Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.89
1.89 ! schwarze 1: /* $Id: mandocdb.c,v 1.88 2014/04/04 18:23:07 schwarze Exp $ */
1.1 schwarze 2: /*
1.47 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.52 schwarze 4: * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 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: *
10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
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: */
1.47 schwarze 18: #include <sys/stat.h>
1.72 schwarze 19: #include <sys/wait.h>
1.1 schwarze 20:
21: #include <assert.h>
1.33 schwarze 22: #include <ctype.h>
1.34 schwarze 23: #include <errno.h>
1.1 schwarze 24: #include <fcntl.h>
1.47 schwarze 25: #include <fts.h>
1.1 schwarze 26: #include <getopt.h>
1.44 schwarze 27: #include <limits.h>
1.47 schwarze 28: #include <stddef.h>
1.1 schwarze 29: #include <stdio.h>
30: #include <stdint.h>
31: #include <stdlib.h>
32: #include <string.h>
1.14 schwarze 33: #include <unistd.h>
1.1 schwarze 34:
1.47 schwarze 35: #include <ohash.h>
36: #include <sqlite3.h>
37:
38: #include "mdoc.h"
1.1 schwarze 39: #include "man.h"
40: #include "mandoc.h"
1.76 schwarze 41: #include "mandoc_aux.h"
1.10 schwarze 42: #include "manpath.h"
1.47 schwarze 43: #include "mansearch.h"
1.1 schwarze 44:
1.68 schwarze 45: extern int mansearch_keymax;
46: extern const char *const mansearch_keynames[];
47:
1.47 schwarze 48: #define SQL_EXEC(_v) \
49: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
50: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
51: #define SQL_BIND_TEXT(_s, _i, _v) \
52: if (SQLITE_OK != sqlite3_bind_text \
53: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
54: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
55: #define SQL_BIND_INT(_s, _i, _v) \
56: if (SQLITE_OK != sqlite3_bind_int \
57: ((_s), (_i)++, (_v))) \
58: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
59: #define SQL_BIND_INT64(_s, _i, _v) \
60: if (SQLITE_OK != sqlite3_bind_int64 \
61: ((_s), (_i)++, (_v))) \
62: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
63: #define SQL_STEP(_s) \
64: if (SQLITE_DONE != sqlite3_step((_s))) \
65: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
1.1 schwarze 66:
1.47 schwarze 67: enum op {
68: OP_DEFAULT = 0, /* new dbs from dir list or default config */
69: OP_CONFFILE, /* new databases from custom config file */
70: OP_UPDATE, /* delete/add entries in existing database */
71: OP_DELETE, /* delete entries from existing database */
72: OP_TEST /* change no databases, report potential problems */
73: };
1.11 schwarze 74:
1.47 schwarze 75: enum form {
76: FORM_NONE, /* format is unknown */
77: FORM_SRC, /* format is -man or -mdoc */
78: FORM_CAT /* format is cat */
79: };
1.28 schwarze 80:
1.47 schwarze 81: struct str {
1.53 schwarze 82: char *rendered; /* key in UTF-8 or ASCII form */
1.47 schwarze 83: const struct mpage *mpage; /* if set, the owning parse */
84: uint64_t mask; /* bitmask in sequence */
1.53 schwarze 85: char key[]; /* may contain escape sequences */
1.28 schwarze 86: };
87:
1.47 schwarze 88: struct inodev {
89: ino_t st_ino;
90: dev_t st_dev;
91: };
1.28 schwarze 92:
1.47 schwarze 93: struct mpage {
94: struct inodev inodev; /* used for hashing routine */
1.75 schwarze 95: int64_t recno; /* id in mpages SQL table */
1.47 schwarze 96: enum form form; /* format from file content */
97: char *sec; /* section from file content */
98: char *arch; /* architecture from file content */
99: char *title; /* title from file content */
100: char *desc; /* description from file content */
101: struct mlink *mlinks; /* singly linked list */
1.28 schwarze 102: };
103:
1.47 schwarze 104: struct mlink {
105: char file[PATH_MAX]; /* filename rel. to manpath */
106: enum form dform; /* format from directory */
107: enum form fform; /* format from file name suffix */
108: char *dsec; /* section from directory */
109: char *arch; /* architecture from directory */
110: char *name; /* name from file name (not empty) */
111: char *fsec; /* section from file name suffix */
112: struct mlink *next; /* singly linked list */
1.75 schwarze 113: struct mpage *mpage; /* parent */
1.81 schwarze 114: int gzip; /* filename has a .gz suffix */
1.2 schwarze 115: };
116:
1.47 schwarze 117: enum stmt {
118: STMT_DELETE_PAGE = 0, /* delete mpage */
119: STMT_INSERT_PAGE, /* insert mpage */
120: STMT_INSERT_LINK, /* insert mlink */
121: STMT_INSERT_KEY, /* insert parsed key */
122: STMT__MAX
1.1 schwarze 123: };
124:
1.47 schwarze 125: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
1.1 schwarze 126:
1.19 schwarze 127: struct mdoc_handler {
1.47 schwarze 128: mdoc_fp fp; /* optional handler */
129: uint64_t mask; /* set unless handler returns 0 */
1.19 schwarze 130: };
131:
1.47 schwarze 132: static void dbclose(int);
1.75 schwarze 133: static void dbadd(struct mpage *, struct mchars *);
134: static void dbadd_mlink(const struct mlink *mlink);
1.47 schwarze 135: static int dbopen(int);
136: static void dbprune(void);
137: static void filescan(const char *);
138: static void *hash_alloc(size_t, void *);
139: static void hash_free(void *, size_t, void *);
140: static void *hash_halloc(size_t, void *);
141: static void mlink_add(struct mlink *, const struct stat *);
1.88 schwarze 142: static void mlink_check(struct mpage *, struct mlink *);
1.47 schwarze 143: static void mlink_free(struct mlink *);
144: static void mlinks_undupe(struct mpage *);
145: static void mpages_free(void);
1.58 schwarze 146: static void mpages_merge(struct mchars *, struct mparse *);
1.87 schwarze 147: static void names_check(void);
1.81 schwarze 148: static void parse_cat(struct mpage *, int);
1.47 schwarze 149: static void parse_man(struct mpage *, const struct man_node *);
150: static void parse_mdoc(struct mpage *, const struct mdoc_node *);
151: static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
152: static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
153: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
154: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
155: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
156: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
157: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
158: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
1.69 schwarze 159: static void putkey(const struct mpage *, char *, uint64_t);
1.47 schwarze 160: static void putkeys(const struct mpage *,
161: const char *, size_t, uint64_t);
162: static void putmdockey(const struct mpage *,
163: const struct mdoc_node *, uint64_t);
1.53 schwarze 164: static void render_key(struct mchars *, struct str *);
1.47 schwarze 165: static void say(const char *, const char *, ...);
166: static int set_basedir(const char *);
167: static int treescan(void);
168: static size_t utf8(unsigned int, char [7]);
169:
1.72 schwarze 170: static char tempfilename[32];
1.47 schwarze 171: static char *progname;
1.59 schwarze 172: static int nodb; /* no database changes */
1.73 schwarze 173: static int mparse_options; /* abort the parse early */
1.47 schwarze 174: static int use_all; /* use all found files */
1.82 schwarze 175: static int debug; /* print what we're doing */
1.47 schwarze 176: static int warnings; /* warn about crap */
1.52 schwarze 177: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 178: static int exitcode; /* to be returned by main */
179: static enum op op; /* operational mode */
180: static char basedir[PATH_MAX]; /* current base directory */
181: static struct ohash mpages; /* table of distinct manual pages */
182: static struct ohash mlinks; /* table of directory entries */
183: static struct ohash strings; /* table of all strings */
184: static sqlite3 *db = NULL; /* current database */
185: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
186:
1.19 schwarze 187: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 188: { NULL, 0 }, /* Ap */
189: { NULL, 0 }, /* Dd */
190: { NULL, 0 }, /* Dt */
191: { NULL, 0 }, /* Os */
192: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
193: { parse_mdoc_head, TYPE_Ss }, /* Ss */
194: { NULL, 0 }, /* Pp */
195: { NULL, 0 }, /* D1 */
196: { NULL, 0 }, /* Dl */
197: { NULL, 0 }, /* Bd */
198: { NULL, 0 }, /* Ed */
199: { NULL, 0 }, /* Bl */
200: { NULL, 0 }, /* El */
201: { NULL, 0 }, /* It */
202: { NULL, 0 }, /* Ad */
203: { NULL, TYPE_An }, /* An */
204: { NULL, TYPE_Ar }, /* Ar */
205: { NULL, TYPE_Cd }, /* Cd */
206: { NULL, TYPE_Cm }, /* Cm */
207: { NULL, TYPE_Dv }, /* Dv */
208: { NULL, TYPE_Er }, /* Er */
209: { NULL, TYPE_Ev }, /* Ev */
210: { NULL, 0 }, /* Ex */
211: { NULL, TYPE_Fa }, /* Fa */
212: { parse_mdoc_Fd, 0 }, /* Fd */
213: { NULL, TYPE_Fl }, /* Fl */
214: { parse_mdoc_Fn, 0 }, /* Fn */
215: { NULL, TYPE_Ft }, /* Ft */
216: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 217: { NULL, TYPE_In }, /* In */
1.47 schwarze 218: { NULL, TYPE_Li }, /* Li */
1.89 ! schwarze 219: { parse_mdoc_Nd, 0 }, /* Nd */
1.86 schwarze 220: { parse_mdoc_Nm, 0 }, /* Nm */
1.47 schwarze 221: { NULL, 0 }, /* Op */
222: { NULL, 0 }, /* Ot */
223: { NULL, TYPE_Pa }, /* Pa */
224: { NULL, 0 }, /* Rv */
1.49 schwarze 225: { NULL, TYPE_St }, /* St */
1.47 schwarze 226: { NULL, TYPE_Va }, /* Va */
227: { parse_mdoc_body, TYPE_Va }, /* Vt */
228: { parse_mdoc_Xr, 0 }, /* Xr */
229: { NULL, 0 }, /* %A */
230: { NULL, 0 }, /* %B */
231: { NULL, 0 }, /* %D */
232: { NULL, 0 }, /* %I */
233: { NULL, 0 }, /* %J */
234: { NULL, 0 }, /* %N */
235: { NULL, 0 }, /* %O */
236: { NULL, 0 }, /* %P */
237: { NULL, 0 }, /* %R */
238: { NULL, 0 }, /* %T */
239: { NULL, 0 }, /* %V */
240: { NULL, 0 }, /* Ac */
241: { NULL, 0 }, /* Ao */
242: { NULL, 0 }, /* Aq */
243: { NULL, TYPE_At }, /* At */
244: { NULL, 0 }, /* Bc */
245: { NULL, 0 }, /* Bf */
246: { NULL, 0 }, /* Bo */
247: { NULL, 0 }, /* Bq */
248: { NULL, TYPE_Bsx }, /* Bsx */
249: { NULL, TYPE_Bx }, /* Bx */
250: { NULL, 0 }, /* Db */
251: { NULL, 0 }, /* Dc */
252: { NULL, 0 }, /* Do */
253: { NULL, 0 }, /* Dq */
254: { NULL, 0 }, /* Ec */
255: { NULL, 0 }, /* Ef */
256: { NULL, TYPE_Em }, /* Em */
257: { NULL, 0 }, /* Eo */
258: { NULL, TYPE_Fx }, /* Fx */
259: { NULL, TYPE_Ms }, /* Ms */
260: { NULL, 0 }, /* No */
261: { NULL, 0 }, /* Ns */
262: { NULL, TYPE_Nx }, /* Nx */
263: { NULL, TYPE_Ox }, /* Ox */
264: { NULL, 0 }, /* Pc */
265: { NULL, 0 }, /* Pf */
266: { NULL, 0 }, /* Po */
267: { NULL, 0 }, /* Pq */
268: { NULL, 0 }, /* Qc */
269: { NULL, 0 }, /* Ql */
270: { NULL, 0 }, /* Qo */
271: { NULL, 0 }, /* Qq */
272: { NULL, 0 }, /* Re */
273: { NULL, 0 }, /* Rs */
274: { NULL, 0 }, /* Sc */
275: { NULL, 0 }, /* So */
276: { NULL, 0 }, /* Sq */
277: { NULL, 0 }, /* Sm */
278: { NULL, 0 }, /* Sx */
279: { NULL, TYPE_Sy }, /* Sy */
280: { NULL, TYPE_Tn }, /* Tn */
281: { NULL, 0 }, /* Ux */
282: { NULL, 0 }, /* Xc */
283: { NULL, 0 }, /* Xo */
284: { parse_mdoc_head, 0 }, /* Fo */
285: { NULL, 0 }, /* Fc */
286: { NULL, 0 }, /* Oo */
287: { NULL, 0 }, /* Oc */
288: { NULL, 0 }, /* Bk */
289: { NULL, 0 }, /* Ek */
290: { NULL, 0 }, /* Bt */
291: { NULL, 0 }, /* Hf */
292: { NULL, 0 }, /* Fr */
293: { NULL, 0 }, /* Ud */
294: { NULL, TYPE_Lb }, /* Lb */
295: { NULL, 0 }, /* Lp */
296: { NULL, TYPE_Lk }, /* Lk */
297: { NULL, TYPE_Mt }, /* Mt */
298: { NULL, 0 }, /* Brq */
299: { NULL, 0 }, /* Bro */
300: { NULL, 0 }, /* Brc */
301: { NULL, 0 }, /* %C */
302: { NULL, 0 }, /* Es */
303: { NULL, 0 }, /* En */
304: { NULL, TYPE_Dx }, /* Dx */
305: { NULL, 0 }, /* %Q */
306: { NULL, 0 }, /* br */
307: { NULL, 0 }, /* sp */
308: { NULL, 0 }, /* %U */
309: { NULL, 0 }, /* Ta */
1.1 schwarze 310: };
311:
312: int
1.3 schwarze 313: mandocdb(int argc, char *argv[])
1.1 schwarze 314: {
1.47 schwarze 315: int ch, i;
316: size_t j, sz;
317: const char *path_arg;
318: struct mchars *mc;
319: struct manpaths dirs;
320: struct mparse *mp;
321: struct ohash_info mpages_info, mlinks_info;
322:
323: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
324: memset(&dirs, 0, sizeof(struct manpaths));
325:
326: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
327: mpages_info.halloc = mlinks_info.halloc = hash_halloc;
328: mpages_info.hfree = mlinks_info.hfree = hash_free;
329:
330: mpages_info.key_offset = offsetof(struct mpage, inodev);
331: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 332:
333: progname = strrchr(argv[0], '/');
334: if (progname == NULL)
335: progname = argv[0];
336: else
337: ++progname;
338:
1.47 schwarze 339: /*
340: * We accept a few different invocations.
341: * The CHECKOP macro makes sure that invocation styles don't
342: * clobber each other.
343: */
344: #define CHECKOP(_op, _ch) do \
345: if (OP_DEFAULT != (_op)) { \
346: fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
347: goto usage; \
348: } while (/*CONSTCOND*/0)
1.10 schwarze 349:
1.47 schwarze 350: path_arg = NULL;
1.28 schwarze 351: op = OP_DEFAULT;
1.1 schwarze 352:
1.83 schwarze 353: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 354: switch (ch) {
1.6 schwarze 355: case ('a'):
356: use_all = 1;
357: break;
1.25 schwarze 358: case ('C'):
1.47 schwarze 359: CHECKOP(op, ch);
360: path_arg = optarg;
1.28 schwarze 361: op = OP_CONFFILE;
1.25 schwarze 362: break;
1.82 schwarze 363: case ('D'):
364: debug++;
365: break;
1.1 schwarze 366: case ('d'):
1.47 schwarze 367: CHECKOP(op, ch);
368: path_arg = optarg;
1.2 schwarze 369: op = OP_UPDATE;
1.1 schwarze 370: break;
1.47 schwarze 371: case ('n'):
372: nodb = 1;
373: break;
1.83 schwarze 374: case ('p'):
375: warnings = 1;
376: break;
1.59 schwarze 377: case ('Q'):
1.73 schwarze 378: mparse_options |= MPARSE_QUICK;
1.59 schwarze 379: break;
1.52 schwarze 380: case ('T'):
381: if (strcmp(optarg, "utf8")) {
382: fprintf(stderr, "-T%s: Unsupported "
383: "output format\n", optarg);
384: goto usage;
385: }
386: write_utf8 = 1;
387: break;
1.28 schwarze 388: case ('t'):
1.47 schwarze 389: CHECKOP(op, ch);
1.28 schwarze 390: dup2(STDOUT_FILENO, STDERR_FILENO);
391: op = OP_TEST;
1.47 schwarze 392: nodb = warnings = 1;
1.28 schwarze 393: break;
1.2 schwarze 394: case ('u'):
1.47 schwarze 395: CHECKOP(op, ch);
396: path_arg = optarg;
1.1 schwarze 397: op = OP_DELETE;
398: break;
1.83 schwarze 399: case ('v'):
400: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 401: break;
1.1 schwarze 402: default:
1.28 schwarze 403: goto usage;
1.1 schwarze 404: }
405:
406: argc -= optind;
407: argv += optind;
408:
1.28 schwarze 409: if (OP_CONFFILE == op && argc > 0) {
1.47 schwarze 410: fprintf(stderr, "-C: Too many arguments\n");
1.28 schwarze 411: goto usage;
412: }
413:
1.47 schwarze 414: exitcode = (int)MANDOCLEVEL_OK;
1.73 schwarze 415: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL, NULL);
1.47 schwarze 416: mc = mchars_alloc();
1.2 schwarze 417:
1.47 schwarze 418: ohash_init(&mpages, 6, &mpages_info);
419: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 420:
1.47 schwarze 421: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
422: /*
423: * Force processing all files.
424: */
425: use_all = 1;
1.2 schwarze 426:
1.47 schwarze 427: /*
428: * All of these deal with a specific directory.
429: * Jump into that directory then collect files specified
430: * on the command-line.
431: */
432: if (0 == set_basedir(path_arg))
433: goto out;
434: for (i = 0; i < argc; i++)
435: filescan(argv[i]);
436: if (0 == dbopen(1))
437: goto out;
438: if (OP_TEST != op)
439: dbprune();
440: if (OP_DELETE != op)
1.58 schwarze 441: mpages_merge(mc, mp);
1.47 schwarze 442: dbclose(1);
443: } else {
444: /*
445: * If we have arguments, use them as our manpaths.
446: * If we don't, grok from manpath(1) or however else
447: * manpath_parse() wants to do it.
448: */
449: if (argc > 0) {
450: dirs.paths = mandoc_calloc
451: (argc, sizeof(char *));
452: dirs.sz = (size_t)argc;
453: for (i = 0; i < argc; i++)
454: dirs.paths[i] = mandoc_strdup(argv[i]);
455: } else
456: manpath_parse(&dirs, path_arg, NULL, NULL);
1.84 schwarze 457:
458: if (0 == dirs.sz) {
459: exitcode = (int)MANDOCLEVEL_BADARG;
460: say("", "Empty manpath");
461: }
1.2 schwarze 462:
1.47 schwarze 463: /*
464: * First scan the tree rooted at a base directory, then
465: * build a new database and finally move it into place.
466: * Ignore zero-length directories and strip trailing
467: * slashes.
468: */
469: for (j = 0; j < dirs.sz; j++) {
470: sz = strlen(dirs.paths[j]);
471: if (sz && '/' == dirs.paths[j][sz - 1])
472: dirs.paths[j][--sz] = '\0';
473: if (0 == sz)
474: continue;
1.2 schwarze 475:
1.47 schwarze 476: if (j) {
477: ohash_init(&mpages, 6, &mpages_info);
478: ohash_init(&mlinks, 6, &mlinks_info);
479: }
1.1 schwarze 480:
1.47 schwarze 481: if (0 == set_basedir(dirs.paths[j]))
482: goto out;
483: if (0 == treescan())
484: goto out;
485: if (0 == set_basedir(dirs.paths[j]))
486: goto out;
487: if (0 == dbopen(0))
488: goto out;
1.41 deraadt 489:
1.58 schwarze 490: mpages_merge(mc, mp);
1.87 schwarze 491: if (warnings &&
492: ! (MPARSE_QUICK & mparse_options))
493: names_check();
1.47 schwarze 494: dbclose(0);
1.28 schwarze 495:
1.47 schwarze 496: if (j + 1 < dirs.sz) {
497: mpages_free();
498: ohash_delete(&mpages);
499: ohash_delete(&mlinks);
500: }
1.2 schwarze 501: }
1.47 schwarze 502: }
503: out:
504: set_basedir(NULL);
505: manpath_free(&dirs);
506: mchars_free(mc);
507: mparse_free(mp);
508: mpages_free();
509: ohash_delete(&mpages);
510: ohash_delete(&mlinks);
511: return(exitcode);
512: usage:
1.83 schwarze 513: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
514: " %s [-aDnpQ] [-Tutf8] dir ...\n"
515: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
516: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 517: " %s [-Q] -t file ...\n",
1.47 schwarze 518: progname, progname, progname,
519: progname, progname);
1.1 schwarze 520:
1.47 schwarze 521: return((int)MANDOCLEVEL_BADARG);
522: }
1.1 schwarze 523:
1.47 schwarze 524: /*
525: * Scan a directory tree rooted at "basedir" for manpages.
526: * We use fts(), scanning directory parts along the way for clues to our
527: * section and architecture.
528: *
529: * If use_all has been specified, grok all files.
530: * If not, sanitise paths to the following:
531: *
532: * [./]man*[/<arch>]/<name>.<section>
533: * or
534: * [./]cat<section>[/<arch>]/<name>.0
535: *
536: * TODO: accomodate for multi-language directories.
537: */
538: static int
539: treescan(void)
540: {
541: FTS *f;
542: FTSENT *ff;
543: struct mlink *mlink;
1.81 schwarze 544: int dform, gzip;
1.51 schwarze 545: char *dsec, *arch, *fsec, *cp;
546: const char *path;
1.47 schwarze 547: const char *argv[2];
1.1 schwarze 548:
1.47 schwarze 549: argv[0] = ".";
550: argv[1] = (char *)NULL;
1.1 schwarze 551:
1.47 schwarze 552: /*
553: * Walk through all components under the directory, using the
554: * logical descent of files.
555: */
556: f = fts_open((char * const *)argv, FTS_LOGICAL, NULL);
557: if (NULL == f) {
558: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 559: say("", "&fts_open");
1.47 schwarze 560: return(0);
561: }
1.2 schwarze 562:
1.47 schwarze 563: dsec = arch = NULL;
564: dform = FORM_NONE;
1.2 schwarze 565:
1.47 schwarze 566: while (NULL != (ff = fts_read(f))) {
567: path = ff->fts_path + 2;
1.15 schwarze 568: /*
1.47 schwarze 569: * If we're a regular file, add an mlink by using the
570: * stored directory data and handling the filename.
1.15 schwarze 571: */
1.47 schwarze 572: if (FTS_F == ff->fts_info) {
573: if (0 == strcmp(path, MANDOC_DB))
574: continue;
575: if ( ! use_all && ff->fts_level < 2) {
576: if (warnings)
577: say(path, "Extraneous file");
578: continue;
1.81 schwarze 579: }
580: gzip = 0;
581: fsec = NULL;
582: while (NULL == fsec) {
583: fsec = strrchr(ff->fts_name, '.');
584: if (NULL == fsec || strcmp(fsec+1, "gz"))
585: break;
586: gzip = 1;
587: *fsec = '\0';
588: fsec = NULL;
589: }
590: if (NULL == fsec) {
1.47 schwarze 591: if ( ! use_all) {
592: if (warnings)
593: say(path,
594: "No filename suffix");
595: continue;
596: }
597: } else if (0 == strcmp(++fsec, "html")) {
598: if (warnings)
599: say(path, "Skip html");
600: continue;
601: } else if (0 == strcmp(fsec, "ps")) {
602: if (warnings)
603: say(path, "Skip ps");
604: continue;
605: } else if (0 == strcmp(fsec, "pdf")) {
606: if (warnings)
607: say(path, "Skip pdf");
608: continue;
609: } else if ( ! use_all &&
610: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
611: (FORM_CAT == dform && strcmp(fsec, "0")))) {
612: if (warnings)
613: say(path, "Wrong filename suffix");
614: continue;
615: } else
616: fsec[-1] = '\0';
1.51 schwarze 617:
1.47 schwarze 618: mlink = mandoc_calloc(1, sizeof(struct mlink));
619: strlcpy(mlink->file, path, sizeof(mlink->file));
620: mlink->dform = dform;
1.51 schwarze 621: mlink->dsec = dsec;
622: mlink->arch = arch;
623: mlink->name = ff->fts_name;
624: mlink->fsec = fsec;
1.81 schwarze 625: mlink->gzip = gzip;
1.47 schwarze 626: mlink_add(mlink, ff->fts_statp);
627: continue;
628: } else if (FTS_D != ff->fts_info &&
629: FTS_DP != ff->fts_info) {
630: if (warnings)
631: say(path, "Not a regular file");
632: continue;
633: }
634:
635: switch (ff->fts_level) {
636: case (0):
637: /* Ignore the root directory. */
638: break;
639: case (1):
640: /*
641: * This might contain manX/ or catX/.
642: * Try to infer this from the name.
643: * If we're not in use_all, enforce it.
644: */
645: cp = ff->fts_name;
646: if (FTS_DP == ff->fts_info)
647: break;
1.15 schwarze 648:
1.47 schwarze 649: if (0 == strncmp(cp, "man", 3)) {
650: dform = FORM_SRC;
651: dsec = cp + 3;
652: } else if (0 == strncmp(cp, "cat", 3)) {
653: dform = FORM_CAT;
654: dsec = cp + 3;
1.51 schwarze 655: } else {
656: dform = FORM_NONE;
657: dsec = NULL;
1.26 schwarze 658: }
1.47 schwarze 659:
660: if (NULL != dsec || use_all)
661: break;
662:
663: if (warnings)
664: say(path, "Unknown directory part");
665: fts_set(f, ff, FTS_SKIP);
666: break;
667: case (2):
668: /*
669: * Possibly our architecture.
670: * If we're descending, keep tabs on it.
671: */
672: if (FTS_DP != ff->fts_info && NULL != dsec)
673: arch = ff->fts_name;
1.51 schwarze 674: else
675: arch = NULL;
1.47 schwarze 676: break;
677: default:
678: if (FTS_DP == ff->fts_info || use_all)
679: break;
680: if (warnings)
681: say(path, "Extraneous directory part");
682: fts_set(f, ff, FTS_SKIP);
683: break;
1.14 schwarze 684: }
1.47 schwarze 685: }
686:
687: fts_close(f);
688: return(1);
689: }
1.1 schwarze 690:
1.47 schwarze 691: /*
692: * Add a file to the mlinks table.
693: * Do not verify that it's a "valid" looking manpage (we'll do that
694: * later).
695: *
696: * Try to infer the manual section, architecture, and page name from the
697: * path, assuming it looks like
698: *
699: * [./]man*[/<arch>]/<name>.<section>
700: * or
701: * [./]cat<section>[/<arch>]/<name>.0
702: *
703: * See treescan() for the fts(3) version of this.
704: */
705: static void
706: filescan(const char *file)
707: {
708: char buf[PATH_MAX];
709: struct stat st;
710: struct mlink *mlink;
711: char *p, *start;
712:
713: assert(use_all);
714:
715: if (0 == strncmp(file, "./", 2))
716: file += 2;
717:
718: if (NULL == realpath(file, buf)) {
719: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 720: say(file, "&realpath");
1.47 schwarze 721: return;
1.63 schwarze 722: }
723:
724: if (strstr(buf, basedir) == buf)
725: start = buf + strlen(basedir) + 1;
726: else if (OP_TEST == op)
727: start = buf;
728: else {
1.47 schwarze 729: exitcode = (int)MANDOCLEVEL_BADARG;
730: say("", "%s: outside base directory", buf);
731: return;
1.63 schwarze 732: }
733:
734: if (-1 == stat(buf, &st)) {
1.47 schwarze 735: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 736: say(file, "&stat");
1.47 schwarze 737: return;
738: } else if ( ! (S_IFREG & st.st_mode)) {
739: exitcode = (int)MANDOCLEVEL_BADARG;
740: say(file, "Not a regular file");
741: return;
1.1 schwarze 742: }
1.63 schwarze 743:
1.47 schwarze 744: mlink = mandoc_calloc(1, sizeof(struct mlink));
745: strlcpy(mlink->file, start, sizeof(mlink->file));
1.1 schwarze 746:
1.10 schwarze 747: /*
1.47 schwarze 748: * First try to guess our directory structure.
749: * If we find a separator, try to look for man* or cat*.
750: * If we find one of these and what's underneath is a directory,
751: * assume it's an architecture.
1.10 schwarze 752: */
1.47 schwarze 753: if (NULL != (p = strchr(start, '/'))) {
754: *p++ = '\0';
755: if (0 == strncmp(start, "man", 3)) {
756: mlink->dform = FORM_SRC;
1.51 schwarze 757: mlink->dsec = start + 3;
1.47 schwarze 758: } else if (0 == strncmp(start, "cat", 3)) {
759: mlink->dform = FORM_CAT;
1.51 schwarze 760: mlink->dsec = start + 3;
1.47 schwarze 761: }
1.10 schwarze 762:
1.47 schwarze 763: start = p;
764: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
765: *p++ = '\0';
1.51 schwarze 766: mlink->arch = start;
1.47 schwarze 767: start = p;
1.41 deraadt 768: }
1.47 schwarze 769: }
1.7 schwarze 770:
1.47 schwarze 771: /*
772: * Now check the file suffix.
773: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
774: */
775: p = strrchr(start, '\0');
776: while (p-- > start && '/' != *p && '.' != *p)
777: /* Loop. */ ;
778:
779: if ('.' == *p) {
780: *p++ = '\0';
1.51 schwarze 781: mlink->fsec = p;
1.47 schwarze 782: }
1.41 deraadt 783:
1.47 schwarze 784: /*
785: * Now try to parse the name.
786: * Use the filename portion of the path.
787: */
788: mlink->name = start;
789: if (NULL != (p = strrchr(start, '/'))) {
790: mlink->name = p + 1;
791: *p = '\0';
792: }
793: mlink_add(mlink, &st);
794: }
1.2 schwarze 795:
1.47 schwarze 796: static void
797: mlink_add(struct mlink *mlink, const struct stat *st)
798: {
799: struct inodev inodev;
800: struct mpage *mpage;
801: unsigned int slot;
802:
803: assert(NULL != mlink->file);
804:
1.51 schwarze 805: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
806: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
807: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
808: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 809:
810: if ('0' == *mlink->fsec) {
811: free(mlink->fsec);
812: mlink->fsec = mandoc_strdup(mlink->dsec);
813: mlink->fform = FORM_CAT;
814: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
815: mlink->fform = FORM_SRC;
816: else
817: mlink->fform = FORM_NONE;
1.1 schwarze 818:
1.47 schwarze 819: slot = ohash_qlookup(&mlinks, mlink->file);
820: assert(NULL == ohash_find(&mlinks, slot));
821: ohash_insert(&mlinks, slot, mlink);
822:
823: inodev.st_ino = st->st_ino;
824: inodev.st_dev = st->st_dev;
825: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
826: sizeof(struct inodev), inodev.st_ino);
827: mpage = ohash_find(&mpages, slot);
828: if (NULL == mpage) {
829: mpage = mandoc_calloc(1, sizeof(struct mpage));
830: mpage->inodev.st_ino = inodev.st_ino;
831: mpage->inodev.st_dev = inodev.st_dev;
832: ohash_insert(&mpages, slot, mpage);
833: } else
834: mlink->next = mpage->mlinks;
835: mpage->mlinks = mlink;
1.75 schwarze 836: mlink->mpage = mpage;
1.47 schwarze 837: }
1.1 schwarze 838:
1.47 schwarze 839: static void
840: mlink_free(struct mlink *mlink)
841: {
1.1 schwarze 842:
1.47 schwarze 843: free(mlink->dsec);
844: free(mlink->arch);
845: free(mlink->name);
846: free(mlink->fsec);
847: free(mlink);
848: }
1.1 schwarze 849:
1.47 schwarze 850: static void
851: mpages_free(void)
852: {
853: struct mpage *mpage;
854: struct mlink *mlink;
855: unsigned int slot;
856:
857: mpage = ohash_first(&mpages, &slot);
858: while (NULL != mpage) {
859: while (NULL != (mlink = mpage->mlinks)) {
860: mpage->mlinks = mlink->next;
861: mlink_free(mlink);
862: }
863: free(mpage->sec);
864: free(mpage->arch);
865: free(mpage->title);
866: free(mpage->desc);
867: free(mpage);
868: mpage = ohash_next(&mpages, &slot);
869: }
870: }
1.1 schwarze 871:
1.47 schwarze 872: /*
873: * For each mlink to the mpage, check whether the path looks like
874: * it is formatted, and if it does, check whether a source manual
875: * exists by the same name, ignoring the suffix.
876: * If both conditions hold, drop the mlink.
877: */
878: static void
879: mlinks_undupe(struct mpage *mpage)
880: {
881: char buf[PATH_MAX];
882: struct mlink **prev;
883: struct mlink *mlink;
884: char *bufp;
885:
886: mpage->form = FORM_CAT;
887: prev = &mpage->mlinks;
888: while (NULL != (mlink = *prev)) {
889: if (FORM_CAT != mlink->dform) {
890: mpage->form = FORM_NONE;
891: goto nextlink;
892: }
893: if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
894: if (warnings)
895: say(mlink->file, "Filename too long");
896: goto nextlink;
1.26 schwarze 897: }
1.47 schwarze 898: bufp = strstr(buf, "cat");
899: assert(NULL != bufp);
900: memcpy(bufp, "man", 3);
901: if (NULL != (bufp = strrchr(buf, '.')))
902: *++bufp = '\0';
903: strlcat(buf, mlink->dsec, PATH_MAX);
904: if (NULL == ohash_find(&mlinks,
905: ohash_qlookup(&mlinks, buf)))
906: goto nextlink;
907: if (warnings)
908: say(mlink->file, "Man source exists: %s", buf);
909: if (use_all)
910: goto nextlink;
911: *prev = mlink->next;
912: mlink_free(mlink);
913: continue;
914: nextlink:
915: prev = &(*prev)->next;
1.1 schwarze 916: }
1.47 schwarze 917: }
1.1 schwarze 918:
1.88 schwarze 919: static void
1.50 schwarze 920: mlink_check(struct mpage *mpage, struct mlink *mlink)
921: {
1.88 schwarze 922: struct str *str;
923: unsigned int slot;
1.50 schwarze 924:
925: /*
926: * Check whether the manual section given in a file
927: * agrees with the directory where the file is located.
928: * Some manuals have suffixes like (3p) on their
929: * section number either inside the file or in the
930: * directory name, some are linked into more than one
931: * section, like encrypt(1) = makekey(8).
932: */
933:
934: if (FORM_SRC == mpage->form &&
1.88 schwarze 935: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 936: say(mlink->file, "Section \"%s\" manual in %s directory",
937: mpage->sec, mlink->dsec);
938:
939: /*
940: * Manual page directories exist for each kernel
941: * architecture as returned by machine(1).
942: * However, many manuals only depend on the
943: * application architecture as returned by arch(1).
944: * For example, some (2/ARM) manuals are shared
945: * across the "armish" and "zaurus" kernel
946: * architectures.
947: * A few manuals are even shared across completely
948: * different architectures, for example fdformat(1)
949: * on amd64, i386, sparc, and sparc64.
950: */
951:
1.88 schwarze 952: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 953: say(mlink->file, "Architecture \"%s\" manual in "
954: "\"%s\" directory", mpage->arch, mlink->arch);
955:
1.88 schwarze 956: /*
957: * XXX
958: * parse_cat() doesn't set TYPE_Nm and TYPE_NAME yet.
959: */
960:
961: if (FORM_CAT == mpage->form)
962: return;
963:
964: /*
965: * Check whether this mlink
966: * appears as a name in the NAME section.
967: */
1.50 schwarze 968:
1.88 schwarze 969: slot = ohash_qlookup(&strings, mlink->name);
970: str = ohash_find(&strings, slot);
971: assert(NULL != str);
972: if ( ! (TYPE_NAME & str->mask))
973: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 974: }
975:
1.47 schwarze 976: /*
977: * Run through the files in the global vector "mpages"
978: * and add them to the database specified in "basedir".
979: *
980: * This handles the parsing scheme itself, using the cues of directory
981: * and filename to determine whether the file is parsable or not.
982: */
983: static void
1.58 schwarze 984: mpages_merge(struct mchars *mc, struct mparse *mp)
1.47 schwarze 985: {
1.70 schwarze 986: char any[] = "any";
1.58 schwarze 987: struct ohash_info str_info;
1.81 schwarze 988: int fd[2];
1.75 schwarze 989: struct mpage *mpage, *mpage_dest;
990: struct mlink *mlink, *mlink_dest;
1.47 schwarze 991: struct mdoc *mdoc;
992: struct man *man;
1.75 schwarze 993: char *sodest;
1.69 schwarze 994: char *cp;
1.81 schwarze 995: pid_t child_pid;
1.88 schwarze 996: int status;
1.58 schwarze 997: unsigned int pslot;
1.47 schwarze 998: enum mandoclevel lvl;
999:
1000: str_info.alloc = hash_alloc;
1001: str_info.halloc = hash_halloc;
1002: str_info.hfree = hash_free;
1003: str_info.key_offset = offsetof(struct str, key);
1004:
1.64 schwarze 1005: if (0 == nodb)
1006: SQL_EXEC("BEGIN TRANSACTION");
1007:
1.47 schwarze 1008: mpage = ohash_first(&mpages, &pslot);
1009: while (NULL != mpage) {
1010: mlinks_undupe(mpage);
1011: if (NULL == mpage->mlinks) {
1012: mpage = ohash_next(&mpages, &pslot);
1013: continue;
1014: }
1.1 schwarze 1015:
1.47 schwarze 1016: ohash_init(&strings, 6, &str_info);
1017: mparse_reset(mp);
1018: mdoc = NULL;
1019: man = NULL;
1.81 schwarze 1020: sodest = NULL;
1021: child_pid = 0;
1022: fd[0] = -1;
1023: fd[1] = -1;
1024:
1025: if (mpage->mlinks->gzip) {
1026: if (-1 == pipe(fd)) {
1027: exitcode = (int)MANDOCLEVEL_SYSERR;
1028: say(mpage->mlinks->file, "&pipe gunzip");
1029: goto nextpage;
1030: }
1031: switch (child_pid = fork()) {
1032: case (-1):
1033: exitcode = (int)MANDOCLEVEL_SYSERR;
1034: say(mpage->mlinks->file, "&fork gunzip");
1035: child_pid = 0;
1036: close(fd[1]);
1037: close(fd[0]);
1038: goto nextpage;
1039: case (0):
1040: close(fd[0]);
1041: if (-1 == dup2(fd[1], STDOUT_FILENO)) {
1042: say(mpage->mlinks->file,
1043: "&dup gunzip");
1044: exit(1);
1045: }
1046: execlp("gunzip", "gunzip", "-c",
1047: mpage->mlinks->file, NULL);
1048: say(mpage->mlinks->file, "&exec gunzip");
1049: exit(1);
1050: default:
1051: close(fd[1]);
1052: break;
1053: }
1054: }
1.11 schwarze 1055:
1056: /*
1.24 schwarze 1057: * Try interpreting the file as mdoc(7) or man(7)
1058: * source code, unless it is already known to be
1059: * formatted. Fall back to formatted mode.
1.11 schwarze 1060: */
1.47 schwarze 1061: if (FORM_CAT != mpage->mlinks->dform ||
1062: FORM_CAT != mpage->mlinks->fform) {
1.81 schwarze 1063: lvl = mparse_readfd(mp, fd[0], mpage->mlinks->file);
1.47 schwarze 1064: if (lvl < MANDOCLEVEL_FATAL)
1.75 schwarze 1065: mparse_result(mp, &mdoc, &man, &sodest);
1.47 schwarze 1066: }
1.11 schwarze 1067:
1.75 schwarze 1068: if (NULL != sodest) {
1069: mlink_dest = ohash_find(&mlinks,
1070: ohash_qlookup(&mlinks, sodest));
1071: if (NULL != mlink_dest) {
1072:
1073: /* The .so target exists. */
1074:
1075: mpage_dest = mlink_dest->mpage;
1076: mlink = mpage->mlinks;
1077: while (1) {
1078: mlink->mpage = mpage_dest;
1079:
1080: /*
1081: * If the target was already
1082: * processed, add the links
1083: * to the database now.
1084: * Otherwise, this will
1085: * happen when we come
1086: * to the target.
1087: */
1088:
1089: if (mpage_dest->recno)
1090: dbadd_mlink(mlink);
1091:
1092: if (NULL == mlink->next)
1093: break;
1094: mlink = mlink->next;
1095: }
1096:
1097: /* Move all links to the target. */
1098:
1099: mlink->next = mlink_dest->next;
1100: mlink_dest->next = mpage->mlinks;
1101: mpage->mlinks = NULL;
1102: }
1.81 schwarze 1103: goto nextpage;
1.75 schwarze 1104: } else if (NULL != mdoc) {
1.47 schwarze 1105: mpage->form = FORM_SRC;
1106: mpage->sec =
1107: mandoc_strdup(mdoc_meta(mdoc)->msec);
1108: mpage->arch = mdoc_meta(mdoc)->arch;
1109: mpage->arch = mandoc_strdup(
1110: NULL == mpage->arch ? "" : mpage->arch);
1111: mpage->title =
1112: mandoc_strdup(mdoc_meta(mdoc)->title);
1.11 schwarze 1113: } else if (NULL != man) {
1.47 schwarze 1114: mpage->form = FORM_SRC;
1115: mpage->sec =
1116: mandoc_strdup(man_meta(man)->msec);
1117: mpage->arch =
1118: mandoc_strdup(mpage->mlinks->arch);
1119: mpage->title =
1120: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1121: } else {
1.47 schwarze 1122: mpage->form = FORM_CAT;
1123: mpage->sec =
1124: mandoc_strdup(mpage->mlinks->dsec);
1125: mpage->arch =
1126: mandoc_strdup(mpage->mlinks->arch);
1127: mpage->title =
1128: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1129: }
1.54 schwarze 1130: putkey(mpage, mpage->sec, TYPE_sec);
1.55 schwarze 1131: putkey(mpage, '\0' == *mpage->arch ?
1.70 schwarze 1132: any : mpage->arch, TYPE_arch);
1.1 schwarze 1133:
1.54 schwarze 1134: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1135: if ('\0' != *mlink->dsec)
1136: putkey(mpage, mlink->dsec, TYPE_sec);
1137: if ('\0' != *mlink->fsec)
1138: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1139: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1140: any : mlink->arch, TYPE_arch);
1.50 schwarze 1141: putkey(mpage, mlink->name, TYPE_Nm);
1.54 schwarze 1142: }
1.41 deraadt 1143:
1.47 schwarze 1144: if (NULL != mdoc) {
1145: if (NULL != (cp = mdoc_meta(mdoc)->name))
1146: putkey(mpage, cp, TYPE_Nm);
1147: assert(NULL == mpage->desc);
1148: parse_mdoc(mpage, mdoc_node(mdoc));
1.89 ! schwarze 1149: if (NULL == mpage->desc)
! 1150: mpage->desc = mandoc_strdup(
! 1151: mpage->mlinks->name);
1.47 schwarze 1152: } else if (NULL != man)
1153: parse_man(mpage, man_node(man));
1154: else
1.81 schwarze 1155: parse_cat(mpage, fd[0]);
1.88 schwarze 1156:
1157: if (warnings && !use_all)
1158: for (mlink = mpage->mlinks; mlink;
1159: mlink = mlink->next)
1160: mlink_check(mpage, mlink);
1.6 schwarze 1161:
1.62 schwarze 1162: dbadd(mpage, mc);
1.81 schwarze 1163:
1164: nextpage:
1165: if (child_pid) {
1166: if (-1 == waitpid(child_pid, &status, 0)) {
1167: exitcode = (int)MANDOCLEVEL_SYSERR;
1168: say(mpage->mlinks->file, "&wait gunzip");
1169: } else if (WIFSIGNALED(status)) {
1170: exitcode = (int)MANDOCLEVEL_SYSERR;
1171: say(mpage->mlinks->file,
1172: "gunzip died from signal %d",
1173: WTERMSIG(status));
1174: } else if (WEXITSTATUS(status)) {
1175: exitcode = (int)MANDOCLEVEL_SYSERR;
1176: say(mpage->mlinks->file,
1177: "gunzip failed with code %d",
1178: WEXITSTATUS(status));
1179: }
1180: }
1.47 schwarze 1181: ohash_delete(&strings);
1182: mpage = ohash_next(&mpages, &pslot);
1183: }
1.64 schwarze 1184:
1185: if (0 == nodb)
1186: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1187: }
1188:
1189: static void
1190: names_check(void)
1191: {
1192: sqlite3_stmt *stmt;
1193: const char *name, *sec, *arch, *key;
1194: size_t i;
1195: int irc;
1196:
1197: sqlite3_prepare_v2(db,
1198: "SELECT name, sec, arch, key FROM ("
1199: "SELECT key, pageid FROM keys "
1200: "WHERE bits & ? AND NOT EXISTS ("
1201: "SELECT pageid FROM mlinks "
1202: "WHERE mlinks.pageid == keys.pageid "
1203: "AND mlinks.name == keys.key"
1204: ")"
1205: ") JOIN ("
1206: "SELECT * FROM mlinks GROUP BY pageid"
1207: ") USING (pageid);",
1208: -1, &stmt, NULL);
1209:
1210: i = 1;
1211: SQL_BIND_INT64(stmt, i, TYPE_NAME);
1212:
1213: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1214: name = sqlite3_column_text(stmt, 0);
1215: sec = sqlite3_column_text(stmt, 1);
1216: arch = sqlite3_column_text(stmt, 2);
1217: key = sqlite3_column_text(stmt, 3);
1218: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1219: '\0' == *arch ? "" : "/",
1220: '\0' == *arch ? "" : arch, key);
1221: }
1222: sqlite3_finalize(stmt);
1.47 schwarze 1223: }
1.6 schwarze 1224:
1.47 schwarze 1225: static void
1.81 schwarze 1226: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1227: {
1228: FILE *stream;
1229: char *line, *p, *title;
1230: size_t len, plen, titlesz;
1.1 schwarze 1231:
1.81 schwarze 1232: stream = (-1 == fd) ?
1233: fopen(mpage->mlinks->file, "r") :
1234: fdopen(fd, "r");
1235: if (NULL == stream) {
1.47 schwarze 1236: if (warnings)
1.80 schwarze 1237: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1238: return;
1239: }
1.1 schwarze 1240:
1.47 schwarze 1241: /* Skip to first blank line. */
1.1 schwarze 1242:
1.47 schwarze 1243: while (NULL != (line = fgetln(stream, &len)))
1244: if ('\n' == *line)
1245: break;
1.1 schwarze 1246:
1.47 schwarze 1247: /*
1248: * Assume the first line that is not indented
1249: * is the first section header. Skip to it.
1250: */
1.1 schwarze 1251:
1.47 schwarze 1252: while (NULL != (line = fgetln(stream, &len)))
1253: if ('\n' != *line && ' ' != *line)
1254: break;
1255:
1256: /*
1257: * Read up until the next section into a buffer.
1258: * Strip the leading and trailing newline from each read line,
1259: * appending a trailing space.
1260: * Ignore empty (whitespace-only) lines.
1261: */
1.28 schwarze 1262:
1.47 schwarze 1263: titlesz = 0;
1264: title = NULL;
1.38 schwarze 1265:
1.47 schwarze 1266: while (NULL != (line = fgetln(stream, &len))) {
1267: if (' ' != *line || '\n' != line[len - 1])
1268: break;
1269: while (len > 0 && isspace((unsigned char)*line)) {
1270: line++;
1271: len--;
1272: }
1273: if (1 == len)
1274: continue;
1275: title = mandoc_realloc(title, titlesz + len);
1276: memcpy(title + titlesz, line, len);
1277: titlesz += len;
1278: title[titlesz - 1] = ' ';
1279: }
1.28 schwarze 1280:
1.47 schwarze 1281: /*
1282: * If no page content can be found, or the input line
1283: * is already the next section header, or there is no
1284: * trailing newline, reuse the page title as the page
1285: * description.
1286: */
1.1 schwarze 1287:
1.47 schwarze 1288: if (NULL == title || '\0' == *title) {
1289: if (warnings)
1290: say(mpage->mlinks->file,
1291: "Cannot find NAME section");
1292: assert(NULL == mpage->desc);
1293: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1294: fclose(stream);
1295: free(title);
1296: return;
1297: }
1.24 schwarze 1298:
1.47 schwarze 1299: title = mandoc_realloc(title, titlesz + 1);
1300: title[titlesz] = '\0';
1.24 schwarze 1301:
1.47 schwarze 1302: /*
1303: * Skip to the first dash.
1304: * Use the remaining line as the description (no more than 70
1305: * bytes).
1306: */
1.28 schwarze 1307:
1.47 schwarze 1308: if (NULL != (p = strstr(title, "- "))) {
1309: for (p += 2; ' ' == *p || '\b' == *p; p++)
1310: /* Skip to next word. */ ;
1311: } else {
1312: if (warnings)
1313: say(mpage->mlinks->file,
1314: "No dash in title line");
1315: p = title;
1316: }
1.1 schwarze 1317:
1.47 schwarze 1318: plen = strlen(p);
1.1 schwarze 1319:
1.47 schwarze 1320: /* Strip backspace-encoding from line. */
1.1 schwarze 1321:
1.47 schwarze 1322: while (NULL != (line = memchr(p, '\b', plen))) {
1323: len = line - p;
1324: if (0 == len) {
1325: memmove(line, line + 1, plen--);
1326: continue;
1327: }
1328: memmove(line - 1, line + 1, plen - len);
1329: plen -= 2;
1330: }
1.1 schwarze 1331:
1.47 schwarze 1332: assert(NULL == mpage->desc);
1333: mpage->desc = mandoc_strdup(p);
1334: fclose(stream);
1335: free(title);
1336: }
1.16 schwarze 1337:
1.47 schwarze 1338: /*
1339: * Put a type/word pair into the word database for this particular file.
1340: */
1341: static void
1.69 schwarze 1342: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1343: {
1.69 schwarze 1344: char *cp;
1.37 schwarze 1345:
1.47 schwarze 1346: assert(NULL != value);
1.69 schwarze 1347: if (TYPE_arch == type)
1348: for (cp = value; *cp; cp++)
1349: if (isupper((unsigned char)*cp))
1350: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1351: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1352: }
1353:
1354: /*
1.47 schwarze 1355: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1356: */
1357: static void
1.47 schwarze 1358: putmdockey(const struct mpage *mpage,
1359: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1360: {
1.16 schwarze 1361:
1.47 schwarze 1362: for ( ; NULL != n; n = n->next) {
1363: if (NULL != n->child)
1364: putmdockey(mpage, n->child, m);
1365: if (MDOC_TEXT == n->type)
1366: putkey(mpage, n->string, m);
1367: }
1368: }
1.16 schwarze 1369:
1.47 schwarze 1370: static void
1371: parse_man(struct mpage *mpage, const struct man_node *n)
1372: {
1373: const struct man_node *head, *body;
1.78 schwarze 1374: char *start, *title;
1.47 schwarze 1375: char byte;
1.78 schwarze 1376: size_t sz;
1.16 schwarze 1377:
1.47 schwarze 1378: if (NULL == n)
1379: return;
1.16 schwarze 1380:
1.47 schwarze 1381: /*
1382: * We're only searching for one thing: the first text child in
1383: * the BODY of a NAME section. Since we don't keep track of
1384: * sections in -man, run some hoops to find out whether we're in
1385: * the correct section or not.
1386: */
1.16 schwarze 1387:
1.47 schwarze 1388: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1389: body = n;
1390: assert(body->parent);
1391: if (NULL != (head = body->parent->head) &&
1392: 1 == head->nchild &&
1393: NULL != (head = (head->child)) &&
1394: MAN_TEXT == head->type &&
1395: 0 == strcmp(head->string, "NAME") &&
1.78 schwarze 1396: NULL != body->child) {
1.2 schwarze 1397:
1.47 schwarze 1398: /*
1399: * Suck the entire NAME section into memory.
1400: * Yes, we might run away.
1401: * But too many manuals have big, spread-out
1402: * NAME sections over many lines.
1403: */
1.2 schwarze 1404:
1.78 schwarze 1405: title = NULL;
1406: man_deroff(&title, body);
1.47 schwarze 1407: if (NULL == title)
1408: return;
1.16 schwarze 1409:
1.47 schwarze 1410: /*
1411: * Go through a special heuristic dance here.
1412: * Conventionally, one or more manual names are
1413: * comma-specified prior to a whitespace, then a
1414: * dash, then a description. Try to puzzle out
1415: * the name parts here.
1416: */
1.16 schwarze 1417:
1.78 schwarze 1418: start = title;
1.47 schwarze 1419: for ( ;; ) {
1420: sz = strcspn(start, " ,");
1421: if ('\0' == start[sz])
1422: break;
1.1 schwarze 1423:
1.47 schwarze 1424: byte = start[sz];
1425: start[sz] = '\0';
1.67 schwarze 1426:
1427: /*
1428: * Assume a stray trailing comma in the
1429: * name list if a name begins with a dash.
1430: */
1431:
1432: if ('-' == start[0] ||
1433: ('\\' == start[0] && '-' == start[1]))
1434: break;
1.1 schwarze 1435:
1.86 schwarze 1436: putkey(mpage, start, TYPE_NAME | TYPE_Nm);
1.1 schwarze 1437:
1.47 schwarze 1438: if (' ' == byte) {
1439: start += sz + 1;
1440: break;
1441: }
1.1 schwarze 1442:
1.47 schwarze 1443: assert(',' == byte);
1444: start += sz + 1;
1445: while (' ' == *start)
1446: start++;
1447: }
1.1 schwarze 1448:
1.78 schwarze 1449: if (start == title) {
1.86 schwarze 1450: putkey(mpage, start, TYPE_NAME | TYPE_Nm);
1.47 schwarze 1451: free(title);
1452: return;
1453: }
1.1 schwarze 1454:
1.47 schwarze 1455: while (isspace((unsigned char)*start))
1456: start++;
1.1 schwarze 1457:
1.47 schwarze 1458: if (0 == strncmp(start, "-", 1))
1459: start += 1;
1460: else if (0 == strncmp(start, "\\-\\-", 4))
1461: start += 4;
1462: else if (0 == strncmp(start, "\\-", 2))
1463: start += 2;
1464: else if (0 == strncmp(start, "\\(en", 4))
1465: start += 4;
1466: else if (0 == strncmp(start, "\\(em", 4))
1467: start += 4;
1.1 schwarze 1468:
1.47 schwarze 1469: while (' ' == *start)
1470: start++;
1.1 schwarze 1471:
1.47 schwarze 1472: assert(NULL == mpage->desc);
1473: mpage->desc = mandoc_strdup(start);
1474: free(title);
1475: return;
1476: }
1477: }
1.1 schwarze 1478:
1.47 schwarze 1479: for (n = n->child; n; n = n->next) {
1480: if (NULL != mpage->desc)
1481: break;
1482: parse_man(mpage, n);
1.1 schwarze 1483: }
1484: }
1485:
1486: static void
1.47 schwarze 1487: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1488: {
1489:
1.47 schwarze 1490: assert(NULL != n);
1491: for (n = n->child; NULL != n; n = n->next) {
1492: switch (n->type) {
1493: case (MDOC_ELEM):
1494: /* FALLTHROUGH */
1495: case (MDOC_BLOCK):
1496: /* FALLTHROUGH */
1497: case (MDOC_HEAD):
1498: /* FALLTHROUGH */
1499: case (MDOC_BODY):
1500: /* FALLTHROUGH */
1501: case (MDOC_TAIL):
1502: if (NULL != mdocs[n->tok].fp)
1503: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1504: break;
1505: if (mdocs[n->tok].mask)
1506: putmdockey(mpage, n->child,
1507: mdocs[n->tok].mask);
1508: break;
1509: default:
1510: assert(MDOC_ROOT != n->type);
1511: continue;
1512: }
1513: if (NULL != n->child)
1514: parse_mdoc(mpage, n);
1.1 schwarze 1515: }
1516: }
1517:
1.19 schwarze 1518: static int
1.47 schwarze 1519: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1520: {
1521: const char *start, *end;
1522: size_t sz;
1.19 schwarze 1523:
1.47 schwarze 1524: if (SEC_SYNOPSIS != n->sec ||
1525: NULL == (n = n->child) ||
1526: MDOC_TEXT != n->type)
1.19 schwarze 1527: return(0);
1.1 schwarze 1528:
1529: /*
1530: * Only consider those `Fd' macro fields that begin with an
1531: * "inclusion" token (versus, e.g., #define).
1532: */
1.47 schwarze 1533:
1.1 schwarze 1534: if (strcmp("#include", n->string))
1.19 schwarze 1535: return(0);
1.1 schwarze 1536:
1537: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1538: return(0);
1.1 schwarze 1539:
1540: /*
1541: * Strip away the enclosing angle brackets and make sure we're
1542: * not zero-length.
1543: */
1544:
1545: start = n->string;
1546: if ('<' == *start || '"' == *start)
1547: start++;
1548:
1549: if (0 == (sz = strlen(start)))
1.19 schwarze 1550: return(0);
1.1 schwarze 1551:
1552: end = &start[(int)sz - 1];
1553: if ('>' == *end || '"' == *end)
1554: end--;
1555:
1.47 schwarze 1556: if (end > start)
1557: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1558: return(0);
1.1 schwarze 1559: }
1560:
1.19 schwarze 1561: static int
1.47 schwarze 1562: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1563: {
1.69 schwarze 1564: char *cp;
1.1 schwarze 1565:
1.47 schwarze 1566: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.19 schwarze 1567: return(0);
1568:
1.47 schwarze 1569: /*
1570: * Parse: .Fn "struct type *name" "char *arg".
1571: * First strip away pointer symbol.
1572: * Then store the function name, then type.
1573: * Finally, store the arguments.
1574: */
1.1 schwarze 1575:
1.47 schwarze 1576: if (NULL == (cp = strrchr(n->string, ' ')))
1577: cp = n->string;
1.1 schwarze 1578:
1579: while ('*' == *cp)
1580: cp++;
1581:
1.47 schwarze 1582: putkey(mpage, cp, TYPE_Fn);
1.19 schwarze 1583:
1.47 schwarze 1584: if (n->string < cp)
1585: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.19 schwarze 1586:
1.47 schwarze 1587: for (n = n->next; NULL != n; n = n->next)
1588: if (MDOC_TEXT == n->type)
1589: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1590:
1591: return(0);
1.1 schwarze 1592: }
1593:
1.19 schwarze 1594: static int
1.47 schwarze 1595: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1596: {
1.47 schwarze 1597: char *cp;
1.1 schwarze 1598:
1599: if (NULL == (n = n->child))
1.19 schwarze 1600: return(0);
1.1 schwarze 1601:
1.47 schwarze 1602: if (NULL == n->next) {
1603: putkey(mpage, n->string, TYPE_Xr);
1604: return(0);
1605: }
1.1 schwarze 1606:
1.77 schwarze 1607: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1608: putkey(mpage, cp, TYPE_Xr);
1609: free(cp);
1610: return(0);
1.1 schwarze 1611: }
1612:
1.19 schwarze 1613: static int
1.47 schwarze 1614: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1615: {
1616:
1.79 schwarze 1617: if (MDOC_BODY == n->type)
1618: mdoc_deroff(&mpage->desc, n);
1619: return(0);
1.1 schwarze 1620: }
1621:
1.19 schwarze 1622: static int
1.47 schwarze 1623: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1624: {
1625:
1.86 schwarze 1626: if (SEC_NAME == n->sec)
1627: putmdockey(mpage, n->child, TYPE_NAME | TYPE_Nm);
1628: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)
1629: putmdockey(mpage, n->child, TYPE_Nm);
1630: return(0);
1.1 schwarze 1631: }
1632:
1.19 schwarze 1633: static int
1.47 schwarze 1634: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1635: {
1636:
1.19 schwarze 1637: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1638: }
1639:
1.47 schwarze 1640: static int
1641: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1642: {
1643:
1.47 schwarze 1644: return(MDOC_HEAD == n->type);
1645: }
1.1 schwarze 1646:
1.47 schwarze 1647: static int
1648: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1649: {
1.1 schwarze 1650:
1.47 schwarze 1651: return(MDOC_BODY == n->type);
1.1 schwarze 1652: }
1653:
1.47 schwarze 1654: /*
1655: * Add a string to the hash table for the current manual.
1656: * Each string has a bitmask telling which macros it belongs to.
1657: * When we finish the manual, we'll dump the table.
1658: */
1.1 schwarze 1659: static void
1.47 schwarze 1660: putkeys(const struct mpage *mpage,
1661: const char *cp, size_t sz, uint64_t v)
1.1 schwarze 1662: {
1.47 schwarze 1663: struct str *s;
1.68 schwarze 1664: const char *end;
1665: uint64_t mask;
1.47 schwarze 1666: unsigned int slot;
1.68 schwarze 1667: int i;
1.1 schwarze 1668:
1.47 schwarze 1669: if (0 == sz)
1670: return;
1.68 schwarze 1671:
1.82 schwarze 1672: if (debug > 1) {
1.68 schwarze 1673: for (i = 0, mask = 1;
1674: i < mansearch_keymax;
1675: i++, mask <<= 1)
1676: if (mask & v)
1677: break;
1678: say(mpage->mlinks->file, "Adding key %s=%*s",
1679: mansearch_keynames[i], sz, cp);
1680: }
1.47 schwarze 1681:
1682: end = cp + sz;
1683: slot = ohash_qlookupi(&strings, cp, &end);
1684: s = ohash_find(&strings, slot);
1.1 schwarze 1685:
1.47 schwarze 1686: if (NULL != s && mpage == s->mpage) {
1687: s->mask |= v;
1.1 schwarze 1688: return;
1.47 schwarze 1689: } else if (NULL == s) {
1690: s = mandoc_calloc(sizeof(struct str) + sz + 1, 1);
1691: memcpy(s->key, cp, sz);
1692: ohash_insert(&strings, slot, s);
1693: }
1694: s->mpage = mpage;
1695: s->mask = v;
1.1 schwarze 1696: }
1697:
1698: /*
1.47 schwarze 1699: * Take a Unicode codepoint and produce its UTF-8 encoding.
1700: * This isn't the best way to do this, but it works.
1701: * The magic numbers are from the UTF-8 packaging.
1702: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1703: */
1.47 schwarze 1704: static size_t
1705: utf8(unsigned int cp, char out[7])
1.1 schwarze 1706: {
1.47 schwarze 1707: size_t rc;
1.1 schwarze 1708:
1.47 schwarze 1709: rc = 0;
1710: if (cp <= 0x0000007F) {
1711: rc = 1;
1712: out[0] = (char)cp;
1713: } else if (cp <= 0x000007FF) {
1714: rc = 2;
1715: out[0] = (cp >> 6 & 31) | 192;
1716: out[1] = (cp & 63) | 128;
1717: } else if (cp <= 0x0000FFFF) {
1718: rc = 3;
1719: out[0] = (cp >> 12 & 15) | 224;
1720: out[1] = (cp >> 6 & 63) | 128;
1721: out[2] = (cp & 63) | 128;
1722: } else if (cp <= 0x001FFFFF) {
1723: rc = 4;
1724: out[0] = (cp >> 18 & 7) | 240;
1725: out[1] = (cp >> 12 & 63) | 128;
1726: out[2] = (cp >> 6 & 63) | 128;
1727: out[3] = (cp & 63) | 128;
1728: } else if (cp <= 0x03FFFFFF) {
1729: rc = 5;
1730: out[0] = (cp >> 24 & 3) | 248;
1731: out[1] = (cp >> 18 & 63) | 128;
1732: out[2] = (cp >> 12 & 63) | 128;
1733: out[3] = (cp >> 6 & 63) | 128;
1734: out[4] = (cp & 63) | 128;
1735: } else if (cp <= 0x7FFFFFFF) {
1736: rc = 6;
1737: out[0] = (cp >> 30 & 1) | 252;
1738: out[1] = (cp >> 24 & 63) | 128;
1739: out[2] = (cp >> 18 & 63) | 128;
1740: out[3] = (cp >> 12 & 63) | 128;
1741: out[4] = (cp >> 6 & 63) | 128;
1742: out[5] = (cp & 63) | 128;
1743: } else
1744: return(0);
1.19 schwarze 1745:
1.47 schwarze 1746: out[rc] = '\0';
1747: return(rc);
1.1 schwarze 1748: }
1749:
1.47 schwarze 1750: /*
1.53 schwarze 1751: * Store the rendered version of a key, or alias the pointer
1752: * if the key contains no escape sequences.
1.47 schwarze 1753: */
1754: static void
1.53 schwarze 1755: render_key(struct mchars *mc, struct str *key)
1.1 schwarze 1756: {
1.47 schwarze 1757: size_t sz, bsz, pos;
1.71 schwarze 1758: char utfbuf[7], res[6];
1.47 schwarze 1759: char *buf;
1760: const char *seq, *cpp, *val;
1761: int len, u;
1762: enum mandoc_esc esc;
1763:
1.53 schwarze 1764: assert(NULL == key->rendered);
1.47 schwarze 1765:
1766: res[0] = '\\';
1767: res[1] = '\t';
1768: res[2] = ASCII_NBRSP;
1769: res[3] = ASCII_HYPH;
1.71 schwarze 1770: res[4] = ASCII_BREAK;
1771: res[5] = '\0';
1.1 schwarze 1772:
1.47 schwarze 1773: val = key->key;
1774: bsz = strlen(val);
1.1 schwarze 1775:
1776: /*
1.47 schwarze 1777: * Pre-check: if we have no stop-characters, then set the
1778: * pointer as ourselvse and get out of here.
1.1 schwarze 1779: */
1.47 schwarze 1780: if (strcspn(val, res) == bsz) {
1.53 schwarze 1781: key->rendered = key->key;
1.47 schwarze 1782: return;
1783: }
1.1 schwarze 1784:
1.47 schwarze 1785: /* Pre-allocate by the length of the input */
1.39 schwarze 1786:
1.47 schwarze 1787: buf = mandoc_malloc(++bsz);
1788: pos = 0;
1.39 schwarze 1789:
1.47 schwarze 1790: while ('\0' != *val) {
1791: /*
1792: * Halt on the first escape sequence.
1793: * This also halts on the end of string, in which case
1794: * we just copy, fallthrough, and exit the loop.
1795: */
1796: if ((sz = strcspn(val, res)) > 0) {
1797: memcpy(&buf[pos], val, sz);
1798: pos += sz;
1799: val += sz;
1800: }
1.39 schwarze 1801:
1.71 schwarze 1802: switch (*val) {
1803: case (ASCII_HYPH):
1.47 schwarze 1804: buf[pos++] = '-';
1805: val++;
1806: continue;
1.71 schwarze 1807: case ('\t'):
1808: /* FALLTHROUGH */
1809: case (ASCII_NBRSP):
1.47 schwarze 1810: buf[pos++] = ' ';
1811: val++;
1.71 schwarze 1812: /* FALLTHROUGH */
1813: case (ASCII_BREAK):
1.47 schwarze 1814: continue;
1.71 schwarze 1815: default:
1816: break;
1817: }
1818: if ('\\' != *val)
1.47 schwarze 1819: break;
1.39 schwarze 1820:
1.47 schwarze 1821: /* Read past the slash. */
1.39 schwarze 1822:
1.47 schwarze 1823: val++;
1.39 schwarze 1824:
1.47 schwarze 1825: /*
1826: * Parse the escape sequence and see if it's a
1827: * predefined character or special character.
1828: */
1.52 schwarze 1829:
1.47 schwarze 1830: esc = mandoc_escape
1831: ((const char **)&val, &seq, &len);
1832: if (ESCAPE_ERROR == esc)
1833: break;
1834: if (ESCAPE_SPECIAL != esc)
1835: continue;
1.39 schwarze 1836:
1.47 schwarze 1837: /*
1.52 schwarze 1838: * Render the special character
1839: * as either UTF-8 or ASCII.
1.47 schwarze 1840: */
1.52 schwarze 1841:
1842: if (write_utf8) {
1843: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1844: continue;
1845: cpp = utfbuf;
1846: if (0 == (sz = utf8(u, utfbuf)))
1847: continue;
1848: sz = strlen(cpp);
1849: } else {
1850: cpp = mchars_spec2str(mc, seq, len, &sz);
1851: if (NULL == cpp)
1852: continue;
1853: if (ASCII_NBRSP == *cpp) {
1854: cpp = " ";
1855: sz = 1;
1856: }
1857: }
1.1 schwarze 1858:
1.47 schwarze 1859: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1860:
1.47 schwarze 1861: bsz += sz;
1862: buf = mandoc_realloc(buf, bsz);
1863: memcpy(&buf[pos], cpp, sz);
1864: pos += sz;
1.1 schwarze 1865: }
1866:
1.47 schwarze 1867: buf[pos] = '\0';
1.53 schwarze 1868: key->rendered = buf;
1.1 schwarze 1869: }
1870:
1.75 schwarze 1871: static void
1872: dbadd_mlink(const struct mlink *mlink)
1873: {
1874: size_t i;
1875:
1876: i = 1;
1877: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1878: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1879: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1880: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->recno);
1881: SQL_STEP(stmts[STMT_INSERT_LINK]);
1882: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1883: }
1884:
1.11 schwarze 1885: /*
1.47 schwarze 1886: * Flush the current page's terms (and their bits) into the database.
1887: * Wrap the entire set of additions in a transaction to make sqlite be a
1888: * little faster.
1.53 schwarze 1889: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 1890: */
1891: static void
1.75 schwarze 1892: dbadd(struct mpage *mpage, struct mchars *mc)
1.11 schwarze 1893: {
1.47 schwarze 1894: struct mlink *mlink;
1895: struct str *key;
1896: size_t i;
1897: unsigned int slot;
1898:
1.85 schwarze 1899: mlink = mpage->mlinks;
1.11 schwarze 1900:
1.85 schwarze 1901: if (nodb) {
1902: while (NULL != mlink) {
1903: fputs(mlink->name, stdout);
1904: if (NULL == mlink->next ||
1905: strcmp(mlink->dsec, mlink->next->dsec) ||
1906: strcmp(mlink->fsec, mlink->next->fsec) ||
1907: strcmp(mlink->arch, mlink->next->arch)) {
1908: putchar('(');
1909: if ('\0' == *mlink->dsec)
1910: fputs(mlink->fsec, stdout);
1911: else
1912: fputs(mlink->dsec, stdout);
1913: if ('\0' != *mlink->arch)
1914: printf("/%s", mlink->arch);
1915: putchar(')');
1916: }
1917: mlink = mlink->next;
1918: if (NULL != mlink)
1919: fputs(", ", stdout);
1920: }
1.89 ! schwarze 1921: printf(" - %s\n", mpage->desc);
1.11 schwarze 1922: return;
1.85 schwarze 1923: }
1924:
1925: if (debug)
1926: say(mlink->file, "Adding to database");
1.47 schwarze 1927:
1928: i = 1;
1.89 ! schwarze 1929: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->desc);
1.47 schwarze 1930: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
1931: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.75 schwarze 1932: mpage->recno = sqlite3_last_insert_rowid(db);
1.47 schwarze 1933: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1934:
1.85 schwarze 1935: while (NULL != mlink) {
1.75 schwarze 1936: dbadd_mlink(mlink);
1.85 schwarze 1937: mlink = mlink->next;
1938: }
1.47 schwarze 1939:
1940: for (key = ohash_first(&strings, &slot); NULL != key;
1941: key = ohash_next(&strings, &slot)) {
1942: assert(key->mpage == mpage);
1.53 schwarze 1943: if (NULL == key->rendered)
1944: render_key(mc, key);
1.47 schwarze 1945: i = 1;
1946: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.53 schwarze 1947: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.75 schwarze 1948: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->recno);
1.47 schwarze 1949: SQL_STEP(stmts[STMT_INSERT_KEY]);
1950: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.53 schwarze 1951: if (key->rendered != key->key)
1952: free(key->rendered);
1.47 schwarze 1953: free(key);
1.33 schwarze 1954: }
1.47 schwarze 1955: }
1.41 deraadt 1956:
1.47 schwarze 1957: static void
1958: dbprune(void)
1959: {
1960: struct mpage *mpage;
1961: struct mlink *mlink;
1962: size_t i;
1963: unsigned int slot;
1.11 schwarze 1964:
1.63 schwarze 1965: if (0 == nodb)
1966: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 1967:
1.63 schwarze 1968: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
1969: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 1970: mlink = mpage->mlinks;
1.82 schwarze 1971: if (debug)
1.63 schwarze 1972: say(mlink->file, "Deleting from database");
1973: if (nodb)
1974: continue;
1975: for ( ; NULL != mlink; mlink = mlink->next) {
1976: i = 1;
1977: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
1978: i, mlink->dsec);
1979: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
1980: i, mlink->arch);
1981: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
1982: i, mlink->name);
1983: SQL_STEP(stmts[STMT_DELETE_PAGE]);
1984: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
1985: }
1.11 schwarze 1986: }
1.63 schwarze 1987:
1988: if (0 == nodb)
1989: SQL_EXEC("END TRANSACTION");
1.47 schwarze 1990: }
1.22 schwarze 1991:
1.47 schwarze 1992: /*
1993: * Close an existing database and its prepared statements.
1994: * If "real" is not set, rename the temporary file into the real one.
1995: */
1996: static void
1997: dbclose(int real)
1998: {
1999: size_t i;
1.72 schwarze 2000: int status;
2001: pid_t child;
1.11 schwarze 2002:
1.47 schwarze 2003: if (nodb)
2004: return;
1.11 schwarze 2005:
1.47 schwarze 2006: for (i = 0; i < STMT__MAX; i++) {
2007: sqlite3_finalize(stmts[i]);
2008: stmts[i] = NULL;
1.28 schwarze 2009: }
1.22 schwarze 2010:
1.47 schwarze 2011: sqlite3_close(db);
2012: db = NULL;
1.11 schwarze 2013:
1.47 schwarze 2014: if (real)
2015: return;
1.22 schwarze 2016:
1.72 schwarze 2017: if ('\0' == *tempfilename) {
2018: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2019: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2020: say(MANDOC_DB, "&rename");
1.72 schwarze 2021: }
2022: return;
2023: }
2024:
2025: switch (child = fork()) {
2026: case (-1):
2027: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2028: say("", "&fork cmp");
1.72 schwarze 2029: return;
2030: case (0):
2031: execlp("cmp", "cmp", "-s",
2032: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2033: say("", "&exec cmp");
1.72 schwarze 2034: exit(0);
2035: default:
2036: break;
2037: }
2038: if (-1 == waitpid(child, &status, 0)) {
2039: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2040: say("", "&wait cmp");
1.72 schwarze 2041: } else if (WIFSIGNALED(status)) {
2042: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2043: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2044: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2045: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2046: say(MANDOC_DB,
2047: "Data changed, but cannot replace database");
2048: }
2049:
2050: *strrchr(tempfilename, '/') = '\0';
2051: switch (child = fork()) {
2052: case (-1):
2053: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2054: say("", "&fork rm");
1.72 schwarze 2055: return;
2056: case (0):
2057: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2058: say("", "&exec rm");
1.72 schwarze 2059: exit((int)MANDOCLEVEL_SYSERR);
2060: default:
2061: break;
2062: }
2063: if (-1 == waitpid(child, &status, 0)) {
2064: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2065: say("", "&wait rm");
1.72 schwarze 2066: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2067: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2068: say("", "%s: Cannot remove temporary directory",
2069: tempfilename);
1.22 schwarze 2070: }
1.11 schwarze 2071: }
2072:
1.47 schwarze 2073: /*
2074: * This is straightforward stuff.
2075: * Open a database connection to a "temporary" database, then open a set
2076: * of prepared statements we'll use over and over again.
2077: * If "real" is set, we use the existing database; if not, we truncate a
2078: * temporary one.
2079: * Must be matched by dbclose().
2080: */
2081: static int
2082: dbopen(int real)
1.2 schwarze 2083: {
1.72 schwarze 2084: const char *sql;
1.47 schwarze 2085: int rc, ofl;
1.6 schwarze 2086:
1.47 schwarze 2087: if (nodb)
2088: return(1);
1.6 schwarze 2089:
1.72 schwarze 2090: *tempfilename = '\0';
1.47 schwarze 2091: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2092:
2093: if (real) {
2094: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2095: if (SQLITE_OK != rc) {
1.47 schwarze 2096: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2097: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.47 schwarze 2098: return(0);
1.28 schwarze 2099: }
1.72 schwarze 2100: goto prepare_statements;
2101: }
2102:
2103: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2104:
1.72 schwarze 2105: remove(MANDOC_DB "~");
2106: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.47 schwarze 2107: if (SQLITE_OK == rc)
1.72 schwarze 2108: goto create_tables;
1.73 schwarze 2109: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2110: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2111: say(MANDOC_DB "~", "%s", sqlite3_errmsg(db));
1.47 schwarze 2112: return(0);
2113: }
1.6 schwarze 2114:
1.72 schwarze 2115: if (strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2116: sizeof(tempfilename)) >= sizeof(tempfilename)) {
2117: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2118: say("", "/tmp/mandocdb.XXXXXX: Filename too long");
1.72 schwarze 2119: return(0);
2120: }
2121: if (NULL == mkdtemp(tempfilename)) {
2122: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2123: say("", "&%s", tempfilename);
1.72 schwarze 2124: return(0);
2125: }
2126: if (strlcat(tempfilename, "/" MANDOC_DB,
2127: sizeof(tempfilename)) >= sizeof(tempfilename)) {
2128: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2129: say("", "%s/" MANDOC_DB ": Filename too long",
2130: tempfilename);
1.72 schwarze 2131: return(0);
2132: }
2133: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2134: if (SQLITE_OK != rc) {
1.47 schwarze 2135: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2136: say("", "%s: %s", tempfilename, sqlite3_errmsg(db));
1.47 schwarze 2137: return(0);
1.2 schwarze 2138: }
2139:
1.72 schwarze 2140: create_tables:
1.47 schwarze 2141: sql = "CREATE TABLE \"mpages\" (\n"
1.89 ! schwarze 2142: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2143: " \"form\" INTEGER NOT NULL,\n"
2144: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
2145: ");\n"
2146: "\n"
2147: "CREATE TABLE \"mlinks\" (\n"
2148: " \"sec\" TEXT NOT NULL,\n"
2149: " \"arch\" TEXT NOT NULL,\n"
2150: " \"name\" TEXT NOT NULL,\n"
2151: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1.66 schwarze 2152: "ON DELETE CASCADE\n"
1.47 schwarze 2153: ");\n"
2154: "\n"
2155: "CREATE TABLE \"keys\" (\n"
2156: " \"bits\" INTEGER NOT NULL,\n"
2157: " \"key\" TEXT NOT NULL,\n"
2158: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1.66 schwarze 2159: "ON DELETE CASCADE\n"
1.65 schwarze 2160: ");\n";
1.47 schwarze 2161:
2162: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2163: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2164: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.47 schwarze 2165: return(0);
1.2 schwarze 2166: }
2167:
1.47 schwarze 2168: prepare_statements:
2169: SQL_EXEC("PRAGMA foreign_keys = ON");
1.63 schwarze 2170: sql = "DELETE FROM mpages WHERE id IN "
2171: "(SELECT pageid FROM mlinks WHERE "
2172: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2173: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2174: sql = "INSERT INTO mpages "
1.89 ! schwarze 2175: "(desc,form) VALUES (?,?)";
1.47 schwarze 2176: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2177: sql = "INSERT INTO mlinks "
1.61 schwarze 2178: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2179: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
2180: sql = "INSERT INTO keys "
2181: "(bits,key,pageid) VALUES (?,?,?)";
2182: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2183:
1.47 schwarze 2184: /*
2185: * When opening a new database, we can turn off
2186: * synchronous mode for much better performance.
2187: */
1.6 schwarze 2188:
1.47 schwarze 2189: if (real)
2190: SQL_EXEC("PRAGMA synchronous = OFF");
1.11 schwarze 2191:
1.47 schwarze 2192: return(1);
2193: }
1.6 schwarze 2194:
1.47 schwarze 2195: static void *
2196: hash_halloc(size_t sz, void *arg)
2197: {
1.6 schwarze 2198:
1.47 schwarze 2199: return(mandoc_calloc(sz, 1));
2200: }
1.2 schwarze 2201:
1.47 schwarze 2202: static void *
2203: hash_alloc(size_t sz, void *arg)
2204: {
1.28 schwarze 2205:
1.47 schwarze 2206: return(mandoc_malloc(sz));
2207: }
1.6 schwarze 2208:
1.47 schwarze 2209: static void
2210: hash_free(void *p, size_t sz, void *arg)
2211: {
1.26 schwarze 2212:
1.47 schwarze 2213: free(p);
2214: }
1.6 schwarze 2215:
1.47 schwarze 2216: static int
2217: set_basedir(const char *targetdir)
2218: {
2219: static char startdir[PATH_MAX];
2220: static int fd;
1.6 schwarze 2221:
1.47 schwarze 2222: /*
2223: * Remember where we started by keeping a fd open to the origin
2224: * path component: throughout this utility, we chdir() a lot to
2225: * handle relative paths, and by doing this, we can return to
2226: * the starting point.
2227: */
2228: if ('\0' == *startdir) {
2229: if (NULL == getcwd(startdir, PATH_MAX)) {
2230: exitcode = (int)MANDOCLEVEL_SYSERR;
2231: if (NULL != targetdir)
1.80 schwarze 2232: say("", "&getcwd");
1.47 schwarze 2233: return(0);
2234: }
2235: if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
2236: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2237: say("", "&open %s", startdir);
1.47 schwarze 2238: return(0);
1.11 schwarze 2239: }
1.47 schwarze 2240: if (NULL == targetdir)
2241: targetdir = startdir;
2242: } else {
2243: if (-1 == fd)
2244: return(0);
2245: if (-1 == fchdir(fd)) {
2246: close(fd);
2247: basedir[0] = '\0';
2248: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2249: say("", "&chdir %s", startdir);
1.47 schwarze 2250: return(0);
1.2 schwarze 2251: }
1.47 schwarze 2252: if (NULL == targetdir) {
2253: close(fd);
2254: return(1);
1.2 schwarze 2255: }
2256: }
1.47 schwarze 2257: if (NULL == realpath(targetdir, basedir)) {
2258: basedir[0] = '\0';
2259: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 2260: say("", "&%s: realpath", targetdir);
1.47 schwarze 2261: return(0);
2262: } else if (-1 == chdir(basedir)) {
2263: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 2264: say("", "&chdir");
1.47 schwarze 2265: return(0);
2266: }
2267: return(1);
1.2 schwarze 2268: }
2269:
2270: static void
1.47 schwarze 2271: say(const char *file, const char *format, ...)
1.2 schwarze 2272: {
1.47 schwarze 2273: va_list ap;
1.80 schwarze 2274: int use_errno;
1.2 schwarze 2275:
1.47 schwarze 2276: if ('\0' != *basedir)
2277: fprintf(stderr, "%s", basedir);
2278: if ('\0' != *basedir && '\0' != *file)
2279: fputs("//", stderr);
2280: if ('\0' != *file)
2281: fprintf(stderr, "%s", file);
1.31 schwarze 2282:
1.80 schwarze 2283: use_errno = 1;
2284: if (NULL != format) {
2285: switch (*format) {
2286: case ('&'):
2287: format++;
2288: break;
2289: case ('\0'):
2290: format = NULL;
2291: break;
2292: default:
2293: use_errno = 0;
2294: break;
2295: }
2296: }
2297: if (NULL != format) {
2298: if ('\0' != *basedir || '\0' != *file)
2299: fputs(": ", stderr);
2300: va_start(ap, format);
2301: vfprintf(stderr, format, ap);
2302: va_end(ap);
2303: }
2304: if (use_errno) {
2305: if ('\0' != *basedir || '\0' != *file || NULL != format)
2306: fputs(": ", stderr);
1.47 schwarze 2307: perror(NULL);
1.80 schwarze 2308: } else
2309: fputc('\n', stderr);
1.1 schwarze 2310: }
![[BACK]](/icons/back.gif){kind=icon}
Return to mandocdb.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc