Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.86
1.86 ! schwarze 1: /* $Id: mandocdb.c,v 1.85 2014/04/04 02:31:01 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 *);
1.50 schwarze 142: static int 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 *);
1.81 schwarze 147: static void parse_cat(struct mpage *, int);
1.47 schwarze 148: static void parse_man(struct mpage *, const struct man_node *);
149: static void parse_mdoc(struct mpage *, const struct mdoc_node *);
150: static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
151: static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
152: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
153: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
154: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
155: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
156: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
157: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
1.69 schwarze 158: static void putkey(const struct mpage *, char *, uint64_t);
1.47 schwarze 159: static void putkeys(const struct mpage *,
160: const char *, size_t, uint64_t);
161: static void putmdockey(const struct mpage *,
162: const struct mdoc_node *, uint64_t);
1.53 schwarze 163: static void render_key(struct mchars *, struct str *);
1.47 schwarze 164: static void say(const char *, const char *, ...);
165: static int set_basedir(const char *);
166: static int treescan(void);
167: static size_t utf8(unsigned int, char [7]);
168:
1.72 schwarze 169: static char tempfilename[32];
1.47 schwarze 170: static char *progname;
1.59 schwarze 171: static int nodb; /* no database changes */
1.73 schwarze 172: static int mparse_options; /* abort the parse early */
1.47 schwarze 173: static int use_all; /* use all found files */
1.82 schwarze 174: static int debug; /* print what we're doing */
1.47 schwarze 175: static int warnings; /* warn about crap */
1.52 schwarze 176: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 177: static int exitcode; /* to be returned by main */
178: static enum op op; /* operational mode */
179: static char basedir[PATH_MAX]; /* current base directory */
180: static struct ohash mpages; /* table of distinct manual pages */
181: static struct ohash mlinks; /* table of directory entries */
182: static struct ohash strings; /* table of all strings */
183: static sqlite3 *db = NULL; /* current database */
184: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
185:
1.19 schwarze 186: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 187: { NULL, 0 }, /* Ap */
188: { NULL, 0 }, /* Dd */
189: { NULL, 0 }, /* Dt */
190: { NULL, 0 }, /* Os */
191: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
192: { parse_mdoc_head, TYPE_Ss }, /* Ss */
193: { NULL, 0 }, /* Pp */
194: { NULL, 0 }, /* D1 */
195: { NULL, 0 }, /* Dl */
196: { NULL, 0 }, /* Bd */
197: { NULL, 0 }, /* Ed */
198: { NULL, 0 }, /* Bl */
199: { NULL, 0 }, /* El */
200: { NULL, 0 }, /* It */
201: { NULL, 0 }, /* Ad */
202: { NULL, TYPE_An }, /* An */
203: { NULL, TYPE_Ar }, /* Ar */
204: { NULL, TYPE_Cd }, /* Cd */
205: { NULL, TYPE_Cm }, /* Cm */
206: { NULL, TYPE_Dv }, /* Dv */
207: { NULL, TYPE_Er }, /* Er */
208: { NULL, TYPE_Ev }, /* Ev */
209: { NULL, 0 }, /* Ex */
210: { NULL, TYPE_Fa }, /* Fa */
211: { parse_mdoc_Fd, 0 }, /* Fd */
212: { NULL, TYPE_Fl }, /* Fl */
213: { parse_mdoc_Fn, 0 }, /* Fn */
214: { NULL, TYPE_Ft }, /* Ft */
215: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 216: { NULL, TYPE_In }, /* In */
1.47 schwarze 217: { NULL, TYPE_Li }, /* Li */
218: { parse_mdoc_Nd, TYPE_Nd }, /* Nd */
1.86 ! schwarze 219: { parse_mdoc_Nm, 0 }, /* Nm */
1.47 schwarze 220: { NULL, 0 }, /* Op */
221: { NULL, 0 }, /* Ot */
222: { NULL, TYPE_Pa }, /* Pa */
223: { NULL, 0 }, /* Rv */
1.49 schwarze 224: { NULL, TYPE_St }, /* St */
1.47 schwarze 225: { NULL, TYPE_Va }, /* Va */
226: { parse_mdoc_body, TYPE_Va }, /* Vt */
227: { parse_mdoc_Xr, 0 }, /* Xr */
228: { NULL, 0 }, /* %A */
229: { NULL, 0 }, /* %B */
230: { NULL, 0 }, /* %D */
231: { NULL, 0 }, /* %I */
232: { NULL, 0 }, /* %J */
233: { NULL, 0 }, /* %N */
234: { NULL, 0 }, /* %O */
235: { NULL, 0 }, /* %P */
236: { NULL, 0 }, /* %R */
237: { NULL, 0 }, /* %T */
238: { NULL, 0 }, /* %V */
239: { NULL, 0 }, /* Ac */
240: { NULL, 0 }, /* Ao */
241: { NULL, 0 }, /* Aq */
242: { NULL, TYPE_At }, /* At */
243: { NULL, 0 }, /* Bc */
244: { NULL, 0 }, /* Bf */
245: { NULL, 0 }, /* Bo */
246: { NULL, 0 }, /* Bq */
247: { NULL, TYPE_Bsx }, /* Bsx */
248: { NULL, TYPE_Bx }, /* Bx */
249: { NULL, 0 }, /* Db */
250: { NULL, 0 }, /* Dc */
251: { NULL, 0 }, /* Do */
252: { NULL, 0 }, /* Dq */
253: { NULL, 0 }, /* Ec */
254: { NULL, 0 }, /* Ef */
255: { NULL, TYPE_Em }, /* Em */
256: { NULL, 0 }, /* Eo */
257: { NULL, TYPE_Fx }, /* Fx */
258: { NULL, TYPE_Ms }, /* Ms */
259: { NULL, 0 }, /* No */
260: { NULL, 0 }, /* Ns */
261: { NULL, TYPE_Nx }, /* Nx */
262: { NULL, TYPE_Ox }, /* Ox */
263: { NULL, 0 }, /* Pc */
264: { NULL, 0 }, /* Pf */
265: { NULL, 0 }, /* Po */
266: { NULL, 0 }, /* Pq */
267: { NULL, 0 }, /* Qc */
268: { NULL, 0 }, /* Ql */
269: { NULL, 0 }, /* Qo */
270: { NULL, 0 }, /* Qq */
271: { NULL, 0 }, /* Re */
272: { NULL, 0 }, /* Rs */
273: { NULL, 0 }, /* Sc */
274: { NULL, 0 }, /* So */
275: { NULL, 0 }, /* Sq */
276: { NULL, 0 }, /* Sm */
277: { NULL, 0 }, /* Sx */
278: { NULL, TYPE_Sy }, /* Sy */
279: { NULL, TYPE_Tn }, /* Tn */
280: { NULL, 0 }, /* Ux */
281: { NULL, 0 }, /* Xc */
282: { NULL, 0 }, /* Xo */
283: { parse_mdoc_head, 0 }, /* Fo */
284: { NULL, 0 }, /* Fc */
285: { NULL, 0 }, /* Oo */
286: { NULL, 0 }, /* Oc */
287: { NULL, 0 }, /* Bk */
288: { NULL, 0 }, /* Ek */
289: { NULL, 0 }, /* Bt */
290: { NULL, 0 }, /* Hf */
291: { NULL, 0 }, /* Fr */
292: { NULL, 0 }, /* Ud */
293: { NULL, TYPE_Lb }, /* Lb */
294: { NULL, 0 }, /* Lp */
295: { NULL, TYPE_Lk }, /* Lk */
296: { NULL, TYPE_Mt }, /* Mt */
297: { NULL, 0 }, /* Brq */
298: { NULL, 0 }, /* Bro */
299: { NULL, 0 }, /* Brc */
300: { NULL, 0 }, /* %C */
301: { NULL, 0 }, /* Es */
302: { NULL, 0 }, /* En */
303: { NULL, TYPE_Dx }, /* Dx */
304: { NULL, 0 }, /* %Q */
305: { NULL, 0 }, /* br */
306: { NULL, 0 }, /* sp */
307: { NULL, 0 }, /* %U */
308: { NULL, 0 }, /* Ta */
1.1 schwarze 309: };
310:
311: int
1.3 schwarze 312: mandocdb(int argc, char *argv[])
1.1 schwarze 313: {
1.47 schwarze 314: int ch, i;
315: size_t j, sz;
316: const char *path_arg;
317: struct mchars *mc;
318: struct manpaths dirs;
319: struct mparse *mp;
320: struct ohash_info mpages_info, mlinks_info;
321:
322: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
323: memset(&dirs, 0, sizeof(struct manpaths));
324:
325: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
326: mpages_info.halloc = mlinks_info.halloc = hash_halloc;
327: mpages_info.hfree = mlinks_info.hfree = hash_free;
328:
329: mpages_info.key_offset = offsetof(struct mpage, inodev);
330: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 331:
332: progname = strrchr(argv[0], '/');
333: if (progname == NULL)
334: progname = argv[0];
335: else
336: ++progname;
337:
1.47 schwarze 338: /*
339: * We accept a few different invocations.
340: * The CHECKOP macro makes sure that invocation styles don't
341: * clobber each other.
342: */
343: #define CHECKOP(_op, _ch) do \
344: if (OP_DEFAULT != (_op)) { \
345: fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
346: goto usage; \
347: } while (/*CONSTCOND*/0)
1.10 schwarze 348:
1.47 schwarze 349: path_arg = NULL;
1.28 schwarze 350: op = OP_DEFAULT;
1.1 schwarze 351:
1.83 schwarze 352: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 353: switch (ch) {
1.6 schwarze 354: case ('a'):
355: use_all = 1;
356: break;
1.25 schwarze 357: case ('C'):
1.47 schwarze 358: CHECKOP(op, ch);
359: path_arg = optarg;
1.28 schwarze 360: op = OP_CONFFILE;
1.25 schwarze 361: break;
1.82 schwarze 362: case ('D'):
363: debug++;
364: break;
1.1 schwarze 365: case ('d'):
1.47 schwarze 366: CHECKOP(op, ch);
367: path_arg = optarg;
1.2 schwarze 368: op = OP_UPDATE;
1.1 schwarze 369: break;
1.47 schwarze 370: case ('n'):
371: nodb = 1;
372: break;
1.83 schwarze 373: case ('p'):
374: warnings = 1;
375: break;
1.59 schwarze 376: case ('Q'):
1.73 schwarze 377: mparse_options |= MPARSE_QUICK;
1.59 schwarze 378: break;
1.52 schwarze 379: case ('T'):
380: if (strcmp(optarg, "utf8")) {
381: fprintf(stderr, "-T%s: Unsupported "
382: "output format\n", optarg);
383: goto usage;
384: }
385: write_utf8 = 1;
386: break;
1.28 schwarze 387: case ('t'):
1.47 schwarze 388: CHECKOP(op, ch);
1.28 schwarze 389: dup2(STDOUT_FILENO, STDERR_FILENO);
390: op = OP_TEST;
1.47 schwarze 391: nodb = warnings = 1;
1.28 schwarze 392: break;
1.2 schwarze 393: case ('u'):
1.47 schwarze 394: CHECKOP(op, ch);
395: path_arg = optarg;
1.1 schwarze 396: op = OP_DELETE;
397: break;
1.83 schwarze 398: case ('v'):
399: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 400: break;
1.1 schwarze 401: default:
1.28 schwarze 402: goto usage;
1.1 schwarze 403: }
404:
405: argc -= optind;
406: argv += optind;
407:
1.28 schwarze 408: if (OP_CONFFILE == op && argc > 0) {
1.47 schwarze 409: fprintf(stderr, "-C: Too many arguments\n");
1.28 schwarze 410: goto usage;
411: }
412:
1.47 schwarze 413: exitcode = (int)MANDOCLEVEL_OK;
1.73 schwarze 414: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL, NULL);
1.47 schwarze 415: mc = mchars_alloc();
1.2 schwarze 416:
1.47 schwarze 417: ohash_init(&mpages, 6, &mpages_info);
418: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 419:
1.47 schwarze 420: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
421: /*
422: * Force processing all files.
423: */
424: use_all = 1;
1.2 schwarze 425:
1.47 schwarze 426: /*
427: * All of these deal with a specific directory.
428: * Jump into that directory then collect files specified
429: * on the command-line.
430: */
431: if (0 == set_basedir(path_arg))
432: goto out;
433: for (i = 0; i < argc; i++)
434: filescan(argv[i]);
435: if (0 == dbopen(1))
436: goto out;
437: if (OP_TEST != op)
438: dbprune();
439: if (OP_DELETE != op)
1.58 schwarze 440: mpages_merge(mc, mp);
1.47 schwarze 441: dbclose(1);
442: } else {
443: /*
444: * If we have arguments, use them as our manpaths.
445: * If we don't, grok from manpath(1) or however else
446: * manpath_parse() wants to do it.
447: */
448: if (argc > 0) {
449: dirs.paths = mandoc_calloc
450: (argc, sizeof(char *));
451: dirs.sz = (size_t)argc;
452: for (i = 0; i < argc; i++)
453: dirs.paths[i] = mandoc_strdup(argv[i]);
454: } else
455: manpath_parse(&dirs, path_arg, NULL, NULL);
1.84 schwarze 456:
457: if (0 == dirs.sz) {
458: exitcode = (int)MANDOCLEVEL_BADARG;
459: say("", "Empty manpath");
460: }
1.2 schwarze 461:
1.47 schwarze 462: /*
463: * First scan the tree rooted at a base directory, then
464: * build a new database and finally move it into place.
465: * Ignore zero-length directories and strip trailing
466: * slashes.
467: */
468: for (j = 0; j < dirs.sz; j++) {
469: sz = strlen(dirs.paths[j]);
470: if (sz && '/' == dirs.paths[j][sz - 1])
471: dirs.paths[j][--sz] = '\0';
472: if (0 == sz)
473: continue;
1.2 schwarze 474:
1.47 schwarze 475: if (j) {
476: ohash_init(&mpages, 6, &mpages_info);
477: ohash_init(&mlinks, 6, &mlinks_info);
478: }
1.1 schwarze 479:
1.47 schwarze 480: if (0 == set_basedir(dirs.paths[j]))
481: goto out;
482: if (0 == treescan())
483: goto out;
484: if (0 == set_basedir(dirs.paths[j]))
485: goto out;
486: if (0 == dbopen(0))
487: goto out;
1.41 deraadt 488:
1.58 schwarze 489: mpages_merge(mc, mp);
1.47 schwarze 490: dbclose(0);
1.28 schwarze 491:
1.47 schwarze 492: if (j + 1 < dirs.sz) {
493: mpages_free();
494: ohash_delete(&mpages);
495: ohash_delete(&mlinks);
496: }
1.2 schwarze 497: }
1.47 schwarze 498: }
499: out:
500: set_basedir(NULL);
501: manpath_free(&dirs);
502: mchars_free(mc);
503: mparse_free(mp);
504: mpages_free();
505: ohash_delete(&mpages);
506: ohash_delete(&mlinks);
507: return(exitcode);
508: usage:
1.83 schwarze 509: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
510: " %s [-aDnpQ] [-Tutf8] dir ...\n"
511: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
512: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 513: " %s [-Q] -t file ...\n",
1.47 schwarze 514: progname, progname, progname,
515: progname, progname);
1.1 schwarze 516:
1.47 schwarze 517: return((int)MANDOCLEVEL_BADARG);
518: }
1.1 schwarze 519:
1.47 schwarze 520: /*
521: * Scan a directory tree rooted at "basedir" for manpages.
522: * We use fts(), scanning directory parts along the way for clues to our
523: * section and architecture.
524: *
525: * If use_all has been specified, grok all files.
526: * If not, sanitise paths to the following:
527: *
528: * [./]man*[/<arch>]/<name>.<section>
529: * or
530: * [./]cat<section>[/<arch>]/<name>.0
531: *
532: * TODO: accomodate for multi-language directories.
533: */
534: static int
535: treescan(void)
536: {
537: FTS *f;
538: FTSENT *ff;
539: struct mlink *mlink;
1.81 schwarze 540: int dform, gzip;
1.51 schwarze 541: char *dsec, *arch, *fsec, *cp;
542: const char *path;
1.47 schwarze 543: const char *argv[2];
1.1 schwarze 544:
1.47 schwarze 545: argv[0] = ".";
546: argv[1] = (char *)NULL;
1.1 schwarze 547:
1.47 schwarze 548: /*
549: * Walk through all components under the directory, using the
550: * logical descent of files.
551: */
552: f = fts_open((char * const *)argv, FTS_LOGICAL, NULL);
553: if (NULL == f) {
554: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 555: say("", "&fts_open");
1.47 schwarze 556: return(0);
557: }
1.2 schwarze 558:
1.47 schwarze 559: dsec = arch = NULL;
560: dform = FORM_NONE;
1.2 schwarze 561:
1.47 schwarze 562: while (NULL != (ff = fts_read(f))) {
563: path = ff->fts_path + 2;
1.15 schwarze 564: /*
1.47 schwarze 565: * If we're a regular file, add an mlink by using the
566: * stored directory data and handling the filename.
1.15 schwarze 567: */
1.47 schwarze 568: if (FTS_F == ff->fts_info) {
569: if (0 == strcmp(path, MANDOC_DB))
570: continue;
571: if ( ! use_all && ff->fts_level < 2) {
572: if (warnings)
573: say(path, "Extraneous file");
574: continue;
1.81 schwarze 575: }
576: gzip = 0;
577: fsec = NULL;
578: while (NULL == fsec) {
579: fsec = strrchr(ff->fts_name, '.');
580: if (NULL == fsec || strcmp(fsec+1, "gz"))
581: break;
582: gzip = 1;
583: *fsec = '\0';
584: fsec = NULL;
585: }
586: if (NULL == fsec) {
1.47 schwarze 587: if ( ! use_all) {
588: if (warnings)
589: say(path,
590: "No filename suffix");
591: continue;
592: }
593: } else if (0 == strcmp(++fsec, "html")) {
594: if (warnings)
595: say(path, "Skip html");
596: continue;
597: } else if (0 == strcmp(fsec, "ps")) {
598: if (warnings)
599: say(path, "Skip ps");
600: continue;
601: } else if (0 == strcmp(fsec, "pdf")) {
602: if (warnings)
603: say(path, "Skip pdf");
604: continue;
605: } else if ( ! use_all &&
606: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
607: (FORM_CAT == dform && strcmp(fsec, "0")))) {
608: if (warnings)
609: say(path, "Wrong filename suffix");
610: continue;
611: } else
612: fsec[-1] = '\0';
1.51 schwarze 613:
1.47 schwarze 614: mlink = mandoc_calloc(1, sizeof(struct mlink));
615: strlcpy(mlink->file, path, sizeof(mlink->file));
616: mlink->dform = dform;
1.51 schwarze 617: mlink->dsec = dsec;
618: mlink->arch = arch;
619: mlink->name = ff->fts_name;
620: mlink->fsec = fsec;
1.81 schwarze 621: mlink->gzip = gzip;
1.47 schwarze 622: mlink_add(mlink, ff->fts_statp);
623: continue;
624: } else if (FTS_D != ff->fts_info &&
625: FTS_DP != ff->fts_info) {
626: if (warnings)
627: say(path, "Not a regular file");
628: continue;
629: }
630:
631: switch (ff->fts_level) {
632: case (0):
633: /* Ignore the root directory. */
634: break;
635: case (1):
636: /*
637: * This might contain manX/ or catX/.
638: * Try to infer this from the name.
639: * If we're not in use_all, enforce it.
640: */
641: cp = ff->fts_name;
642: if (FTS_DP == ff->fts_info)
643: break;
1.15 schwarze 644:
1.47 schwarze 645: if (0 == strncmp(cp, "man", 3)) {
646: dform = FORM_SRC;
647: dsec = cp + 3;
648: } else if (0 == strncmp(cp, "cat", 3)) {
649: dform = FORM_CAT;
650: dsec = cp + 3;
1.51 schwarze 651: } else {
652: dform = FORM_NONE;
653: dsec = NULL;
1.26 schwarze 654: }
1.47 schwarze 655:
656: if (NULL != dsec || use_all)
657: break;
658:
659: if (warnings)
660: say(path, "Unknown directory part");
661: fts_set(f, ff, FTS_SKIP);
662: break;
663: case (2):
664: /*
665: * Possibly our architecture.
666: * If we're descending, keep tabs on it.
667: */
668: if (FTS_DP != ff->fts_info && NULL != dsec)
669: arch = ff->fts_name;
1.51 schwarze 670: else
671: arch = NULL;
1.47 schwarze 672: break;
673: default:
674: if (FTS_DP == ff->fts_info || use_all)
675: break;
676: if (warnings)
677: say(path, "Extraneous directory part");
678: fts_set(f, ff, FTS_SKIP);
679: break;
1.14 schwarze 680: }
1.47 schwarze 681: }
682:
683: fts_close(f);
684: return(1);
685: }
1.1 schwarze 686:
1.47 schwarze 687: /*
688: * Add a file to the mlinks table.
689: * Do not verify that it's a "valid" looking manpage (we'll do that
690: * later).
691: *
692: * Try to infer the manual section, architecture, and page name from the
693: * path, assuming it looks like
694: *
695: * [./]man*[/<arch>]/<name>.<section>
696: * or
697: * [./]cat<section>[/<arch>]/<name>.0
698: *
699: * See treescan() for the fts(3) version of this.
700: */
701: static void
702: filescan(const char *file)
703: {
704: char buf[PATH_MAX];
705: struct stat st;
706: struct mlink *mlink;
707: char *p, *start;
708:
709: assert(use_all);
710:
711: if (0 == strncmp(file, "./", 2))
712: file += 2;
713:
714: if (NULL == realpath(file, buf)) {
715: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 716: say(file, "&realpath");
1.47 schwarze 717: return;
1.63 schwarze 718: }
719:
720: if (strstr(buf, basedir) == buf)
721: start = buf + strlen(basedir) + 1;
722: else if (OP_TEST == op)
723: start = buf;
724: else {
1.47 schwarze 725: exitcode = (int)MANDOCLEVEL_BADARG;
726: say("", "%s: outside base directory", buf);
727: return;
1.63 schwarze 728: }
729:
730: if (-1 == stat(buf, &st)) {
1.47 schwarze 731: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 732: say(file, "&stat");
1.47 schwarze 733: return;
734: } else if ( ! (S_IFREG & st.st_mode)) {
735: exitcode = (int)MANDOCLEVEL_BADARG;
736: say(file, "Not a regular file");
737: return;
1.1 schwarze 738: }
1.63 schwarze 739:
1.47 schwarze 740: mlink = mandoc_calloc(1, sizeof(struct mlink));
741: strlcpy(mlink->file, start, sizeof(mlink->file));
1.1 schwarze 742:
1.10 schwarze 743: /*
1.47 schwarze 744: * First try to guess our directory structure.
745: * If we find a separator, try to look for man* or cat*.
746: * If we find one of these and what's underneath is a directory,
747: * assume it's an architecture.
1.10 schwarze 748: */
1.47 schwarze 749: if (NULL != (p = strchr(start, '/'))) {
750: *p++ = '\0';
751: if (0 == strncmp(start, "man", 3)) {
752: mlink->dform = FORM_SRC;
1.51 schwarze 753: mlink->dsec = start + 3;
1.47 schwarze 754: } else if (0 == strncmp(start, "cat", 3)) {
755: mlink->dform = FORM_CAT;
1.51 schwarze 756: mlink->dsec = start + 3;
1.47 schwarze 757: }
1.10 schwarze 758:
1.47 schwarze 759: start = p;
760: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
761: *p++ = '\0';
1.51 schwarze 762: mlink->arch = start;
1.47 schwarze 763: start = p;
1.41 deraadt 764: }
1.47 schwarze 765: }
1.7 schwarze 766:
1.47 schwarze 767: /*
768: * Now check the file suffix.
769: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
770: */
771: p = strrchr(start, '\0');
772: while (p-- > start && '/' != *p && '.' != *p)
773: /* Loop. */ ;
774:
775: if ('.' == *p) {
776: *p++ = '\0';
1.51 schwarze 777: mlink->fsec = p;
1.47 schwarze 778: }
1.41 deraadt 779:
1.47 schwarze 780: /*
781: * Now try to parse the name.
782: * Use the filename portion of the path.
783: */
784: mlink->name = start;
785: if (NULL != (p = strrchr(start, '/'))) {
786: mlink->name = p + 1;
787: *p = '\0';
788: }
789: mlink_add(mlink, &st);
790: }
1.2 schwarze 791:
1.47 schwarze 792: static void
793: mlink_add(struct mlink *mlink, const struct stat *st)
794: {
795: struct inodev inodev;
796: struct mpage *mpage;
797: unsigned int slot;
798:
799: assert(NULL != mlink->file);
800:
1.51 schwarze 801: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
802: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
803: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
804: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 805:
806: if ('0' == *mlink->fsec) {
807: free(mlink->fsec);
808: mlink->fsec = mandoc_strdup(mlink->dsec);
809: mlink->fform = FORM_CAT;
810: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
811: mlink->fform = FORM_SRC;
812: else
813: mlink->fform = FORM_NONE;
1.1 schwarze 814:
1.47 schwarze 815: slot = ohash_qlookup(&mlinks, mlink->file);
816: assert(NULL == ohash_find(&mlinks, slot));
817: ohash_insert(&mlinks, slot, mlink);
818:
819: inodev.st_ino = st->st_ino;
820: inodev.st_dev = st->st_dev;
821: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
822: sizeof(struct inodev), inodev.st_ino);
823: mpage = ohash_find(&mpages, slot);
824: if (NULL == mpage) {
825: mpage = mandoc_calloc(1, sizeof(struct mpage));
826: mpage->inodev.st_ino = inodev.st_ino;
827: mpage->inodev.st_dev = inodev.st_dev;
828: ohash_insert(&mpages, slot, mpage);
829: } else
830: mlink->next = mpage->mlinks;
831: mpage->mlinks = mlink;
1.75 schwarze 832: mlink->mpage = mpage;
1.47 schwarze 833: }
1.1 schwarze 834:
1.47 schwarze 835: static void
836: mlink_free(struct mlink *mlink)
837: {
1.1 schwarze 838:
1.47 schwarze 839: free(mlink->dsec);
840: free(mlink->arch);
841: free(mlink->name);
842: free(mlink->fsec);
843: free(mlink);
844: }
1.1 schwarze 845:
1.47 schwarze 846: static void
847: mpages_free(void)
848: {
849: struct mpage *mpage;
850: struct mlink *mlink;
851: unsigned int slot;
852:
853: mpage = ohash_first(&mpages, &slot);
854: while (NULL != mpage) {
855: while (NULL != (mlink = mpage->mlinks)) {
856: mpage->mlinks = mlink->next;
857: mlink_free(mlink);
858: }
859: free(mpage->sec);
860: free(mpage->arch);
861: free(mpage->title);
862: free(mpage->desc);
863: free(mpage);
864: mpage = ohash_next(&mpages, &slot);
865: }
866: }
1.1 schwarze 867:
1.47 schwarze 868: /*
869: * For each mlink to the mpage, check whether the path looks like
870: * it is formatted, and if it does, check whether a source manual
871: * exists by the same name, ignoring the suffix.
872: * If both conditions hold, drop the mlink.
873: */
874: static void
875: mlinks_undupe(struct mpage *mpage)
876: {
877: char buf[PATH_MAX];
878: struct mlink **prev;
879: struct mlink *mlink;
880: char *bufp;
881:
882: mpage->form = FORM_CAT;
883: prev = &mpage->mlinks;
884: while (NULL != (mlink = *prev)) {
885: if (FORM_CAT != mlink->dform) {
886: mpage->form = FORM_NONE;
887: goto nextlink;
888: }
889: if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
890: if (warnings)
891: say(mlink->file, "Filename too long");
892: goto nextlink;
1.26 schwarze 893: }
1.47 schwarze 894: bufp = strstr(buf, "cat");
895: assert(NULL != bufp);
896: memcpy(bufp, "man", 3);
897: if (NULL != (bufp = strrchr(buf, '.')))
898: *++bufp = '\0';
899: strlcat(buf, mlink->dsec, PATH_MAX);
900: if (NULL == ohash_find(&mlinks,
901: ohash_qlookup(&mlinks, buf)))
902: goto nextlink;
903: if (warnings)
904: say(mlink->file, "Man source exists: %s", buf);
905: if (use_all)
906: goto nextlink;
907: *prev = mlink->next;
908: mlink_free(mlink);
909: continue;
910: nextlink:
911: prev = &(*prev)->next;
1.1 schwarze 912: }
1.47 schwarze 913: }
1.1 schwarze 914:
1.50 schwarze 915: static int
916: mlink_check(struct mpage *mpage, struct mlink *mlink)
917: {
918: int match;
919:
920: match = 1;
921:
922: /*
923: * Check whether the manual section given in a file
924: * agrees with the directory where the file is located.
925: * Some manuals have suffixes like (3p) on their
926: * section number either inside the file or in the
927: * directory name, some are linked into more than one
928: * section, like encrypt(1) = makekey(8).
929: */
930:
931: if (FORM_SRC == mpage->form &&
932: strcasecmp(mpage->sec, mlink->dsec)) {
933: match = 0;
934: say(mlink->file, "Section \"%s\" manual in %s directory",
935: mpage->sec, mlink->dsec);
936: }
937:
938: /*
939: * Manual page directories exist for each kernel
940: * architecture as returned by machine(1).
941: * However, many manuals only depend on the
942: * application architecture as returned by arch(1).
943: * For example, some (2/ARM) manuals are shared
944: * across the "armish" and "zaurus" kernel
945: * architectures.
946: * A few manuals are even shared across completely
947: * different architectures, for example fdformat(1)
948: * on amd64, i386, sparc, and sparc64.
949: */
950:
951: if (strcasecmp(mpage->arch, mlink->arch)) {
952: match = 0;
953: say(mlink->file, "Architecture \"%s\" manual in "
954: "\"%s\" directory", mpage->arch, mlink->arch);
955: }
956:
957: if (strcasecmp(mpage->title, mlink->name))
958: match = 0;
959:
960: return(match);
961: }
962:
1.47 schwarze 963: /*
964: * Run through the files in the global vector "mpages"
965: * and add them to the database specified in "basedir".
966: *
967: * This handles the parsing scheme itself, using the cues of directory
968: * and filename to determine whether the file is parsable or not.
969: */
970: static void
1.58 schwarze 971: mpages_merge(struct mchars *mc, struct mparse *mp)
1.47 schwarze 972: {
1.70 schwarze 973: char any[] = "any";
1.58 schwarze 974: struct ohash_info str_info;
1.81 schwarze 975: int fd[2];
1.75 schwarze 976: struct mpage *mpage, *mpage_dest;
977: struct mlink *mlink, *mlink_dest;
1.47 schwarze 978: struct mdoc *mdoc;
979: struct man *man;
1.75 schwarze 980: char *sodest;
1.69 schwarze 981: char *cp;
1.81 schwarze 982: pid_t child_pid;
983: int match, status;
1.58 schwarze 984: unsigned int pslot;
1.47 schwarze 985: enum mandoclevel lvl;
986:
987: str_info.alloc = hash_alloc;
988: str_info.halloc = hash_halloc;
989: str_info.hfree = hash_free;
990: str_info.key_offset = offsetof(struct str, key);
991:
1.64 schwarze 992: if (0 == nodb)
993: SQL_EXEC("BEGIN TRANSACTION");
994:
1.47 schwarze 995: mpage = ohash_first(&mpages, &pslot);
996: while (NULL != mpage) {
997: mlinks_undupe(mpage);
998: if (NULL == mpage->mlinks) {
999: mpage = ohash_next(&mpages, &pslot);
1000: continue;
1001: }
1.1 schwarze 1002:
1.47 schwarze 1003: ohash_init(&strings, 6, &str_info);
1004: mparse_reset(mp);
1005: mdoc = NULL;
1006: man = NULL;
1.81 schwarze 1007: sodest = NULL;
1008: child_pid = 0;
1009: fd[0] = -1;
1010: fd[1] = -1;
1011:
1012: if (mpage->mlinks->gzip) {
1013: if (-1 == pipe(fd)) {
1014: exitcode = (int)MANDOCLEVEL_SYSERR;
1015: say(mpage->mlinks->file, "&pipe gunzip");
1016: goto nextpage;
1017: }
1018: switch (child_pid = fork()) {
1019: case (-1):
1020: exitcode = (int)MANDOCLEVEL_SYSERR;
1021: say(mpage->mlinks->file, "&fork gunzip");
1022: child_pid = 0;
1023: close(fd[1]);
1024: close(fd[0]);
1025: goto nextpage;
1026: case (0):
1027: close(fd[0]);
1028: if (-1 == dup2(fd[1], STDOUT_FILENO)) {
1029: say(mpage->mlinks->file,
1030: "&dup gunzip");
1031: exit(1);
1032: }
1033: execlp("gunzip", "gunzip", "-c",
1034: mpage->mlinks->file, NULL);
1035: say(mpage->mlinks->file, "&exec gunzip");
1036: exit(1);
1037: default:
1038: close(fd[1]);
1039: break;
1040: }
1041: }
1.11 schwarze 1042:
1043: /*
1.24 schwarze 1044: * Try interpreting the file as mdoc(7) or man(7)
1045: * source code, unless it is already known to be
1046: * formatted. Fall back to formatted mode.
1.11 schwarze 1047: */
1.47 schwarze 1048: if (FORM_CAT != mpage->mlinks->dform ||
1049: FORM_CAT != mpage->mlinks->fform) {
1.81 schwarze 1050: lvl = mparse_readfd(mp, fd[0], mpage->mlinks->file);
1.47 schwarze 1051: if (lvl < MANDOCLEVEL_FATAL)
1.75 schwarze 1052: mparse_result(mp, &mdoc, &man, &sodest);
1.47 schwarze 1053: }
1.11 schwarze 1054:
1.75 schwarze 1055: if (NULL != sodest) {
1056: mlink_dest = ohash_find(&mlinks,
1057: ohash_qlookup(&mlinks, sodest));
1058: if (NULL != mlink_dest) {
1059:
1060: /* The .so target exists. */
1061:
1062: mpage_dest = mlink_dest->mpage;
1063: mlink = mpage->mlinks;
1064: while (1) {
1065: mlink->mpage = mpage_dest;
1066:
1067: /*
1068: * If the target was already
1069: * processed, add the links
1070: * to the database now.
1071: * Otherwise, this will
1072: * happen when we come
1073: * to the target.
1074: */
1075:
1076: if (mpage_dest->recno)
1077: dbadd_mlink(mlink);
1078:
1079: if (NULL == mlink->next)
1080: break;
1081: mlink = mlink->next;
1082: }
1083:
1084: /* Move all links to the target. */
1085:
1086: mlink->next = mlink_dest->next;
1087: mlink_dest->next = mpage->mlinks;
1088: mpage->mlinks = NULL;
1089: }
1.81 schwarze 1090: goto nextpage;
1.75 schwarze 1091: } else if (NULL != mdoc) {
1.47 schwarze 1092: mpage->form = FORM_SRC;
1093: mpage->sec =
1094: mandoc_strdup(mdoc_meta(mdoc)->msec);
1095: mpage->arch = mdoc_meta(mdoc)->arch;
1096: mpage->arch = mandoc_strdup(
1097: NULL == mpage->arch ? "" : mpage->arch);
1098: mpage->title =
1099: mandoc_strdup(mdoc_meta(mdoc)->title);
1.11 schwarze 1100: } else if (NULL != man) {
1.47 schwarze 1101: mpage->form = FORM_SRC;
1102: mpage->sec =
1103: mandoc_strdup(man_meta(man)->msec);
1104: mpage->arch =
1105: mandoc_strdup(mpage->mlinks->arch);
1106: mpage->title =
1107: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1108: } else {
1.47 schwarze 1109: mpage->form = FORM_CAT;
1110: mpage->sec =
1111: mandoc_strdup(mpage->mlinks->dsec);
1112: mpage->arch =
1113: mandoc_strdup(mpage->mlinks->arch);
1114: mpage->title =
1115: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1116: }
1.54 schwarze 1117: putkey(mpage, mpage->sec, TYPE_sec);
1.55 schwarze 1118: putkey(mpage, '\0' == *mpage->arch ?
1.70 schwarze 1119: any : mpage->arch, TYPE_arch);
1.1 schwarze 1120:
1.54 schwarze 1121: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1122: if ('\0' != *mlink->dsec)
1123: putkey(mpage, mlink->dsec, TYPE_sec);
1124: if ('\0' != *mlink->fsec)
1125: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1126: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1127: any : mlink->arch, TYPE_arch);
1.50 schwarze 1128: putkey(mpage, mlink->name, TYPE_Nm);
1.54 schwarze 1129: }
1.41 deraadt 1130:
1.50 schwarze 1131: if (warnings && !use_all) {
1.47 schwarze 1132: match = 0;
1.50 schwarze 1133: for (mlink = mpage->mlinks; mlink;
1134: mlink = mlink->next)
1135: if (mlink_check(mpage, mlink))
1136: match = 1;
1137: } else
1138: match = 1;
1.6 schwarze 1139:
1.47 schwarze 1140: if (NULL != mdoc) {
1141: if (NULL != (cp = mdoc_meta(mdoc)->name))
1142: putkey(mpage, cp, TYPE_Nm);
1143: assert(NULL == mpage->desc);
1144: parse_mdoc(mpage, mdoc_node(mdoc));
1145: putkey(mpage, NULL != mpage->desc ?
1146: mpage->desc : mpage->mlinks->name, TYPE_Nd);
1147: } else if (NULL != man)
1148: parse_man(mpage, man_node(man));
1149: else
1.81 schwarze 1150: parse_cat(mpage, fd[0]);
1.6 schwarze 1151:
1.62 schwarze 1152: dbadd(mpage, mc);
1.81 schwarze 1153:
1154: nextpage:
1155: if (child_pid) {
1156: if (-1 == waitpid(child_pid, &status, 0)) {
1157: exitcode = (int)MANDOCLEVEL_SYSERR;
1158: say(mpage->mlinks->file, "&wait gunzip");
1159: } else if (WIFSIGNALED(status)) {
1160: exitcode = (int)MANDOCLEVEL_SYSERR;
1161: say(mpage->mlinks->file,
1162: "gunzip died from signal %d",
1163: WTERMSIG(status));
1164: } else if (WEXITSTATUS(status)) {
1165: exitcode = (int)MANDOCLEVEL_SYSERR;
1166: say(mpage->mlinks->file,
1167: "gunzip failed with code %d",
1168: WEXITSTATUS(status));
1169: }
1170: }
1.47 schwarze 1171: ohash_delete(&strings);
1172: mpage = ohash_next(&mpages, &pslot);
1173: }
1.64 schwarze 1174:
1175: if (0 == nodb)
1176: SQL_EXEC("END TRANSACTION");
1.47 schwarze 1177: }
1.6 schwarze 1178:
1.47 schwarze 1179: static void
1.81 schwarze 1180: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1181: {
1182: FILE *stream;
1183: char *line, *p, *title;
1184: size_t len, plen, titlesz;
1.1 schwarze 1185:
1.81 schwarze 1186: stream = (-1 == fd) ?
1187: fopen(mpage->mlinks->file, "r") :
1188: fdopen(fd, "r");
1189: if (NULL == stream) {
1.47 schwarze 1190: if (warnings)
1.80 schwarze 1191: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1192: return;
1193: }
1.1 schwarze 1194:
1.47 schwarze 1195: /* Skip to first blank line. */
1.1 schwarze 1196:
1.47 schwarze 1197: while (NULL != (line = fgetln(stream, &len)))
1198: if ('\n' == *line)
1199: break;
1.1 schwarze 1200:
1.47 schwarze 1201: /*
1202: * Assume the first line that is not indented
1203: * is the first section header. Skip to it.
1204: */
1.1 schwarze 1205:
1.47 schwarze 1206: while (NULL != (line = fgetln(stream, &len)))
1207: if ('\n' != *line && ' ' != *line)
1208: break;
1209:
1210: /*
1211: * Read up until the next section into a buffer.
1212: * Strip the leading and trailing newline from each read line,
1213: * appending a trailing space.
1214: * Ignore empty (whitespace-only) lines.
1215: */
1.28 schwarze 1216:
1.47 schwarze 1217: titlesz = 0;
1218: title = NULL;
1.38 schwarze 1219:
1.47 schwarze 1220: while (NULL != (line = fgetln(stream, &len))) {
1221: if (' ' != *line || '\n' != line[len - 1])
1222: break;
1223: while (len > 0 && isspace((unsigned char)*line)) {
1224: line++;
1225: len--;
1226: }
1227: if (1 == len)
1228: continue;
1229: title = mandoc_realloc(title, titlesz + len);
1230: memcpy(title + titlesz, line, len);
1231: titlesz += len;
1232: title[titlesz - 1] = ' ';
1233: }
1.28 schwarze 1234:
1.47 schwarze 1235: /*
1236: * If no page content can be found, or the input line
1237: * is already the next section header, or there is no
1238: * trailing newline, reuse the page title as the page
1239: * description.
1240: */
1.1 schwarze 1241:
1.47 schwarze 1242: if (NULL == title || '\0' == *title) {
1243: if (warnings)
1244: say(mpage->mlinks->file,
1245: "Cannot find NAME section");
1246: assert(NULL == mpage->desc);
1247: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1248: putkey(mpage, mpage->mlinks->name, TYPE_Nd);
1249: fclose(stream);
1250: free(title);
1251: return;
1252: }
1.24 schwarze 1253:
1.47 schwarze 1254: title = mandoc_realloc(title, titlesz + 1);
1255: title[titlesz] = '\0';
1.24 schwarze 1256:
1.47 schwarze 1257: /*
1258: * Skip to the first dash.
1259: * Use the remaining line as the description (no more than 70
1260: * bytes).
1261: */
1.28 schwarze 1262:
1.47 schwarze 1263: if (NULL != (p = strstr(title, "- "))) {
1264: for (p += 2; ' ' == *p || '\b' == *p; p++)
1265: /* Skip to next word. */ ;
1266: } else {
1267: if (warnings)
1268: say(mpage->mlinks->file,
1269: "No dash in title line");
1270: p = title;
1271: }
1.1 schwarze 1272:
1.47 schwarze 1273: plen = strlen(p);
1.1 schwarze 1274:
1.47 schwarze 1275: /* Strip backspace-encoding from line. */
1.1 schwarze 1276:
1.47 schwarze 1277: while (NULL != (line = memchr(p, '\b', plen))) {
1278: len = line - p;
1279: if (0 == len) {
1280: memmove(line, line + 1, plen--);
1281: continue;
1282: }
1283: memmove(line - 1, line + 1, plen - len);
1284: plen -= 2;
1285: }
1.1 schwarze 1286:
1.47 schwarze 1287: assert(NULL == mpage->desc);
1288: mpage->desc = mandoc_strdup(p);
1289: putkey(mpage, mpage->desc, TYPE_Nd);
1290: fclose(stream);
1291: free(title);
1292: }
1.16 schwarze 1293:
1.47 schwarze 1294: /*
1295: * Put a type/word pair into the word database for this particular file.
1296: */
1297: static void
1.69 schwarze 1298: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1299: {
1.69 schwarze 1300: char *cp;
1.37 schwarze 1301:
1.47 schwarze 1302: assert(NULL != value);
1.69 schwarze 1303: if (TYPE_arch == type)
1304: for (cp = value; *cp; cp++)
1305: if (isupper((unsigned char)*cp))
1306: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1307: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1308: }
1309:
1310: /*
1.47 schwarze 1311: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1312: */
1313: static void
1.47 schwarze 1314: putmdockey(const struct mpage *mpage,
1315: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1316: {
1.16 schwarze 1317:
1.47 schwarze 1318: for ( ; NULL != n; n = n->next) {
1319: if (NULL != n->child)
1320: putmdockey(mpage, n->child, m);
1321: if (MDOC_TEXT == n->type)
1322: putkey(mpage, n->string, m);
1323: }
1324: }
1.16 schwarze 1325:
1.47 schwarze 1326: static void
1327: parse_man(struct mpage *mpage, const struct man_node *n)
1328: {
1329: const struct man_node *head, *body;
1.78 schwarze 1330: char *start, *title;
1.47 schwarze 1331: char byte;
1.78 schwarze 1332: size_t sz;
1.16 schwarze 1333:
1.47 schwarze 1334: if (NULL == n)
1335: return;
1.16 schwarze 1336:
1.47 schwarze 1337: /*
1338: * We're only searching for one thing: the first text child in
1339: * the BODY of a NAME section. Since we don't keep track of
1340: * sections in -man, run some hoops to find out whether we're in
1341: * the correct section or not.
1342: */
1.16 schwarze 1343:
1.47 schwarze 1344: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1345: body = n;
1346: assert(body->parent);
1347: if (NULL != (head = body->parent->head) &&
1348: 1 == head->nchild &&
1349: NULL != (head = (head->child)) &&
1350: MAN_TEXT == head->type &&
1351: 0 == strcmp(head->string, "NAME") &&
1.78 schwarze 1352: NULL != body->child) {
1.2 schwarze 1353:
1.47 schwarze 1354: /*
1355: * Suck the entire NAME section into memory.
1356: * Yes, we might run away.
1357: * But too many manuals have big, spread-out
1358: * NAME sections over many lines.
1359: */
1.2 schwarze 1360:
1.78 schwarze 1361: title = NULL;
1362: man_deroff(&title, body);
1.47 schwarze 1363: if (NULL == title)
1364: return;
1.16 schwarze 1365:
1.47 schwarze 1366: /*
1367: * Go through a special heuristic dance here.
1368: * Conventionally, one or more manual names are
1369: * comma-specified prior to a whitespace, then a
1370: * dash, then a description. Try to puzzle out
1371: * the name parts here.
1372: */
1.16 schwarze 1373:
1.78 schwarze 1374: start = title;
1.47 schwarze 1375: for ( ;; ) {
1376: sz = strcspn(start, " ,");
1377: if ('\0' == start[sz])
1378: break;
1.1 schwarze 1379:
1.47 schwarze 1380: byte = start[sz];
1381: start[sz] = '\0';
1.67 schwarze 1382:
1383: /*
1384: * Assume a stray trailing comma in the
1385: * name list if a name begins with a dash.
1386: */
1387:
1388: if ('-' == start[0] ||
1389: ('\\' == start[0] && '-' == start[1]))
1390: break;
1.1 schwarze 1391:
1.86 ! schwarze 1392: putkey(mpage, start, TYPE_NAME | TYPE_Nm);
1.1 schwarze 1393:
1.47 schwarze 1394: if (' ' == byte) {
1395: start += sz + 1;
1396: break;
1397: }
1.1 schwarze 1398:
1.47 schwarze 1399: assert(',' == byte);
1400: start += sz + 1;
1401: while (' ' == *start)
1402: start++;
1403: }
1.1 schwarze 1404:
1.78 schwarze 1405: if (start == title) {
1.86 ! schwarze 1406: putkey(mpage, start, TYPE_NAME | TYPE_Nm);
1.47 schwarze 1407: free(title);
1408: return;
1409: }
1.1 schwarze 1410:
1.47 schwarze 1411: while (isspace((unsigned char)*start))
1412: start++;
1.1 schwarze 1413:
1.47 schwarze 1414: if (0 == strncmp(start, "-", 1))
1415: start += 1;
1416: else if (0 == strncmp(start, "\\-\\-", 4))
1417: start += 4;
1418: else if (0 == strncmp(start, "\\-", 2))
1419: start += 2;
1420: else if (0 == strncmp(start, "\\(en", 4))
1421: start += 4;
1422: else if (0 == strncmp(start, "\\(em", 4))
1423: start += 4;
1.1 schwarze 1424:
1.47 schwarze 1425: while (' ' == *start)
1426: start++;
1.1 schwarze 1427:
1.47 schwarze 1428: assert(NULL == mpage->desc);
1429: mpage->desc = mandoc_strdup(start);
1430: putkey(mpage, mpage->desc, TYPE_Nd);
1431: free(title);
1432: return;
1433: }
1434: }
1.1 schwarze 1435:
1.47 schwarze 1436: for (n = n->child; n; n = n->next) {
1437: if (NULL != mpage->desc)
1438: break;
1439: parse_man(mpage, n);
1.1 schwarze 1440: }
1441: }
1442:
1443: static void
1.47 schwarze 1444: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1445: {
1446:
1.47 schwarze 1447: assert(NULL != n);
1448: for (n = n->child; NULL != n; n = n->next) {
1449: switch (n->type) {
1450: case (MDOC_ELEM):
1451: /* FALLTHROUGH */
1452: case (MDOC_BLOCK):
1453: /* FALLTHROUGH */
1454: case (MDOC_HEAD):
1455: /* FALLTHROUGH */
1456: case (MDOC_BODY):
1457: /* FALLTHROUGH */
1458: case (MDOC_TAIL):
1459: if (NULL != mdocs[n->tok].fp)
1460: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1461: break;
1462: if (mdocs[n->tok].mask)
1463: putmdockey(mpage, n->child,
1464: mdocs[n->tok].mask);
1465: break;
1466: default:
1467: assert(MDOC_ROOT != n->type);
1468: continue;
1469: }
1470: if (NULL != n->child)
1471: parse_mdoc(mpage, n);
1.1 schwarze 1472: }
1473: }
1474:
1.19 schwarze 1475: static int
1.47 schwarze 1476: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1477: {
1478: const char *start, *end;
1479: size_t sz;
1.19 schwarze 1480:
1.47 schwarze 1481: if (SEC_SYNOPSIS != n->sec ||
1482: NULL == (n = n->child) ||
1483: MDOC_TEXT != n->type)
1.19 schwarze 1484: return(0);
1.1 schwarze 1485:
1486: /*
1487: * Only consider those `Fd' macro fields that begin with an
1488: * "inclusion" token (versus, e.g., #define).
1489: */
1.47 schwarze 1490:
1.1 schwarze 1491: if (strcmp("#include", n->string))
1.19 schwarze 1492: return(0);
1.1 schwarze 1493:
1494: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1495: return(0);
1.1 schwarze 1496:
1497: /*
1498: * Strip away the enclosing angle brackets and make sure we're
1499: * not zero-length.
1500: */
1501:
1502: start = n->string;
1503: if ('<' == *start || '"' == *start)
1504: start++;
1505:
1506: if (0 == (sz = strlen(start)))
1.19 schwarze 1507: return(0);
1.1 schwarze 1508:
1509: end = &start[(int)sz - 1];
1510: if ('>' == *end || '"' == *end)
1511: end--;
1512:
1.47 schwarze 1513: if (end > start)
1514: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1515: return(0);
1.1 schwarze 1516: }
1517:
1.19 schwarze 1518: static int
1.47 schwarze 1519: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1520: {
1.69 schwarze 1521: char *cp;
1.1 schwarze 1522:
1.47 schwarze 1523: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.19 schwarze 1524: return(0);
1525:
1.47 schwarze 1526: /*
1527: * Parse: .Fn "struct type *name" "char *arg".
1528: * First strip away pointer symbol.
1529: * Then store the function name, then type.
1530: * Finally, store the arguments.
1531: */
1.1 schwarze 1532:
1.47 schwarze 1533: if (NULL == (cp = strrchr(n->string, ' ')))
1534: cp = n->string;
1.1 schwarze 1535:
1536: while ('*' == *cp)
1537: cp++;
1538:
1.47 schwarze 1539: putkey(mpage, cp, TYPE_Fn);
1.19 schwarze 1540:
1.47 schwarze 1541: if (n->string < cp)
1542: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.19 schwarze 1543:
1.47 schwarze 1544: for (n = n->next; NULL != n; n = n->next)
1545: if (MDOC_TEXT == n->type)
1546: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1547:
1548: return(0);
1.1 schwarze 1549: }
1550:
1.19 schwarze 1551: static int
1.47 schwarze 1552: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1553: {
1.47 schwarze 1554: char *cp;
1.1 schwarze 1555:
1556: if (NULL == (n = n->child))
1.19 schwarze 1557: return(0);
1.1 schwarze 1558:
1.47 schwarze 1559: if (NULL == n->next) {
1560: putkey(mpage, n->string, TYPE_Xr);
1561: return(0);
1562: }
1.1 schwarze 1563:
1.77 schwarze 1564: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1565: putkey(mpage, cp, TYPE_Xr);
1566: free(cp);
1567: return(0);
1.1 schwarze 1568: }
1569:
1.19 schwarze 1570: static int
1.47 schwarze 1571: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1572: {
1573:
1.79 schwarze 1574: if (MDOC_BODY == n->type)
1575: mdoc_deroff(&mpage->desc, n);
1576: return(0);
1.1 schwarze 1577: }
1578:
1.19 schwarze 1579: static int
1.47 schwarze 1580: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1581: {
1582:
1.86 ! schwarze 1583: if (SEC_NAME == n->sec)
! 1584: putmdockey(mpage, n->child, TYPE_NAME | TYPE_Nm);
! 1585: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)
! 1586: putmdockey(mpage, n->child, TYPE_Nm);
! 1587: return(0);
1.1 schwarze 1588: }
1589:
1.19 schwarze 1590: static int
1.47 schwarze 1591: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1592: {
1593:
1.19 schwarze 1594: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1595: }
1596:
1.47 schwarze 1597: static int
1598: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1599: {
1600:
1.47 schwarze 1601: return(MDOC_HEAD == n->type);
1602: }
1.1 schwarze 1603:
1.47 schwarze 1604: static int
1605: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1606: {
1.1 schwarze 1607:
1.47 schwarze 1608: return(MDOC_BODY == n->type);
1.1 schwarze 1609: }
1610:
1.47 schwarze 1611: /*
1612: * Add a string to the hash table for the current manual.
1613: * Each string has a bitmask telling which macros it belongs to.
1614: * When we finish the manual, we'll dump the table.
1615: */
1.1 schwarze 1616: static void
1.47 schwarze 1617: putkeys(const struct mpage *mpage,
1618: const char *cp, size_t sz, uint64_t v)
1.1 schwarze 1619: {
1.47 schwarze 1620: struct str *s;
1.68 schwarze 1621: const char *end;
1622: uint64_t mask;
1.47 schwarze 1623: unsigned int slot;
1.68 schwarze 1624: int i;
1.1 schwarze 1625:
1.47 schwarze 1626: if (0 == sz)
1627: return;
1.68 schwarze 1628:
1.82 schwarze 1629: if (debug > 1) {
1.68 schwarze 1630: for (i = 0, mask = 1;
1631: i < mansearch_keymax;
1632: i++, mask <<= 1)
1633: if (mask & v)
1634: break;
1635: say(mpage->mlinks->file, "Adding key %s=%*s",
1636: mansearch_keynames[i], sz, cp);
1637: }
1.47 schwarze 1638:
1639: end = cp + sz;
1640: slot = ohash_qlookupi(&strings, cp, &end);
1641: s = ohash_find(&strings, slot);
1.1 schwarze 1642:
1.47 schwarze 1643: if (NULL != s && mpage == s->mpage) {
1644: s->mask |= v;
1.1 schwarze 1645: return;
1.47 schwarze 1646: } else if (NULL == s) {
1647: s = mandoc_calloc(sizeof(struct str) + sz + 1, 1);
1648: memcpy(s->key, cp, sz);
1649: ohash_insert(&strings, slot, s);
1650: }
1651: s->mpage = mpage;
1652: s->mask = v;
1.1 schwarze 1653: }
1654:
1655: /*
1.47 schwarze 1656: * Take a Unicode codepoint and produce its UTF-8 encoding.
1657: * This isn't the best way to do this, but it works.
1658: * The magic numbers are from the UTF-8 packaging.
1659: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1660: */
1.47 schwarze 1661: static size_t
1662: utf8(unsigned int cp, char out[7])
1.1 schwarze 1663: {
1.47 schwarze 1664: size_t rc;
1.1 schwarze 1665:
1.47 schwarze 1666: rc = 0;
1667: if (cp <= 0x0000007F) {
1668: rc = 1;
1669: out[0] = (char)cp;
1670: } else if (cp <= 0x000007FF) {
1671: rc = 2;
1672: out[0] = (cp >> 6 & 31) | 192;
1673: out[1] = (cp & 63) | 128;
1674: } else if (cp <= 0x0000FFFF) {
1675: rc = 3;
1676: out[0] = (cp >> 12 & 15) | 224;
1677: out[1] = (cp >> 6 & 63) | 128;
1678: out[2] = (cp & 63) | 128;
1679: } else if (cp <= 0x001FFFFF) {
1680: rc = 4;
1681: out[0] = (cp >> 18 & 7) | 240;
1682: out[1] = (cp >> 12 & 63) | 128;
1683: out[2] = (cp >> 6 & 63) | 128;
1684: out[3] = (cp & 63) | 128;
1685: } else if (cp <= 0x03FFFFFF) {
1686: rc = 5;
1687: out[0] = (cp >> 24 & 3) | 248;
1688: out[1] = (cp >> 18 & 63) | 128;
1689: out[2] = (cp >> 12 & 63) | 128;
1690: out[3] = (cp >> 6 & 63) | 128;
1691: out[4] = (cp & 63) | 128;
1692: } else if (cp <= 0x7FFFFFFF) {
1693: rc = 6;
1694: out[0] = (cp >> 30 & 1) | 252;
1695: out[1] = (cp >> 24 & 63) | 128;
1696: out[2] = (cp >> 18 & 63) | 128;
1697: out[3] = (cp >> 12 & 63) | 128;
1698: out[4] = (cp >> 6 & 63) | 128;
1699: out[5] = (cp & 63) | 128;
1700: } else
1701: return(0);
1.19 schwarze 1702:
1.47 schwarze 1703: out[rc] = '\0';
1704: return(rc);
1.1 schwarze 1705: }
1706:
1.47 schwarze 1707: /*
1.53 schwarze 1708: * Store the rendered version of a key, or alias the pointer
1709: * if the key contains no escape sequences.
1.47 schwarze 1710: */
1711: static void
1.53 schwarze 1712: render_key(struct mchars *mc, struct str *key)
1.1 schwarze 1713: {
1.47 schwarze 1714: size_t sz, bsz, pos;
1.71 schwarze 1715: char utfbuf[7], res[6];
1.47 schwarze 1716: char *buf;
1717: const char *seq, *cpp, *val;
1718: int len, u;
1719: enum mandoc_esc esc;
1720:
1.53 schwarze 1721: assert(NULL == key->rendered);
1.47 schwarze 1722:
1723: res[0] = '\\';
1724: res[1] = '\t';
1725: res[2] = ASCII_NBRSP;
1726: res[3] = ASCII_HYPH;
1.71 schwarze 1727: res[4] = ASCII_BREAK;
1728: res[5] = '\0';
1.1 schwarze 1729:
1.47 schwarze 1730: val = key->key;
1731: bsz = strlen(val);
1.1 schwarze 1732:
1733: /*
1.47 schwarze 1734: * Pre-check: if we have no stop-characters, then set the
1735: * pointer as ourselvse and get out of here.
1.1 schwarze 1736: */
1.47 schwarze 1737: if (strcspn(val, res) == bsz) {
1.53 schwarze 1738: key->rendered = key->key;
1.47 schwarze 1739: return;
1740: }
1.1 schwarze 1741:
1.47 schwarze 1742: /* Pre-allocate by the length of the input */
1.39 schwarze 1743:
1.47 schwarze 1744: buf = mandoc_malloc(++bsz);
1745: pos = 0;
1.39 schwarze 1746:
1.47 schwarze 1747: while ('\0' != *val) {
1748: /*
1749: * Halt on the first escape sequence.
1750: * This also halts on the end of string, in which case
1751: * we just copy, fallthrough, and exit the loop.
1752: */
1753: if ((sz = strcspn(val, res)) > 0) {
1754: memcpy(&buf[pos], val, sz);
1755: pos += sz;
1756: val += sz;
1757: }
1.39 schwarze 1758:
1.71 schwarze 1759: switch (*val) {
1760: case (ASCII_HYPH):
1.47 schwarze 1761: buf[pos++] = '-';
1762: val++;
1763: continue;
1.71 schwarze 1764: case ('\t'):
1765: /* FALLTHROUGH */
1766: case (ASCII_NBRSP):
1.47 schwarze 1767: buf[pos++] = ' ';
1768: val++;
1.71 schwarze 1769: /* FALLTHROUGH */
1770: case (ASCII_BREAK):
1.47 schwarze 1771: continue;
1.71 schwarze 1772: default:
1773: break;
1774: }
1775: if ('\\' != *val)
1.47 schwarze 1776: break;
1.39 schwarze 1777:
1.47 schwarze 1778: /* Read past the slash. */
1.39 schwarze 1779:
1.47 schwarze 1780: val++;
1.39 schwarze 1781:
1.47 schwarze 1782: /*
1783: * Parse the escape sequence and see if it's a
1784: * predefined character or special character.
1785: */
1.52 schwarze 1786:
1.47 schwarze 1787: esc = mandoc_escape
1788: ((const char **)&val, &seq, &len);
1789: if (ESCAPE_ERROR == esc)
1790: break;
1791: if (ESCAPE_SPECIAL != esc)
1792: continue;
1.39 schwarze 1793:
1.47 schwarze 1794: /*
1.52 schwarze 1795: * Render the special character
1796: * as either UTF-8 or ASCII.
1.47 schwarze 1797: */
1.52 schwarze 1798:
1799: if (write_utf8) {
1800: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1801: continue;
1802: cpp = utfbuf;
1803: if (0 == (sz = utf8(u, utfbuf)))
1804: continue;
1805: sz = strlen(cpp);
1806: } else {
1807: cpp = mchars_spec2str(mc, seq, len, &sz);
1808: if (NULL == cpp)
1809: continue;
1810: if (ASCII_NBRSP == *cpp) {
1811: cpp = " ";
1812: sz = 1;
1813: }
1814: }
1.1 schwarze 1815:
1.47 schwarze 1816: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1817:
1.47 schwarze 1818: bsz += sz;
1819: buf = mandoc_realloc(buf, bsz);
1820: memcpy(&buf[pos], cpp, sz);
1821: pos += sz;
1.1 schwarze 1822: }
1823:
1.47 schwarze 1824: buf[pos] = '\0';
1.53 schwarze 1825: key->rendered = buf;
1.1 schwarze 1826: }
1827:
1.75 schwarze 1828: static void
1829: dbadd_mlink(const struct mlink *mlink)
1830: {
1831: size_t i;
1832:
1833: i = 1;
1834: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1835: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1836: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1837: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->recno);
1838: SQL_STEP(stmts[STMT_INSERT_LINK]);
1839: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1840: }
1841:
1.11 schwarze 1842: /*
1.47 schwarze 1843: * Flush the current page's terms (and their bits) into the database.
1844: * Wrap the entire set of additions in a transaction to make sqlite be a
1845: * little faster.
1.53 schwarze 1846: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 1847: */
1848: static void
1.75 schwarze 1849: dbadd(struct mpage *mpage, struct mchars *mc)
1.11 schwarze 1850: {
1.47 schwarze 1851: struct mlink *mlink;
1852: struct str *key;
1853: size_t i;
1854: unsigned int slot;
1855:
1.85 schwarze 1856: mlink = mpage->mlinks;
1.11 schwarze 1857:
1.85 schwarze 1858: if (nodb) {
1859: while (NULL != mlink) {
1860: fputs(mlink->name, stdout);
1861: if (NULL == mlink->next ||
1862: strcmp(mlink->dsec, mlink->next->dsec) ||
1863: strcmp(mlink->fsec, mlink->next->fsec) ||
1864: strcmp(mlink->arch, mlink->next->arch)) {
1865: putchar('(');
1866: if ('\0' == *mlink->dsec)
1867: fputs(mlink->fsec, stdout);
1868: else
1869: fputs(mlink->dsec, stdout);
1870: if ('\0' != *mlink->arch)
1871: printf("/%s", mlink->arch);
1872: putchar(')');
1873: }
1874: mlink = mlink->next;
1875: if (NULL != mlink)
1876: fputs(", ", stdout);
1877: }
1878: for (key = ohash_first(&strings, &slot); NULL != key;
1879: key = ohash_next(&strings, &slot)) {
1880: if (TYPE_Nd & key->mask) {
1881: if (NULL == key->rendered)
1882: render_key(mc, key);
1883: printf(" - %s", key->rendered);
1884: break;
1885: }
1886: }
1887: putchar('\n');
1.11 schwarze 1888: return;
1.85 schwarze 1889: }
1890:
1891: if (debug)
1892: say(mlink->file, "Adding to database");
1.47 schwarze 1893:
1894: i = 1;
1895: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
1896: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.75 schwarze 1897: mpage->recno = sqlite3_last_insert_rowid(db);
1.47 schwarze 1898: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1899:
1.85 schwarze 1900: while (NULL != mlink) {
1.75 schwarze 1901: dbadd_mlink(mlink);
1.85 schwarze 1902: mlink = mlink->next;
1903: }
1.47 schwarze 1904:
1905: for (key = ohash_first(&strings, &slot); NULL != key;
1906: key = ohash_next(&strings, &slot)) {
1907: assert(key->mpage == mpage);
1.53 schwarze 1908: if (NULL == key->rendered)
1909: render_key(mc, key);
1.47 schwarze 1910: i = 1;
1911: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.53 schwarze 1912: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.75 schwarze 1913: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->recno);
1.47 schwarze 1914: SQL_STEP(stmts[STMT_INSERT_KEY]);
1915: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.53 schwarze 1916: if (key->rendered != key->key)
1917: free(key->rendered);
1.47 schwarze 1918: free(key);
1.33 schwarze 1919: }
1.47 schwarze 1920: }
1.41 deraadt 1921:
1.47 schwarze 1922: static void
1923: dbprune(void)
1924: {
1925: struct mpage *mpage;
1926: struct mlink *mlink;
1927: size_t i;
1928: unsigned int slot;
1.11 schwarze 1929:
1.63 schwarze 1930: if (0 == nodb)
1931: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 1932:
1.63 schwarze 1933: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
1934: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 1935: mlink = mpage->mlinks;
1.82 schwarze 1936: if (debug)
1.63 schwarze 1937: say(mlink->file, "Deleting from database");
1938: if (nodb)
1939: continue;
1940: for ( ; NULL != mlink; mlink = mlink->next) {
1941: i = 1;
1942: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
1943: i, mlink->dsec);
1944: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
1945: i, mlink->arch);
1946: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
1947: i, mlink->name);
1948: SQL_STEP(stmts[STMT_DELETE_PAGE]);
1949: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
1950: }
1.11 schwarze 1951: }
1.63 schwarze 1952:
1953: if (0 == nodb)
1954: SQL_EXEC("END TRANSACTION");
1.47 schwarze 1955: }
1.22 schwarze 1956:
1.47 schwarze 1957: /*
1958: * Close an existing database and its prepared statements.
1959: * If "real" is not set, rename the temporary file into the real one.
1960: */
1961: static void
1962: dbclose(int real)
1963: {
1964: size_t i;
1.72 schwarze 1965: int status;
1966: pid_t child;
1.11 schwarze 1967:
1.47 schwarze 1968: if (nodb)
1969: return;
1.11 schwarze 1970:
1.47 schwarze 1971: for (i = 0; i < STMT__MAX; i++) {
1972: sqlite3_finalize(stmts[i]);
1973: stmts[i] = NULL;
1.28 schwarze 1974: }
1.22 schwarze 1975:
1.47 schwarze 1976: sqlite3_close(db);
1977: db = NULL;
1.11 schwarze 1978:
1.47 schwarze 1979: if (real)
1980: return;
1.22 schwarze 1981:
1.72 schwarze 1982: if ('\0' == *tempfilename) {
1983: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
1984: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 1985: say(MANDOC_DB, "&rename");
1.72 schwarze 1986: }
1987: return;
1988: }
1989:
1990: switch (child = fork()) {
1991: case (-1):
1992: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 1993: say("", "&fork cmp");
1.72 schwarze 1994: return;
1995: case (0):
1996: execlp("cmp", "cmp", "-s",
1997: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 1998: say("", "&exec cmp");
1.72 schwarze 1999: exit(0);
2000: default:
2001: break;
2002: }
2003: if (-1 == waitpid(child, &status, 0)) {
2004: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2005: say("", "&wait cmp");
1.72 schwarze 2006: } else if (WIFSIGNALED(status)) {
2007: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2008: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2009: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2010: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2011: say(MANDOC_DB,
2012: "Data changed, but cannot replace database");
2013: }
2014:
2015: *strrchr(tempfilename, '/') = '\0';
2016: switch (child = fork()) {
2017: case (-1):
2018: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2019: say("", "&fork rm");
1.72 schwarze 2020: return;
2021: case (0):
2022: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2023: say("", "&exec rm");
1.72 schwarze 2024: exit((int)MANDOCLEVEL_SYSERR);
2025: default:
2026: break;
2027: }
2028: if (-1 == waitpid(child, &status, 0)) {
2029: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2030: say("", "&wait rm");
1.72 schwarze 2031: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2032: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2033: say("", "%s: Cannot remove temporary directory",
2034: tempfilename);
1.22 schwarze 2035: }
1.11 schwarze 2036: }
2037:
1.47 schwarze 2038: /*
2039: * This is straightforward stuff.
2040: * Open a database connection to a "temporary" database, then open a set
2041: * of prepared statements we'll use over and over again.
2042: * If "real" is set, we use the existing database; if not, we truncate a
2043: * temporary one.
2044: * Must be matched by dbclose().
2045: */
2046: static int
2047: dbopen(int real)
1.2 schwarze 2048: {
1.72 schwarze 2049: const char *sql;
1.47 schwarze 2050: int rc, ofl;
1.6 schwarze 2051:
1.47 schwarze 2052: if (nodb)
2053: return(1);
1.6 schwarze 2054:
1.72 schwarze 2055: *tempfilename = '\0';
1.47 schwarze 2056: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2057:
2058: if (real) {
2059: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2060: if (SQLITE_OK != rc) {
1.47 schwarze 2061: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2062: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.47 schwarze 2063: return(0);
1.28 schwarze 2064: }
1.72 schwarze 2065: goto prepare_statements;
2066: }
2067:
2068: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2069:
1.72 schwarze 2070: remove(MANDOC_DB "~");
2071: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.47 schwarze 2072: if (SQLITE_OK == rc)
1.72 schwarze 2073: goto create_tables;
1.73 schwarze 2074: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2075: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2076: say(MANDOC_DB "~", "%s", sqlite3_errmsg(db));
1.47 schwarze 2077: return(0);
2078: }
1.6 schwarze 2079:
1.72 schwarze 2080: if (strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2081: sizeof(tempfilename)) >= sizeof(tempfilename)) {
2082: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2083: say("", "/tmp/mandocdb.XXXXXX: Filename too long");
1.72 schwarze 2084: return(0);
2085: }
2086: if (NULL == mkdtemp(tempfilename)) {
2087: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2088: say("", "&%s", tempfilename);
1.72 schwarze 2089: return(0);
2090: }
2091: if (strlcat(tempfilename, "/" MANDOC_DB,
2092: sizeof(tempfilename)) >= sizeof(tempfilename)) {
2093: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2094: say("", "%s/" MANDOC_DB ": Filename too long",
2095: tempfilename);
1.72 schwarze 2096: return(0);
2097: }
2098: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2099: if (SQLITE_OK != rc) {
1.47 schwarze 2100: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2101: say("", "%s: %s", tempfilename, sqlite3_errmsg(db));
1.47 schwarze 2102: return(0);
1.2 schwarze 2103: }
2104:
1.72 schwarze 2105: create_tables:
1.47 schwarze 2106: sql = "CREATE TABLE \"mpages\" (\n"
2107: " \"form\" INTEGER NOT NULL,\n"
2108: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
2109: ");\n"
2110: "\n"
2111: "CREATE TABLE \"mlinks\" (\n"
2112: " \"sec\" TEXT NOT NULL,\n"
2113: " \"arch\" TEXT NOT NULL,\n"
2114: " \"name\" TEXT NOT NULL,\n"
2115: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1.66 schwarze 2116: "ON DELETE CASCADE\n"
1.47 schwarze 2117: ");\n"
2118: "\n"
2119: "CREATE TABLE \"keys\" (\n"
2120: " \"bits\" INTEGER NOT NULL,\n"
2121: " \"key\" TEXT NOT NULL,\n"
2122: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
1.66 schwarze 2123: "ON DELETE CASCADE\n"
1.65 schwarze 2124: ");\n";
1.47 schwarze 2125:
2126: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2127: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2128: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.47 schwarze 2129: return(0);
1.2 schwarze 2130: }
2131:
1.47 schwarze 2132: prepare_statements:
2133: SQL_EXEC("PRAGMA foreign_keys = ON");
1.63 schwarze 2134: sql = "DELETE FROM mpages WHERE id IN "
2135: "(SELECT pageid FROM mlinks WHERE "
2136: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2137: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2138: sql = "INSERT INTO mpages "
1.60 schwarze 2139: "(form) VALUES (?)";
1.47 schwarze 2140: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2141: sql = "INSERT INTO mlinks "
1.61 schwarze 2142: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2143: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
2144: sql = "INSERT INTO keys "
2145: "(bits,key,pageid) VALUES (?,?,?)";
2146: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2147:
1.47 schwarze 2148: /*
2149: * When opening a new database, we can turn off
2150: * synchronous mode for much better performance.
2151: */
1.6 schwarze 2152:
1.47 schwarze 2153: if (real)
2154: SQL_EXEC("PRAGMA synchronous = OFF");
1.11 schwarze 2155:
1.47 schwarze 2156: return(1);
2157: }
1.6 schwarze 2158:
1.47 schwarze 2159: static void *
2160: hash_halloc(size_t sz, void *arg)
2161: {
1.6 schwarze 2162:
1.47 schwarze 2163: return(mandoc_calloc(sz, 1));
2164: }
1.2 schwarze 2165:
1.47 schwarze 2166: static void *
2167: hash_alloc(size_t sz, void *arg)
2168: {
1.28 schwarze 2169:
1.47 schwarze 2170: return(mandoc_malloc(sz));
2171: }
1.6 schwarze 2172:
1.47 schwarze 2173: static void
2174: hash_free(void *p, size_t sz, void *arg)
2175: {
1.26 schwarze 2176:
1.47 schwarze 2177: free(p);
2178: }
1.6 schwarze 2179:
1.47 schwarze 2180: static int
2181: set_basedir(const char *targetdir)
2182: {
2183: static char startdir[PATH_MAX];
2184: static int fd;
1.6 schwarze 2185:
1.47 schwarze 2186: /*
2187: * Remember where we started by keeping a fd open to the origin
2188: * path component: throughout this utility, we chdir() a lot to
2189: * handle relative paths, and by doing this, we can return to
2190: * the starting point.
2191: */
2192: if ('\0' == *startdir) {
2193: if (NULL == getcwd(startdir, PATH_MAX)) {
2194: exitcode = (int)MANDOCLEVEL_SYSERR;
2195: if (NULL != targetdir)
1.80 schwarze 2196: say("", "&getcwd");
1.47 schwarze 2197: return(0);
2198: }
2199: if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
2200: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2201: say("", "&open %s", startdir);
1.47 schwarze 2202: return(0);
1.11 schwarze 2203: }
1.47 schwarze 2204: if (NULL == targetdir)
2205: targetdir = startdir;
2206: } else {
2207: if (-1 == fd)
2208: return(0);
2209: if (-1 == fchdir(fd)) {
2210: close(fd);
2211: basedir[0] = '\0';
2212: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2213: say("", "&chdir %s", startdir);
1.47 schwarze 2214: return(0);
1.2 schwarze 2215: }
1.47 schwarze 2216: if (NULL == targetdir) {
2217: close(fd);
2218: return(1);
1.2 schwarze 2219: }
2220: }
1.47 schwarze 2221: if (NULL == realpath(targetdir, basedir)) {
2222: basedir[0] = '\0';
2223: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 2224: say("", "&%s: realpath", targetdir);
1.47 schwarze 2225: return(0);
2226: } else if (-1 == chdir(basedir)) {
2227: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 2228: say("", "&chdir");
1.47 schwarze 2229: return(0);
2230: }
2231: return(1);
1.2 schwarze 2232: }
2233:
2234: static void
1.47 schwarze 2235: say(const char *file, const char *format, ...)
1.2 schwarze 2236: {
1.47 schwarze 2237: va_list ap;
1.80 schwarze 2238: int use_errno;
1.2 schwarze 2239:
1.47 schwarze 2240: if ('\0' != *basedir)
2241: fprintf(stderr, "%s", basedir);
2242: if ('\0' != *basedir && '\0' != *file)
2243: fputs("//", stderr);
2244: if ('\0' != *file)
2245: fprintf(stderr, "%s", file);
1.31 schwarze 2246:
1.80 schwarze 2247: use_errno = 1;
2248: if (NULL != format) {
2249: switch (*format) {
2250: case ('&'):
2251: format++;
2252: break;
2253: case ('\0'):
2254: format = NULL;
2255: break;
2256: default:
2257: use_errno = 0;
2258: break;
2259: }
2260: }
2261: if (NULL != format) {
2262: if ('\0' != *basedir || '\0' != *file)
2263: fputs(": ", stderr);
2264: va_start(ap, format);
2265: vfprintf(stderr, format, ap);
2266: va_end(ap);
2267: }
2268: if (use_errno) {
2269: if ('\0' != *basedir || '\0' != *file || NULL != format)
2270: fputs(": ", stderr);
1.47 schwarze 2271: perror(NULL);
1.80 schwarze 2272: } else
2273: fputc('\n', stderr);
1.1 schwarze 2274: }
![[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