: return 0;
1.1 kristaps 1670:
1.67 schwarze 1671: if (NULL == n->next) {
1.78 schwarze 1672: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1673: return 0;
1.67 schwarze 1674: }
1675:
1.120 schwarze 1676: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1677: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1678: free(cp);
1.197 schwarze 1679: return 0;
1.1 kristaps 1680: }
1681:
1.25 schwarze 1682: static int
1.190 schwarze 1683: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1684: const struct roff_node *n)
1.1 kristaps 1685: {
1686:
1.188 schwarze 1687: if (n->type == ROFFT_BODY)
1.194 schwarze 1688: deroff(&mpage->desc, n);
1.197 schwarze 1689: return 0;
1.1 kristaps 1690: }
1691:
1.25 schwarze 1692: static int
1.190 schwarze 1693: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1694: const struct roff_node *n)
1.1 kristaps 1695: {
1696:
1.129 schwarze 1697: if (SEC_NAME == n->sec)
1.133 schwarze 1698: putmdockey(mpage, n->child, NAME_TITLE);
1.188 schwarze 1699: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1700: if (n->child == NULL)
1701: putkey(mpage, meta->name, NAME_SYN);
1702: else
1703: putmdockey(mpage, n->child, NAME_SYN);
1.174 schwarze 1704: }
1705: if ( ! (mpage->name_head_done ||
1706: n->child == NULL || n->child->string == NULL ||
1707: strcasecmp(n->child->string, meta->title))) {
1.188 schwarze 1708: putkey(mpage, n->child->string, ROFFT_HEAD);
1.174 schwarze 1709: mpage->name_head_done = 1;
1.172 schwarze 1710: }
1.197 schwarze 1711: return 0;
1.1 kristaps 1712: }
1713:
1.25 schwarze 1714: static int
1.190 schwarze 1715: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1716: const struct roff_node *n)
1.1 kristaps 1717: {
1718:
1.197 schwarze 1719: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1720: }
1721:
1.50 kristaps 1722: static int
1.190 schwarze 1723: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1724: const struct roff_node *n)
1.1 kristaps 1725: {
1726:
1.197 schwarze 1727: return n->type == ROFFT_HEAD;
1.50 kristaps 1728: }
1.1 kristaps 1729:
1.50 kristaps 1730: static int
1.190 schwarze 1731: parse_mdoc_body(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1732: const struct roff_node *n)
1.50 kristaps 1733: {
1.1 kristaps 1734:
1.197 schwarze 1735: return n->type == ROFFT_BODY;
1.1 kristaps 1736: }
1737:
1.50 kristaps 1738: /*
1.66 schwarze 1739: * Add a string to the hash table for the current manual.
1740: * Each string has a bitmask telling which macros it belongs to.
1741: * When we finish the manual, we'll dump the table.
1.50 kristaps 1742: */
1743: static void
1.176 schwarze 1744: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1745: {
1.133 schwarze 1746: struct ohash *htab;
1.50 kristaps 1747: struct str *s;
1.111 schwarze 1748: const char *end;
1.65 schwarze 1749: unsigned int slot;
1.176 schwarze 1750: int i, mustfree;
1.25 schwarze 1751:
1.50 kristaps 1752: if (0 == sz)
1753: return;
1.111 schwarze 1754:
1.176 schwarze 1755: mustfree = render_string(&cp, &sz);
1756:
1.133 schwarze 1757: if (TYPE_Nm & v) {
1758: htab = &names;
1759: v &= name_mask;
1.169 schwarze 1760: if (v & NAME_FIRST)
1761: name_mask &= ~NAME_FIRST;
1.133 schwarze 1762: if (debug > 1)
1763: say(mpage->mlinks->file,
1.175 schwarze 1764: "Adding name %*s, bits=%d", sz, cp, v);
1.133 schwarze 1765: } else {
1766: htab = &strings;
1767: if (debug > 1)
1768: for (i = 0; i < mansearch_keymax; i++)
1.163 schwarze 1769: if ((uint64_t)1 << i & v)
1.133 schwarze 1770: say(mpage->mlinks->file,
1771: "Adding key %s=%*s",
1772: mansearch_keynames[i], sz, cp);
1.111 schwarze 1773: }
1.25 schwarze 1774:
1.65 schwarze 1775: end = cp + sz;
1.133 schwarze 1776: slot = ohash_qlookupi(htab, cp, &end);
1777: s = ohash_find(htab, slot);
1.25 schwarze 1778:
1.78 schwarze 1779: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1780: s->mask |= v;
1781: return;
1782: } else if (NULL == s) {
1.144 schwarze 1783: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1784: memcpy(s->key, cp, sz);
1.133 schwarze 1785: ohash_insert(htab, slot, s);
1.1 kristaps 1786: }
1.78 schwarze 1787: s->mpage = mpage;
1.50 kristaps 1788: s->mask = v;
1.176 schwarze 1789:
1790: if (mustfree)
1791: free(cp);
1.1 kristaps 1792: }
1793:
1.50 kristaps 1794: /*
1795: * Take a Unicode codepoint and produce its UTF-8 encoding.
1796: * This isn't the best way to do this, but it works.
1797: * The magic numbers are from the UTF-8 packaging.
1798: * They're not as scary as they seem: read the UTF-8 spec for details.
1799: */
1800: static size_t
1801: utf8(unsigned int cp, char out[7])
1.1 kristaps 1802: {
1.50 kristaps 1803: size_t rc;
1.1 kristaps 1804:
1.50 kristaps 1805: rc = 0;
1806: if (cp <= 0x0000007F) {
1807: rc = 1;
1808: out[0] = (char)cp;
1809: } else if (cp <= 0x000007FF) {
1810: rc = 2;
1811: out[0] = (cp >> 6 & 31) | 192;
1812: out[1] = (cp & 63) | 128;
1813: } else if (cp <= 0x0000FFFF) {
1814: rc = 3;
1815: out[0] = (cp >> 12 & 15) | 224;
1816: out[1] = (cp >> 6 & 63) | 128;
1817: out[2] = (cp & 63) | 128;
1818: } else if (cp <= 0x001FFFFF) {
1819: rc = 4;
1820: out[0] = (cp >> 18 & 7) | 240;
1821: out[1] = (cp >> 12 & 63) | 128;
1822: out[2] = (cp >> 6 & 63) | 128;
1823: out[3] = (cp & 63) | 128;
1824: } else if (cp <= 0x03FFFFFF) {
1825: rc = 5;
1826: out[0] = (cp >> 24 & 3) | 248;
1827: out[1] = (cp >> 18 & 63) | 128;
1828: out[2] = (cp >> 12 & 63) | 128;
1829: out[3] = (cp >> 6 & 63) | 128;
1830: out[4] = (cp & 63) | 128;
1831: } else if (cp <= 0x7FFFFFFF) {
1832: rc = 6;
1833: out[0] = (cp >> 30 & 1) | 252;
1834: out[1] = (cp >> 24 & 63) | 128;
1835: out[2] = (cp >> 18 & 63) | 128;
1836: out[3] = (cp >> 12 & 63) | 128;
1837: out[4] = (cp >> 6 & 63) | 128;
1838: out[5] = (cp & 63) | 128;
1839: } else
1.197 schwarze 1840: return 0;
1.1 kristaps 1841:
1.50 kristaps 1842: out[rc] = '\0';
1.197 schwarze 1843: return rc;
1.50 kristaps 1844: }
1.1 kristaps 1845:
1.50 kristaps 1846: /*
1.176 schwarze 1847: * If the string contains escape sequences,
1848: * replace it with an allocated rendering and return 1,
1849: * such that the caller can free it after use.
1850: * Otherwise, do nothing and return 0.
1.50 kristaps 1851: */
1.176 schwarze 1852: static int
1853: render_string(char **public, size_t *psz)
1.50 kristaps 1854: {
1.176 schwarze 1855: const char *src, *scp, *addcp, *seq;
1856: char *dst;
1857: size_t ssz, dsz, addsz;
1.114 schwarze 1858: char utfbuf[7], res[6];
1.176 schwarze 1859: int seqlen, unicode;
1.50 kristaps 1860:
1861: res[0] = '\\';
1862: res[1] = '\t';
1863: res[2] = ASCII_NBRSP;
1864: res[3] = ASCII_HYPH;
1.114 schwarze 1865: res[4] = ASCII_BREAK;
1866: res[5] = '\0';
1.1 kristaps 1867:
1.176 schwarze 1868: src = scp = *public;
1869: ssz = *psz;
1870: dst = NULL;
1871: dsz = 0;
1872:
1873: while (scp < src + *psz) {
1874:
1875: /* Leave normal characters unchanged. */
1876:
1877: if (strchr(res, *scp) == NULL) {
1878: if (dst != NULL)
1879: dst[dsz++] = *scp;
1880: scp++;
1881: continue;
1882: }
1.46 kristaps 1883:
1.50 kristaps 1884: /*
1.176 schwarze 1885: * Found something that requires replacing,
1886: * make sure we have a destination buffer.
1.50 kristaps 1887: */
1.176 schwarze 1888:
1889: if (dst == NULL) {
1890: dst = mandoc_malloc(ssz + 1);
1891: dsz = scp - src;
1892: memcpy(dst, src, dsz);
1.50 kristaps 1893: }
1.46 kristaps 1894:
1.176 schwarze 1895: /* Handle single-char special characters. */
1896:
1897: switch (*scp) {
1898: case '\\':
1899: break;
1.141 schwarze 1900: case '\t':
1901: case ASCII_NBRSP:
1.176 schwarze 1902: dst[dsz++] = ' ';
1903: scp++;
1904: continue;
1905: case ASCII_HYPH:
1906: dst[dsz++] = '-';
1.114 schwarze 1907: /* FALLTHROUGH */
1.141 schwarze 1908: case ASCII_BREAK:
1.176 schwarze 1909: scp++;
1.50 kristaps 1910: continue;
1.114 schwarze 1911: default:
1.176 schwarze 1912: abort();
1.114 schwarze 1913: }
1.46 kristaps 1914:
1.50 kristaps 1915: /*
1.176 schwarze 1916: * Found an escape sequence.
1917: * Read past the slash, then parse it.
1918: * Ignore everything except characters.
1.50 kristaps 1919: */
1.95 schwarze 1920:
1.176 schwarze 1921: scp++;
1922: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1923: continue;
1.1 kristaps 1924:
1.50 kristaps 1925: /*
1.95 schwarze 1926: * Render the special character
1927: * as either UTF-8 or ASCII.
1.50 kristaps 1928: */
1.95 schwarze 1929:
1930: if (write_utf8) {
1.176 schwarze 1931: unicode = mchars_spec2cp(mchars, seq, seqlen);
1932: if (unicode <= 0)
1.95 schwarze 1933: continue;
1.176 schwarze 1934: addsz = utf8(unicode, utfbuf);
1935: if (addsz == 0)
1.95 schwarze 1936: continue;
1.176 schwarze 1937: addcp = utfbuf;
1.95 schwarze 1938: } else {
1.176 schwarze 1939: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1940: if (addcp == NULL)
1.95 schwarze 1941: continue;
1.176 schwarze 1942: if (*addcp == ASCII_NBRSP) {
1943: addcp = " ";
1944: addsz = 1;
1.95 schwarze 1945: }
1946: }
1.1 kristaps 1947:
1.50 kristaps 1948: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1949:
1.176 schwarze 1950: ssz += addsz;
1951: dst = mandoc_realloc(dst, ssz + 1);
1952: memcpy(dst + dsz, addcp, addsz);
1953: dsz += addsz;
1954: }
1955: if (dst != NULL) {
1956: *public = dst;
1957: *psz = dsz;
1958: }
1959:
1960: /* Trim trailing whitespace and NUL-terminate. */
1961:
1962: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1963: --*psz;
1964: if (dst != NULL) {
1965: (*public)[*psz] = '\0';
1.197 schwarze 1966: return 1;
1.176 schwarze 1967: } else
1.197 schwarze 1968: return 0;
1.1 kristaps 1969: }
1970:
1.118 schwarze 1971: static void
1972: dbadd_mlink(const struct mlink *mlink)
1973: {
1974: size_t i;
1975:
1976: i = 1;
1977: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1978: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1979: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 1980: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 1981: SQL_STEP(stmts[STMT_INSERT_LINK]);
1982: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.169 schwarze 1983: }
1984:
1985: static void
1986: dbadd_mlink_name(const struct mlink *mlink)
1987: {
1.175 schwarze 1988: uint64_t bits;
1.169 schwarze 1989: size_t i;
1990:
1991: dbadd_mlink(mlink);
1.160 schwarze 1992:
1993: i = 1;
1.175 schwarze 1994: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
1995: bits = NAME_FILE & NAME_MASK;
1996: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
1997: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
1998: sqlite3_reset(stmts[STMT_SELECT_NAME]);
1999: }
2000:
2001: i = 1;
2002: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.160 schwarze 2003: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2004: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2005: SQL_STEP(stmts[STMT_INSERT_NAME]);
2006: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.118 schwarze 2007: }
2008:
1.14 schwarze 2009: /*
1.50 kristaps 2010: * Flush the current page's terms (and their bits) into the database.
2011: * Wrap the entire set of additions in a transaction to make sqlite be a
2012: * little faster.
1.96 schwarze 2013: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2014: */
2015: static void
1.176 schwarze 2016: dbadd(struct mpage *mpage)
1.14 schwarze 2017: {
1.87 schwarze 2018: struct mlink *mlink;
1.50 kristaps 2019: struct str *key;
1.176 schwarze 2020: char *cp;
1.52 kristaps 2021: size_t i;
1.66 schwarze 2022: unsigned int slot;
1.176 schwarze 2023: int mustfree;
1.14 schwarze 2024:
1.128 schwarze 2025: mlink = mpage->mlinks;
1.43 kristaps 2026:
1.128 schwarze 2027: if (nodb) {
1.147 schwarze 2028: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2029: key = ohash_next(&names, &slot))
1.147 schwarze 2030: free(key);
2031: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2032: key = ohash_next(&strings, &slot))
1.147 schwarze 2033: free(key);
1.145 schwarze 2034: if (0 == debug)
2035: return;
1.128 schwarze 2036: while (NULL != mlink) {
2037: fputs(mlink->name, stdout);
2038: if (NULL == mlink->next ||
2039: strcmp(mlink->dsec, mlink->next->dsec) ||
2040: strcmp(mlink->fsec, mlink->next->fsec) ||
2041: strcmp(mlink->arch, mlink->next->arch)) {
2042: putchar('(');
2043: if ('\0' == *mlink->dsec)
2044: fputs(mlink->fsec, stdout);
2045: else
2046: fputs(mlink->dsec, stdout);
2047: if ('\0' != *mlink->arch)
2048: printf("/%s", mlink->arch);
2049: putchar(')');
2050: }
2051: mlink = mlink->next;
2052: if (NULL != mlink)
2053: fputs(", ", stdout);
2054: }
1.132 schwarze 2055: printf(" - %s\n", mpage->desc);
1.14 schwarze 2056: return;
1.128 schwarze 2057: }
2058:
2059: if (debug)
2060: say(mlink->file, "Adding to database");
1.28 kristaps 2061:
1.176 schwarze 2062: cp = mpage->desc;
2063: i = strlen(cp);
2064: mustfree = render_string(&cp, &i);
1.52 kristaps 2065: i = 1;
1.176 schwarze 2066: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.161 schwarze 2067: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.81 schwarze 2068: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2069: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2070: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.176 schwarze 2071: if (mustfree)
2072: free(cp);
1.81 schwarze 2073:
1.128 schwarze 2074: while (NULL != mlink) {
1.118 schwarze 2075: dbadd_mlink(mlink);
1.128 schwarze 2076: mlink = mlink->next;
2077: }
1.134 schwarze 2078: mlink = mpage->mlinks;
1.50 kristaps 2079:
1.133 schwarze 2080: for (key = ohash_first(&names, &slot); NULL != key;
2081: key = ohash_next(&names, &slot)) {
2082: assert(key->mpage == mpage);
2083: i = 1;
2084: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.176 schwarze 2085: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.137 schwarze 2086: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2087: SQL_STEP(stmts[STMT_INSERT_NAME]);
2088: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2089: free(key);
2090: }
1.66 schwarze 2091: for (key = ohash_first(&strings, &slot); NULL != key;
2092: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2093: assert(key->mpage == mpage);
1.52 kristaps 2094: i = 1;
2095: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.176 schwarze 2096: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.137 schwarze 2097: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2098: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2099: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 2100: free(key);
1.38 schwarze 2101: }
1.14 schwarze 2102: }
2103:
1.5 kristaps 2104: static void
1.59 schwarze 2105: dbprune(void)
1.5 kristaps 2106: {
1.78 schwarze 2107: struct mpage *mpage;
1.82 schwarze 2108: struct mlink *mlink;
1.52 kristaps 2109: size_t i;
1.80 schwarze 2110: unsigned int slot;
1.5 kristaps 2111:
1.106 schwarze 2112: if (0 == nodb)
2113: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2114:
1.106 schwarze 2115: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2116: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2117: mlink = mpage->mlinks;
1.125 schwarze 2118: if (debug)
1.106 schwarze 2119: say(mlink->file, "Deleting from database");
2120: if (nodb)
2121: continue;
2122: for ( ; NULL != mlink; mlink = mlink->next) {
2123: i = 1;
2124: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2125: i, mlink->dsec);
2126: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2127: i, mlink->arch);
2128: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2129: i, mlink->name);
2130: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2131: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2132: }
1.5 kristaps 2133: }
1.106 schwarze 2134:
2135: if (0 == nodb)
2136: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2137: }
2138:
1.4 kristaps 2139: /*
1.50 kristaps 2140: * Close an existing database and its prepared statements.
2141: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2142: */
1.35 kristaps 2143: static void
1.59 schwarze 2144: dbclose(int real)
1.4 kristaps 2145: {
1.50 kristaps 2146: size_t i;
1.115 schwarze 2147: int status;
2148: pid_t child;
1.4 kristaps 2149:
1.50 kristaps 2150: if (nodb)
1.38 schwarze 2151: return;
1.50 kristaps 2152:
2153: for (i = 0; i < STMT__MAX; i++) {
2154: sqlite3_finalize(stmts[i]);
2155: stmts[i] = NULL;
1.4 kristaps 2156: }
2157:
1.50 kristaps 2158: sqlite3_close(db);
2159: db = NULL;
1.12 schwarze 2160:
1.50 kristaps 2161: if (real)
2162: return;
1.12 schwarze 2163:
1.115 schwarze 2164: if ('\0' == *tempfilename) {
2165: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2166: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2167: say(MANDOC_DB, "&rename");
1.115 schwarze 2168: }
2169: return;
2170: }
2171:
2172: switch (child = fork()) {
1.141 schwarze 2173: case -1:
1.115 schwarze 2174: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2175: say("", "&fork cmp");
1.115 schwarze 2176: return;
1.141 schwarze 2177: case 0:
1.115 schwarze 2178: execlp("cmp", "cmp", "-s",
1.196 schwarze 2179: tempfilename, MANDOC_DB, (char *)NULL);
1.123 schwarze 2180: say("", "&exec cmp");
1.115 schwarze 2181: exit(0);
2182: default:
2183: break;
2184: }
2185: if (-1 == waitpid(child, &status, 0)) {
2186: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2187: say("", "&wait cmp");
1.115 schwarze 2188: } else if (WIFSIGNALED(status)) {
2189: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2190: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2191: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2192: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2193: say(MANDOC_DB,
2194: "Data changed, but cannot replace database");
2195: }
2196:
2197: *strrchr(tempfilename, '/') = '\0';
2198: switch (child = fork()) {
1.141 schwarze 2199: case -1:
1.115 schwarze 2200: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2201: say("", "&fork rm");
1.115 schwarze 2202: return;
1.141 schwarze 2203: case 0:
1.196 schwarze 2204: execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
1.123 schwarze 2205: say("", "&exec rm");
1.115 schwarze 2206: exit((int)MANDOCLEVEL_SYSERR);
2207: default:
2208: break;
2209: }
2210: if (-1 == waitpid(child, &status, 0)) {
2211: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2212: say("", "&wait rm");
1.115 schwarze 2213: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2214: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2215: say("", "%s: Cannot remove temporary directory",
2216: tempfilename);
1.59 schwarze 2217: }
1.50 kristaps 2218: }
1.14 schwarze 2219:
1.50 kristaps 2220: /*
2221: * This is straightforward stuff.
2222: * Open a database connection to a "temporary" database, then open a set
2223: * of prepared statements we'll use over and over again.
2224: * If "real" is set, we use the existing database; if not, we truncate a
2225: * temporary one.
2226: * Must be matched by dbclose().
2227: */
2228: static int
1.59 schwarze 2229: dbopen(int real)
1.50 kristaps 2230: {
1.115 schwarze 2231: const char *sql;
1.50 kristaps 2232: int rc, ofl;
1.12 schwarze 2233:
1.141 schwarze 2234: if (nodb)
1.197 schwarze 2235: return 1;
1.12 schwarze 2236:
1.115 schwarze 2237: *tempfilename = '\0';
1.63 schwarze 2238: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2239:
2240: if (real) {
2241: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2242: if (SQLITE_OK != rc) {
1.63 schwarze 2243: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2244: if (SQLITE_CANTOPEN != rc)
2245: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.197 schwarze 2246: return 0;
1.63 schwarze 2247: }
1.115 schwarze 2248: goto prepare_statements;
2249: }
2250:
2251: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2252:
1.115 schwarze 2253: remove(MANDOC_DB "~");
2254: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2255: if (SQLITE_OK == rc)
1.115 schwarze 2256: goto create_tables;
1.116 schwarze 2257: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2258: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2259: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.197 schwarze 2260: return 0;
1.50 kristaps 2261: }
1.12 schwarze 2262:
1.143 schwarze 2263: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2264: sizeof(tempfilename));
1.115 schwarze 2265: if (NULL == mkdtemp(tempfilename)) {
2266: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2267: say("", "&%s", tempfilename);
1.197 schwarze 2268: return 0;
1.115 schwarze 2269: }
1.143 schwarze 2270: (void)strlcat(tempfilename, "/" MANDOC_DB,
2271: sizeof(tempfilename));
1.115 schwarze 2272: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2273: if (SQLITE_OK != rc) {
1.59 schwarze 2274: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2275: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.197 schwarze 2276: return 0;
1.50 kristaps 2277: }
1.12 schwarze 2278:
1.115 schwarze 2279: create_tables:
1.81 schwarze 2280: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2281: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2282: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2283: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2284: ");\n"
2285: "\n"
1.81 schwarze 2286: "CREATE TABLE \"mlinks\" (\n"
2287: " \"sec\" TEXT NOT NULL,\n"
2288: " \"arch\" TEXT NOT NULL,\n"
2289: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2290: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2291: "ON DELETE CASCADE\n"
1.81 schwarze 2292: ");\n"
1.136 schwarze 2293: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2294: "\n"
1.133 schwarze 2295: "CREATE TABLE \"names\" (\n"
2296: " \"bits\" INTEGER NOT NULL,\n"
2297: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2298: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.175 schwarze 2299: "ON DELETE CASCADE,\n"
2300: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.133 schwarze 2301: ");\n"
2302: "\n"
1.50 kristaps 2303: "CREATE TABLE \"keys\" (\n"
2304: " \"bits\" INTEGER NOT NULL,\n"
2305: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2306: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2307: "ON DELETE CASCADE\n"
1.136 schwarze 2308: ");\n"
2309: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2310:
1.50 kristaps 2311: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2312: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2313: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2314: sqlite3_close(db);
1.197 schwarze 2315: return 0;
1.50 kristaps 2316: }
1.4 kristaps 2317:
1.57 schwarze 2318: prepare_statements:
1.146 schwarze 2319: if (SQLITE_OK != sqlite3_exec(db,
2320: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2321: exitcode = (int)MANDOCLEVEL_SYSERR;
2322: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2323: sqlite3_errmsg(db));
2324: sqlite3_close(db);
1.197 schwarze 2325: return 0;
1.146 schwarze 2326: }
2327:
1.137 schwarze 2328: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2329: "(SELECT pageid FROM mlinks WHERE "
2330: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2331: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2332: sql = "INSERT INTO mpages "
1.132 schwarze 2333: "(desc,form) VALUES (?,?)";
1.81 schwarze 2334: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2335: sql = "INSERT INTO mlinks "
1.104 schwarze 2336: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2337: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.175 schwarze 2338: sql = "SELECT bits FROM names where pageid = ?";
2339: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.133 schwarze 2340: sql = "INSERT INTO names "
2341: "(bits,name,pageid) VALUES (?,?,?)";
2342: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2343: sql = "INSERT INTO keys "
1.81 schwarze 2344: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2345: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2346:
2347: #ifndef __APPLE__
2348: /*
2349: * When opening a new database, we can turn off
2350: * synchronous mode for much better performance.
2351: */
2352:
1.146 schwarze 2353: if (real && SQLITE_OK != sqlite3_exec(db,
2354: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2355: exitcode = (int)MANDOCLEVEL_SYSERR;
2356: say(MANDOC_DB, "PRAGMA synchronous: %s",
1.162 schwarze 2357: sqlite3_errmsg(db));
1.146 schwarze 2358: sqlite3_close(db);
1.197 schwarze 2359: return 0;
1.146 schwarze 2360: }
1.70 schwarze 2361: #endif
2362:
1.197 schwarze 2363: return 1;
1.4 kristaps 2364: }
2365:
1.50 kristaps 2366: static int
1.165 schwarze 2367: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2368: {
1.59 schwarze 2369: static char startdir[PATH_MAX];
1.151 schwarze 2370: static int getcwd_status; /* 1 = ok, 2 = failure */
2371: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2372: char *cp;
1.4 kristaps 2373:
1.59 schwarze 2374: /*
1.151 schwarze 2375: * Remember the original working directory, if possible.
2376: * This will be needed if the second or a later directory
2377: * on the command line is given as a relative path.
2378: * Do not error out if the current directory is not
2379: * searchable: Maybe it won't be needed after all.
2380: */
2381: if (0 == getcwd_status) {
2382: if (NULL == getcwd(startdir, sizeof(startdir))) {
2383: getcwd_status = 2;
2384: (void)strlcpy(startdir, strerror(errno),
2385: sizeof(startdir));
2386: } else
2387: getcwd_status = 1;
2388: }
2389:
2390: /*
2391: * We are leaving the old base directory.
2392: * Do not use it any longer, not even for messages.
2393: */
2394: *basedir = '\0';
2395:
2396: /*
2397: * If and only if the directory was changed earlier and
2398: * the next directory to process is given as a relative path,
2399: * first go back, or bail out if that is impossible.
1.59 schwarze 2400: */
1.151 schwarze 2401: if (chdir_status && '/' != *targetdir) {
2402: if (2 == getcwd_status) {
1.59 schwarze 2403: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2404: say("", "getcwd: %s", startdir);
1.197 schwarze 2405: return 0;
1.59 schwarze 2406: }
1.151 schwarze 2407: if (-1 == chdir(startdir)) {
1.59 schwarze 2408: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2409: say("", "&chdir %s", startdir);
1.197 schwarze 2410: return 0;
1.59 schwarze 2411: }
2412: }
1.151 schwarze 2413:
2414: /*
2415: * Always resolve basedir to the canonicalized absolute
2416: * pathname and append a trailing slash, such that
2417: * we can reliably check whether files are inside.
2418: */
1.59 schwarze 2419: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2420: if (report_baddir || errno != ENOENT) {
2421: exitcode = (int)MANDOCLEVEL_BADARG;
2422: say("", "&%s: realpath", targetdir);
2423: }
1.197 schwarze 2424: return 0;
1.59 schwarze 2425: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2426: if (report_baddir || errno != ENOENT) {
2427: exitcode = (int)MANDOCLEVEL_BADARG;
2428: say("", "&chdir");
2429: }
1.197 schwarze 2430: return 0;
1.150 schwarze 2431: }
1.151 schwarze 2432: chdir_status = 1;
1.150 schwarze 2433: cp = strchr(basedir, '\0');
2434: if ('/' != cp[-1]) {
2435: if (cp - basedir >= PATH_MAX - 1) {
2436: exitcode = (int)MANDOCLEVEL_SYSERR;
2437: say("", "Filename too long");
1.197 schwarze 2438: return 0;
1.150 schwarze 2439: }
2440: *cp++ = '/';
2441: *cp = '\0';
1.4 kristaps 2442: }
1.197 schwarze 2443: return 1;
1.56 schwarze 2444: }
2445:
2446: static void
1.59 schwarze 2447: say(const char *file, const char *format, ...)
1.56 schwarze 2448: {
2449: va_list ap;
1.123 schwarze 2450: int use_errno;
1.56 schwarze 2451:
1.59 schwarze 2452: if ('\0' != *basedir)
2453: fprintf(stderr, "%s", basedir);
2454: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2455: fputc('/', stderr);
1.56 schwarze 2456: if ('\0' != *file)
1.59 schwarze 2457: fprintf(stderr, "%s", file);
2458:
1.123 schwarze 2459: use_errno = 1;
2460: if (NULL != format) {
2461: switch (*format) {
1.141 schwarze 2462: case '&':
1.123 schwarze 2463: format++;
2464: break;
1.141 schwarze 2465: case '\0':
1.123 schwarze 2466: format = NULL;
2467: break;
2468: default:
2469: use_errno = 0;
2470: break;
2471: }
2472: }
2473: if (NULL != format) {
2474: if ('\0' != *basedir || '\0' != *file)
2475: fputs(": ", stderr);
2476: va_start(ap, format);
2477: vfprintf(stderr, format, ap);
2478: va_end(ap);
2479: }
2480: if (use_errno) {
2481: if ('\0' != *basedir || '\0' != *file || NULL != format)
2482: fputs(": ", stderr);
1.59 schwarze 2483: perror(NULL);
1.123 schwarze 2484: } else
2485: fputc('\n', stderr);
1.1 kristaps 2486: }
CVSweb