Annotation of mandoc/mandocdb.c, Revision 1.220.2.4
1.220.2.4! schwarze 1: /* $Id: mandocdb.c,v 1.234 2016/10/22 10:09:27 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>
1.220.2.3 schwarze 5: * Copyright (c) 2016 Ed Maste <emaste@freebsd.org>
1.1 kristaps 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
1.187 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.187 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
19: #include "config.h"
20:
1.162 schwarze 21: #include <sys/types.h>
1.50 kristaps 22: #include <sys/stat.h>
1.115 schwarze 23: #include <sys/wait.h>
1.1 kristaps 24:
25: #include <assert.h>
1.43 kristaps 26: #include <ctype.h>
1.208 schwarze 27: #if HAVE_ERR
1.198 schwarze 28: #include <err.h>
1.208 schwarze 29: #endif
1.63 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.158 schwarze 32: #if HAVE_FTS
1.50 kristaps 33: #include <fts.h>
1.157 schwarze 34: #else
35: #include "compat_fts.h"
36: #endif
1.58 schwarze 37: #include <limits.h>
1.218 kristaps 38: #if HAVE_SANDBOX_INIT
39: #include <sandbox.h>
40: #endif
1.50 kristaps 41: #include <stddef.h>
1.59 schwarze 42: #include <stdio.h>
1.1 kristaps 43: #include <stdint.h>
44: #include <stdlib.h>
45: #include <string.h>
1.17 schwarze 46: #include <unistd.h>
1.1 kristaps 47:
1.50 kristaps 48: #include <sqlite3.h>
1.1 kristaps 49:
1.188 schwarze 50: #include "mandoc_aux.h"
1.202 schwarze 51: #include "mandoc_ohash.h"
1.188 schwarze 52: #include "mandoc.h"
53: #include "roff.h"
1.50 kristaps 54: #include "mdoc.h"
1.1 kristaps 55: #include "man.h"
1.187 schwarze 56: #include "manconf.h"
1.55 kristaps 57: #include "mansearch.h"
1.1 kristaps 58:
1.111 schwarze 59: extern int mansearch_keymax;
60: extern const char *const mansearch_keynames[];
61:
1.52 kristaps 62: #define SQL_EXEC(_v) \
63: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.134 schwarze 64: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.52 kristaps 65: #define SQL_BIND_TEXT(_s, _i, _v) \
66: if (SQLITE_OK != sqlite3_bind_text \
67: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.134 schwarze 68: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 69: #define SQL_BIND_INT(_s, _i, _v) \
70: if (SQLITE_OK != sqlite3_bind_int \
71: ((_s), (_i)++, (_v))) \
1.134 schwarze 72: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 73: #define SQL_BIND_INT64(_s, _i, _v) \
74: if (SQLITE_OK != sqlite3_bind_int64 \
75: ((_s), (_i)++, (_v))) \
1.134 schwarze 76: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 77: #define SQL_STEP(_s) \
78: if (SQLITE_DONE != sqlite3_step((_s))) \
1.134 schwarze 79: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 80:
1.50 kristaps 81: enum op {
82: OP_DEFAULT = 0, /* new dbs from dir list or default config */
83: OP_CONFFILE, /* new databases from custom config file */
84: OP_UPDATE, /* delete/add entries in existing database */
85: OP_DELETE, /* delete entries from existing database */
86: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 87: };
88:
1.50 kristaps 89: struct str {
1.78 schwarze 90: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 91: uint64_t mask; /* bitmask in sequence */
1.176 schwarze 92: char key[]; /* rendered text */
1.38 schwarze 93: };
94:
1.79 schwarze 95: struct inodev {
96: ino_t st_ino;
97: dev_t st_dev;
1.50 kristaps 98: };
1.5 kristaps 99:
1.78 schwarze 100: struct mpage {
1.79 schwarze 101: struct inodev inodev; /* used for hashing routine */
1.137 schwarze 102: int64_t pageid; /* pageid in mpages SQL table */
1.82 schwarze 103: char *sec; /* section from file content */
104: char *arch; /* architecture from file content */
105: char *title; /* title from file content */
106: char *desc; /* description from file content */
1.220.2.3 schwarze 107: struct mpage *next; /* singly linked list */
1.82 schwarze 108: struct mlink *mlinks; /* singly linked list */
1.161 schwarze 109: int form; /* format from file content */
1.174 schwarze 110: int name_head_done;
1.82 schwarze 111: };
112:
113: struct mlink {
1.58 schwarze 114: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 115: char *dsec; /* section from directory */
116: char *arch; /* architecture from directory */
117: char *name; /* name from file name (not empty) */
118: char *fsec; /* section from file name suffix */
1.85 schwarze 119: struct mlink *next; /* singly linked list */
1.118 schwarze 120: struct mpage *mpage; /* parent */
1.161 schwarze 121: int dform; /* format from directory */
122: int fform; /* format from file name suffix */
1.124 schwarze 123: int gzip; /* filename has a .gz suffix */
1.3 kristaps 124: };
125:
1.50 kristaps 126: enum stmt {
1.81 schwarze 127: STMT_DELETE_PAGE = 0, /* delete mpage */
128: STMT_INSERT_PAGE, /* insert mpage */
129: STMT_INSERT_LINK, /* insert mlink */
1.133 schwarze 130: STMT_INSERT_NAME, /* insert name */
1.175 schwarze 131: STMT_SELECT_NAME, /* retrieve existing name flags */
1.81 schwarze 132: STMT_INSERT_KEY, /* insert parsed key */
1.50 kristaps 133: STMT__MAX
1.1 kristaps 134: };
135:
1.190 schwarze 136: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.189 schwarze 137: const struct roff_node *);
1.1 kristaps 138:
1.50 kristaps 139: struct mdoc_handler {
140: mdoc_fp fp; /* optional handler */
141: uint64_t mask; /* set unless handler returns 0 */
1.1 kristaps 142: };
1.219 schwarze 143:
144:
145: int mandocdb(int, char *[]);
1.1 kristaps 146:
1.59 schwarze 147: static void dbclose(int);
1.176 schwarze 148: static void dbadd(struct mpage *);
1.118 schwarze 149: static void dbadd_mlink(const struct mlink *mlink);
1.169 schwarze 150: static void dbadd_mlink_name(const struct mlink *mlink);
1.59 schwarze 151: static int dbopen(int);
152: static void dbprune(void);
153: static void filescan(const char *);
1.220.2.3 schwarze 154: #if HAVE_FTS_COMPARE_CONST
155: static int fts_compare(const FTSENT *const *, const FTSENT *const *);
156: #else
157: static int fts_compare(const FTSENT **, const FTSENT **);
158: #endif
1.84 schwarze 159: static void mlink_add(struct mlink *, const struct stat *);
1.131 schwarze 160: static void mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 161: static void mlink_free(struct mlink *);
1.89 schwarze 162: static void mlinks_undupe(struct mpage *);
1.78 schwarze 163: static void mpages_free(void);
1.176 schwarze 164: static void mpages_merge(struct mparse *);
1.130 schwarze 165: static void names_check(void);
1.124 schwarze 166: static void parse_cat(struct mpage *, int);
1.190 schwarze 167: static void parse_man(struct mpage *, const struct roff_meta *,
1.189 schwarze 168: const struct roff_node *);
1.190 schwarze 169: static void parse_mdoc(struct mpage *, const struct roff_meta *,
1.189 schwarze 170: const struct roff_node *);
1.190 schwarze 171: static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
1.189 schwarze 172: const struct roff_node *);
1.190 schwarze 173: static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
1.189 schwarze 174: const struct roff_node *);
175: static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
1.190 schwarze 176: static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
1.189 schwarze 177: const struct roff_node *);
1.190 schwarze 178: static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
1.189 schwarze 179: const struct roff_node *);
1.190 schwarze 180: static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
1.189 schwarze 181: const struct roff_node *);
1.190 schwarze 182: static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
1.189 schwarze 183: const struct roff_node *);
1.190 schwarze 184: static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
1.189 schwarze 185: const struct roff_node *);
1.211 schwarze 186: static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
187: const struct roff_node *);
1.190 schwarze 188: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.189 schwarze 189: const struct roff_node *);
1.112 schwarze 190: static void putkey(const struct mpage *, char *, uint64_t);
1.176 schwarze 191: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.78 schwarze 192: static void putmdockey(const struct mpage *,
1.189 schwarze 193: const struct roff_node *, uint64_t);
1.176 schwarze 194: static int render_string(char **, size_t *);
1.220 schwarze 195: static void say(const char *, const char *, ...)
196: __attribute__((__format__ (printf, 2, 3)));
1.165 schwarze 197: static int set_basedir(const char *, int);
1.59 schwarze 198: static int treescan(void);
1.50 kristaps 199: static size_t utf8(unsigned int, char [7]);
200:
1.115 schwarze 201: static char tempfilename[32];
1.102 schwarze 202: static int nodb; /* no database changes */
1.116 schwarze 203: static int mparse_options; /* abort the parse early */
1.141 schwarze 204: static int use_all; /* use all found files */
205: static int debug; /* print what we're doing */
206: static int warnings; /* warn about crap */
1.95 schwarze 207: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 208: static int exitcode; /* to be returned by main */
1.141 schwarze 209: static enum op op; /* operational mode */
1.59 schwarze 210: static char basedir[PATH_MAX]; /* current base directory */
1.220.2.3 schwarze 211: static struct mpage *mpage_head; /* list of distinct manual pages */
1.78 schwarze 212: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 213: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 214: static struct ohash names; /* table of all names */
1.50 kristaps 215: static struct ohash strings; /* table of all strings */
216: static sqlite3 *db = NULL; /* current database */
217: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.133 schwarze 218: static uint64_t name_mask;
1.25 schwarze 219:
220: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 221: { NULL, 0 }, /* Ap */
222: { NULL, 0 }, /* Dd */
223: { NULL, 0 }, /* Dt */
224: { NULL, 0 }, /* Os */
225: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
226: { parse_mdoc_head, TYPE_Ss }, /* Ss */
227: { NULL, 0 }, /* Pp */
228: { NULL, 0 }, /* D1 */
229: { NULL, 0 }, /* Dl */
230: { NULL, 0 }, /* Bd */
231: { NULL, 0 }, /* Ed */
232: { NULL, 0 }, /* Bl */
233: { NULL, 0 }, /* El */
234: { NULL, 0 }, /* It */
235: { NULL, 0 }, /* Ad */
236: { NULL, TYPE_An }, /* An */
237: { NULL, TYPE_Ar }, /* Ar */
238: { NULL, TYPE_Cd }, /* Cd */
239: { NULL, TYPE_Cm }, /* Cm */
240: { NULL, TYPE_Dv }, /* Dv */
241: { NULL, TYPE_Er }, /* Er */
242: { NULL, TYPE_Ev }, /* Ev */
243: { NULL, 0 }, /* Ex */
244: { NULL, TYPE_Fa }, /* Fa */
245: { parse_mdoc_Fd, 0 }, /* Fd */
246: { NULL, TYPE_Fl }, /* Fl */
247: { parse_mdoc_Fn, 0 }, /* Fn */
248: { NULL, TYPE_Ft }, /* Ft */
249: { NULL, TYPE_Ic }, /* Ic */
1.92 schwarze 250: { NULL, TYPE_In }, /* In */
1.68 schwarze 251: { NULL, TYPE_Li }, /* Li */
1.132 schwarze 252: { parse_mdoc_Nd, 0 }, /* Nd */
1.129 schwarze 253: { parse_mdoc_Nm, 0 }, /* Nm */
1.68 schwarze 254: { NULL, 0 }, /* Op */
255: { NULL, 0 }, /* Ot */
256: { NULL, TYPE_Pa }, /* Pa */
257: { NULL, 0 }, /* Rv */
1.92 schwarze 258: { NULL, TYPE_St }, /* St */
1.211 schwarze 259: { parse_mdoc_Va, TYPE_Va }, /* Va */
260: { parse_mdoc_Va, TYPE_Vt }, /* Vt */
1.68 schwarze 261: { parse_mdoc_Xr, 0 }, /* Xr */
262: { NULL, 0 }, /* %A */
263: { NULL, 0 }, /* %B */
264: { NULL, 0 }, /* %D */
265: { NULL, 0 }, /* %I */
266: { NULL, 0 }, /* %J */
267: { NULL, 0 }, /* %N */
268: { NULL, 0 }, /* %O */
269: { NULL, 0 }, /* %P */
270: { NULL, 0 }, /* %R */
271: { NULL, 0 }, /* %T */
272: { NULL, 0 }, /* %V */
273: { NULL, 0 }, /* Ac */
274: { NULL, 0 }, /* Ao */
275: { NULL, 0 }, /* Aq */
276: { NULL, TYPE_At }, /* At */
277: { NULL, 0 }, /* Bc */
278: { NULL, 0 }, /* Bf */
279: { NULL, 0 }, /* Bo */
280: { NULL, 0 }, /* Bq */
281: { NULL, TYPE_Bsx }, /* Bsx */
282: { NULL, TYPE_Bx }, /* Bx */
283: { NULL, 0 }, /* Db */
284: { NULL, 0 }, /* Dc */
285: { NULL, 0 }, /* Do */
286: { NULL, 0 }, /* Dq */
287: { NULL, 0 }, /* Ec */
288: { NULL, 0 }, /* Ef */
289: { NULL, TYPE_Em }, /* Em */
290: { NULL, 0 }, /* Eo */
291: { NULL, TYPE_Fx }, /* Fx */
292: { NULL, TYPE_Ms }, /* Ms */
293: { NULL, 0 }, /* No */
294: { NULL, 0 }, /* Ns */
295: { NULL, TYPE_Nx }, /* Nx */
296: { NULL, TYPE_Ox }, /* Ox */
297: { NULL, 0 }, /* Pc */
298: { NULL, 0 }, /* Pf */
299: { NULL, 0 }, /* Po */
300: { NULL, 0 }, /* Pq */
301: { NULL, 0 }, /* Qc */
302: { NULL, 0 }, /* Ql */
303: { NULL, 0 }, /* Qo */
304: { NULL, 0 }, /* Qq */
305: { NULL, 0 }, /* Re */
306: { NULL, 0 }, /* Rs */
307: { NULL, 0 }, /* Sc */
308: { NULL, 0 }, /* So */
309: { NULL, 0 }, /* Sq */
310: { NULL, 0 }, /* Sm */
311: { NULL, 0 }, /* Sx */
312: { NULL, TYPE_Sy }, /* Sy */
313: { NULL, TYPE_Tn }, /* Tn */
314: { NULL, 0 }, /* Ux */
315: { NULL, 0 }, /* Xc */
316: { NULL, 0 }, /* Xo */
1.173 schwarze 317: { parse_mdoc_Fo, 0 }, /* Fo */
1.68 schwarze 318: { NULL, 0 }, /* Fc */
319: { NULL, 0 }, /* Oo */
320: { NULL, 0 }, /* Oc */
321: { NULL, 0 }, /* Bk */
322: { NULL, 0 }, /* Ek */
323: { NULL, 0 }, /* Bt */
324: { NULL, 0 }, /* Hf */
325: { NULL, 0 }, /* Fr */
326: { NULL, 0 }, /* Ud */
327: { NULL, TYPE_Lb }, /* Lb */
328: { NULL, 0 }, /* Lp */
329: { NULL, TYPE_Lk }, /* Lk */
330: { NULL, TYPE_Mt }, /* Mt */
331: { NULL, 0 }, /* Brq */
332: { NULL, 0 }, /* Bro */
333: { NULL, 0 }, /* Brc */
334: { NULL, 0 }, /* %C */
335: { NULL, 0 }, /* Es */
336: { NULL, 0 }, /* En */
337: { NULL, TYPE_Dx }, /* Dx */
338: { NULL, 0 }, /* %Q */
339: { NULL, 0 }, /* br */
340: { NULL, 0 }, /* sp */
341: { NULL, 0 }, /* %U */
342: { NULL, 0 }, /* Ta */
1.166 schwarze 343: { NULL, 0 }, /* ll */
1.1 kristaps 344: };
345:
1.141 schwarze 346:
1.1 kristaps 347: int
1.179 schwarze 348: mandocdb(int argc, char *argv[])
1.1 kristaps 349: {
1.187 schwarze 350: struct manconf conf;
351: struct mparse *mp;
1.206 schwarze 352: const char *path_arg, *progname;
1.187 schwarze 353: size_t j, sz;
1.59 schwarze 354: int ch, i;
1.50 kristaps 355:
1.207 schwarze 356: #if HAVE_PLEDGE
357: if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
1.214 schwarze 358: warn("pledge");
1.218 kristaps 359: return (int)MANDOCLEVEL_SYSERR;
360: }
361: #endif
362:
363: #if HAVE_SANDBOX_INIT
364: if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) {
365: warnx("sandbox_init");
1.207 schwarze 366: return (int)MANDOCLEVEL_SYSERR;
367: }
368: #endif
369:
1.187 schwarze 370: memset(&conf, 0, sizeof(conf));
1.50 kristaps 371: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
372:
373: /*
1.141 schwarze 374: * We accept a few different invocations.
1.50 kristaps 375: * The CHECKOP macro makes sure that invocation styles don't
376: * clobber each other.
377: */
378: #define CHECKOP(_op, _ch) do \
379: if (OP_DEFAULT != (_op)) { \
1.198 schwarze 380: warnx("-%c: Conflicting option", (_ch)); \
1.50 kristaps 381: goto usage; \
382: } while (/*CONSTCOND*/0)
1.10 kristaps 383:
1.59 schwarze 384: path_arg = NULL;
1.38 schwarze 385: op = OP_DEFAULT;
1.1 kristaps 386:
1.126 schwarze 387: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 388: switch (ch) {
1.141 schwarze 389: case 'a':
1.12 schwarze 390: use_all = 1;
391: break;
1.141 schwarze 392: case 'C':
1.50 kristaps 393: CHECKOP(op, ch);
1.59 schwarze 394: path_arg = optarg;
1.38 schwarze 395: op = OP_CONFFILE;
1.34 schwarze 396: break;
1.141 schwarze 397: case 'D':
1.125 schwarze 398: debug++;
399: break;
1.141 schwarze 400: case 'd':
1.50 kristaps 401: CHECKOP(op, ch);
1.59 schwarze 402: path_arg = optarg;
1.5 kristaps 403: op = OP_UPDATE;
404: break;
1.141 schwarze 405: case 'n':
1.50 kristaps 406: nodb = 1;
407: break;
1.141 schwarze 408: case 'p':
1.126 schwarze 409: warnings = 1;
410: break;
1.141 schwarze 411: case 'Q':
1.116 schwarze 412: mparse_options |= MPARSE_QUICK;
1.102 schwarze 413: break;
1.141 schwarze 414: case 'T':
1.95 schwarze 415: if (strcmp(optarg, "utf8")) {
1.198 schwarze 416: warnx("-T%s: Unsupported output format",
417: optarg);
1.95 schwarze 418: goto usage;
419: }
420: write_utf8 = 1;
421: break;
1.141 schwarze 422: case 't':
1.50 kristaps 423: CHECKOP(op, ch);
1.38 schwarze 424: dup2(STDOUT_FILENO, STDERR_FILENO);
425: op = OP_TEST;
1.50 kristaps 426: nodb = warnings = 1;
1.38 schwarze 427: break;
1.141 schwarze 428: case 'u':
1.50 kristaps 429: CHECKOP(op, ch);
1.59 schwarze 430: path_arg = optarg;
1.5 kristaps 431: op = OP_DELETE;
432: break;
1.141 schwarze 433: case 'v':
1.126 schwarze 434: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 435: break;
1.1 kristaps 436: default:
1.38 schwarze 437: goto usage;
1.1 kristaps 438: }
439:
440: argc -= optind;
441: argv += optind;
442:
1.207 schwarze 443: #if HAVE_PLEDGE
1.210 schwarze 444: if (nodb) {
445: if (pledge("stdio rpath", NULL) == -1) {
1.214 schwarze 446: warn("pledge");
1.210 schwarze 447: return (int)MANDOCLEVEL_SYSERR;
448: }
1.207 schwarze 449: }
450: #endif
451:
1.38 schwarze 452: if (OP_CONFFILE == op && argc > 0) {
1.198 schwarze 453: warnx("-C: Too many arguments");
1.38 schwarze 454: goto usage;
455: }
456:
1.59 schwarze 457: exitcode = (int)MANDOCLEVEL_OK;
1.203 schwarze 458: mchars_alloc();
459: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
1.202 schwarze 460: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
461: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.50 kristaps 462:
463: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 464:
1.50 kristaps 465: /*
1.150 schwarze 466: * Most of these deal with a specific directory.
1.140 schwarze 467: * Jump into that directory first.
1.50 kristaps 468: */
1.165 schwarze 469: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.50 kristaps 470: goto out;
1.140 schwarze 471:
1.138 schwarze 472: if (dbopen(1)) {
1.140 schwarze 473: /*
474: * The existing database is usable. Process
475: * all files specified on the command-line.
476: */
1.207 schwarze 477: #if HAVE_PLEDGE
1.210 schwarze 478: if (!nodb) {
479: if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
1.214 schwarze 480: warn("pledge");
1.210 schwarze 481: exitcode = (int)MANDOCLEVEL_SYSERR;
482: goto out;
483: }
1.207 schwarze 484: }
485: #endif
1.140 schwarze 486: use_all = 1;
1.138 schwarze 487: for (i = 0; i < argc; i++)
488: filescan(argv[i]);
489: if (OP_TEST != op)
490: dbprune();
491: } else {
492: /*
493: * Database missing or corrupt.
494: * Recreate from scratch.
495: */
1.140 schwarze 496: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 497: op = OP_DEFAULT;
498: if (0 == treescan())
499: goto out;
500: if (0 == dbopen(0))
501: goto out;
502: }
1.50 kristaps 503: if (OP_DELETE != op)
1.176 schwarze 504: mpages_merge(mp);
1.138 schwarze 505: dbclose(OP_DEFAULT == op ? 0 : 1);
1.50 kristaps 506: } else {
507: /*
508: * If we have arguments, use them as our manpaths.
509: * If we don't, grok from manpath(1) or however else
1.187 schwarze 510: * manconf_parse() wants to do it.
1.50 kristaps 511: */
512: if (argc > 0) {
1.187 schwarze 513: conf.manpath.paths = mandoc_reallocarray(NULL,
1.144 schwarze 514: argc, sizeof(char *));
1.187 schwarze 515: conf.manpath.sz = (size_t)argc;
1.50 kristaps 516: for (i = 0; i < argc; i++)
1.187 schwarze 517: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.50 kristaps 518: } else
1.187 schwarze 519: manconf_parse(&conf, path_arg, NULL, NULL);
1.127 schwarze 520:
1.187 schwarze 521: if (conf.manpath.sz == 0) {
1.127 schwarze 522: exitcode = (int)MANDOCLEVEL_BADARG;
523: say("", "Empty manpath");
524: }
1.50 kristaps 525:
526: /*
1.69 schwarze 527: * First scan the tree rooted at a base directory, then
528: * build a new database and finally move it into place.
1.50 kristaps 529: * Ignore zero-length directories and strip trailing
530: * slashes.
531: */
1.187 schwarze 532: for (j = 0; j < conf.manpath.sz; j++) {
533: sz = strlen(conf.manpath.paths[j]);
534: if (sz && conf.manpath.paths[j][sz - 1] == '/')
535: conf.manpath.paths[j][--sz] = '\0';
1.50 kristaps 536: if (0 == sz)
537: continue;
1.66 schwarze 538:
539: if (j) {
1.202 schwarze 540: mandoc_ohash_init(&mpages, 6,
541: offsetof(struct mpage, inodev));
542: mandoc_ohash_init(&mlinks, 6,
543: offsetof(struct mlink, file));
1.66 schwarze 544: }
545:
1.187 schwarze 546: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.165 schwarze 547: continue;
1.59 schwarze 548: if (0 == treescan())
1.165 schwarze 549: continue;
1.59 schwarze 550: if (0 == dbopen(0))
1.165 schwarze 551: continue;
1.52 kristaps 552:
1.176 schwarze 553: mpages_merge(mp);
1.147 schwarze 554: if (warnings && !nodb &&
1.130 schwarze 555: ! (MPARSE_QUICK & mparse_options))
556: names_check();
1.59 schwarze 557: dbclose(0);
1.66 schwarze 558:
1.187 schwarze 559: if (j + 1 < conf.manpath.sz) {
1.80 schwarze 560: mpages_free();
1.78 schwarze 561: ohash_delete(&mpages);
1.83 schwarze 562: ohash_delete(&mlinks);
1.66 schwarze 563: }
1.50 kristaps 564: }
565: }
566: out:
1.187 schwarze 567: manconf_free(&conf);
1.168 schwarze 568: mparse_free(mp);
1.203 schwarze 569: mchars_free();
1.80 schwarze 570: mpages_free();
1.78 schwarze 571: ohash_delete(&mpages);
1.83 schwarze 572: ohash_delete(&mlinks);
1.197 schwarze 573: return exitcode;
1.50 kristaps 574: usage:
1.206 schwarze 575: progname = getprogname();
1.126 schwarze 576: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
577: " %s [-aDnpQ] [-Tutf8] dir ...\n"
578: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
579: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 580: " %s [-Q] -t file ...\n",
1.206 schwarze 581: progname, progname, progname, progname, progname);
1.50 kristaps 582:
1.197 schwarze 583: return (int)MANDOCLEVEL_BADARG;
1.50 kristaps 584: }
585:
586: /*
1.220.2.3 schwarze 587: * To get a singly linked list in alpha order while inserting entries
588: * at the beginning, process directory entries in reverse alpha order.
589: */
590: static int
591: #if HAVE_FTS_COMPARE_CONST
592: fts_compare(const FTSENT *const *a, const FTSENT *const *b)
593: #else
594: fts_compare(const FTSENT **a, const FTSENT **b)
595: #endif
596: {
597: return -strcmp((*a)->fts_name, (*b)->fts_name);
598: }
599:
600: /*
1.59 schwarze 601: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 602: * We use fts(), scanning directory parts along the way for clues to our
603: * section and architecture.
604: *
605: * If use_all has been specified, grok all files.
606: * If not, sanitise paths to the following:
607: *
1.141 schwarze 608: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 609: * or
610: * [./]cat<section>[/<arch>]/<name>.0
611: *
1.216 schwarze 612: * TODO: accommodate for multi-language directories.
1.50 kristaps 613: */
614: static int
1.59 schwarze 615: treescan(void)
1.50 kristaps 616: {
1.139 schwarze 617: char buf[PATH_MAX];
1.50 kristaps 618: FTS *f;
619: FTSENT *ff;
1.84 schwarze 620: struct mlink *mlink;
1.124 schwarze 621: int dform, gzip;
1.94 schwarze 622: char *dsec, *arch, *fsec, *cp;
623: const char *path;
1.50 kristaps 624: const char *argv[2];
625:
626: argv[0] = ".";
627: argv[1] = (char *)NULL;
628:
1.220.2.3 schwarze 629: f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR,
630: fts_compare);
1.200 schwarze 631: if (f == NULL) {
1.59 schwarze 632: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 633: say("", "&fts_open");
1.197 schwarze 634: return 0;
1.50 kristaps 635: }
636:
637: dsec = arch = NULL;
638: dform = FORM_NONE;
639:
1.200 schwarze 640: while ((ff = fts_read(f)) != NULL) {
1.50 kristaps 641: path = ff->fts_path + 2;
1.139 schwarze 642: switch (ff->fts_info) {
643:
644: /*
645: * Symbolic links require various sanity checks,
646: * then get handled just like regular files.
647: */
1.141 schwarze 648: case FTS_SL:
1.200 schwarze 649: if (realpath(path, buf) == NULL) {
1.139 schwarze 650: if (warnings)
651: say(path, "&realpath");
652: continue;
653: }
1.184 schwarze 654: if (strstr(buf, basedir) != buf
655: #ifdef HOMEBREWDIR
656: && strstr(buf, HOMEBREWDIR) != buf
657: #endif
658: ) {
1.139 schwarze 659: if (warnings) say("",
660: "%s: outside base directory", buf);
661: continue;
662: }
663: /* Use logical inode to avoid mpages dupe. */
1.200 schwarze 664: if (stat(path, ff->fts_statp) == -1) {
1.139 schwarze 665: if (warnings)
666: say(path, "&stat");
667: continue;
668: }
669: /* FALLTHROUGH */
670:
1.50 kristaps 671: /*
1.83 schwarze 672: * If we're a regular file, add an mlink by using the
1.50 kristaps 673: * stored directory data and handling the filename.
674: */
1.141 schwarze 675: case FTS_F:
1.200 schwarze 676: if ( ! strcmp(path, MANDOC_DB))
1.60 schwarze 677: continue;
1.50 kristaps 678: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 679: if (warnings)
1.59 schwarze 680: say(path, "Extraneous file");
1.50 kristaps 681: continue;
1.124 schwarze 682: }
683: gzip = 0;
684: fsec = NULL;
1.200 schwarze 685: while (fsec == NULL) {
1.124 schwarze 686: fsec = strrchr(ff->fts_name, '.');
1.200 schwarze 687: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.124 schwarze 688: break;
689: gzip = 1;
690: *fsec = '\0';
691: fsec = NULL;
692: }
1.200 schwarze 693: if (fsec == NULL) {
1.60 schwarze 694: if ( ! use_all) {
1.56 schwarze 695: if (warnings)
1.60 schwarze 696: say(path,
697: "No filename suffix");
1.50 kristaps 698: continue;
699: }
1.200 schwarze 700: } else if ( ! strcmp(++fsec, "html")) {
1.60 schwarze 701: if (warnings)
702: say(path, "Skip html");
703: continue;
1.200 schwarze 704: } else if ( ! strcmp(fsec, "ps")) {
1.60 schwarze 705: if (warnings)
706: say(path, "Skip ps");
707: continue;
1.200 schwarze 708: } else if ( ! strcmp(fsec, "pdf")) {
1.60 schwarze 709: if (warnings)
710: say(path, "Skip pdf");
711: continue;
712: } else if ( ! use_all &&
1.200 schwarze 713: ((dform == FORM_SRC &&
1.185 schwarze 714: strncmp(fsec, dsec, strlen(dsec))) ||
1.200 schwarze 715: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.60 schwarze 716: if (warnings)
717: say(path, "Wrong filename suffix");
718: continue;
1.66 schwarze 719: } else
1.84 schwarze 720: fsec[-1] = '\0';
1.94 schwarze 721:
1.84 schwarze 722: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 723: if (strlcpy(mlink->file, path,
724: sizeof(mlink->file)) >=
725: sizeof(mlink->file)) {
726: say(path, "Filename too long");
727: free(mlink);
728: continue;
729: }
1.84 schwarze 730: mlink->dform = dform;
1.94 schwarze 731: mlink->dsec = dsec;
732: mlink->arch = arch;
733: mlink->name = ff->fts_name;
734: mlink->fsec = fsec;
1.124 schwarze 735: mlink->gzip = gzip;
1.84 schwarze 736: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 737: continue;
1.139 schwarze 738:
1.141 schwarze 739: case FTS_D:
740: case FTS_DP:
1.139 schwarze 741: break;
742:
743: default:
1.60 schwarze 744: if (warnings)
745: say(path, "Not a regular file");
1.50 kristaps 746: continue;
1.60 schwarze 747: }
1.1 kristaps 748:
1.50 kristaps 749: switch (ff->fts_level) {
1.141 schwarze 750: case 0:
1.50 kristaps 751: /* Ignore the root directory. */
752: break;
1.141 schwarze 753: case 1:
1.50 kristaps 754: /*
755: * This might contain manX/ or catX/.
756: * Try to infer this from the name.
757: * If we're not in use_all, enforce it.
758: */
759: cp = ff->fts_name;
1.200 schwarze 760: if (ff->fts_info == FTS_DP) {
761: dform = FORM_NONE;
762: dsec = NULL;
1.50 kristaps 763: break;
1.200 schwarze 764: }
1.1 kristaps 765:
1.200 schwarze 766: if ( ! strncmp(cp, "man", 3)) {
1.50 kristaps 767: dform = FORM_SRC;
1.66 schwarze 768: dsec = cp + 3;
1.200 schwarze 769: } else if ( ! strncmp(cp, "cat", 3)) {
1.50 kristaps 770: dform = FORM_CAT;
1.66 schwarze 771: dsec = cp + 3;
1.94 schwarze 772: } else {
773: dform = FORM_NONE;
774: dsec = NULL;
1.50 kristaps 775: }
1.1 kristaps 776:
1.200 schwarze 777: if (dsec != NULL || use_all)
1.50 kristaps 778: break;
1.1 kristaps 779:
1.56 schwarze 780: if (warnings)
1.59 schwarze 781: say(path, "Unknown directory part");
1.50 kristaps 782: fts_set(f, ff, FTS_SKIP);
783: break;
1.141 schwarze 784: case 2:
1.50 kristaps 785: /*
786: * Possibly our architecture.
787: * If we're descending, keep tabs on it.
788: */
1.200 schwarze 789: if (ff->fts_info != FTS_DP && dsec != NULL)
1.66 schwarze 790: arch = ff->fts_name;
1.94 schwarze 791: else
792: arch = NULL;
1.50 kristaps 793: break;
794: default:
1.200 schwarze 795: if (ff->fts_info == FTS_DP || use_all)
1.50 kristaps 796: break;
1.56 schwarze 797: if (warnings)
1.59 schwarze 798: say(path, "Extraneous directory part");
1.50 kristaps 799: fts_set(f, ff, FTS_SKIP);
800: break;
1.5 kristaps 801: }
1.50 kristaps 802: }
803:
804: fts_close(f);
1.197 schwarze 805: return 1;
1.50 kristaps 806: }
1.5 kristaps 807:
1.50 kristaps 808: /*
1.89 schwarze 809: * Add a file to the mlinks table.
1.50 kristaps 810: * Do not verify that it's a "valid" looking manpage (we'll do that
811: * later).
812: *
813: * Try to infer the manual section, architecture, and page name from the
814: * path, assuming it looks like
815: *
1.141 schwarze 816: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 817: * or
818: * [./]cat<section>[/<arch>]/<name>.0
819: *
820: * See treescan() for the fts(3) version of this.
821: */
822: static void
1.59 schwarze 823: filescan(const char *file)
1.50 kristaps 824: {
1.59 schwarze 825: char buf[PATH_MAX];
1.84 schwarze 826: struct stat st;
827: struct mlink *mlink;
1.59 schwarze 828: char *p, *start;
1.5 kristaps 829:
1.50 kristaps 830: assert(use_all);
1.5 kristaps 831:
1.50 kristaps 832: if (0 == strncmp(file, "./", 2))
833: file += 2;
1.5 kristaps 834:
1.139 schwarze 835: /*
836: * We have to do lstat(2) before realpath(3) loses
837: * the information whether this is a symbolic link.
838: * We need to know that because for symbolic links,
839: * we want to use the orginal file name, while for
840: * regular files, we want to use the real path.
841: */
842: if (-1 == lstat(file, &st)) {
843: exitcode = (int)MANDOCLEVEL_BADARG;
844: say(file, "&lstat");
845: return;
846: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
847: exitcode = (int)MANDOCLEVEL_BADARG;
848: say(file, "Not a regular file");
849: return;
850: }
851:
852: /*
853: * We have to resolve the file name to the real path
854: * in any case for the base directory check.
855: */
1.59 schwarze 856: if (NULL == realpath(file, buf)) {
857: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 858: say(file, "&realpath");
1.59 schwarze 859: return;
1.106 schwarze 860: }
861:
1.150 schwarze 862: if (OP_TEST == op)
1.106 schwarze 863: start = buf;
1.150 schwarze 864: else if (strstr(buf, basedir) == buf)
865: start = buf + strlen(basedir);
1.184 schwarze 866: #ifdef HOMEBREWDIR
867: else if (strstr(buf, HOMEBREWDIR) == buf)
868: start = buf;
869: #endif
1.106 schwarze 870: else {
1.59 schwarze 871: exitcode = (int)MANDOCLEVEL_BADARG;
872: say("", "%s: outside base directory", buf);
873: return;
1.106 schwarze 874: }
875:
1.139 schwarze 876: /*
877: * Now we are sure the file is inside our tree.
878: * If it is a symbolic link, ignore the real path
879: * and use the original name.
880: * This implies passing stuff like "cat1/../man1/foo.1"
881: * on the command line won't work. So don't do that.
882: * Note the stat(2) can still fail if the link target
883: * doesn't exist.
884: */
885: if (S_IFLNK & st.st_mode) {
886: if (-1 == stat(buf, &st)) {
887: exitcode = (int)MANDOCLEVEL_BADARG;
888: say(file, "&stat");
889: return;
890: }
1.143 schwarze 891: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
892: say(file, "Filename too long");
893: return;
894: }
1.150 schwarze 895: start = buf;
896: if (OP_TEST != op && strstr(buf, basedir) == buf)
897: start += strlen(basedir);
1.50 kristaps 898: }
1.106 schwarze 899:
1.84 schwarze 900: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 901: mlink->dform = FORM_NONE;
1.143 schwarze 902: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
903: sizeof(mlink->file)) {
904: say(start, "Filename too long");
1.180 schwarze 905: free(mlink);
1.143 schwarze 906: return;
907: }
1.17 schwarze 908:
1.50 kristaps 909: /*
910: * First try to guess our directory structure.
911: * If we find a separator, try to look for man* or cat*.
912: * If we find one of these and what's underneath is a directory,
913: * assume it's an architecture.
914: */
915: if (NULL != (p = strchr(start, '/'))) {
916: *p++ = '\0';
917: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 918: mlink->dform = FORM_SRC;
1.94 schwarze 919: mlink->dsec = start + 3;
1.50 kristaps 920: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 921: mlink->dform = FORM_CAT;
1.94 schwarze 922: mlink->dsec = start + 3;
1.17 schwarze 923: }
1.5 kristaps 924:
1.50 kristaps 925: start = p;
1.84 schwarze 926: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 927: *p++ = '\0';
1.94 schwarze 928: mlink->arch = start;
1.50 kristaps 929: start = p;
1.84 schwarze 930: }
1.50 kristaps 931: }
932:
933: /*
934: * Now check the file suffix.
935: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
936: */
937: p = strrchr(start, '\0');
938: while (p-- > start && '/' != *p && '.' != *p)
939: /* Loop. */ ;
940:
941: if ('.' == *p) {
942: *p++ = '\0';
1.94 schwarze 943: mlink->fsec = p;
1.5 kristaps 944: }
945:
1.10 kristaps 946: /*
1.50 kristaps 947: * Now try to parse the name.
948: * Use the filename portion of the path.
1.10 kristaps 949: */
1.84 schwarze 950: mlink->name = start;
1.50 kristaps 951: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 952: mlink->name = p + 1;
1.50 kristaps 953: *p = '\0';
1.84 schwarze 954: }
955: mlink_add(mlink, &st);
1.50 kristaps 956: }
1.5 kristaps 957:
1.50 kristaps 958: static void
1.84 schwarze 959: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 960: {
1.83 schwarze 961: struct inodev inodev;
1.78 schwarze 962: struct mpage *mpage;
1.83 schwarze 963: unsigned int slot;
1.50 kristaps 964:
1.84 schwarze 965: assert(NULL != mlink->file);
1.50 kristaps 966:
1.94 schwarze 967: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
968: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
969: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
970: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 971:
972: if ('0' == *mlink->fsec) {
973: free(mlink->fsec);
974: mlink->fsec = mandoc_strdup(mlink->dsec);
975: mlink->fform = FORM_CAT;
976: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
977: mlink->fform = FORM_SRC;
1.74 schwarze 978: else
1.84 schwarze 979: mlink->fform = FORM_NONE;
1.82 schwarze 980:
1.83 schwarze 981: slot = ohash_qlookup(&mlinks, mlink->file);
982: assert(NULL == ohash_find(&mlinks, slot));
983: ohash_insert(&mlinks, slot, mlink);
984:
1.186 schwarze 985: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 986: inodev.st_ino = st->st_ino;
987: inodev.st_dev = st->st_dev;
988: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
989: sizeof(struct inodev), inodev.st_ino);
990: mpage = ohash_find(&mpages, slot);
991: if (NULL == mpage) {
992: mpage = mandoc_calloc(1, sizeof(struct mpage));
993: mpage->inodev.st_ino = inodev.st_ino;
994: mpage->inodev.st_dev = inodev.st_dev;
1.220.2.3 schwarze 995: mpage->next = mpage_head;
996: mpage_head = mpage;
1.83 schwarze 997: ohash_insert(&mpages, slot, mpage);
998: } else
1.85 schwarze 999: mlink->next = mpage->mlinks;
1.82 schwarze 1000: mpage->mlinks = mlink;
1.118 schwarze 1001: mlink->mpage = mpage;
1.82 schwarze 1002: }
1003:
1004: static void
1005: mlink_free(struct mlink *mlink)
1006: {
1007:
1008: free(mlink->dsec);
1009: free(mlink->arch);
1010: free(mlink->name);
1011: free(mlink->fsec);
1012: free(mlink);
1.50 kristaps 1013: }
1.3 kristaps 1014:
1.50 kristaps 1015: static void
1.78 schwarze 1016: mpages_free(void)
1.50 kristaps 1017: {
1.78 schwarze 1018: struct mpage *mpage;
1.82 schwarze 1019: struct mlink *mlink;
1.3 kristaps 1020:
1.220.2.3 schwarze 1021: while ((mpage = mpage_head) != NULL) {
1022: while ((mlink = mpage->mlinks) != NULL) {
1.85 schwarze 1023: mpage->mlinks = mlink->next;
1.82 schwarze 1024: mlink_free(mlink);
1025: }
1.220.2.3 schwarze 1026: mpage_head = mpage->next;
1.78 schwarze 1027: free(mpage->sec);
1028: free(mpage->arch);
1.82 schwarze 1029: free(mpage->title);
1030: free(mpage->desc);
1.78 schwarze 1031: free(mpage);
1.50 kristaps 1032: }
1033: }
1.38 schwarze 1034:
1.50 kristaps 1035: /*
1.89 schwarze 1036: * For each mlink to the mpage, check whether the path looks like
1037: * it is formatted, and if it does, check whether a source manual
1038: * exists by the same name, ignoring the suffix.
1039: * If both conditions hold, drop the mlink.
1040: */
1041: static void
1042: mlinks_undupe(struct mpage *mpage)
1043: {
1044: char buf[PATH_MAX];
1045: struct mlink **prev;
1046: struct mlink *mlink;
1047: char *bufp;
1048:
1049: mpage->form = FORM_CAT;
1.90 schwarze 1050: prev = &mpage->mlinks;
1051: while (NULL != (mlink = *prev)) {
1.89 schwarze 1052: if (FORM_CAT != mlink->dform) {
1053: mpage->form = FORM_NONE;
1.90 schwarze 1054: goto nextlink;
1.89 schwarze 1055: }
1.143 schwarze 1056: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1057: bufp = strstr(buf, "cat");
1058: assert(NULL != bufp);
1059: memcpy(bufp, "man", 3);
1060: if (NULL != (bufp = strrchr(buf, '.')))
1061: *++bufp = '\0';
1.143 schwarze 1062: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1063: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1064: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1065: goto nextlink;
1.89 schwarze 1066: if (warnings)
1067: say(mlink->file, "Man source exists: %s", buf);
1068: if (use_all)
1.90 schwarze 1069: goto nextlink;
1.89 schwarze 1070: *prev = mlink->next;
1071: mlink_free(mlink);
1.90 schwarze 1072: continue;
1073: nextlink:
1074: prev = &(*prev)->next;
1.89 schwarze 1075: }
1076: }
1077:
1.131 schwarze 1078: static void
1.93 schwarze 1079: mlink_check(struct mpage *mpage, struct mlink *mlink)
1080: {
1.131 schwarze 1081: struct str *str;
1082: unsigned int slot;
1.93 schwarze 1083:
1084: /*
1085: * Check whether the manual section given in a file
1086: * agrees with the directory where the file is located.
1087: * Some manuals have suffixes like (3p) on their
1088: * section number either inside the file or in the
1089: * directory name, some are linked into more than one
1090: * section, like encrypt(1) = makekey(8).
1091: */
1092:
1093: if (FORM_SRC == mpage->form &&
1.131 schwarze 1094: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1095: say(mlink->file, "Section \"%s\" manual in %s directory",
1096: mpage->sec, mlink->dsec);
1097:
1098: /*
1099: * Manual page directories exist for each kernel
1100: * architecture as returned by machine(1).
1101: * However, many manuals only depend on the
1102: * application architecture as returned by arch(1).
1103: * For example, some (2/ARM) manuals are shared
1104: * across the "armish" and "zaurus" kernel
1105: * architectures.
1106: * A few manuals are even shared across completely
1107: * different architectures, for example fdformat(1)
1108: * on amd64, i386, sparc, and sparc64.
1109: */
1110:
1.131 schwarze 1111: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1112: say(mlink->file, "Architecture \"%s\" manual in "
1113: "\"%s\" directory", mpage->arch, mlink->arch);
1114:
1.131 schwarze 1115: /*
1116: * XXX
1.133 schwarze 1117: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1118: */
1119:
1120: if (FORM_CAT == mpage->form)
1121: return;
1122:
1123: /*
1124: * Check whether this mlink
1125: * appears as a name in the NAME section.
1126: */
1.93 schwarze 1127:
1.133 schwarze 1128: slot = ohash_qlookup(&names, mlink->name);
1129: str = ohash_find(&names, slot);
1.131 schwarze 1130: assert(NULL != str);
1.133 schwarze 1131: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1132: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1133: }
1134:
1.89 schwarze 1135: /*
1.80 schwarze 1136: * Run through the files in the global vector "mpages"
1137: * and add them to the database specified in "basedir".
1.50 kristaps 1138: *
1139: * This handles the parsing scheme itself, using the cues of directory
1140: * and filename to determine whether the file is parsable or not.
1141: */
1.59 schwarze 1142: static void
1.176 schwarze 1143: mpages_merge(struct mparse *mp)
1.50 kristaps 1144: {
1.113 schwarze 1145: char any[] = "any";
1.118 schwarze 1146: struct mpage *mpage, *mpage_dest;
1147: struct mlink *mlink, *mlink_dest;
1.191 schwarze 1148: struct roff_man *man;
1.118 schwarze 1149: char *sodest;
1.112 schwarze 1150: char *cp;
1.162 schwarze 1151: int fd;
1.69 schwarze 1152:
1.171 schwarze 1153: if ( ! nodb)
1.107 schwarze 1154: SQL_EXEC("BEGIN TRANSACTION");
1155:
1.220.2.3 schwarze 1156: for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) {
1.89 schwarze 1157: mlinks_undupe(mpage);
1.220.2.3 schwarze 1158: if ((mlink = mpage->mlinks) == NULL)
1.89 schwarze 1159: continue;
1.50 kristaps 1160:
1.133 schwarze 1161: name_mask = NAME_MASK;
1.202 schwarze 1162: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1163: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.50 kristaps 1164: mparse_reset(mp);
1165: man = NULL;
1.124 schwarze 1166: sodest = NULL;
1167:
1.213 schwarze 1168: if ((fd = mparse_open(mp, mlink->file)) == -1) {
1.182 schwarze 1169: say(mlink->file, "&open");
1.162 schwarze 1170: goto nextpage;
1.124 schwarze 1171: }
1.14 schwarze 1172:
1173: /*
1.183 schwarze 1174: * Interpret the file as mdoc(7) or man(7) source
1175: * code, unless it is known to be formatted.
1.14 schwarze 1176: */
1.182 schwarze 1177: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1178: mparse_readfd(mp, fd, mlink->file);
1.212 schwarze 1179: close(fd);
1.192 schwarze 1180: mparse_result(mp, &man, &sodest);
1.76 schwarze 1181: }
1.14 schwarze 1182:
1.171 schwarze 1183: if (sodest != NULL) {
1.118 schwarze 1184: mlink_dest = ohash_find(&mlinks,
1185: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1186: if (mlink_dest == NULL) {
1187: mandoc_asprintf(&cp, "%s.gz", sodest);
1188: mlink_dest = ohash_find(&mlinks,
1189: ohash_qlookup(&mlinks, cp));
1190: free(cp);
1191: }
1192: if (mlink_dest != NULL) {
1.118 schwarze 1193:
1194: /* The .so target exists. */
1195:
1196: mpage_dest = mlink_dest->mpage;
1197: while (1) {
1198: mlink->mpage = mpage_dest;
1199:
1200: /*
1201: * If the target was already
1202: * processed, add the links
1203: * to the database now.
1204: * Otherwise, this will
1205: * happen when we come
1206: * to the target.
1207: */
1208:
1.137 schwarze 1209: if (mpage_dest->pageid)
1.169 schwarze 1210: dbadd_mlink_name(mlink);
1.118 schwarze 1211:
1.171 schwarze 1212: if (mlink->next == NULL)
1.118 schwarze 1213: break;
1214: mlink = mlink->next;
1215: }
1216:
1217: /* Move all links to the target. */
1218:
1219: mlink->next = mlink_dest->next;
1220: mlink_dest->next = mpage->mlinks;
1221: mpage->mlinks = NULL;
1222: }
1.124 schwarze 1223: goto nextpage;
1.192 schwarze 1224: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.204 schwarze 1225: mdoc_validate(man);
1.82 schwarze 1226: mpage->form = FORM_SRC;
1.193 schwarze 1227: mpage->sec = man->meta.msec;
1.155 schwarze 1228: mpage->sec = mandoc_strdup(
1.171 schwarze 1229: mpage->sec == NULL ? "" : mpage->sec);
1.193 schwarze 1230: mpage->arch = man->meta.arch;
1.84 schwarze 1231: mpage->arch = mandoc_strdup(
1.171 schwarze 1232: mpage->arch == NULL ? "" : mpage->arch);
1.193 schwarze 1233: mpage->title = mandoc_strdup(man->meta.title);
1.192 schwarze 1234: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.205 schwarze 1235: man_validate(man);
1.82 schwarze 1236: mpage->form = FORM_SRC;
1.193 schwarze 1237: mpage->sec = mandoc_strdup(man->meta.msec);
1.182 schwarze 1238: mpage->arch = mandoc_strdup(mlink->arch);
1.193 schwarze 1239: mpage->title = mandoc_strdup(man->meta.title);
1.76 schwarze 1240: } else {
1.82 schwarze 1241: mpage->form = FORM_CAT;
1.182 schwarze 1242: mpage->sec = mandoc_strdup(mlink->dsec);
1243: mpage->arch = mandoc_strdup(mlink->arch);
1244: mpage->title = mandoc_strdup(mlink->name);
1.76 schwarze 1245: }
1.97 schwarze 1246: putkey(mpage, mpage->sec, TYPE_sec);
1.164 schwarze 1247: if (*mpage->arch != '\0')
1248: putkey(mpage, mpage->arch, TYPE_arch);
1.50 kristaps 1249:
1.182 schwarze 1250: for ( ; mlink != NULL; mlink = mlink->next) {
1.97 schwarze 1251: if ('\0' != *mlink->dsec)
1252: putkey(mpage, mlink->dsec, TYPE_sec);
1253: if ('\0' != *mlink->fsec)
1254: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1255: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1256: any : mlink->arch, TYPE_arch);
1.133 schwarze 1257: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1258: }
1.12 schwarze 1259:
1.174 schwarze 1260: assert(mpage->desc == NULL);
1.192 schwarze 1261: if (man != NULL && man->macroset == MACROSET_MDOC)
1.193 schwarze 1262: parse_mdoc(mpage, &man->meta, man->first);
1.174 schwarze 1263: else if (man != NULL)
1.193 schwarze 1264: parse_man(mpage, &man->meta, man->first);
1.50 kristaps 1265: else
1.162 schwarze 1266: parse_cat(mpage, fd);
1.174 schwarze 1267: if (mpage->desc == NULL)
1.135 schwarze 1268: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1269:
1270: if (warnings && !use_all)
1271: for (mlink = mpage->mlinks; mlink;
1272: mlink = mlink->next)
1273: mlink_check(mpage, mlink);
1.44 kristaps 1274:
1.176 schwarze 1275: dbadd(mpage);
1.182 schwarze 1276: mlink = mpage->mlinks;
1.124 schwarze 1277:
1278: nextpage:
1.66 schwarze 1279: ohash_delete(&strings);
1.133 schwarze 1280: ohash_delete(&names);
1.50 kristaps 1281: }
1.107 schwarze 1282:
1283: if (0 == nodb)
1284: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1285: }
1286:
1287: static void
1288: names_check(void)
1289: {
1290: sqlite3_stmt *stmt;
1291: const char *name, *sec, *arch, *key;
1292:
1293: sqlite3_prepare_v2(db,
1294: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1295: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1296: "WHERE bits & ? AND NOT EXISTS ("
1297: "SELECT pageid FROM mlinks "
1.133 schwarze 1298: "WHERE mlinks.pageid == names.pageid "
1299: "AND mlinks.name == names.name"
1.130 schwarze 1300: ")"
1301: ") JOIN ("
1.142 schwarze 1302: "SELECT sec, arch, name, pageid FROM mlinks "
1303: "GROUP BY pageid"
1.130 schwarze 1304: ") USING (pageid);",
1305: -1, &stmt, NULL);
1306:
1.201 schwarze 1307: if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK)
1.134 schwarze 1308: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1309:
1.201 schwarze 1310: while (sqlite3_step(stmt) == SQLITE_ROW) {
1.154 schwarze 1311: name = (const char *)sqlite3_column_text(stmt, 0);
1312: sec = (const char *)sqlite3_column_text(stmt, 1);
1313: arch = (const char *)sqlite3_column_text(stmt, 2);
1314: key = (const char *)sqlite3_column_text(stmt, 3);
1.130 schwarze 1315: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1316: '\0' == *arch ? "" : "/",
1317: '\0' == *arch ? "" : arch, key);
1318: }
1319: sqlite3_finalize(stmt);
1.50 kristaps 1320: }
1.12 schwarze 1321:
1.50 kristaps 1322: static void
1.124 schwarze 1323: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1324: {
1325: FILE *stream;
1326: char *line, *p, *title;
1.209 schwarze 1327: size_t linesz, plen, titlesz;
1328: ssize_t len;
1329: int offs;
1.12 schwarze 1330:
1.124 schwarze 1331: stream = (-1 == fd) ?
1332: fopen(mpage->mlinks->file, "r") :
1333: fdopen(fd, "r");
1334: if (NULL == stream) {
1.159 schwarze 1335: if (-1 != fd)
1336: close(fd);
1.56 schwarze 1337: if (warnings)
1.123 schwarze 1338: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1339: return;
1340: }
1.1 kristaps 1341:
1.209 schwarze 1342: line = NULL;
1343: linesz = 0;
1344:
1.50 kristaps 1345: /* Skip to first blank line. */
1.1 kristaps 1346:
1.209 schwarze 1347: while (getline(&line, &linesz, stream) != -1)
1348: if (*line == '\n')
1.50 kristaps 1349: break;
1.1 kristaps 1350:
1.50 kristaps 1351: /*
1352: * Assume the first line that is not indented
1353: * is the first section header. Skip to it.
1354: */
1.1 kristaps 1355:
1.209 schwarze 1356: while (getline(&line, &linesz, stream) != -1)
1357: if (*line != '\n' && *line != ' ')
1.50 kristaps 1358: break;
1.141 schwarze 1359:
1.50 kristaps 1360: /*
1361: * Read up until the next section into a buffer.
1362: * Strip the leading and trailing newline from each read line,
1363: * appending a trailing space.
1364: * Ignore empty (whitespace-only) lines.
1365: */
1.1 kristaps 1366:
1.50 kristaps 1367: titlesz = 0;
1368: title = NULL;
1.38 schwarze 1369:
1.209 schwarze 1370: while ((len = getline(&line, &linesz, stream)) != -1) {
1371: if (*line != ' ')
1.50 kristaps 1372: break;
1.209 schwarze 1373: offs = 0;
1374: while (isspace((unsigned char)line[offs]))
1375: offs++;
1376: if (line[offs] == '\0')
1.38 schwarze 1377: continue;
1.209 schwarze 1378: title = mandoc_realloc(title, titlesz + len - offs);
1379: memcpy(title + titlesz, line + offs, len - offs);
1380: titlesz += len - offs;
1.50 kristaps 1381: title[titlesz - 1] = ' ';
1382: }
1.209 schwarze 1383: free(line);
1.38 schwarze 1384:
1.50 kristaps 1385: /*
1386: * If no page content can be found, or the input line
1387: * is already the next section header, or there is no
1388: * trailing newline, reuse the page title as the page
1389: * description.
1390: */
1.44 kristaps 1391:
1.50 kristaps 1392: if (NULL == title || '\0' == *title) {
1.56 schwarze 1393: if (warnings)
1.82 schwarze 1394: say(mpage->mlinks->file,
1395: "Cannot find NAME section");
1.50 kristaps 1396: fclose(stream);
1397: free(title);
1398: return;
1399: }
1.1 kristaps 1400:
1.209 schwarze 1401: title[titlesz - 1] = '\0';
1.33 schwarze 1402:
1.50 kristaps 1403: /*
1404: * Skip to the first dash.
1405: * Use the remaining line as the description (no more than 70
1406: * bytes).
1407: */
1.33 schwarze 1408:
1.50 kristaps 1409: if (NULL != (p = strstr(title, "- "))) {
1410: for (p += 2; ' ' == *p || '\b' == *p; p++)
1411: /* Skip to next word. */ ;
1412: } else {
1.56 schwarze 1413: if (warnings)
1.82 schwarze 1414: say(mpage->mlinks->file,
1415: "No dash in title line");
1.50 kristaps 1416: p = title;
1417: }
1.38 schwarze 1418:
1.50 kristaps 1419: plen = strlen(p);
1.1 kristaps 1420:
1.50 kristaps 1421: /* Strip backspace-encoding from line. */
1.1 kristaps 1422:
1.50 kristaps 1423: while (NULL != (line = memchr(p, '\b', plen))) {
1424: len = line - p;
1425: if (0 == len) {
1426: memmove(line, line + 1, plen--);
1427: continue;
1.141 schwarze 1428: }
1.50 kristaps 1429: memmove(line - 1, line + 1, plen - len);
1430: plen -= 2;
1431: }
1.1 kristaps 1432:
1.78 schwarze 1433: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1434: fclose(stream);
1435: free(title);
1436: }
1.1 kristaps 1437:
1.50 kristaps 1438: /*
1439: * Put a type/word pair into the word database for this particular file.
1440: */
1441: static void
1.112 schwarze 1442: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1443: {
1.112 schwarze 1444: char *cp;
1.18 kristaps 1445:
1.50 kristaps 1446: assert(NULL != value);
1.112 schwarze 1447: if (TYPE_arch == type)
1448: for (cp = value; *cp; cp++)
1449: if (isupper((unsigned char)*cp))
1450: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1451: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1452: }
1453:
1454: /*
1.50 kristaps 1455: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1456: */
1457: static void
1.78 schwarze 1458: putmdockey(const struct mpage *mpage,
1.189 schwarze 1459: const struct roff_node *n, uint64_t m)
1.3 kristaps 1460: {
1.18 kristaps 1461:
1.50 kristaps 1462: for ( ; NULL != n; n = n->next) {
1463: if (NULL != n->child)
1.78 schwarze 1464: putmdockey(mpage, n->child, m);
1.188 schwarze 1465: if (n->type == ROFFT_TEXT)
1.78 schwarze 1466: putkey(mpage, n->string, m);
1.50 kristaps 1467: }
1468: }
1.18 kristaps 1469:
1.61 schwarze 1470: static void
1.190 schwarze 1471: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1472: const struct roff_node *n)
1.50 kristaps 1473: {
1.189 schwarze 1474: const struct roff_node *head, *body;
1.121 schwarze 1475: char *start, *title;
1.50 kristaps 1476: char byte;
1.121 schwarze 1477: size_t sz;
1.18 kristaps 1478:
1.215 schwarze 1479: if (n == NULL)
1.61 schwarze 1480: return;
1.18 kristaps 1481:
1.50 kristaps 1482: /*
1483: * We're only searching for one thing: the first text child in
1484: * the BODY of a NAME section. Since we don't keep track of
1485: * sections in -man, run some hoops to find out whether we're in
1486: * the correct section or not.
1487: */
1.18 kristaps 1488:
1.188 schwarze 1489: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.50 kristaps 1490: body = n;
1.215 schwarze 1491: if ((head = body->parent->head) != NULL &&
1492: (head = head->child) != NULL &&
1493: head->next == NULL &&
1.188 schwarze 1494: head->type == ROFFT_TEXT &&
1.215 schwarze 1495: strcmp(head->string, "NAME") == 0 &&
1496: body->child != NULL) {
1.3 kristaps 1497:
1.50 kristaps 1498: /*
1499: * Suck the entire NAME section into memory.
1500: * Yes, we might run away.
1501: * But too many manuals have big, spread-out
1502: * NAME sections over many lines.
1503: */
1.3 kristaps 1504:
1.121 schwarze 1505: title = NULL;
1.194 schwarze 1506: deroff(&title, body);
1.50 kristaps 1507: if (NULL == title)
1.61 schwarze 1508: return;
1.18 kristaps 1509:
1.141 schwarze 1510: /*
1.50 kristaps 1511: * Go through a special heuristic dance here.
1512: * Conventionally, one or more manual names are
1513: * comma-specified prior to a whitespace, then a
1514: * dash, then a description. Try to puzzle out
1515: * the name parts here.
1516: */
1.18 kristaps 1517:
1.121 schwarze 1518: start = title;
1.50 kristaps 1519: for ( ;; ) {
1520: sz = strcspn(start, " ,");
1521: if ('\0' == start[sz])
1522: break;
1.1 kristaps 1523:
1.50 kristaps 1524: byte = start[sz];
1525: start[sz] = '\0';
1.110 schwarze 1526:
1527: /*
1528: * Assume a stray trailing comma in the
1529: * name list if a name begins with a dash.
1530: */
1531:
1532: if ('-' == start[0] ||
1533: ('\\' == start[0] && '-' == start[1]))
1534: break;
1.1 kristaps 1535:
1.133 schwarze 1536: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1537: if ( ! (mpage->name_head_done ||
1538: strcasecmp(start, meta->title))) {
1539: putkey(mpage, start, NAME_HEAD);
1540: mpage->name_head_done = 1;
1541: }
1.1 kristaps 1542:
1.50 kristaps 1543: if (' ' == byte) {
1544: start += sz + 1;
1545: break;
1546: }
1.1 kristaps 1547:
1.50 kristaps 1548: assert(',' == byte);
1549: start += sz + 1;
1550: while (' ' == *start)
1551: start++;
1552: }
1.1 kristaps 1553:
1.121 schwarze 1554: if (start == title) {
1.133 schwarze 1555: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1556: if ( ! (mpage->name_head_done ||
1557: strcasecmp(start, meta->title))) {
1558: putkey(mpage, start, NAME_HEAD);
1559: mpage->name_head_done = 1;
1560: }
1.50 kristaps 1561: free(title);
1.61 schwarze 1562: return;
1.50 kristaps 1563: }
1.1 kristaps 1564:
1.50 kristaps 1565: while (isspace((unsigned char)*start))
1566: start++;
1.1 kristaps 1567:
1.50 kristaps 1568: if (0 == strncmp(start, "-", 1))
1569: start += 1;
1570: else if (0 == strncmp(start, "\\-\\-", 4))
1571: start += 4;
1572: else if (0 == strncmp(start, "\\-", 2))
1573: start += 2;
1574: else if (0 == strncmp(start, "\\(en", 4))
1575: start += 4;
1576: else if (0 == strncmp(start, "\\(em", 4))
1577: start += 4;
1.1 kristaps 1578:
1.50 kristaps 1579: while (' ' == *start)
1580: start++;
1.1 kristaps 1581:
1.78 schwarze 1582: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1583: free(title);
1.61 schwarze 1584: return;
1.50 kristaps 1585: }
1586: }
1.1 kristaps 1587:
1.77 schwarze 1588: for (n = n->child; n; n = n->next) {
1.78 schwarze 1589: if (NULL != mpage->desc)
1.77 schwarze 1590: break;
1.174 schwarze 1591: parse_man(mpage, meta, n);
1.77 schwarze 1592: }
1.1 kristaps 1593: }
1594:
1595: static void
1.190 schwarze 1596: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1597: const struct roff_node *n)
1.1 kristaps 1598: {
1599:
1.50 kristaps 1600: assert(NULL != n);
1601: for (n = n->child; NULL != n; n = n->next) {
1602: switch (n->type) {
1.188 schwarze 1603: case ROFFT_ELEM:
1604: case ROFFT_BLOCK:
1605: case ROFFT_HEAD:
1606: case ROFFT_BODY:
1607: case ROFFT_TAIL:
1.50 kristaps 1608: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1609: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1610: break;
1.68 schwarze 1611: if (mdocs[n->tok].mask)
1.78 schwarze 1612: putmdockey(mpage, n->child,
1613: mdocs[n->tok].mask);
1.50 kristaps 1614: break;
1615: default:
1.188 schwarze 1616: assert(n->type != ROFFT_ROOT);
1.50 kristaps 1617: continue;
1618: }
1619: if (NULL != n->child)
1.172 schwarze 1620: parse_mdoc(mpage, meta, n);
1.1 kristaps 1621: }
1622: }
1623:
1.25 schwarze 1624: static int
1.190 schwarze 1625: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1626: const struct roff_node *n)
1.1 kristaps 1627: {
1.176 schwarze 1628: char *start, *end;
1.1 kristaps 1629: size_t sz;
1.25 schwarze 1630:
1.50 kristaps 1631: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1632: NULL == (n = n->child) ||
1.188 schwarze 1633: n->type != ROFFT_TEXT)
1.197 schwarze 1634: return 0;
1.1 kristaps 1635:
1636: /*
1637: * Only consider those `Fd' macro fields that begin with an
1638: * "inclusion" token (versus, e.g., #define).
1639: */
1.50 kristaps 1640:
1.1 kristaps 1641: if (strcmp("#include", n->string))
1.197 schwarze 1642: return 0;
1.1 kristaps 1643:
1.188 schwarze 1644: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1645: return 0;
1.1 kristaps 1646:
1647: /*
1648: * Strip away the enclosing angle brackets and make sure we're
1649: * not zero-length.
1650: */
1651:
1652: start = n->string;
1653: if ('<' == *start || '"' == *start)
1654: start++;
1655:
1656: if (0 == (sz = strlen(start)))
1.197 schwarze 1657: return 0;
1.1 kristaps 1658:
1659: end = &start[(int)sz - 1];
1660: if ('>' == *end || '"' == *end)
1661: end--;
1662:
1.50 kristaps 1663: if (end > start)
1.78 schwarze 1664: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1665: return 0;
1.1 kristaps 1666: }
1667:
1.178 schwarze 1668: static void
1.189 schwarze 1669: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1670: {
1.112 schwarze 1671: char *cp;
1.178 schwarze 1672: size_t sz;
1.1 kristaps 1673:
1.188 schwarze 1674: if (n->type != ROFFT_TEXT)
1.178 schwarze 1675: return;
1.25 schwarze 1676:
1.178 schwarze 1677: /* Skip function pointer punctuation. */
1.1 kristaps 1678:
1.178 schwarze 1679: cp = n->string;
1680: while (*cp == '(' || *cp == '*')
1.1 kristaps 1681: cp++;
1.178 schwarze 1682: sz = strcspn(cp, "()");
1.1 kristaps 1683:
1.178 schwarze 1684: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1685: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1686: putkeys(mpage, cp, sz, NAME_SYN);
1687: }
1.25 schwarze 1688:
1.178 schwarze 1689: static int
1.190 schwarze 1690: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1691: const struct roff_node *n)
1.178 schwarze 1692: {
1693:
1694: if (n->child == NULL)
1.197 schwarze 1695: return 0;
1.178 schwarze 1696:
1697: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1698:
1.178 schwarze 1699: for (n = n->child->next; n != NULL; n = n->next)
1.188 schwarze 1700: if (n->type == ROFFT_TEXT)
1.78 schwarze 1701: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1702:
1.197 schwarze 1703: return 0;
1.173 schwarze 1704: }
1705:
1706: static int
1.190 schwarze 1707: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1708: const struct roff_node *n)
1.173 schwarze 1709: {
1.177 schwarze 1710:
1.188 schwarze 1711: if (n->type != ROFFT_HEAD)
1.197 schwarze 1712: return 1;
1.173 schwarze 1713:
1.178 schwarze 1714: if (n->child != NULL)
1715: parse_mdoc_fname(mpage, n->child);
1716:
1.197 schwarze 1717: return 0;
1.1 kristaps 1718: }
1719:
1.25 schwarze 1720: static int
1.211 schwarze 1721: parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1722: const struct roff_node *n)
1723: {
1724: char *cp;
1725:
1726: if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1727: return 0;
1728:
1.215 schwarze 1729: if (n->child != NULL &&
1730: n->child->next == NULL &&
1731: n->child->type == ROFFT_TEXT)
1.211 schwarze 1732: return 1;
1733:
1734: cp = NULL;
1735: deroff(&cp, n);
1736: if (cp != NULL) {
1737: putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1738: n->type == ROFFT_BODY ? TYPE_Va : 0));
1739: free(cp);
1740: }
1741:
1742: return 0;
1743: }
1744:
1745: static int
1.190 schwarze 1746: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1747: const struct roff_node *n)
1.1 kristaps 1748: {
1.67 schwarze 1749: char *cp;
1.1 kristaps 1750:
1751: if (NULL == (n = n->child))
1.197 schwarze 1752: return 0;
1.1 kristaps 1753:
1.67 schwarze 1754: if (NULL == n->next) {
1.78 schwarze 1755: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1756: return 0;
1.67 schwarze 1757: }
1758:
1.120 schwarze 1759: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1760: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1761: free(cp);
1.197 schwarze 1762: return 0;
1.1 kristaps 1763: }
1764:
1.25 schwarze 1765: static int
1.190 schwarze 1766: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1767: const struct roff_node *n)
1.1 kristaps 1768: {
1769:
1.188 schwarze 1770: if (n->type == ROFFT_BODY)
1.194 schwarze 1771: deroff(&mpage->desc, n);
1.197 schwarze 1772: return 0;
1.1 kristaps 1773: }
1774:
1.25 schwarze 1775: static int
1.190 schwarze 1776: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1777: const struct roff_node *n)
1.1 kristaps 1778: {
1779:
1.129 schwarze 1780: if (SEC_NAME == n->sec)
1.133 schwarze 1781: putmdockey(mpage, n->child, NAME_TITLE);
1.188 schwarze 1782: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1783: if (n->child == NULL)
1784: putkey(mpage, meta->name, NAME_SYN);
1785: else
1786: putmdockey(mpage, n->child, NAME_SYN);
1.174 schwarze 1787: }
1788: if ( ! (mpage->name_head_done ||
1789: n->child == NULL || n->child->string == NULL ||
1790: strcasecmp(n->child->string, meta->title))) {
1.220.2.1 schwarze 1791: putkey(mpage, n->child->string, NAME_HEAD);
1.174 schwarze 1792: mpage->name_head_done = 1;
1.172 schwarze 1793: }
1.197 schwarze 1794: return 0;
1.1 kristaps 1795: }
1796:
1.25 schwarze 1797: static int
1.190 schwarze 1798: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1799: const struct roff_node *n)
1.1 kristaps 1800: {
1801:
1.197 schwarze 1802: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1803: }
1804:
1.50 kristaps 1805: static int
1.190 schwarze 1806: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1807: const struct roff_node *n)
1.1 kristaps 1808: {
1809:
1.197 schwarze 1810: return n->type == ROFFT_HEAD;
1.1 kristaps 1811: }
1812:
1.50 kristaps 1813: /*
1.66 schwarze 1814: * Add a string to the hash table for the current manual.
1815: * Each string has a bitmask telling which macros it belongs to.
1816: * When we finish the manual, we'll dump the table.
1.50 kristaps 1817: */
1818: static void
1.176 schwarze 1819: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1820: {
1.133 schwarze 1821: struct ohash *htab;
1.50 kristaps 1822: struct str *s;
1.111 schwarze 1823: const char *end;
1.65 schwarze 1824: unsigned int slot;
1.176 schwarze 1825: int i, mustfree;
1.25 schwarze 1826:
1.50 kristaps 1827: if (0 == sz)
1828: return;
1.111 schwarze 1829:
1.176 schwarze 1830: mustfree = render_string(&cp, &sz);
1831:
1.133 schwarze 1832: if (TYPE_Nm & v) {
1833: htab = &names;
1834: v &= name_mask;
1.169 schwarze 1835: if (v & NAME_FIRST)
1836: name_mask &= ~NAME_FIRST;
1.133 schwarze 1837: if (debug > 1)
1838: say(mpage->mlinks->file,
1.220.2.4! schwarze 1839: "Adding name %*s, bits=0x%llx", (int)sz, cp,
! 1840: (unsigned long long)v);
1.133 schwarze 1841: } else {
1842: htab = &strings;
1843: if (debug > 1)
1844: for (i = 0; i < mansearch_keymax; i++)
1.163 schwarze 1845: if ((uint64_t)1 << i & v)
1.133 schwarze 1846: say(mpage->mlinks->file,
1847: "Adding key %s=%*s",
1.220 schwarze 1848: mansearch_keynames[i], (int)sz, cp);
1.111 schwarze 1849: }
1.25 schwarze 1850:
1.65 schwarze 1851: end = cp + sz;
1.133 schwarze 1852: slot = ohash_qlookupi(htab, cp, &end);
1853: s = ohash_find(htab, slot);
1.25 schwarze 1854:
1.78 schwarze 1855: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1856: s->mask |= v;
1857: return;
1858: } else if (NULL == s) {
1.144 schwarze 1859: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1860: memcpy(s->key, cp, sz);
1.133 schwarze 1861: ohash_insert(htab, slot, s);
1.1 kristaps 1862: }
1.78 schwarze 1863: s->mpage = mpage;
1.50 kristaps 1864: s->mask = v;
1.176 schwarze 1865:
1866: if (mustfree)
1867: free(cp);
1.1 kristaps 1868: }
1869:
1.50 kristaps 1870: /*
1871: * Take a Unicode codepoint and produce its UTF-8 encoding.
1872: * This isn't the best way to do this, but it works.
1873: * The magic numbers are from the UTF-8 packaging.
1874: * They're not as scary as they seem: read the UTF-8 spec for details.
1875: */
1876: static size_t
1877: utf8(unsigned int cp, char out[7])
1.1 kristaps 1878: {
1.50 kristaps 1879: size_t rc;
1.1 kristaps 1880:
1.50 kristaps 1881: rc = 0;
1882: if (cp <= 0x0000007F) {
1883: rc = 1;
1884: out[0] = (char)cp;
1885: } else if (cp <= 0x000007FF) {
1886: rc = 2;
1887: out[0] = (cp >> 6 & 31) | 192;
1888: out[1] = (cp & 63) | 128;
1889: } else if (cp <= 0x0000FFFF) {
1890: rc = 3;
1891: out[0] = (cp >> 12 & 15) | 224;
1892: out[1] = (cp >> 6 & 63) | 128;
1893: out[2] = (cp & 63) | 128;
1894: } else if (cp <= 0x001FFFFF) {
1895: rc = 4;
1896: out[0] = (cp >> 18 & 7) | 240;
1897: out[1] = (cp >> 12 & 63) | 128;
1898: out[2] = (cp >> 6 & 63) | 128;
1899: out[3] = (cp & 63) | 128;
1900: } else if (cp <= 0x03FFFFFF) {
1901: rc = 5;
1902: out[0] = (cp >> 24 & 3) | 248;
1903: out[1] = (cp >> 18 & 63) | 128;
1904: out[2] = (cp >> 12 & 63) | 128;
1905: out[3] = (cp >> 6 & 63) | 128;
1906: out[4] = (cp & 63) | 128;
1907: } else if (cp <= 0x7FFFFFFF) {
1908: rc = 6;
1909: out[0] = (cp >> 30 & 1) | 252;
1910: out[1] = (cp >> 24 & 63) | 128;
1911: out[2] = (cp >> 18 & 63) | 128;
1912: out[3] = (cp >> 12 & 63) | 128;
1913: out[4] = (cp >> 6 & 63) | 128;
1914: out[5] = (cp & 63) | 128;
1915: } else
1.197 schwarze 1916: return 0;
1.1 kristaps 1917:
1.50 kristaps 1918: out[rc] = '\0';
1.197 schwarze 1919: return rc;
1.50 kristaps 1920: }
1.1 kristaps 1921:
1.50 kristaps 1922: /*
1.176 schwarze 1923: * If the string contains escape sequences,
1924: * replace it with an allocated rendering and return 1,
1925: * such that the caller can free it after use.
1926: * Otherwise, do nothing and return 0.
1.50 kristaps 1927: */
1.176 schwarze 1928: static int
1929: render_string(char **public, size_t *psz)
1.50 kristaps 1930: {
1.176 schwarze 1931: const char *src, *scp, *addcp, *seq;
1932: char *dst;
1933: size_t ssz, dsz, addsz;
1.114 schwarze 1934: char utfbuf[7], res[6];
1.176 schwarze 1935: int seqlen, unicode;
1.50 kristaps 1936:
1937: res[0] = '\\';
1938: res[1] = '\t';
1939: res[2] = ASCII_NBRSP;
1940: res[3] = ASCII_HYPH;
1.114 schwarze 1941: res[4] = ASCII_BREAK;
1942: res[5] = '\0';
1.1 kristaps 1943:
1.176 schwarze 1944: src = scp = *public;
1945: ssz = *psz;
1946: dst = NULL;
1947: dsz = 0;
1948:
1949: while (scp < src + *psz) {
1950:
1951: /* Leave normal characters unchanged. */
1952:
1953: if (strchr(res, *scp) == NULL) {
1954: if (dst != NULL)
1955: dst[dsz++] = *scp;
1956: scp++;
1957: continue;
1958: }
1.46 kristaps 1959:
1.50 kristaps 1960: /*
1.176 schwarze 1961: * Found something that requires replacing,
1962: * make sure we have a destination buffer.
1.50 kristaps 1963: */
1.176 schwarze 1964:
1965: if (dst == NULL) {
1966: dst = mandoc_malloc(ssz + 1);
1967: dsz = scp - src;
1968: memcpy(dst, src, dsz);
1.50 kristaps 1969: }
1.46 kristaps 1970:
1.176 schwarze 1971: /* Handle single-char special characters. */
1972:
1973: switch (*scp) {
1974: case '\\':
1975: break;
1.141 schwarze 1976: case '\t':
1977: case ASCII_NBRSP:
1.176 schwarze 1978: dst[dsz++] = ' ';
1979: scp++;
1980: continue;
1981: case ASCII_HYPH:
1982: dst[dsz++] = '-';
1.114 schwarze 1983: /* FALLTHROUGH */
1.141 schwarze 1984: case ASCII_BREAK:
1.176 schwarze 1985: scp++;
1.50 kristaps 1986: continue;
1.114 schwarze 1987: default:
1.176 schwarze 1988: abort();
1.114 schwarze 1989: }
1.46 kristaps 1990:
1.50 kristaps 1991: /*
1.176 schwarze 1992: * Found an escape sequence.
1993: * Read past the slash, then parse it.
1994: * Ignore everything except characters.
1.50 kristaps 1995: */
1.95 schwarze 1996:
1.176 schwarze 1997: scp++;
1998: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1999: continue;
1.1 kristaps 2000:
1.50 kristaps 2001: /*
1.95 schwarze 2002: * Render the special character
2003: * as either UTF-8 or ASCII.
1.50 kristaps 2004: */
1.95 schwarze 2005:
2006: if (write_utf8) {
1.203 schwarze 2007: unicode = mchars_spec2cp(seq, seqlen);
1.176 schwarze 2008: if (unicode <= 0)
1.95 schwarze 2009: continue;
1.176 schwarze 2010: addsz = utf8(unicode, utfbuf);
2011: if (addsz == 0)
1.95 schwarze 2012: continue;
1.176 schwarze 2013: addcp = utfbuf;
1.95 schwarze 2014: } else {
1.203 schwarze 2015: addcp = mchars_spec2str(seq, seqlen, &addsz);
1.176 schwarze 2016: if (addcp == NULL)
1.95 schwarze 2017: continue;
1.176 schwarze 2018: if (*addcp == ASCII_NBRSP) {
2019: addcp = " ";
2020: addsz = 1;
1.95 schwarze 2021: }
2022: }
1.1 kristaps 2023:
1.50 kristaps 2024: /* Copy the rendered glyph into the stream. */
1.1 kristaps 2025:
1.176 schwarze 2026: ssz += addsz;
2027: dst = mandoc_realloc(dst, ssz + 1);
2028: memcpy(dst + dsz, addcp, addsz);
2029: dsz += addsz;
2030: }
2031: if (dst != NULL) {
2032: *public = dst;
2033: *psz = dsz;
2034: }
2035:
2036: /* Trim trailing whitespace and NUL-terminate. */
2037:
2038: while (*psz > 0 && (*public)[*psz - 1] == ' ')
2039: --*psz;
2040: if (dst != NULL) {
2041: (*public)[*psz] = '\0';
1.197 schwarze 2042: return 1;
1.176 schwarze 2043: } else
1.197 schwarze 2044: return 0;
1.1 kristaps 2045: }
2046:
1.118 schwarze 2047: static void
2048: dbadd_mlink(const struct mlink *mlink)
2049: {
2050: size_t i;
2051:
2052: i = 1;
2053: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2054: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2055: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 2056: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 2057: SQL_STEP(stmts[STMT_INSERT_LINK]);
2058: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.169 schwarze 2059: }
2060:
2061: static void
2062: dbadd_mlink_name(const struct mlink *mlink)
2063: {
1.175 schwarze 2064: uint64_t bits;
1.169 schwarze 2065: size_t i;
2066:
2067: dbadd_mlink(mlink);
1.160 schwarze 2068:
2069: i = 1;
1.175 schwarze 2070: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2071: bits = NAME_FILE & NAME_MASK;
2072: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2073: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2074: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2075: }
2076:
2077: i = 1;
2078: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.160 schwarze 2079: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2080: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2081: SQL_STEP(stmts[STMT_INSERT_NAME]);
2082: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.118 schwarze 2083: }
2084:
1.14 schwarze 2085: /*
1.50 kristaps 2086: * Flush the current page's terms (and their bits) into the database.
2087: * Wrap the entire set of additions in a transaction to make sqlite be a
2088: * little faster.
1.96 schwarze 2089: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2090: */
2091: static void
1.176 schwarze 2092: dbadd(struct mpage *mpage)
1.14 schwarze 2093: {
1.87 schwarze 2094: struct mlink *mlink;
1.50 kristaps 2095: struct str *key;
1.176 schwarze 2096: char *cp;
1.52 kristaps 2097: size_t i;
1.66 schwarze 2098: unsigned int slot;
1.176 schwarze 2099: int mustfree;
1.14 schwarze 2100:
1.128 schwarze 2101: mlink = mpage->mlinks;
1.43 kristaps 2102:
1.128 schwarze 2103: if (nodb) {
1.147 schwarze 2104: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2105: key = ohash_next(&names, &slot))
1.147 schwarze 2106: free(key);
2107: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2108: key = ohash_next(&strings, &slot))
1.147 schwarze 2109: free(key);
1.145 schwarze 2110: if (0 == debug)
2111: return;
1.128 schwarze 2112: while (NULL != mlink) {
2113: fputs(mlink->name, stdout);
2114: if (NULL == mlink->next ||
2115: strcmp(mlink->dsec, mlink->next->dsec) ||
2116: strcmp(mlink->fsec, mlink->next->fsec) ||
2117: strcmp(mlink->arch, mlink->next->arch)) {
2118: putchar('(');
2119: if ('\0' == *mlink->dsec)
2120: fputs(mlink->fsec, stdout);
2121: else
2122: fputs(mlink->dsec, stdout);
2123: if ('\0' != *mlink->arch)
2124: printf("/%s", mlink->arch);
2125: putchar(')');
2126: }
2127: mlink = mlink->next;
2128: if (NULL != mlink)
2129: fputs(", ", stdout);
2130: }
1.132 schwarze 2131: printf(" - %s\n", mpage->desc);
1.14 schwarze 2132: return;
1.128 schwarze 2133: }
2134:
2135: if (debug)
2136: say(mlink->file, "Adding to database");
1.28 kristaps 2137:
1.176 schwarze 2138: cp = mpage->desc;
2139: i = strlen(cp);
2140: mustfree = render_string(&cp, &i);
1.52 kristaps 2141: i = 1;
1.176 schwarze 2142: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.161 schwarze 2143: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.81 schwarze 2144: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2145: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2146: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.176 schwarze 2147: if (mustfree)
2148: free(cp);
1.81 schwarze 2149:
1.128 schwarze 2150: while (NULL != mlink) {
1.118 schwarze 2151: dbadd_mlink(mlink);
1.128 schwarze 2152: mlink = mlink->next;
2153: }
1.134 schwarze 2154: mlink = mpage->mlinks;
1.50 kristaps 2155:
1.133 schwarze 2156: for (key = ohash_first(&names, &slot); NULL != key;
2157: key = ohash_next(&names, &slot)) {
2158: assert(key->mpage == mpage);
2159: i = 1;
2160: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.176 schwarze 2161: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.137 schwarze 2162: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2163: SQL_STEP(stmts[STMT_INSERT_NAME]);
2164: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2165: free(key);
2166: }
1.66 schwarze 2167: for (key = ohash_first(&strings, &slot); NULL != key;
2168: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2169: assert(key->mpage == mpage);
1.52 kristaps 2170: i = 1;
2171: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.176 schwarze 2172: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.137 schwarze 2173: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2174: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2175: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 2176: free(key);
1.38 schwarze 2177: }
1.14 schwarze 2178: }
2179:
1.5 kristaps 2180: static void
1.59 schwarze 2181: dbprune(void)
1.5 kristaps 2182: {
1.78 schwarze 2183: struct mpage *mpage;
1.82 schwarze 2184: struct mlink *mlink;
1.52 kristaps 2185: size_t i;
1.80 schwarze 2186: unsigned int slot;
1.5 kristaps 2187:
1.106 schwarze 2188: if (0 == nodb)
2189: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2190:
1.106 schwarze 2191: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2192: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2193: mlink = mpage->mlinks;
1.125 schwarze 2194: if (debug)
1.106 schwarze 2195: say(mlink->file, "Deleting from database");
2196: if (nodb)
2197: continue;
2198: for ( ; NULL != mlink; mlink = mlink->next) {
2199: i = 1;
2200: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2201: i, mlink->dsec);
2202: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2203: i, mlink->arch);
2204: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2205: i, mlink->name);
2206: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2207: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2208: }
1.5 kristaps 2209: }
1.106 schwarze 2210:
2211: if (0 == nodb)
2212: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2213: }
2214:
1.4 kristaps 2215: /*
1.50 kristaps 2216: * Close an existing database and its prepared statements.
2217: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2218: */
1.35 kristaps 2219: static void
1.59 schwarze 2220: dbclose(int real)
1.4 kristaps 2221: {
1.50 kristaps 2222: size_t i;
1.115 schwarze 2223: int status;
2224: pid_t child;
1.4 kristaps 2225:
1.50 kristaps 2226: if (nodb)
1.38 schwarze 2227: return;
1.50 kristaps 2228:
2229: for (i = 0; i < STMT__MAX; i++) {
2230: sqlite3_finalize(stmts[i]);
2231: stmts[i] = NULL;
1.4 kristaps 2232: }
2233:
1.50 kristaps 2234: sqlite3_close(db);
2235: db = NULL;
1.12 schwarze 2236:
1.50 kristaps 2237: if (real)
2238: return;
1.12 schwarze 2239:
1.115 schwarze 2240: if ('\0' == *tempfilename) {
2241: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2242: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2243: say(MANDOC_DB, "&rename");
1.115 schwarze 2244: }
2245: return;
2246: }
2247:
2248: switch (child = fork()) {
1.141 schwarze 2249: case -1:
1.115 schwarze 2250: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2251: say("", "&fork cmp");
1.115 schwarze 2252: return;
1.141 schwarze 2253: case 0:
1.115 schwarze 2254: execlp("cmp", "cmp", "-s",
1.196 schwarze 2255: tempfilename, MANDOC_DB, (char *)NULL);
1.123 schwarze 2256: say("", "&exec cmp");
1.115 schwarze 2257: exit(0);
2258: default:
2259: break;
2260: }
2261: if (-1 == waitpid(child, &status, 0)) {
2262: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2263: say("", "&wait cmp");
1.115 schwarze 2264: } else if (WIFSIGNALED(status)) {
2265: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2266: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2267: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2268: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2269: say(MANDOC_DB,
2270: "Data changed, but cannot replace database");
2271: }
2272:
2273: *strrchr(tempfilename, '/') = '\0';
2274: switch (child = fork()) {
1.141 schwarze 2275: case -1:
1.115 schwarze 2276: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2277: say("", "&fork rm");
1.115 schwarze 2278: return;
1.141 schwarze 2279: case 0:
1.196 schwarze 2280: execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
1.123 schwarze 2281: say("", "&exec rm");
1.115 schwarze 2282: exit((int)MANDOCLEVEL_SYSERR);
2283: default:
2284: break;
2285: }
2286: if (-1 == waitpid(child, &status, 0)) {
2287: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2288: say("", "&wait rm");
1.115 schwarze 2289: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2290: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2291: say("", "%s: Cannot remove temporary directory",
2292: tempfilename);
1.59 schwarze 2293: }
1.50 kristaps 2294: }
1.14 schwarze 2295:
1.50 kristaps 2296: /*
2297: * This is straightforward stuff.
2298: * Open a database connection to a "temporary" database, then open a set
2299: * of prepared statements we'll use over and over again.
2300: * If "real" is set, we use the existing database; if not, we truncate a
2301: * temporary one.
2302: * Must be matched by dbclose().
2303: */
2304: static int
1.59 schwarze 2305: dbopen(int real)
1.50 kristaps 2306: {
1.115 schwarze 2307: const char *sql;
1.50 kristaps 2308: int rc, ofl;
1.12 schwarze 2309:
1.141 schwarze 2310: if (nodb)
1.197 schwarze 2311: return 1;
1.12 schwarze 2312:
1.115 schwarze 2313: *tempfilename = '\0';
1.63 schwarze 2314: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2315:
2316: if (real) {
2317: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2318: if (SQLITE_OK != rc) {
1.63 schwarze 2319: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2320: if (SQLITE_CANTOPEN != rc)
2321: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.197 schwarze 2322: return 0;
1.63 schwarze 2323: }
1.115 schwarze 2324: goto prepare_statements;
2325: }
2326:
2327: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2328:
1.115 schwarze 2329: remove(MANDOC_DB "~");
2330: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2331: if (SQLITE_OK == rc)
1.115 schwarze 2332: goto create_tables;
1.116 schwarze 2333: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2334: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2335: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.197 schwarze 2336: return 0;
1.50 kristaps 2337: }
1.12 schwarze 2338:
1.143 schwarze 2339: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2340: sizeof(tempfilename));
1.115 schwarze 2341: if (NULL == mkdtemp(tempfilename)) {
2342: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2343: say("", "&%s", tempfilename);
1.197 schwarze 2344: return 0;
1.115 schwarze 2345: }
1.143 schwarze 2346: (void)strlcat(tempfilename, "/" MANDOC_DB,
2347: sizeof(tempfilename));
1.115 schwarze 2348: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2349: if (SQLITE_OK != rc) {
1.59 schwarze 2350: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2351: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.197 schwarze 2352: return 0;
1.50 kristaps 2353: }
1.12 schwarze 2354:
1.115 schwarze 2355: create_tables:
1.81 schwarze 2356: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2357: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2358: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2359: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2360: ");\n"
2361: "\n"
1.81 schwarze 2362: "CREATE TABLE \"mlinks\" (\n"
2363: " \"sec\" TEXT NOT NULL,\n"
2364: " \"arch\" TEXT NOT NULL,\n"
2365: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2366: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2367: "ON DELETE CASCADE\n"
1.81 schwarze 2368: ");\n"
1.136 schwarze 2369: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2370: "\n"
1.133 schwarze 2371: "CREATE TABLE \"names\" (\n"
2372: " \"bits\" INTEGER NOT NULL,\n"
2373: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2374: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.175 schwarze 2375: "ON DELETE CASCADE,\n"
2376: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.133 schwarze 2377: ");\n"
2378: "\n"
1.50 kristaps 2379: "CREATE TABLE \"keys\" (\n"
2380: " \"bits\" INTEGER NOT NULL,\n"
2381: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2382: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2383: "ON DELETE CASCADE\n"
1.136 schwarze 2384: ");\n"
2385: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2386:
1.50 kristaps 2387: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2388: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2389: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2390: sqlite3_close(db);
1.197 schwarze 2391: return 0;
1.50 kristaps 2392: }
1.4 kristaps 2393:
1.57 schwarze 2394: prepare_statements:
1.146 schwarze 2395: if (SQLITE_OK != sqlite3_exec(db,
2396: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2397: exitcode = (int)MANDOCLEVEL_SYSERR;
2398: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2399: sqlite3_errmsg(db));
2400: sqlite3_close(db);
1.197 schwarze 2401: return 0;
1.146 schwarze 2402: }
2403:
1.137 schwarze 2404: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2405: "(SELECT pageid FROM mlinks WHERE "
2406: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2407: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2408: sql = "INSERT INTO mpages "
1.132 schwarze 2409: "(desc,form) VALUES (?,?)";
1.81 schwarze 2410: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2411: sql = "INSERT INTO mlinks "
1.104 schwarze 2412: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2413: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.175 schwarze 2414: sql = "SELECT bits FROM names where pageid = ?";
2415: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.133 schwarze 2416: sql = "INSERT INTO names "
2417: "(bits,name,pageid) VALUES (?,?,?)";
2418: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2419: sql = "INSERT INTO keys "
1.81 schwarze 2420: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2421: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2422:
2423: #ifndef __APPLE__
2424: /*
2425: * When opening a new database, we can turn off
2426: * synchronous mode for much better performance.
2427: */
2428:
1.146 schwarze 2429: if (real && SQLITE_OK != sqlite3_exec(db,
2430: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2431: exitcode = (int)MANDOCLEVEL_SYSERR;
2432: say(MANDOC_DB, "PRAGMA synchronous: %s",
1.162 schwarze 2433: sqlite3_errmsg(db));
1.146 schwarze 2434: sqlite3_close(db);
1.197 schwarze 2435: return 0;
1.146 schwarze 2436: }
1.70 schwarze 2437: #endif
2438:
1.197 schwarze 2439: return 1;
1.4 kristaps 2440: }
2441:
1.50 kristaps 2442: static int
1.165 schwarze 2443: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2444: {
1.59 schwarze 2445: static char startdir[PATH_MAX];
1.151 schwarze 2446: static int getcwd_status; /* 1 = ok, 2 = failure */
2447: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2448: char *cp;
1.4 kristaps 2449:
1.59 schwarze 2450: /*
1.151 schwarze 2451: * Remember the original working directory, if possible.
2452: * This will be needed if the second or a later directory
2453: * on the command line is given as a relative path.
2454: * Do not error out if the current directory is not
2455: * searchable: Maybe it won't be needed after all.
2456: */
2457: if (0 == getcwd_status) {
2458: if (NULL == getcwd(startdir, sizeof(startdir))) {
2459: getcwd_status = 2;
2460: (void)strlcpy(startdir, strerror(errno),
2461: sizeof(startdir));
2462: } else
2463: getcwd_status = 1;
2464: }
2465:
2466: /*
2467: * We are leaving the old base directory.
2468: * Do not use it any longer, not even for messages.
2469: */
2470: *basedir = '\0';
2471:
2472: /*
2473: * If and only if the directory was changed earlier and
2474: * the next directory to process is given as a relative path,
2475: * first go back, or bail out if that is impossible.
1.59 schwarze 2476: */
1.151 schwarze 2477: if (chdir_status && '/' != *targetdir) {
2478: if (2 == getcwd_status) {
1.59 schwarze 2479: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2480: say("", "getcwd: %s", startdir);
1.197 schwarze 2481: return 0;
1.59 schwarze 2482: }
1.151 schwarze 2483: if (-1 == chdir(startdir)) {
1.59 schwarze 2484: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2485: say("", "&chdir %s", startdir);
1.197 schwarze 2486: return 0;
1.59 schwarze 2487: }
2488: }
1.151 schwarze 2489:
2490: /*
2491: * Always resolve basedir to the canonicalized absolute
2492: * pathname and append a trailing slash, such that
2493: * we can reliably check whether files are inside.
2494: */
1.59 schwarze 2495: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2496: if (report_baddir || errno != ENOENT) {
2497: exitcode = (int)MANDOCLEVEL_BADARG;
2498: say("", "&%s: realpath", targetdir);
2499: }
1.197 schwarze 2500: return 0;
1.59 schwarze 2501: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2502: if (report_baddir || errno != ENOENT) {
2503: exitcode = (int)MANDOCLEVEL_BADARG;
2504: say("", "&chdir");
2505: }
1.197 schwarze 2506: return 0;
1.150 schwarze 2507: }
1.151 schwarze 2508: chdir_status = 1;
1.150 schwarze 2509: cp = strchr(basedir, '\0');
2510: if ('/' != cp[-1]) {
2511: if (cp - basedir >= PATH_MAX - 1) {
2512: exitcode = (int)MANDOCLEVEL_SYSERR;
2513: say("", "Filename too long");
1.197 schwarze 2514: return 0;
1.150 schwarze 2515: }
2516: *cp++ = '/';
2517: *cp = '\0';
1.4 kristaps 2518: }
1.197 schwarze 2519: return 1;
1.56 schwarze 2520: }
2521:
2522: static void
1.59 schwarze 2523: say(const char *file, const char *format, ...)
1.56 schwarze 2524: {
2525: va_list ap;
1.123 schwarze 2526: int use_errno;
1.56 schwarze 2527:
1.59 schwarze 2528: if ('\0' != *basedir)
2529: fprintf(stderr, "%s", basedir);
2530: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2531: fputc('/', stderr);
1.56 schwarze 2532: if ('\0' != *file)
1.59 schwarze 2533: fprintf(stderr, "%s", file);
2534:
1.123 schwarze 2535: use_errno = 1;
2536: if (NULL != format) {
2537: switch (*format) {
1.141 schwarze 2538: case '&':
1.123 schwarze 2539: format++;
2540: break;
1.141 schwarze 2541: case '\0':
1.123 schwarze 2542: format = NULL;
2543: break;
2544: default:
2545: use_errno = 0;
2546: break;
2547: }
2548: }
2549: if (NULL != format) {
2550: if ('\0' != *basedir || '\0' != *file)
2551: fputs(": ", stderr);
2552: va_start(ap, format);
2553: vfprintf(stderr, format, ap);
2554: va_end(ap);
2555: }
2556: if (use_errno) {
2557: if ('\0' != *basedir || '\0' != *file || NULL != format)
2558: fputs(": ", stderr);
1.59 schwarze 2559: perror(NULL);
1.123 schwarze 2560: } else
2561: fputc('\n', stderr);
1.1 kristaps 2562: }
CVSweb