Annotation of mandoc/mandocdb.c, Revision 1.220.2.10
1.220.2.10! schwarze 1: /* $Id: mandocdb.c,v 1.220.2.9 2017/01/27 14:18:42 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.220.2.5 schwarze 4: * Copyright (c) 2011-2017 Ingo Schwarze <schwarze@openbsd.org>
1.220.2.3 schwarze 5: * Copyright (c) 2016 Ed Maste <emaste@freebsd.org>
1.1 kristaps 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
1.187 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.187 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
19: #include "config.h"
20:
1.162 schwarze 21: #include <sys/types.h>
1.50 kristaps 22: #include <sys/stat.h>
1.115 schwarze 23: #include <sys/wait.h>
1.1 kristaps 24:
25: #include <assert.h>
1.43 kristaps 26: #include <ctype.h>
1.208 schwarze 27: #if HAVE_ERR
1.198 schwarze 28: #include <err.h>
1.208 schwarze 29: #endif
1.63 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.158 schwarze 32: #if HAVE_FTS
1.50 kristaps 33: #include <fts.h>
1.157 schwarze 34: #else
35: #include "compat_fts.h"
36: #endif
1.58 schwarze 37: #include <limits.h>
1.218 kristaps 38: #if HAVE_SANDBOX_INIT
39: #include <sandbox.h>
40: #endif
1.50 kristaps 41: #include <stddef.h>
1.59 schwarze 42: #include <stdio.h>
1.1 kristaps 43: #include <stdint.h>
44: #include <stdlib.h>
45: #include <string.h>
1.17 schwarze 46: #include <unistd.h>
1.1 kristaps 47:
1.50 kristaps 48: #include <sqlite3.h>
1.1 kristaps 49:
1.188 schwarze 50: #include "mandoc_aux.h"
1.202 schwarze 51: #include "mandoc_ohash.h"
1.188 schwarze 52: #include "mandoc.h"
53: #include "roff.h"
1.50 kristaps 54: #include "mdoc.h"
1.1 kristaps 55: #include "man.h"
1.187 schwarze 56: #include "manconf.h"
1.55 kristaps 57: #include "mansearch.h"
1.1 kristaps 58:
1.111 schwarze 59: extern int mansearch_keymax;
60: extern const char *const mansearch_keynames[];
61:
1.52 kristaps 62: #define SQL_EXEC(_v) \
63: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.134 schwarze 64: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.52 kristaps 65: #define SQL_BIND_TEXT(_s, _i, _v) \
66: if (SQLITE_OK != sqlite3_bind_text \
67: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.134 schwarze 68: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 69: #define SQL_BIND_INT(_s, _i, _v) \
70: if (SQLITE_OK != sqlite3_bind_int \
71: ((_s), (_i)++, (_v))) \
1.134 schwarze 72: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 73: #define SQL_BIND_INT64(_s, _i, _v) \
74: if (SQLITE_OK != sqlite3_bind_int64 \
75: ((_s), (_i)++, (_v))) \
1.134 schwarze 76: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 77: #define SQL_STEP(_s) \
78: if (SQLITE_DONE != sqlite3_step((_s))) \
1.134 schwarze 79: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 80:
1.50 kristaps 81: enum op {
82: OP_DEFAULT = 0, /* new dbs from dir list or default config */
83: OP_CONFFILE, /* new databases from custom config file */
84: OP_UPDATE, /* delete/add entries in existing database */
85: OP_DELETE, /* delete entries from existing database */
86: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 87: };
88:
1.50 kristaps 89: struct str {
1.78 schwarze 90: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 91: uint64_t mask; /* bitmask in sequence */
1.176 schwarze 92: char key[]; /* rendered text */
1.38 schwarze 93: };
94:
1.79 schwarze 95: struct inodev {
96: ino_t st_ino;
97: dev_t st_dev;
1.50 kristaps 98: };
1.5 kristaps 99:
1.78 schwarze 100: struct mpage {
1.79 schwarze 101: struct inodev inodev; /* used for hashing routine */
1.137 schwarze 102: int64_t pageid; /* pageid in mpages SQL table */
1.82 schwarze 103: char *sec; /* section from file content */
104: char *arch; /* architecture from file content */
105: char *title; /* title from file content */
106: char *desc; /* description from file content */
1.220.2.3 schwarze 107: struct mpage *next; /* singly linked list */
1.82 schwarze 108: struct mlink *mlinks; /* singly linked list */
1.161 schwarze 109: int form; /* format from file content */
1.174 schwarze 110: int name_head_done;
1.82 schwarze 111: };
112:
113: struct mlink {
1.58 schwarze 114: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 115: char *dsec; /* section from directory */
116: char *arch; /* architecture from directory */
117: char *name; /* name from file name (not empty) */
118: char *fsec; /* section from file name suffix */
1.85 schwarze 119: struct mlink *next; /* singly linked list */
1.118 schwarze 120: struct mpage *mpage; /* parent */
1.161 schwarze 121: int dform; /* format from directory */
122: int fform; /* format from file name suffix */
1.124 schwarze 123: int gzip; /* filename has a .gz suffix */
1.3 kristaps 124: };
125:
1.50 kristaps 126: enum stmt {
1.81 schwarze 127: STMT_DELETE_PAGE = 0, /* delete mpage */
128: STMT_INSERT_PAGE, /* insert mpage */
129: STMT_INSERT_LINK, /* insert mlink */
1.133 schwarze 130: STMT_INSERT_NAME, /* insert name */
1.175 schwarze 131: STMT_SELECT_NAME, /* retrieve existing name flags */
1.81 schwarze 132: STMT_INSERT_KEY, /* insert parsed key */
1.50 kristaps 133: STMT__MAX
1.1 kristaps 134: };
135:
1.190 schwarze 136: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.189 schwarze 137: const struct roff_node *);
1.1 kristaps 138:
1.50 kristaps 139: struct mdoc_handler {
140: mdoc_fp fp; /* optional handler */
141: uint64_t mask; /* set unless handler returns 0 */
1.220.2.5 schwarze 142: int taboo; /* node flags that must not be set */
1.1 kristaps 143: };
1.219 schwarze 144:
145:
146: int mandocdb(int, char *[]);
1.1 kristaps 147:
1.59 schwarze 148: static void dbclose(int);
1.176 schwarze 149: static void dbadd(struct mpage *);
1.118 schwarze 150: static void dbadd_mlink(const struct mlink *mlink);
1.169 schwarze 151: static void dbadd_mlink_name(const struct mlink *mlink);
1.59 schwarze 152: static int dbopen(int);
153: static void dbprune(void);
154: static void filescan(const char *);
1.220.2.3 schwarze 155: #if HAVE_FTS_COMPARE_CONST
156: static int fts_compare(const FTSENT *const *, const FTSENT *const *);
157: #else
158: static int fts_compare(const FTSENT **, const FTSENT **);
159: #endif
1.84 schwarze 160: static void mlink_add(struct mlink *, const struct stat *);
1.131 schwarze 161: static void mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 162: static void mlink_free(struct mlink *);
1.89 schwarze 163: static void mlinks_undupe(struct mpage *);
1.78 schwarze 164: static void mpages_free(void);
1.176 schwarze 165: static void mpages_merge(struct mparse *);
1.130 schwarze 166: static void names_check(void);
1.124 schwarze 167: static void parse_cat(struct mpage *, int);
1.190 schwarze 168: static void parse_man(struct mpage *, const struct roff_meta *,
1.189 schwarze 169: const struct roff_node *);
1.190 schwarze 170: static void parse_mdoc(struct mpage *, const struct roff_meta *,
1.189 schwarze 171: const struct roff_node *);
1.190 schwarze 172: static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
1.189 schwarze 173: const struct roff_node *);
1.190 schwarze 174: static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
1.189 schwarze 175: const struct roff_node *);
176: static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
1.190 schwarze 177: static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
1.189 schwarze 178: const struct roff_node *);
1.190 schwarze 179: static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
1.189 schwarze 180: const struct roff_node *);
1.190 schwarze 181: static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
1.189 schwarze 182: const struct roff_node *);
1.190 schwarze 183: static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
1.189 schwarze 184: const struct roff_node *);
1.190 schwarze 185: static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
1.189 schwarze 186: const struct roff_node *);
1.211 schwarze 187: static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
188: const struct roff_node *);
1.190 schwarze 189: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.189 schwarze 190: const struct roff_node *);
1.112 schwarze 191: static void putkey(const struct mpage *, char *, uint64_t);
1.176 schwarze 192: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.78 schwarze 193: static void putmdockey(const struct mpage *,
1.220.2.5 schwarze 194: const struct roff_node *, uint64_t, int);
1.176 schwarze 195: static int render_string(char **, size_t *);
1.220 schwarze 196: static void say(const char *, const char *, ...)
197: __attribute__((__format__ (printf, 2, 3)));
1.165 schwarze 198: static int set_basedir(const char *, int);
1.59 schwarze 199: static int treescan(void);
1.50 kristaps 200: static size_t utf8(unsigned int, char [7]);
201:
1.115 schwarze 202: static char tempfilename[32];
1.102 schwarze 203: static int nodb; /* no database changes */
1.116 schwarze 204: static int mparse_options; /* abort the parse early */
1.141 schwarze 205: static int use_all; /* use all found files */
206: static int debug; /* print what we're doing */
207: static int warnings; /* warn about crap */
1.95 schwarze 208: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 209: static int exitcode; /* to be returned by main */
1.141 schwarze 210: static enum op op; /* operational mode */
1.59 schwarze 211: static char basedir[PATH_MAX]; /* current base directory */
1.220.2.3 schwarze 212: static struct mpage *mpage_head; /* list of distinct manual pages */
1.78 schwarze 213: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 214: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 215: static struct ohash names; /* table of all names */
1.50 kristaps 216: static struct ohash strings; /* table of all strings */
217: static sqlite3 *db = NULL; /* current database */
218: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.133 schwarze 219: static uint64_t name_mask;
1.25 schwarze 220:
221: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.220.2.5 schwarze 222: { NULL, 0, 0 }, /* Ap */
223: { NULL, 0, NODE_NOPRT }, /* Dd */
224: { NULL, 0, NODE_NOPRT }, /* Dt */
225: { NULL, 0, NODE_NOPRT }, /* Os */
226: { parse_mdoc_Sh, TYPE_Sh, 0 }, /* Sh */
227: { parse_mdoc_head, TYPE_Ss, 0 }, /* Ss */
228: { NULL, 0, 0 }, /* Pp */
229: { NULL, 0, 0 }, /* D1 */
230: { NULL, 0, 0 }, /* Dl */
231: { NULL, 0, 0 }, /* Bd */
232: { NULL, 0, 0 }, /* Ed */
233: { NULL, 0, 0 }, /* Bl */
234: { NULL, 0, 0 }, /* El */
235: { NULL, 0, 0 }, /* It */
236: { NULL, 0, 0 }, /* Ad */
237: { NULL, TYPE_An, 0 }, /* An */
238: { NULL, TYPE_Ar, 0 }, /* Ar */
239: { NULL, TYPE_Cd, 0 }, /* Cd */
240: { NULL, TYPE_Cm, 0 }, /* Cm */
241: { NULL, TYPE_Dv, 0 }, /* Dv */
242: { NULL, TYPE_Er, 0 }, /* Er */
243: { NULL, TYPE_Ev, 0 }, /* Ev */
244: { NULL, 0, 0 }, /* Ex */
245: { NULL, TYPE_Fa, 0 }, /* Fa */
246: { parse_mdoc_Fd, 0, 0 }, /* Fd */
247: { NULL, TYPE_Fl, 0 }, /* Fl */
248: { parse_mdoc_Fn, 0, 0 }, /* Fn */
249: { NULL, TYPE_Ft, 0 }, /* Ft */
250: { NULL, TYPE_Ic, 0 }, /* Ic */
251: { NULL, TYPE_In, 0 }, /* In */
252: { NULL, TYPE_Li, 0 }, /* Li */
253: { parse_mdoc_Nd, 0, 0 }, /* Nd */
254: { parse_mdoc_Nm, 0, 0 }, /* Nm */
255: { NULL, 0, 0 }, /* Op */
256: { NULL, 0, 0 }, /* Ot */
257: { NULL, TYPE_Pa, NODE_NOSRC }, /* Pa */
258: { NULL, 0, 0 }, /* Rv */
259: { NULL, TYPE_St, 0 }, /* St */
260: { parse_mdoc_Va, TYPE_Va, 0 }, /* Va */
261: { parse_mdoc_Va, TYPE_Vt, 0 }, /* Vt */
262: { parse_mdoc_Xr, 0, 0 }, /* Xr */
263: { NULL, 0, 0 }, /* %A */
264: { NULL, 0, 0 }, /* %B */
265: { NULL, 0, 0 }, /* %D */
266: { NULL, 0, 0 }, /* %I */
267: { NULL, 0, 0 }, /* %J */
268: { NULL, 0, 0 }, /* %N */
269: { NULL, 0, 0 }, /* %O */
270: { NULL, 0, 0 }, /* %P */
271: { NULL, 0, 0 }, /* %R */
272: { NULL, 0, 0 }, /* %T */
273: { NULL, 0, 0 }, /* %V */
274: { NULL, 0, 0 }, /* Ac */
275: { NULL, 0, 0 }, /* Ao */
276: { NULL, 0, 0 }, /* Aq */
1.220.2.7 schwarze 277: { NULL, TYPE_At, 0 }, /* At */
1.220.2.5 schwarze 278: { NULL, 0, 0 }, /* Bc */
279: { NULL, 0, 0 }, /* Bf */
280: { NULL, 0, 0 }, /* Bo */
281: { NULL, 0, 0 }, /* Bq */
282: { NULL, TYPE_Bsx, NODE_NOSRC }, /* Bsx */
1.220.2.6 schwarze 283: { NULL, TYPE_Bx, NODE_NOSRC }, /* Bx */
1.220.2.5 schwarze 284: { NULL, 0, 0 }, /* Db */
285: { NULL, 0, 0 }, /* Dc */
286: { NULL, 0, 0 }, /* Do */
287: { NULL, 0, 0 }, /* Dq */
288: { NULL, 0, 0 }, /* Ec */
289: { NULL, 0, 0 }, /* Ef */
290: { NULL, TYPE_Em, 0 }, /* Em */
291: { NULL, 0, 0 }, /* Eo */
292: { NULL, TYPE_Fx, NODE_NOSRC }, /* Fx */
293: { NULL, TYPE_Ms, 0 }, /* Ms */
294: { NULL, 0, 0 }, /* No */
295: { NULL, 0, 0 }, /* Ns */
296: { NULL, TYPE_Nx, NODE_NOSRC }, /* Nx */
297: { NULL, TYPE_Ox, NODE_NOSRC }, /* Ox */
298: { NULL, 0, 0 }, /* Pc */
299: { NULL, 0, 0 }, /* Pf */
300: { NULL, 0, 0 }, /* Po */
301: { NULL, 0, 0 }, /* Pq */
302: { NULL, 0, 0 }, /* Qc */
303: { NULL, 0, 0 }, /* Ql */
304: { NULL, 0, 0 }, /* Qo */
305: { NULL, 0, 0 }, /* Qq */
306: { NULL, 0, 0 }, /* Re */
307: { NULL, 0, 0 }, /* Rs */
308: { NULL, 0, 0 }, /* Sc */
309: { NULL, 0, 0 }, /* So */
310: { NULL, 0, 0 }, /* Sq */
311: { NULL, 0, 0 }, /* Sm */
312: { NULL, 0, 0 }, /* Sx */
313: { NULL, TYPE_Sy, 0 }, /* Sy */
314: { NULL, TYPE_Tn, 0 }, /* Tn */
315: { NULL, 0, NODE_NOSRC }, /* Ux */
316: { NULL, 0, 0 }, /* Xc */
317: { NULL, 0, 0 }, /* Xo */
318: { parse_mdoc_Fo, 0, 0 }, /* Fo */
319: { NULL, 0, 0 }, /* Fc */
320: { NULL, 0, 0 }, /* Oo */
321: { NULL, 0, 0 }, /* Oc */
322: { NULL, 0, 0 }, /* Bk */
323: { NULL, 0, 0 }, /* Ek */
324: { NULL, 0, 0 }, /* Bt */
325: { NULL, 0, 0 }, /* Hf */
326: { NULL, 0, 0 }, /* Fr */
327: { NULL, 0, 0 }, /* Ud */
1.220.2.7 schwarze 328: { NULL, TYPE_Lb, NODE_NOSRC }, /* Lb */
1.220.2.5 schwarze 329: { NULL, 0, 0 }, /* Lp */
330: { NULL, TYPE_Lk, 0 }, /* Lk */
331: { NULL, TYPE_Mt, NODE_NOSRC }, /* Mt */
332: { NULL, 0, 0 }, /* Brq */
333: { NULL, 0, 0 }, /* Bro */
334: { NULL, 0, 0 }, /* Brc */
335: { NULL, 0, 0 }, /* %C */
336: { NULL, 0, 0 }, /* Es */
337: { NULL, 0, 0 }, /* En */
338: { NULL, TYPE_Dx, NODE_NOSRC }, /* Dx */
339: { NULL, 0, 0 }, /* %Q */
340: { NULL, 0, 0 }, /* br */
341: { NULL, 0, 0 }, /* sp */
342: { NULL, 0, 0 }, /* %U */
343: { NULL, 0, 0 }, /* Ta */
344: { NULL, 0, 0 }, /* ll */
1.1 kristaps 345: };
346:
1.141 schwarze 347:
1.1 kristaps 348: int
1.179 schwarze 349: mandocdb(int argc, char *argv[])
1.1 kristaps 350: {
1.187 schwarze 351: struct manconf conf;
352: struct mparse *mp;
1.206 schwarze 353: const char *path_arg, *progname;
1.187 schwarze 354: size_t j, sz;
1.59 schwarze 355: int ch, i;
1.50 kristaps 356:
1.207 schwarze 357: #if HAVE_PLEDGE
358: if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
1.214 schwarze 359: warn("pledge");
1.218 kristaps 360: return (int)MANDOCLEVEL_SYSERR;
361: }
362: #endif
363:
364: #if HAVE_SANDBOX_INIT
365: if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) {
366: warnx("sandbox_init");
1.207 schwarze 367: return (int)MANDOCLEVEL_SYSERR;
368: }
369: #endif
370:
1.187 schwarze 371: memset(&conf, 0, sizeof(conf));
1.50 kristaps 372: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
373:
374: /*
1.141 schwarze 375: * We accept a few different invocations.
1.50 kristaps 376: * The CHECKOP macro makes sure that invocation styles don't
377: * clobber each other.
378: */
379: #define CHECKOP(_op, _ch) do \
380: if (OP_DEFAULT != (_op)) { \
1.198 schwarze 381: warnx("-%c: Conflicting option", (_ch)); \
1.50 kristaps 382: goto usage; \
383: } while (/*CONSTCOND*/0)
1.10 kristaps 384:
1.59 schwarze 385: path_arg = NULL;
1.38 schwarze 386: op = OP_DEFAULT;
1.1 kristaps 387:
1.126 schwarze 388: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 389: switch (ch) {
1.141 schwarze 390: case 'a':
1.12 schwarze 391: use_all = 1;
392: break;
1.141 schwarze 393: case 'C':
1.50 kristaps 394: CHECKOP(op, ch);
1.59 schwarze 395: path_arg = optarg;
1.38 schwarze 396: op = OP_CONFFILE;
1.34 schwarze 397: break;
1.141 schwarze 398: case 'D':
1.125 schwarze 399: debug++;
400: break;
1.141 schwarze 401: case 'd':
1.50 kristaps 402: CHECKOP(op, ch);
1.59 schwarze 403: path_arg = optarg;
1.5 kristaps 404: op = OP_UPDATE;
405: break;
1.141 schwarze 406: case 'n':
1.50 kristaps 407: nodb = 1;
408: break;
1.141 schwarze 409: case 'p':
1.126 schwarze 410: warnings = 1;
411: break;
1.141 schwarze 412: case 'Q':
1.116 schwarze 413: mparse_options |= MPARSE_QUICK;
1.102 schwarze 414: break;
1.141 schwarze 415: case 'T':
1.95 schwarze 416: if (strcmp(optarg, "utf8")) {
1.198 schwarze 417: warnx("-T%s: Unsupported output format",
418: optarg);
1.95 schwarze 419: goto usage;
420: }
421: write_utf8 = 1;
422: break;
1.141 schwarze 423: case 't':
1.50 kristaps 424: CHECKOP(op, ch);
1.38 schwarze 425: dup2(STDOUT_FILENO, STDERR_FILENO);
426: op = OP_TEST;
1.50 kristaps 427: nodb = warnings = 1;
1.38 schwarze 428: break;
1.141 schwarze 429: case 'u':
1.50 kristaps 430: CHECKOP(op, ch);
1.59 schwarze 431: path_arg = optarg;
1.5 kristaps 432: op = OP_DELETE;
433: break;
1.141 schwarze 434: case 'v':
1.126 schwarze 435: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 436: break;
1.1 kristaps 437: default:
1.38 schwarze 438: goto usage;
1.1 kristaps 439: }
440:
441: argc -= optind;
442: argv += optind;
443:
1.207 schwarze 444: #if HAVE_PLEDGE
1.210 schwarze 445: if (nodb) {
446: if (pledge("stdio rpath", NULL) == -1) {
1.214 schwarze 447: warn("pledge");
1.210 schwarze 448: return (int)MANDOCLEVEL_SYSERR;
449: }
1.207 schwarze 450: }
451: #endif
452:
1.38 schwarze 453: if (OP_CONFFILE == op && argc > 0) {
1.198 schwarze 454: warnx("-C: Too many arguments");
1.38 schwarze 455: goto usage;
456: }
457:
1.59 schwarze 458: exitcode = (int)MANDOCLEVEL_OK;
1.203 schwarze 459: mchars_alloc();
460: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
1.202 schwarze 461: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
462: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.50 kristaps 463:
464: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 465:
1.50 kristaps 466: /*
1.150 schwarze 467: * Most of these deal with a specific directory.
1.140 schwarze 468: * Jump into that directory first.
1.50 kristaps 469: */
1.165 schwarze 470: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.50 kristaps 471: goto out;
1.140 schwarze 472:
1.138 schwarze 473: if (dbopen(1)) {
1.140 schwarze 474: /*
475: * The existing database is usable. Process
476: * all files specified on the command-line.
477: */
1.207 schwarze 478: #if HAVE_PLEDGE
1.210 schwarze 479: if (!nodb) {
480: if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
1.214 schwarze 481: warn("pledge");
1.210 schwarze 482: exitcode = (int)MANDOCLEVEL_SYSERR;
483: goto out;
484: }
1.207 schwarze 485: }
486: #endif
1.140 schwarze 487: use_all = 1;
1.138 schwarze 488: for (i = 0; i < argc; i++)
489: filescan(argv[i]);
490: if (OP_TEST != op)
491: dbprune();
492: } else {
493: /*
494: * Database missing or corrupt.
495: * Recreate from scratch.
496: */
1.140 schwarze 497: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 498: op = OP_DEFAULT;
499: if (0 == treescan())
500: goto out;
501: if (0 == dbopen(0))
502: goto out;
503: }
1.50 kristaps 504: if (OP_DELETE != op)
1.176 schwarze 505: mpages_merge(mp);
1.138 schwarze 506: dbclose(OP_DEFAULT == op ? 0 : 1);
1.50 kristaps 507: } else {
508: /*
509: * If we have arguments, use them as our manpaths.
510: * If we don't, grok from manpath(1) or however else
1.187 schwarze 511: * manconf_parse() wants to do it.
1.50 kristaps 512: */
513: if (argc > 0) {
1.187 schwarze 514: conf.manpath.paths = mandoc_reallocarray(NULL,
1.144 schwarze 515: argc, sizeof(char *));
1.187 schwarze 516: conf.manpath.sz = (size_t)argc;
1.50 kristaps 517: for (i = 0; i < argc; i++)
1.187 schwarze 518: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.50 kristaps 519: } else
1.187 schwarze 520: manconf_parse(&conf, path_arg, NULL, NULL);
1.127 schwarze 521:
1.187 schwarze 522: if (conf.manpath.sz == 0) {
1.127 schwarze 523: exitcode = (int)MANDOCLEVEL_BADARG;
524: say("", "Empty manpath");
525: }
1.50 kristaps 526:
527: /*
1.69 schwarze 528: * First scan the tree rooted at a base directory, then
529: * build a new database and finally move it into place.
1.50 kristaps 530: * Ignore zero-length directories and strip trailing
531: * slashes.
532: */
1.187 schwarze 533: for (j = 0; j < conf.manpath.sz; j++) {
534: sz = strlen(conf.manpath.paths[j]);
535: if (sz && conf.manpath.paths[j][sz - 1] == '/')
536: conf.manpath.paths[j][--sz] = '\0';
1.50 kristaps 537: if (0 == sz)
538: continue;
1.66 schwarze 539:
540: if (j) {
1.202 schwarze 541: mandoc_ohash_init(&mpages, 6,
542: offsetof(struct mpage, inodev));
543: mandoc_ohash_init(&mlinks, 6,
544: offsetof(struct mlink, file));
1.66 schwarze 545: }
546:
1.187 schwarze 547: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.165 schwarze 548: continue;
1.59 schwarze 549: if (0 == treescan())
1.165 schwarze 550: continue;
1.59 schwarze 551: if (0 == dbopen(0))
1.165 schwarze 552: continue;
1.52 kristaps 553:
1.176 schwarze 554: mpages_merge(mp);
1.147 schwarze 555: if (warnings && !nodb &&
1.130 schwarze 556: ! (MPARSE_QUICK & mparse_options))
557: names_check();
1.59 schwarze 558: dbclose(0);
1.66 schwarze 559:
1.187 schwarze 560: if (j + 1 < conf.manpath.sz) {
1.80 schwarze 561: mpages_free();
1.78 schwarze 562: ohash_delete(&mpages);
1.83 schwarze 563: ohash_delete(&mlinks);
1.66 schwarze 564: }
1.50 kristaps 565: }
566: }
567: out:
1.187 schwarze 568: manconf_free(&conf);
1.168 schwarze 569: mparse_free(mp);
1.203 schwarze 570: mchars_free();
1.80 schwarze 571: mpages_free();
1.78 schwarze 572: ohash_delete(&mpages);
1.83 schwarze 573: ohash_delete(&mlinks);
1.197 schwarze 574: return exitcode;
1.50 kristaps 575: usage:
1.206 schwarze 576: progname = getprogname();
1.126 schwarze 577: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
578: " %s [-aDnpQ] [-Tutf8] dir ...\n"
579: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
580: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 581: " %s [-Q] -t file ...\n",
1.206 schwarze 582: progname, progname, progname, progname, progname);
1.50 kristaps 583:
1.197 schwarze 584: return (int)MANDOCLEVEL_BADARG;
1.50 kristaps 585: }
586:
587: /*
1.220.2.3 schwarze 588: * To get a singly linked list in alpha order while inserting entries
589: * at the beginning, process directory entries in reverse alpha order.
590: */
591: static int
592: #if HAVE_FTS_COMPARE_CONST
593: fts_compare(const FTSENT *const *a, const FTSENT *const *b)
594: #else
595: fts_compare(const FTSENT **a, const FTSENT **b)
596: #endif
597: {
598: return -strcmp((*a)->fts_name, (*b)->fts_name);
599: }
600:
601: /*
1.59 schwarze 602: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 603: * We use fts(), scanning directory parts along the way for clues to our
604: * section and architecture.
605: *
606: * If use_all has been specified, grok all files.
607: * If not, sanitise paths to the following:
608: *
1.141 schwarze 609: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 610: * or
611: * [./]cat<section>[/<arch>]/<name>.0
612: *
1.216 schwarze 613: * TODO: accommodate for multi-language directories.
1.50 kristaps 614: */
615: static int
1.59 schwarze 616: treescan(void)
1.50 kristaps 617: {
1.139 schwarze 618: char buf[PATH_MAX];
1.50 kristaps 619: FTS *f;
620: FTSENT *ff;
1.84 schwarze 621: struct mlink *mlink;
1.124 schwarze 622: int dform, gzip;
1.94 schwarze 623: char *dsec, *arch, *fsec, *cp;
624: const char *path;
1.50 kristaps 625: const char *argv[2];
626:
627: argv[0] = ".";
628: argv[1] = (char *)NULL;
629:
1.220.2.3 schwarze 630: f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR,
631: fts_compare);
1.200 schwarze 632: if (f == NULL) {
1.59 schwarze 633: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 634: say("", "&fts_open");
1.197 schwarze 635: return 0;
1.50 kristaps 636: }
637:
638: dsec = arch = NULL;
639: dform = FORM_NONE;
640:
1.200 schwarze 641: while ((ff = fts_read(f)) != NULL) {
1.50 kristaps 642: path = ff->fts_path + 2;
1.139 schwarze 643: switch (ff->fts_info) {
644:
645: /*
646: * Symbolic links require various sanity checks,
647: * then get handled just like regular files.
648: */
1.141 schwarze 649: case FTS_SL:
1.200 schwarze 650: if (realpath(path, buf) == NULL) {
1.139 schwarze 651: if (warnings)
652: say(path, "&realpath");
653: continue;
654: }
1.184 schwarze 655: if (strstr(buf, basedir) != buf
656: #ifdef HOMEBREWDIR
657: && strstr(buf, HOMEBREWDIR) != buf
658: #endif
659: ) {
1.139 schwarze 660: if (warnings) say("",
661: "%s: outside base directory", buf);
662: continue;
663: }
664: /* Use logical inode to avoid mpages dupe. */
1.200 schwarze 665: if (stat(path, ff->fts_statp) == -1) {
1.139 schwarze 666: if (warnings)
667: say(path, "&stat");
668: continue;
669: }
670: /* FALLTHROUGH */
671:
1.50 kristaps 672: /*
1.83 schwarze 673: * If we're a regular file, add an mlink by using the
1.50 kristaps 674: * stored directory data and handling the filename.
675: */
1.141 schwarze 676: case FTS_F:
1.200 schwarze 677: if ( ! strcmp(path, MANDOC_DB))
1.60 schwarze 678: continue;
1.50 kristaps 679: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 680: if (warnings)
1.59 schwarze 681: say(path, "Extraneous file");
1.50 kristaps 682: continue;
1.124 schwarze 683: }
684: gzip = 0;
685: fsec = NULL;
1.200 schwarze 686: while (fsec == NULL) {
1.124 schwarze 687: fsec = strrchr(ff->fts_name, '.');
1.200 schwarze 688: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.124 schwarze 689: break;
690: gzip = 1;
691: *fsec = '\0';
692: fsec = NULL;
693: }
1.200 schwarze 694: if (fsec == NULL) {
1.60 schwarze 695: if ( ! use_all) {
1.56 schwarze 696: if (warnings)
1.60 schwarze 697: say(path,
698: "No filename suffix");
1.50 kristaps 699: continue;
700: }
1.200 schwarze 701: } else if ( ! strcmp(++fsec, "html")) {
1.60 schwarze 702: if (warnings)
703: say(path, "Skip html");
704: continue;
1.200 schwarze 705: } else if ( ! strcmp(fsec, "ps")) {
1.60 schwarze 706: if (warnings)
707: say(path, "Skip ps");
708: continue;
1.200 schwarze 709: } else if ( ! strcmp(fsec, "pdf")) {
1.60 schwarze 710: if (warnings)
711: say(path, "Skip pdf");
712: continue;
713: } else if ( ! use_all &&
1.200 schwarze 714: ((dform == FORM_SRC &&
1.185 schwarze 715: strncmp(fsec, dsec, strlen(dsec))) ||
1.200 schwarze 716: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.60 schwarze 717: if (warnings)
718: say(path, "Wrong filename suffix");
719: continue;
1.66 schwarze 720: } else
1.84 schwarze 721: fsec[-1] = '\0';
1.94 schwarze 722:
1.84 schwarze 723: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 724: if (strlcpy(mlink->file, path,
725: sizeof(mlink->file)) >=
726: sizeof(mlink->file)) {
727: say(path, "Filename too long");
728: free(mlink);
729: continue;
730: }
1.84 schwarze 731: mlink->dform = dform;
1.94 schwarze 732: mlink->dsec = dsec;
733: mlink->arch = arch;
734: mlink->name = ff->fts_name;
735: mlink->fsec = fsec;
1.124 schwarze 736: mlink->gzip = gzip;
1.84 schwarze 737: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 738: continue;
1.139 schwarze 739:
1.141 schwarze 740: case FTS_D:
741: case FTS_DP:
1.139 schwarze 742: break;
743:
744: default:
1.60 schwarze 745: if (warnings)
746: say(path, "Not a regular file");
1.50 kristaps 747: continue;
1.60 schwarze 748: }
1.1 kristaps 749:
1.50 kristaps 750: switch (ff->fts_level) {
1.141 schwarze 751: case 0:
1.50 kristaps 752: /* Ignore the root directory. */
753: break;
1.141 schwarze 754: case 1:
1.50 kristaps 755: /*
756: * This might contain manX/ or catX/.
757: * Try to infer this from the name.
758: * If we're not in use_all, enforce it.
759: */
760: cp = ff->fts_name;
1.200 schwarze 761: if (ff->fts_info == FTS_DP) {
762: dform = FORM_NONE;
763: dsec = NULL;
1.50 kristaps 764: break;
1.200 schwarze 765: }
1.1 kristaps 766:
1.200 schwarze 767: if ( ! strncmp(cp, "man", 3)) {
1.50 kristaps 768: dform = FORM_SRC;
1.66 schwarze 769: dsec = cp + 3;
1.200 schwarze 770: } else if ( ! strncmp(cp, "cat", 3)) {
1.50 kristaps 771: dform = FORM_CAT;
1.66 schwarze 772: dsec = cp + 3;
1.94 schwarze 773: } else {
774: dform = FORM_NONE;
775: dsec = NULL;
1.50 kristaps 776: }
1.1 kristaps 777:
1.200 schwarze 778: if (dsec != NULL || use_all)
1.50 kristaps 779: break;
1.1 kristaps 780:
1.56 schwarze 781: if (warnings)
1.59 schwarze 782: say(path, "Unknown directory part");
1.50 kristaps 783: fts_set(f, ff, FTS_SKIP);
784: break;
1.141 schwarze 785: case 2:
1.50 kristaps 786: /*
787: * Possibly our architecture.
788: * If we're descending, keep tabs on it.
789: */
1.200 schwarze 790: if (ff->fts_info != FTS_DP && dsec != NULL)
1.66 schwarze 791: arch = ff->fts_name;
1.94 schwarze 792: else
793: arch = NULL;
1.50 kristaps 794: break;
795: default:
1.200 schwarze 796: if (ff->fts_info == FTS_DP || use_all)
1.50 kristaps 797: break;
1.56 schwarze 798: if (warnings)
1.59 schwarze 799: say(path, "Extraneous directory part");
1.50 kristaps 800: fts_set(f, ff, FTS_SKIP);
801: break;
1.5 kristaps 802: }
1.50 kristaps 803: }
804:
805: fts_close(f);
1.197 schwarze 806: return 1;
1.50 kristaps 807: }
1.5 kristaps 808:
1.50 kristaps 809: /*
1.89 schwarze 810: * Add a file to the mlinks table.
1.50 kristaps 811: * Do not verify that it's a "valid" looking manpage (we'll do that
812: * later).
813: *
814: * Try to infer the manual section, architecture, and page name from the
815: * path, assuming it looks like
816: *
1.141 schwarze 817: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 818: * or
819: * [./]cat<section>[/<arch>]/<name>.0
820: *
821: * See treescan() for the fts(3) version of this.
822: */
823: static void
1.59 schwarze 824: filescan(const char *file)
1.50 kristaps 825: {
1.59 schwarze 826: char buf[PATH_MAX];
1.84 schwarze 827: struct stat st;
828: struct mlink *mlink;
1.59 schwarze 829: char *p, *start;
1.5 kristaps 830:
1.50 kristaps 831: assert(use_all);
1.5 kristaps 832:
1.50 kristaps 833: if (0 == strncmp(file, "./", 2))
834: file += 2;
1.5 kristaps 835:
1.139 schwarze 836: /*
837: * We have to do lstat(2) before realpath(3) loses
838: * the information whether this is a symbolic link.
839: * We need to know that because for symbolic links,
840: * we want to use the orginal file name, while for
841: * regular files, we want to use the real path.
842: */
843: if (-1 == lstat(file, &st)) {
844: exitcode = (int)MANDOCLEVEL_BADARG;
845: say(file, "&lstat");
846: return;
847: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
848: exitcode = (int)MANDOCLEVEL_BADARG;
849: say(file, "Not a regular file");
850: return;
851: }
852:
853: /*
854: * We have to resolve the file name to the real path
855: * in any case for the base directory check.
856: */
1.59 schwarze 857: if (NULL == realpath(file, buf)) {
858: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 859: say(file, "&realpath");
1.59 schwarze 860: return;
1.106 schwarze 861: }
862:
1.150 schwarze 863: if (OP_TEST == op)
1.106 schwarze 864: start = buf;
1.150 schwarze 865: else if (strstr(buf, basedir) == buf)
866: start = buf + strlen(basedir);
1.184 schwarze 867: #ifdef HOMEBREWDIR
868: else if (strstr(buf, HOMEBREWDIR) == buf)
869: start = buf;
870: #endif
1.106 schwarze 871: else {
1.59 schwarze 872: exitcode = (int)MANDOCLEVEL_BADARG;
873: say("", "%s: outside base directory", buf);
874: return;
1.106 schwarze 875: }
876:
1.139 schwarze 877: /*
878: * Now we are sure the file is inside our tree.
879: * If it is a symbolic link, ignore the real path
880: * and use the original name.
881: * This implies passing stuff like "cat1/../man1/foo.1"
882: * on the command line won't work. So don't do that.
883: * Note the stat(2) can still fail if the link target
884: * doesn't exist.
885: */
886: if (S_IFLNK & st.st_mode) {
887: if (-1 == stat(buf, &st)) {
888: exitcode = (int)MANDOCLEVEL_BADARG;
889: say(file, "&stat");
890: return;
891: }
1.143 schwarze 892: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
893: say(file, "Filename too long");
894: return;
895: }
1.150 schwarze 896: start = buf;
897: if (OP_TEST != op && strstr(buf, basedir) == buf)
898: start += strlen(basedir);
1.50 kristaps 899: }
1.106 schwarze 900:
1.84 schwarze 901: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 902: mlink->dform = FORM_NONE;
1.143 schwarze 903: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
904: sizeof(mlink->file)) {
905: say(start, "Filename too long");
1.180 schwarze 906: free(mlink);
1.143 schwarze 907: return;
1.220.2.8 schwarze 908: }
909:
910: /*
911: * In test mode or when the original name is absolute
912: * but outside our tree, guess the base directory.
913: */
914:
915: if (op == OP_TEST || (start == buf && *start == '/')) {
916: if (strncmp(buf, "man/", 4) == 0)
917: start = buf + 4;
918: else if ((start = strstr(buf, "/man/")) != NULL)
919: start += 5;
920: else
921: start = buf;
1.143 schwarze 922: }
1.17 schwarze 923:
1.50 kristaps 924: /*
925: * First try to guess our directory structure.
926: * If we find a separator, try to look for man* or cat*.
927: * If we find one of these and what's underneath is a directory,
928: * assume it's an architecture.
929: */
930: if (NULL != (p = strchr(start, '/'))) {
931: *p++ = '\0';
932: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 933: mlink->dform = FORM_SRC;
1.94 schwarze 934: mlink->dsec = start + 3;
1.50 kristaps 935: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 936: mlink->dform = FORM_CAT;
1.94 schwarze 937: mlink->dsec = start + 3;
1.17 schwarze 938: }
1.5 kristaps 939:
1.50 kristaps 940: start = p;
1.84 schwarze 941: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 942: *p++ = '\0';
1.94 schwarze 943: mlink->arch = start;
1.50 kristaps 944: start = p;
1.84 schwarze 945: }
1.50 kristaps 946: }
947:
948: /*
949: * Now check the file suffix.
950: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
951: */
952: p = strrchr(start, '\0');
953: while (p-- > start && '/' != *p && '.' != *p)
954: /* Loop. */ ;
955:
956: if ('.' == *p) {
957: *p++ = '\0';
1.94 schwarze 958: mlink->fsec = p;
1.5 kristaps 959: }
960:
1.10 kristaps 961: /*
1.50 kristaps 962: * Now try to parse the name.
963: * Use the filename portion of the path.
1.10 kristaps 964: */
1.84 schwarze 965: mlink->name = start;
1.50 kristaps 966: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 967: mlink->name = p + 1;
1.50 kristaps 968: *p = '\0';
1.84 schwarze 969: }
970: mlink_add(mlink, &st);
1.50 kristaps 971: }
1.5 kristaps 972:
1.50 kristaps 973: static void
1.84 schwarze 974: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 975: {
1.83 schwarze 976: struct inodev inodev;
1.78 schwarze 977: struct mpage *mpage;
1.83 schwarze 978: unsigned int slot;
1.50 kristaps 979:
1.84 schwarze 980: assert(NULL != mlink->file);
1.50 kristaps 981:
1.94 schwarze 982: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
983: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
984: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
985: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 986:
987: if ('0' == *mlink->fsec) {
988: free(mlink->fsec);
989: mlink->fsec = mandoc_strdup(mlink->dsec);
990: mlink->fform = FORM_CAT;
991: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
992: mlink->fform = FORM_SRC;
1.74 schwarze 993: else
1.84 schwarze 994: mlink->fform = FORM_NONE;
1.82 schwarze 995:
1.83 schwarze 996: slot = ohash_qlookup(&mlinks, mlink->file);
997: assert(NULL == ohash_find(&mlinks, slot));
998: ohash_insert(&mlinks, slot, mlink);
999:
1.186 schwarze 1000: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 1001: inodev.st_ino = st->st_ino;
1002: inodev.st_dev = st->st_dev;
1003: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
1004: sizeof(struct inodev), inodev.st_ino);
1005: mpage = ohash_find(&mpages, slot);
1006: if (NULL == mpage) {
1007: mpage = mandoc_calloc(1, sizeof(struct mpage));
1008: mpage->inodev.st_ino = inodev.st_ino;
1009: mpage->inodev.st_dev = inodev.st_dev;
1.220.2.3 schwarze 1010: mpage->next = mpage_head;
1011: mpage_head = mpage;
1.83 schwarze 1012: ohash_insert(&mpages, slot, mpage);
1013: } else
1.85 schwarze 1014: mlink->next = mpage->mlinks;
1.82 schwarze 1015: mpage->mlinks = mlink;
1.118 schwarze 1016: mlink->mpage = mpage;
1.82 schwarze 1017: }
1018:
1019: static void
1020: mlink_free(struct mlink *mlink)
1021: {
1022:
1023: free(mlink->dsec);
1024: free(mlink->arch);
1025: free(mlink->name);
1026: free(mlink->fsec);
1027: free(mlink);
1.50 kristaps 1028: }
1.3 kristaps 1029:
1.50 kristaps 1030: static void
1.78 schwarze 1031: mpages_free(void)
1.50 kristaps 1032: {
1.78 schwarze 1033: struct mpage *mpage;
1.82 schwarze 1034: struct mlink *mlink;
1.3 kristaps 1035:
1.220.2.3 schwarze 1036: while ((mpage = mpage_head) != NULL) {
1037: while ((mlink = mpage->mlinks) != NULL) {
1.85 schwarze 1038: mpage->mlinks = mlink->next;
1.82 schwarze 1039: mlink_free(mlink);
1040: }
1.220.2.3 schwarze 1041: mpage_head = mpage->next;
1.78 schwarze 1042: free(mpage->sec);
1043: free(mpage->arch);
1.82 schwarze 1044: free(mpage->title);
1045: free(mpage->desc);
1.78 schwarze 1046: free(mpage);
1.50 kristaps 1047: }
1048: }
1.38 schwarze 1049:
1.50 kristaps 1050: /*
1.89 schwarze 1051: * For each mlink to the mpage, check whether the path looks like
1052: * it is formatted, and if it does, check whether a source manual
1053: * exists by the same name, ignoring the suffix.
1054: * If both conditions hold, drop the mlink.
1055: */
1056: static void
1057: mlinks_undupe(struct mpage *mpage)
1058: {
1059: char buf[PATH_MAX];
1060: struct mlink **prev;
1061: struct mlink *mlink;
1062: char *bufp;
1063:
1064: mpage->form = FORM_CAT;
1.90 schwarze 1065: prev = &mpage->mlinks;
1066: while (NULL != (mlink = *prev)) {
1.89 schwarze 1067: if (FORM_CAT != mlink->dform) {
1068: mpage->form = FORM_NONE;
1.90 schwarze 1069: goto nextlink;
1.89 schwarze 1070: }
1.143 schwarze 1071: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1072: bufp = strstr(buf, "cat");
1073: assert(NULL != bufp);
1074: memcpy(bufp, "man", 3);
1075: if (NULL != (bufp = strrchr(buf, '.')))
1076: *++bufp = '\0';
1.143 schwarze 1077: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1078: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1079: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1080: goto nextlink;
1.89 schwarze 1081: if (warnings)
1082: say(mlink->file, "Man source exists: %s", buf);
1083: if (use_all)
1.90 schwarze 1084: goto nextlink;
1.89 schwarze 1085: *prev = mlink->next;
1086: mlink_free(mlink);
1.90 schwarze 1087: continue;
1088: nextlink:
1089: prev = &(*prev)->next;
1.89 schwarze 1090: }
1091: }
1092:
1.131 schwarze 1093: static void
1.93 schwarze 1094: mlink_check(struct mpage *mpage, struct mlink *mlink)
1095: {
1.131 schwarze 1096: struct str *str;
1097: unsigned int slot;
1.93 schwarze 1098:
1099: /*
1100: * Check whether the manual section given in a file
1101: * agrees with the directory where the file is located.
1102: * Some manuals have suffixes like (3p) on their
1103: * section number either inside the file or in the
1104: * directory name, some are linked into more than one
1105: * section, like encrypt(1) = makekey(8).
1106: */
1107:
1108: if (FORM_SRC == mpage->form &&
1.131 schwarze 1109: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1110: say(mlink->file, "Section \"%s\" manual in %s directory",
1111: mpage->sec, mlink->dsec);
1112:
1113: /*
1114: * Manual page directories exist for each kernel
1115: * architecture as returned by machine(1).
1116: * However, many manuals only depend on the
1117: * application architecture as returned by arch(1).
1118: * For example, some (2/ARM) manuals are shared
1119: * across the "armish" and "zaurus" kernel
1120: * architectures.
1121: * A few manuals are even shared across completely
1122: * different architectures, for example fdformat(1)
1123: * on amd64, i386, sparc, and sparc64.
1124: */
1125:
1.131 schwarze 1126: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1127: say(mlink->file, "Architecture \"%s\" manual in "
1128: "\"%s\" directory", mpage->arch, mlink->arch);
1129:
1.131 schwarze 1130: /*
1131: * XXX
1.133 schwarze 1132: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1133: */
1134:
1135: if (FORM_CAT == mpage->form)
1136: return;
1137:
1138: /*
1139: * Check whether this mlink
1140: * appears as a name in the NAME section.
1141: */
1.93 schwarze 1142:
1.133 schwarze 1143: slot = ohash_qlookup(&names, mlink->name);
1144: str = ohash_find(&names, slot);
1.131 schwarze 1145: assert(NULL != str);
1.133 schwarze 1146: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1147: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1148: }
1149:
1.89 schwarze 1150: /*
1.80 schwarze 1151: * Run through the files in the global vector "mpages"
1152: * and add them to the database specified in "basedir".
1.50 kristaps 1153: *
1154: * This handles the parsing scheme itself, using the cues of directory
1155: * and filename to determine whether the file is parsable or not.
1156: */
1.59 schwarze 1157: static void
1.176 schwarze 1158: mpages_merge(struct mparse *mp)
1.50 kristaps 1159: {
1.113 schwarze 1160: char any[] = "any";
1.118 schwarze 1161: struct mpage *mpage, *mpage_dest;
1162: struct mlink *mlink, *mlink_dest;
1.191 schwarze 1163: struct roff_man *man;
1.118 schwarze 1164: char *sodest;
1.112 schwarze 1165: char *cp;
1.162 schwarze 1166: int fd;
1.69 schwarze 1167:
1.171 schwarze 1168: if ( ! nodb)
1.107 schwarze 1169: SQL_EXEC("BEGIN TRANSACTION");
1170:
1.220.2.3 schwarze 1171: for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) {
1.89 schwarze 1172: mlinks_undupe(mpage);
1.220.2.3 schwarze 1173: if ((mlink = mpage->mlinks) == NULL)
1.89 schwarze 1174: continue;
1.50 kristaps 1175:
1.133 schwarze 1176: name_mask = NAME_MASK;
1.202 schwarze 1177: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1178: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.50 kristaps 1179: mparse_reset(mp);
1180: man = NULL;
1.124 schwarze 1181: sodest = NULL;
1182:
1.213 schwarze 1183: if ((fd = mparse_open(mp, mlink->file)) == -1) {
1.182 schwarze 1184: say(mlink->file, "&open");
1.162 schwarze 1185: goto nextpage;
1.124 schwarze 1186: }
1.14 schwarze 1187:
1188: /*
1.183 schwarze 1189: * Interpret the file as mdoc(7) or man(7) source
1190: * code, unless it is known to be formatted.
1.14 schwarze 1191: */
1.182 schwarze 1192: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1193: mparse_readfd(mp, fd, mlink->file);
1.212 schwarze 1194: close(fd);
1.220.2.9 schwarze 1195: fd = -1;
1.192 schwarze 1196: mparse_result(mp, &man, &sodest);
1.76 schwarze 1197: }
1.14 schwarze 1198:
1.171 schwarze 1199: if (sodest != NULL) {
1.118 schwarze 1200: mlink_dest = ohash_find(&mlinks,
1201: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1202: if (mlink_dest == NULL) {
1203: mandoc_asprintf(&cp, "%s.gz", sodest);
1204: mlink_dest = ohash_find(&mlinks,
1205: ohash_qlookup(&mlinks, cp));
1206: free(cp);
1207: }
1208: if (mlink_dest != NULL) {
1.118 schwarze 1209:
1210: /* The .so target exists. */
1211:
1212: mpage_dest = mlink_dest->mpage;
1213: while (1) {
1214: mlink->mpage = mpage_dest;
1215:
1216: /*
1217: * If the target was already
1218: * processed, add the links
1219: * to the database now.
1220: * Otherwise, this will
1221: * happen when we come
1222: * to the target.
1223: */
1224:
1.137 schwarze 1225: if (mpage_dest->pageid)
1.169 schwarze 1226: dbadd_mlink_name(mlink);
1.118 schwarze 1227:
1.171 schwarze 1228: if (mlink->next == NULL)
1.118 schwarze 1229: break;
1230: mlink = mlink->next;
1231: }
1232:
1233: /* Move all links to the target. */
1234:
1235: mlink->next = mlink_dest->next;
1236: mlink_dest->next = mpage->mlinks;
1237: mpage->mlinks = NULL;
1238: }
1.124 schwarze 1239: goto nextpage;
1.192 schwarze 1240: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.204 schwarze 1241: mdoc_validate(man);
1.82 schwarze 1242: mpage->form = FORM_SRC;
1.193 schwarze 1243: mpage->sec = man->meta.msec;
1.155 schwarze 1244: mpage->sec = mandoc_strdup(
1.171 schwarze 1245: mpage->sec == NULL ? "" : mpage->sec);
1.193 schwarze 1246: mpage->arch = man->meta.arch;
1.84 schwarze 1247: mpage->arch = mandoc_strdup(
1.171 schwarze 1248: mpage->arch == NULL ? "" : mpage->arch);
1.193 schwarze 1249: mpage->title = mandoc_strdup(man->meta.title);
1.192 schwarze 1250: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.205 schwarze 1251: man_validate(man);
1.220.2.9 schwarze 1252: if (*man->meta.msec != '\0' ||
1253: *man->meta.msec != '\0') {
1254: mpage->form = FORM_SRC;
1255: mpage->sec = mandoc_strdup(man->meta.msec);
1256: mpage->arch = mandoc_strdup(mlink->arch);
1257: mpage->title = mandoc_strdup(man->meta.title);
1258: } else
1259: man = NULL;
1260: }
1261:
1262: assert(mpage->desc == NULL);
1263: if (man == NULL) {
1.82 schwarze 1264: mpage->form = FORM_CAT;
1.182 schwarze 1265: mpage->sec = mandoc_strdup(mlink->dsec);
1266: mpage->arch = mandoc_strdup(mlink->arch);
1267: mpage->title = mandoc_strdup(mlink->name);
1.220.2.9 schwarze 1268: parse_cat(mpage, fd);
1269: } else if (man->macroset == MACROSET_MDOC)
1270: parse_mdoc(mpage, &man->meta, man->first);
1271: else
1272: parse_man(mpage, &man->meta, man->first);
1273:
1.97 schwarze 1274: putkey(mpage, mpage->sec, TYPE_sec);
1.164 schwarze 1275: if (*mpage->arch != '\0')
1276: putkey(mpage, mpage->arch, TYPE_arch);
1.50 kristaps 1277:
1.182 schwarze 1278: for ( ; mlink != NULL; mlink = mlink->next) {
1.97 schwarze 1279: if ('\0' != *mlink->dsec)
1280: putkey(mpage, mlink->dsec, TYPE_sec);
1281: if ('\0' != *mlink->fsec)
1282: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1283: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1284: any : mlink->arch, TYPE_arch);
1.133 schwarze 1285: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1286: }
1.12 schwarze 1287:
1.174 schwarze 1288: if (mpage->desc == NULL)
1.135 schwarze 1289: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1290:
1291: if (warnings && !use_all)
1292: for (mlink = mpage->mlinks; mlink;
1293: mlink = mlink->next)
1294: mlink_check(mpage, mlink);
1.44 kristaps 1295:
1.176 schwarze 1296: dbadd(mpage);
1.124 schwarze 1297:
1298: nextpage:
1.66 schwarze 1299: ohash_delete(&strings);
1.133 schwarze 1300: ohash_delete(&names);
1.50 kristaps 1301: }
1.107 schwarze 1302:
1303: if (0 == nodb)
1304: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1305: }
1306:
1307: static void
1308: names_check(void)
1309: {
1310: sqlite3_stmt *stmt;
1311: const char *name, *sec, *arch, *key;
1312:
1313: sqlite3_prepare_v2(db,
1314: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1315: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1316: "WHERE bits & ? AND NOT EXISTS ("
1317: "SELECT pageid FROM mlinks "
1.133 schwarze 1318: "WHERE mlinks.pageid == names.pageid "
1319: "AND mlinks.name == names.name"
1.130 schwarze 1320: ")"
1321: ") JOIN ("
1.142 schwarze 1322: "SELECT sec, arch, name, pageid FROM mlinks "
1323: "GROUP BY pageid"
1.130 schwarze 1324: ") USING (pageid);",
1325: -1, &stmt, NULL);
1326:
1.201 schwarze 1327: if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK)
1.134 schwarze 1328: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1329:
1.201 schwarze 1330: while (sqlite3_step(stmt) == SQLITE_ROW) {
1.154 schwarze 1331: name = (const char *)sqlite3_column_text(stmt, 0);
1332: sec = (const char *)sqlite3_column_text(stmt, 1);
1333: arch = (const char *)sqlite3_column_text(stmt, 2);
1334: key = (const char *)sqlite3_column_text(stmt, 3);
1.130 schwarze 1335: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1336: '\0' == *arch ? "" : "/",
1337: '\0' == *arch ? "" : arch, key);
1338: }
1339: sqlite3_finalize(stmt);
1.50 kristaps 1340: }
1.12 schwarze 1341:
1.50 kristaps 1342: static void
1.124 schwarze 1343: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1344: {
1345: FILE *stream;
1.220.2.10! schwarze 1346: struct mlink *mlink;
! 1347: char *line, *p, *title, *sec;
1.209 schwarze 1348: size_t linesz, plen, titlesz;
1349: ssize_t len;
1350: int offs;
1.12 schwarze 1351:
1.220.2.10! schwarze 1352: mlink = mpage->mlinks;
! 1353: stream = fd == -1 ? fopen(mlink->file, "r") : fdopen(fd, "r");
! 1354: if (stream == NULL) {
! 1355: if (fd != -1)
1.159 schwarze 1356: close(fd);
1.56 schwarze 1357: if (warnings)
1.220.2.10! schwarze 1358: say(mlink->file, "&fopen");
1.50 kristaps 1359: return;
1360: }
1.1 kristaps 1361:
1.209 schwarze 1362: line = NULL;
1363: linesz = 0;
1364:
1.220.2.10! schwarze 1365: /* Parse the section number from the header line. */
1.1 kristaps 1366:
1.220.2.10! schwarze 1367: while (getline(&line, &linesz, stream) != -1) {
1.209 schwarze 1368: if (*line == '\n')
1.220.2.10! schwarze 1369: continue;
! 1370: if ((sec = strchr(line, '(')) == NULL)
! 1371: break;
! 1372: if ((p = strchr(++sec, ')')) == NULL)
! 1373: break;
! 1374: free(mpage->sec);
! 1375: mpage->sec = mandoc_strndup(sec, p - sec);
! 1376: if (warnings && *mlink->dsec != '\0' &&
! 1377: strcasecmp(mpage->sec, mlink->dsec))
! 1378: say(mlink->file,
! 1379: "Section \"%s\" manual in %s directory",
! 1380: mpage->sec, mlink->dsec);
! 1381: break;
! 1382: }
! 1383:
! 1384: /* Skip to first blank line. */
! 1385:
! 1386: while (line == NULL || *line != '\n')
! 1387: if (getline(&line, &linesz, stream) == -1)
1.50 kristaps 1388: break;
1.1 kristaps 1389:
1.50 kristaps 1390: /*
1391: * Assume the first line that is not indented
1392: * is the first section header. Skip to it.
1393: */
1.1 kristaps 1394:
1.209 schwarze 1395: while (getline(&line, &linesz, stream) != -1)
1396: if (*line != '\n' && *line != ' ')
1.50 kristaps 1397: break;
1.141 schwarze 1398:
1.50 kristaps 1399: /*
1400: * Read up until the next section into a buffer.
1401: * Strip the leading and trailing newline from each read line,
1402: * appending a trailing space.
1403: * Ignore empty (whitespace-only) lines.
1404: */
1.1 kristaps 1405:
1.50 kristaps 1406: titlesz = 0;
1407: title = NULL;
1.38 schwarze 1408:
1.209 schwarze 1409: while ((len = getline(&line, &linesz, stream)) != -1) {
1410: if (*line != ' ')
1.50 kristaps 1411: break;
1.209 schwarze 1412: offs = 0;
1413: while (isspace((unsigned char)line[offs]))
1414: offs++;
1415: if (line[offs] == '\0')
1.38 schwarze 1416: continue;
1.209 schwarze 1417: title = mandoc_realloc(title, titlesz + len - offs);
1418: memcpy(title + titlesz, line + offs, len - offs);
1419: titlesz += len - offs;
1.50 kristaps 1420: title[titlesz - 1] = ' ';
1421: }
1.209 schwarze 1422: free(line);
1.38 schwarze 1423:
1.50 kristaps 1424: /*
1425: * If no page content can be found, or the input line
1426: * is already the next section header, or there is no
1427: * trailing newline, reuse the page title as the page
1428: * description.
1429: */
1.44 kristaps 1430:
1.50 kristaps 1431: if (NULL == title || '\0' == *title) {
1.56 schwarze 1432: if (warnings)
1.220.2.10! schwarze 1433: say(mlink->file, "Cannot find NAME section");
1.50 kristaps 1434: fclose(stream);
1435: free(title);
1436: return;
1437: }
1.1 kristaps 1438:
1.209 schwarze 1439: title[titlesz - 1] = '\0';
1.33 schwarze 1440:
1.50 kristaps 1441: /*
1442: * Skip to the first dash.
1443: * Use the remaining line as the description (no more than 70
1444: * bytes).
1445: */
1.33 schwarze 1446:
1.50 kristaps 1447: if (NULL != (p = strstr(title, "- "))) {
1448: for (p += 2; ' ' == *p || '\b' == *p; p++)
1449: /* Skip to next word. */ ;
1450: } else {
1.56 schwarze 1451: if (warnings)
1.220.2.10! schwarze 1452: say(mlink->file, "No dash in title line");
1.50 kristaps 1453: p = title;
1454: }
1.38 schwarze 1455:
1.50 kristaps 1456: plen = strlen(p);
1.1 kristaps 1457:
1.50 kristaps 1458: /* Strip backspace-encoding from line. */
1.1 kristaps 1459:
1.50 kristaps 1460: while (NULL != (line = memchr(p, '\b', plen))) {
1461: len = line - p;
1462: if (0 == len) {
1463: memmove(line, line + 1, plen--);
1464: continue;
1.141 schwarze 1465: }
1.50 kristaps 1466: memmove(line - 1, line + 1, plen - len);
1467: plen -= 2;
1468: }
1.1 kristaps 1469:
1.78 schwarze 1470: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1471: fclose(stream);
1472: free(title);
1473: }
1.1 kristaps 1474:
1.50 kristaps 1475: /*
1476: * Put a type/word pair into the word database for this particular file.
1477: */
1478: static void
1.112 schwarze 1479: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1480: {
1.112 schwarze 1481: char *cp;
1.18 kristaps 1482:
1.50 kristaps 1483: assert(NULL != value);
1.112 schwarze 1484: if (TYPE_arch == type)
1485: for (cp = value; *cp; cp++)
1486: if (isupper((unsigned char)*cp))
1487: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1488: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1489: }
1490:
1491: /*
1.50 kristaps 1492: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1493: */
1494: static void
1.78 schwarze 1495: putmdockey(const struct mpage *mpage,
1.220.2.5 schwarze 1496: const struct roff_node *n, uint64_t m, int taboo)
1.3 kristaps 1497: {
1.18 kristaps 1498:
1.50 kristaps 1499: for ( ; NULL != n; n = n->next) {
1.220.2.5 schwarze 1500: if (n->flags & taboo)
1501: continue;
1.50 kristaps 1502: if (NULL != n->child)
1.220.2.5 schwarze 1503: putmdockey(mpage, n->child, m, taboo);
1.188 schwarze 1504: if (n->type == ROFFT_TEXT)
1.78 schwarze 1505: putkey(mpage, n->string, m);
1.50 kristaps 1506: }
1507: }
1.18 kristaps 1508:
1.61 schwarze 1509: static void
1.190 schwarze 1510: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1511: const struct roff_node *n)
1.50 kristaps 1512: {
1.189 schwarze 1513: const struct roff_node *head, *body;
1.121 schwarze 1514: char *start, *title;
1.50 kristaps 1515: char byte;
1.121 schwarze 1516: size_t sz;
1.18 kristaps 1517:
1.215 schwarze 1518: if (n == NULL)
1.61 schwarze 1519: return;
1.18 kristaps 1520:
1.50 kristaps 1521: /*
1522: * We're only searching for one thing: the first text child in
1523: * the BODY of a NAME section. Since we don't keep track of
1524: * sections in -man, run some hoops to find out whether we're in
1525: * the correct section or not.
1526: */
1.18 kristaps 1527:
1.188 schwarze 1528: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.50 kristaps 1529: body = n;
1.215 schwarze 1530: if ((head = body->parent->head) != NULL &&
1531: (head = head->child) != NULL &&
1532: head->next == NULL &&
1.188 schwarze 1533: head->type == ROFFT_TEXT &&
1.215 schwarze 1534: strcmp(head->string, "NAME") == 0 &&
1535: body->child != NULL) {
1.3 kristaps 1536:
1.50 kristaps 1537: /*
1538: * Suck the entire NAME section into memory.
1539: * Yes, we might run away.
1540: * But too many manuals have big, spread-out
1541: * NAME sections over many lines.
1542: */
1.3 kristaps 1543:
1.121 schwarze 1544: title = NULL;
1.194 schwarze 1545: deroff(&title, body);
1.50 kristaps 1546: if (NULL == title)
1.61 schwarze 1547: return;
1.18 kristaps 1548:
1.141 schwarze 1549: /*
1.50 kristaps 1550: * Go through a special heuristic dance here.
1551: * Conventionally, one or more manual names are
1552: * comma-specified prior to a whitespace, then a
1553: * dash, then a description. Try to puzzle out
1554: * the name parts here.
1555: */
1.18 kristaps 1556:
1.121 schwarze 1557: start = title;
1.50 kristaps 1558: for ( ;; ) {
1559: sz = strcspn(start, " ,");
1560: if ('\0' == start[sz])
1561: break;
1.1 kristaps 1562:
1.50 kristaps 1563: byte = start[sz];
1564: start[sz] = '\0';
1.110 schwarze 1565:
1566: /*
1567: * Assume a stray trailing comma in the
1568: * name list if a name begins with a dash.
1569: */
1570:
1571: if ('-' == start[0] ||
1572: ('\\' == start[0] && '-' == start[1]))
1573: break;
1.1 kristaps 1574:
1.133 schwarze 1575: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1576: if ( ! (mpage->name_head_done ||
1577: strcasecmp(start, meta->title))) {
1578: putkey(mpage, start, NAME_HEAD);
1579: mpage->name_head_done = 1;
1580: }
1.1 kristaps 1581:
1.50 kristaps 1582: if (' ' == byte) {
1583: start += sz + 1;
1584: break;
1585: }
1.1 kristaps 1586:
1.50 kristaps 1587: assert(',' == byte);
1588: start += sz + 1;
1589: while (' ' == *start)
1590: start++;
1591: }
1.1 kristaps 1592:
1.121 schwarze 1593: if (start == title) {
1.133 schwarze 1594: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1595: if ( ! (mpage->name_head_done ||
1596: strcasecmp(start, meta->title))) {
1597: putkey(mpage, start, NAME_HEAD);
1598: mpage->name_head_done = 1;
1599: }
1.50 kristaps 1600: free(title);
1.61 schwarze 1601: return;
1.50 kristaps 1602: }
1.1 kristaps 1603:
1.50 kristaps 1604: while (isspace((unsigned char)*start))
1605: start++;
1.1 kristaps 1606:
1.50 kristaps 1607: if (0 == strncmp(start, "-", 1))
1608: start += 1;
1609: else if (0 == strncmp(start, "\\-\\-", 4))
1610: start += 4;
1611: else if (0 == strncmp(start, "\\-", 2))
1612: start += 2;
1613: else if (0 == strncmp(start, "\\(en", 4))
1614: start += 4;
1615: else if (0 == strncmp(start, "\\(em", 4))
1616: start += 4;
1.1 kristaps 1617:
1.50 kristaps 1618: while (' ' == *start)
1619: start++;
1.1 kristaps 1620:
1.78 schwarze 1621: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1622: free(title);
1.61 schwarze 1623: return;
1.50 kristaps 1624: }
1625: }
1.1 kristaps 1626:
1.77 schwarze 1627: for (n = n->child; n; n = n->next) {
1.78 schwarze 1628: if (NULL != mpage->desc)
1.77 schwarze 1629: break;
1.174 schwarze 1630: parse_man(mpage, meta, n);
1.77 schwarze 1631: }
1.1 kristaps 1632: }
1633:
1634: static void
1.190 schwarze 1635: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1636: const struct roff_node *n)
1.1 kristaps 1637: {
1638:
1.50 kristaps 1639: assert(NULL != n);
1640: for (n = n->child; NULL != n; n = n->next) {
1.220.2.5 schwarze 1641: if (n->flags & mdocs[n->tok].taboo)
1642: continue;
1.50 kristaps 1643: switch (n->type) {
1.188 schwarze 1644: case ROFFT_ELEM:
1645: case ROFFT_BLOCK:
1646: case ROFFT_HEAD:
1647: case ROFFT_BODY:
1648: case ROFFT_TAIL:
1.50 kristaps 1649: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1650: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1651: break;
1.68 schwarze 1652: if (mdocs[n->tok].mask)
1.78 schwarze 1653: putmdockey(mpage, n->child,
1.220.2.5 schwarze 1654: mdocs[n->tok].mask, mdocs[n->tok].taboo);
1.50 kristaps 1655: break;
1656: default:
1.188 schwarze 1657: assert(n->type != ROFFT_ROOT);
1.50 kristaps 1658: continue;
1659: }
1660: if (NULL != n->child)
1.172 schwarze 1661: parse_mdoc(mpage, meta, n);
1.1 kristaps 1662: }
1663: }
1664:
1.25 schwarze 1665: static int
1.190 schwarze 1666: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1667: const struct roff_node *n)
1.1 kristaps 1668: {
1.176 schwarze 1669: char *start, *end;
1.1 kristaps 1670: size_t sz;
1.25 schwarze 1671:
1.50 kristaps 1672: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1673: NULL == (n = n->child) ||
1.188 schwarze 1674: n->type != ROFFT_TEXT)
1.197 schwarze 1675: return 0;
1.1 kristaps 1676:
1677: /*
1678: * Only consider those `Fd' macro fields that begin with an
1679: * "inclusion" token (versus, e.g., #define).
1680: */
1.50 kristaps 1681:
1.1 kristaps 1682: if (strcmp("#include", n->string))
1.197 schwarze 1683: return 0;
1.1 kristaps 1684:
1.188 schwarze 1685: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1686: return 0;
1.1 kristaps 1687:
1688: /*
1689: * Strip away the enclosing angle brackets and make sure we're
1690: * not zero-length.
1691: */
1692:
1693: start = n->string;
1694: if ('<' == *start || '"' == *start)
1695: start++;
1696:
1697: if (0 == (sz = strlen(start)))
1.197 schwarze 1698: return 0;
1.1 kristaps 1699:
1700: end = &start[(int)sz - 1];
1701: if ('>' == *end || '"' == *end)
1702: end--;
1703:
1.50 kristaps 1704: if (end > start)
1.78 schwarze 1705: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1706: return 0;
1.1 kristaps 1707: }
1708:
1.178 schwarze 1709: static void
1.189 schwarze 1710: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1711: {
1.112 schwarze 1712: char *cp;
1.178 schwarze 1713: size_t sz;
1.1 kristaps 1714:
1.188 schwarze 1715: if (n->type != ROFFT_TEXT)
1.178 schwarze 1716: return;
1.25 schwarze 1717:
1.178 schwarze 1718: /* Skip function pointer punctuation. */
1.1 kristaps 1719:
1.178 schwarze 1720: cp = n->string;
1721: while (*cp == '(' || *cp == '*')
1.1 kristaps 1722: cp++;
1.178 schwarze 1723: sz = strcspn(cp, "()");
1.1 kristaps 1724:
1.178 schwarze 1725: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1726: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1727: putkeys(mpage, cp, sz, NAME_SYN);
1728: }
1.25 schwarze 1729:
1.178 schwarze 1730: static int
1.190 schwarze 1731: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1732: const struct roff_node *n)
1.178 schwarze 1733: {
1734:
1735: if (n->child == NULL)
1.197 schwarze 1736: return 0;
1.178 schwarze 1737:
1738: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1739:
1.178 schwarze 1740: for (n = n->child->next; n != NULL; n = n->next)
1.188 schwarze 1741: if (n->type == ROFFT_TEXT)
1.78 schwarze 1742: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1743:
1.197 schwarze 1744: return 0;
1.173 schwarze 1745: }
1746:
1747: static int
1.190 schwarze 1748: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1749: const struct roff_node *n)
1.173 schwarze 1750: {
1.177 schwarze 1751:
1.188 schwarze 1752: if (n->type != ROFFT_HEAD)
1.197 schwarze 1753: return 1;
1.173 schwarze 1754:
1.178 schwarze 1755: if (n->child != NULL)
1756: parse_mdoc_fname(mpage, n->child);
1757:
1.197 schwarze 1758: return 0;
1.1 kristaps 1759: }
1760:
1.25 schwarze 1761: static int
1.211 schwarze 1762: parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1763: const struct roff_node *n)
1764: {
1765: char *cp;
1766:
1767: if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1768: return 0;
1769:
1.215 schwarze 1770: if (n->child != NULL &&
1771: n->child->next == NULL &&
1772: n->child->type == ROFFT_TEXT)
1.211 schwarze 1773: return 1;
1774:
1775: cp = NULL;
1776: deroff(&cp, n);
1777: if (cp != NULL) {
1778: putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1779: n->type == ROFFT_BODY ? TYPE_Va : 0));
1780: free(cp);
1781: }
1782:
1783: return 0;
1784: }
1785:
1786: static int
1.190 schwarze 1787: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1788: const struct roff_node *n)
1.1 kristaps 1789: {
1.67 schwarze 1790: char *cp;
1.1 kristaps 1791:
1792: if (NULL == (n = n->child))
1.197 schwarze 1793: return 0;
1.1 kristaps 1794:
1.67 schwarze 1795: if (NULL == n->next) {
1.78 schwarze 1796: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1797: return 0;
1.67 schwarze 1798: }
1799:
1.120 schwarze 1800: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1801: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1802: free(cp);
1.197 schwarze 1803: return 0;
1.1 kristaps 1804: }
1805:
1.25 schwarze 1806: static int
1.190 schwarze 1807: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1808: const struct roff_node *n)
1.1 kristaps 1809: {
1810:
1.188 schwarze 1811: if (n->type == ROFFT_BODY)
1.194 schwarze 1812: deroff(&mpage->desc, n);
1.197 schwarze 1813: return 0;
1.1 kristaps 1814: }
1815:
1.25 schwarze 1816: static int
1.190 schwarze 1817: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1818: const struct roff_node *n)
1.1 kristaps 1819: {
1820:
1.129 schwarze 1821: if (SEC_NAME == n->sec)
1.220.2.5 schwarze 1822: putmdockey(mpage, n->child, NAME_TITLE, 0);
1.188 schwarze 1823: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1824: if (n->child == NULL)
1825: putkey(mpage, meta->name, NAME_SYN);
1826: else
1.220.2.5 schwarze 1827: putmdockey(mpage, n->child, NAME_SYN, 0);
1.174 schwarze 1828: }
1829: if ( ! (mpage->name_head_done ||
1830: n->child == NULL || n->child->string == NULL ||
1831: strcasecmp(n->child->string, meta->title))) {
1.220.2.1 schwarze 1832: putkey(mpage, n->child->string, NAME_HEAD);
1.174 schwarze 1833: mpage->name_head_done = 1;
1.172 schwarze 1834: }
1.197 schwarze 1835: return 0;
1.1 kristaps 1836: }
1837:
1.25 schwarze 1838: static int
1.190 schwarze 1839: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1840: const struct roff_node *n)
1.1 kristaps 1841: {
1842:
1.197 schwarze 1843: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1844: }
1845:
1.50 kristaps 1846: static int
1.190 schwarze 1847: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1848: const struct roff_node *n)
1.1 kristaps 1849: {
1850:
1.197 schwarze 1851: return n->type == ROFFT_HEAD;
1.1 kristaps 1852: }
1853:
1.50 kristaps 1854: /*
1.66 schwarze 1855: * Add a string to the hash table for the current manual.
1856: * Each string has a bitmask telling which macros it belongs to.
1857: * When we finish the manual, we'll dump the table.
1.50 kristaps 1858: */
1859: static void
1.176 schwarze 1860: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1861: {
1.133 schwarze 1862: struct ohash *htab;
1.50 kristaps 1863: struct str *s;
1.111 schwarze 1864: const char *end;
1.65 schwarze 1865: unsigned int slot;
1.176 schwarze 1866: int i, mustfree;
1.25 schwarze 1867:
1.50 kristaps 1868: if (0 == sz)
1869: return;
1.111 schwarze 1870:
1.176 schwarze 1871: mustfree = render_string(&cp, &sz);
1872:
1.133 schwarze 1873: if (TYPE_Nm & v) {
1874: htab = &names;
1875: v &= name_mask;
1.169 schwarze 1876: if (v & NAME_FIRST)
1877: name_mask &= ~NAME_FIRST;
1.133 schwarze 1878: if (debug > 1)
1879: say(mpage->mlinks->file,
1.220.2.4 schwarze 1880: "Adding name %*s, bits=0x%llx", (int)sz, cp,
1881: (unsigned long long)v);
1.133 schwarze 1882: } else {
1883: htab = &strings;
1884: if (debug > 1)
1885: for (i = 0; i < mansearch_keymax; i++)
1.163 schwarze 1886: if ((uint64_t)1 << i & v)
1.133 schwarze 1887: say(mpage->mlinks->file,
1888: "Adding key %s=%*s",
1.220 schwarze 1889: mansearch_keynames[i], (int)sz, cp);
1.111 schwarze 1890: }
1.25 schwarze 1891:
1.65 schwarze 1892: end = cp + sz;
1.133 schwarze 1893: slot = ohash_qlookupi(htab, cp, &end);
1894: s = ohash_find(htab, slot);
1.25 schwarze 1895:
1.78 schwarze 1896: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1897: s->mask |= v;
1898: return;
1899: } else if (NULL == s) {
1.144 schwarze 1900: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1901: memcpy(s->key, cp, sz);
1.133 schwarze 1902: ohash_insert(htab, slot, s);
1.1 kristaps 1903: }
1.78 schwarze 1904: s->mpage = mpage;
1.50 kristaps 1905: s->mask = v;
1.176 schwarze 1906:
1907: if (mustfree)
1908: free(cp);
1.1 kristaps 1909: }
1910:
1.50 kristaps 1911: /*
1912: * Take a Unicode codepoint and produce its UTF-8 encoding.
1913: * This isn't the best way to do this, but it works.
1914: * The magic numbers are from the UTF-8 packaging.
1915: * They're not as scary as they seem: read the UTF-8 spec for details.
1916: */
1917: static size_t
1918: utf8(unsigned int cp, char out[7])
1.1 kristaps 1919: {
1.50 kristaps 1920: size_t rc;
1.1 kristaps 1921:
1.50 kristaps 1922: rc = 0;
1923: if (cp <= 0x0000007F) {
1924: rc = 1;
1925: out[0] = (char)cp;
1926: } else if (cp <= 0x000007FF) {
1927: rc = 2;
1928: out[0] = (cp >> 6 & 31) | 192;
1929: out[1] = (cp & 63) | 128;
1930: } else if (cp <= 0x0000FFFF) {
1931: rc = 3;
1932: out[0] = (cp >> 12 & 15) | 224;
1933: out[1] = (cp >> 6 & 63) | 128;
1934: out[2] = (cp & 63) | 128;
1935: } else if (cp <= 0x001FFFFF) {
1936: rc = 4;
1937: out[0] = (cp >> 18 & 7) | 240;
1938: out[1] = (cp >> 12 & 63) | 128;
1939: out[2] = (cp >> 6 & 63) | 128;
1940: out[3] = (cp & 63) | 128;
1941: } else if (cp <= 0x03FFFFFF) {
1942: rc = 5;
1943: out[0] = (cp >> 24 & 3) | 248;
1944: out[1] = (cp >> 18 & 63) | 128;
1945: out[2] = (cp >> 12 & 63) | 128;
1946: out[3] = (cp >> 6 & 63) | 128;
1947: out[4] = (cp & 63) | 128;
1948: } else if (cp <= 0x7FFFFFFF) {
1949: rc = 6;
1950: out[0] = (cp >> 30 & 1) | 252;
1951: out[1] = (cp >> 24 & 63) | 128;
1952: out[2] = (cp >> 18 & 63) | 128;
1953: out[3] = (cp >> 12 & 63) | 128;
1954: out[4] = (cp >> 6 & 63) | 128;
1955: out[5] = (cp & 63) | 128;
1956: } else
1.197 schwarze 1957: return 0;
1.1 kristaps 1958:
1.50 kristaps 1959: out[rc] = '\0';
1.197 schwarze 1960: return rc;
1.50 kristaps 1961: }
1.1 kristaps 1962:
1.50 kristaps 1963: /*
1.176 schwarze 1964: * If the string contains escape sequences,
1965: * replace it with an allocated rendering and return 1,
1966: * such that the caller can free it after use.
1967: * Otherwise, do nothing and return 0.
1.50 kristaps 1968: */
1.176 schwarze 1969: static int
1970: render_string(char **public, size_t *psz)
1.50 kristaps 1971: {
1.176 schwarze 1972: const char *src, *scp, *addcp, *seq;
1973: char *dst;
1974: size_t ssz, dsz, addsz;
1.114 schwarze 1975: char utfbuf[7], res[6];
1.176 schwarze 1976: int seqlen, unicode;
1.50 kristaps 1977:
1978: res[0] = '\\';
1979: res[1] = '\t';
1980: res[2] = ASCII_NBRSP;
1981: res[3] = ASCII_HYPH;
1.114 schwarze 1982: res[4] = ASCII_BREAK;
1983: res[5] = '\0';
1.1 kristaps 1984:
1.176 schwarze 1985: src = scp = *public;
1986: ssz = *psz;
1987: dst = NULL;
1988: dsz = 0;
1989:
1990: while (scp < src + *psz) {
1991:
1992: /* Leave normal characters unchanged. */
1993:
1994: if (strchr(res, *scp) == NULL) {
1995: if (dst != NULL)
1996: dst[dsz++] = *scp;
1997: scp++;
1998: continue;
1999: }
1.46 kristaps 2000:
1.50 kristaps 2001: /*
1.176 schwarze 2002: * Found something that requires replacing,
2003: * make sure we have a destination buffer.
1.50 kristaps 2004: */
1.176 schwarze 2005:
2006: if (dst == NULL) {
2007: dst = mandoc_malloc(ssz + 1);
2008: dsz = scp - src;
2009: memcpy(dst, src, dsz);
1.50 kristaps 2010: }
1.46 kristaps 2011:
1.176 schwarze 2012: /* Handle single-char special characters. */
2013:
2014: switch (*scp) {
2015: case '\\':
2016: break;
1.141 schwarze 2017: case '\t':
2018: case ASCII_NBRSP:
1.176 schwarze 2019: dst[dsz++] = ' ';
2020: scp++;
2021: continue;
2022: case ASCII_HYPH:
2023: dst[dsz++] = '-';
1.114 schwarze 2024: /* FALLTHROUGH */
1.141 schwarze 2025: case ASCII_BREAK:
1.176 schwarze 2026: scp++;
1.50 kristaps 2027: continue;
1.114 schwarze 2028: default:
1.176 schwarze 2029: abort();
1.114 schwarze 2030: }
1.46 kristaps 2031:
1.50 kristaps 2032: /*
1.176 schwarze 2033: * Found an escape sequence.
2034: * Read past the slash, then parse it.
2035: * Ignore everything except characters.
1.50 kristaps 2036: */
1.95 schwarze 2037:
1.176 schwarze 2038: scp++;
2039: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 2040: continue;
1.1 kristaps 2041:
1.50 kristaps 2042: /*
1.95 schwarze 2043: * Render the special character
2044: * as either UTF-8 or ASCII.
1.50 kristaps 2045: */
1.95 schwarze 2046:
2047: if (write_utf8) {
1.203 schwarze 2048: unicode = mchars_spec2cp(seq, seqlen);
1.176 schwarze 2049: if (unicode <= 0)
1.95 schwarze 2050: continue;
1.176 schwarze 2051: addsz = utf8(unicode, utfbuf);
2052: if (addsz == 0)
1.95 schwarze 2053: continue;
1.176 schwarze 2054: addcp = utfbuf;
1.95 schwarze 2055: } else {
1.203 schwarze 2056: addcp = mchars_spec2str(seq, seqlen, &addsz);
1.176 schwarze 2057: if (addcp == NULL)
1.95 schwarze 2058: continue;
1.176 schwarze 2059: if (*addcp == ASCII_NBRSP) {
2060: addcp = " ";
2061: addsz = 1;
1.95 schwarze 2062: }
2063: }
1.1 kristaps 2064:
1.50 kristaps 2065: /* Copy the rendered glyph into the stream. */
1.1 kristaps 2066:
1.176 schwarze 2067: ssz += addsz;
2068: dst = mandoc_realloc(dst, ssz + 1);
2069: memcpy(dst + dsz, addcp, addsz);
2070: dsz += addsz;
2071: }
2072: if (dst != NULL) {
2073: *public = dst;
2074: *psz = dsz;
2075: }
2076:
2077: /* Trim trailing whitespace and NUL-terminate. */
2078:
2079: while (*psz > 0 && (*public)[*psz - 1] == ' ')
2080: --*psz;
2081: if (dst != NULL) {
2082: (*public)[*psz] = '\0';
1.197 schwarze 2083: return 1;
1.176 schwarze 2084: } else
1.197 schwarze 2085: return 0;
1.1 kristaps 2086: }
2087:
1.118 schwarze 2088: static void
2089: dbadd_mlink(const struct mlink *mlink)
2090: {
2091: size_t i;
2092:
2093: i = 1;
2094: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2095: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2096: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 2097: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 2098: SQL_STEP(stmts[STMT_INSERT_LINK]);
2099: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.169 schwarze 2100: }
2101:
2102: static void
2103: dbadd_mlink_name(const struct mlink *mlink)
2104: {
1.175 schwarze 2105: uint64_t bits;
1.169 schwarze 2106: size_t i;
2107:
2108: dbadd_mlink(mlink);
1.160 schwarze 2109:
2110: i = 1;
1.175 schwarze 2111: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2112: bits = NAME_FILE & NAME_MASK;
2113: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2114: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2115: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2116: }
2117:
2118: i = 1;
2119: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.160 schwarze 2120: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2121: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2122: SQL_STEP(stmts[STMT_INSERT_NAME]);
2123: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.118 schwarze 2124: }
2125:
1.14 schwarze 2126: /*
1.50 kristaps 2127: * Flush the current page's terms (and their bits) into the database.
2128: * Wrap the entire set of additions in a transaction to make sqlite be a
2129: * little faster.
1.96 schwarze 2130: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2131: */
2132: static void
1.176 schwarze 2133: dbadd(struct mpage *mpage)
1.14 schwarze 2134: {
1.87 schwarze 2135: struct mlink *mlink;
1.50 kristaps 2136: struct str *key;
1.176 schwarze 2137: char *cp;
1.52 kristaps 2138: size_t i;
1.66 schwarze 2139: unsigned int slot;
1.176 schwarze 2140: int mustfree;
1.14 schwarze 2141:
1.128 schwarze 2142: mlink = mpage->mlinks;
1.43 kristaps 2143:
1.128 schwarze 2144: if (nodb) {
1.147 schwarze 2145: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2146: key = ohash_next(&names, &slot))
1.147 schwarze 2147: free(key);
2148: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2149: key = ohash_next(&strings, &slot))
1.147 schwarze 2150: free(key);
1.145 schwarze 2151: if (0 == debug)
2152: return;
1.128 schwarze 2153: while (NULL != mlink) {
2154: fputs(mlink->name, stdout);
2155: if (NULL == mlink->next ||
2156: strcmp(mlink->dsec, mlink->next->dsec) ||
2157: strcmp(mlink->fsec, mlink->next->fsec) ||
2158: strcmp(mlink->arch, mlink->next->arch)) {
2159: putchar('(');
2160: if ('\0' == *mlink->dsec)
2161: fputs(mlink->fsec, stdout);
2162: else
2163: fputs(mlink->dsec, stdout);
2164: if ('\0' != *mlink->arch)
2165: printf("/%s", mlink->arch);
2166: putchar(')');
2167: }
2168: mlink = mlink->next;
2169: if (NULL != mlink)
2170: fputs(", ", stdout);
2171: }
1.132 schwarze 2172: printf(" - %s\n", mpage->desc);
1.14 schwarze 2173: return;
1.128 schwarze 2174: }
2175:
2176: if (debug)
2177: say(mlink->file, "Adding to database");
1.28 kristaps 2178:
1.176 schwarze 2179: cp = mpage->desc;
2180: i = strlen(cp);
2181: mustfree = render_string(&cp, &i);
1.52 kristaps 2182: i = 1;
1.176 schwarze 2183: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.161 schwarze 2184: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.81 schwarze 2185: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2186: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2187: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.176 schwarze 2188: if (mustfree)
2189: free(cp);
1.81 schwarze 2190:
1.128 schwarze 2191: while (NULL != mlink) {
1.118 schwarze 2192: dbadd_mlink(mlink);
1.128 schwarze 2193: mlink = mlink->next;
2194: }
1.134 schwarze 2195: mlink = mpage->mlinks;
1.50 kristaps 2196:
1.133 schwarze 2197: for (key = ohash_first(&names, &slot); NULL != key;
2198: key = ohash_next(&names, &slot)) {
2199: assert(key->mpage == mpage);
2200: i = 1;
2201: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.176 schwarze 2202: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.137 schwarze 2203: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2204: SQL_STEP(stmts[STMT_INSERT_NAME]);
2205: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2206: free(key);
2207: }
1.66 schwarze 2208: for (key = ohash_first(&strings, &slot); NULL != key;
2209: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2210: assert(key->mpage == mpage);
1.52 kristaps 2211: i = 1;
2212: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.176 schwarze 2213: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.137 schwarze 2214: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2215: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2216: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 2217: free(key);
1.38 schwarze 2218: }
1.14 schwarze 2219: }
2220:
1.5 kristaps 2221: static void
1.59 schwarze 2222: dbprune(void)
1.5 kristaps 2223: {
1.78 schwarze 2224: struct mpage *mpage;
1.82 schwarze 2225: struct mlink *mlink;
1.52 kristaps 2226: size_t i;
1.80 schwarze 2227: unsigned int slot;
1.5 kristaps 2228:
1.106 schwarze 2229: if (0 == nodb)
2230: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2231:
1.106 schwarze 2232: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2233: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2234: mlink = mpage->mlinks;
1.125 schwarze 2235: if (debug)
1.106 schwarze 2236: say(mlink->file, "Deleting from database");
2237: if (nodb)
2238: continue;
2239: for ( ; NULL != mlink; mlink = mlink->next) {
2240: i = 1;
2241: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2242: i, mlink->dsec);
2243: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2244: i, mlink->arch);
2245: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2246: i, mlink->name);
2247: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2248: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2249: }
1.5 kristaps 2250: }
1.106 schwarze 2251:
2252: if (0 == nodb)
2253: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2254: }
2255:
1.4 kristaps 2256: /*
1.50 kristaps 2257: * Close an existing database and its prepared statements.
2258: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2259: */
1.35 kristaps 2260: static void
1.59 schwarze 2261: dbclose(int real)
1.4 kristaps 2262: {
1.50 kristaps 2263: size_t i;
1.115 schwarze 2264: int status;
2265: pid_t child;
1.4 kristaps 2266:
1.50 kristaps 2267: if (nodb)
1.38 schwarze 2268: return;
1.50 kristaps 2269:
2270: for (i = 0; i < STMT__MAX; i++) {
2271: sqlite3_finalize(stmts[i]);
2272: stmts[i] = NULL;
1.4 kristaps 2273: }
2274:
1.50 kristaps 2275: sqlite3_close(db);
2276: db = NULL;
1.12 schwarze 2277:
1.50 kristaps 2278: if (real)
2279: return;
1.12 schwarze 2280:
1.115 schwarze 2281: if ('\0' == *tempfilename) {
2282: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2283: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2284: say(MANDOC_DB, "&rename");
1.115 schwarze 2285: }
2286: return;
2287: }
2288:
2289: switch (child = fork()) {
1.141 schwarze 2290: case -1:
1.115 schwarze 2291: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2292: say("", "&fork cmp");
1.115 schwarze 2293: return;
1.141 schwarze 2294: case 0:
1.115 schwarze 2295: execlp("cmp", "cmp", "-s",
1.196 schwarze 2296: tempfilename, MANDOC_DB, (char *)NULL);
1.123 schwarze 2297: say("", "&exec cmp");
1.115 schwarze 2298: exit(0);
2299: default:
2300: break;
2301: }
2302: if (-1 == waitpid(child, &status, 0)) {
2303: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2304: say("", "&wait cmp");
1.115 schwarze 2305: } else if (WIFSIGNALED(status)) {
2306: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2307: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2308: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2309: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2310: say(MANDOC_DB,
2311: "Data changed, but cannot replace database");
2312: }
2313:
2314: *strrchr(tempfilename, '/') = '\0';
2315: switch (child = fork()) {
1.141 schwarze 2316: case -1:
1.115 schwarze 2317: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2318: say("", "&fork rm");
1.115 schwarze 2319: return;
1.141 schwarze 2320: case 0:
1.196 schwarze 2321: execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
1.123 schwarze 2322: say("", "&exec rm");
1.115 schwarze 2323: exit((int)MANDOCLEVEL_SYSERR);
2324: default:
2325: break;
2326: }
2327: if (-1 == waitpid(child, &status, 0)) {
2328: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2329: say("", "&wait rm");
1.115 schwarze 2330: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2331: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2332: say("", "%s: Cannot remove temporary directory",
2333: tempfilename);
1.59 schwarze 2334: }
1.50 kristaps 2335: }
1.14 schwarze 2336:
1.50 kristaps 2337: /*
2338: * This is straightforward stuff.
2339: * Open a database connection to a "temporary" database, then open a set
2340: * of prepared statements we'll use over and over again.
2341: * If "real" is set, we use the existing database; if not, we truncate a
2342: * temporary one.
2343: * Must be matched by dbclose().
2344: */
2345: static int
1.59 schwarze 2346: dbopen(int real)
1.50 kristaps 2347: {
1.115 schwarze 2348: const char *sql;
1.50 kristaps 2349: int rc, ofl;
1.12 schwarze 2350:
1.141 schwarze 2351: if (nodb)
1.197 schwarze 2352: return 1;
1.12 schwarze 2353:
1.115 schwarze 2354: *tempfilename = '\0';
1.63 schwarze 2355: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2356:
2357: if (real) {
2358: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2359: if (SQLITE_OK != rc) {
1.63 schwarze 2360: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2361: if (SQLITE_CANTOPEN != rc)
2362: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.197 schwarze 2363: return 0;
1.63 schwarze 2364: }
1.115 schwarze 2365: goto prepare_statements;
2366: }
2367:
2368: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2369:
1.115 schwarze 2370: remove(MANDOC_DB "~");
2371: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2372: if (SQLITE_OK == rc)
1.115 schwarze 2373: goto create_tables;
1.116 schwarze 2374: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2375: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2376: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.197 schwarze 2377: return 0;
1.50 kristaps 2378: }
1.12 schwarze 2379:
1.143 schwarze 2380: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2381: sizeof(tempfilename));
1.115 schwarze 2382: if (NULL == mkdtemp(tempfilename)) {
2383: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2384: say("", "&%s", tempfilename);
1.197 schwarze 2385: return 0;
1.115 schwarze 2386: }
1.143 schwarze 2387: (void)strlcat(tempfilename, "/" MANDOC_DB,
2388: sizeof(tempfilename));
1.115 schwarze 2389: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2390: if (SQLITE_OK != rc) {
1.59 schwarze 2391: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2392: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.197 schwarze 2393: return 0;
1.50 kristaps 2394: }
1.12 schwarze 2395:
1.115 schwarze 2396: create_tables:
1.81 schwarze 2397: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2398: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2399: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2400: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2401: ");\n"
2402: "\n"
1.81 schwarze 2403: "CREATE TABLE \"mlinks\" (\n"
2404: " \"sec\" TEXT NOT NULL,\n"
2405: " \"arch\" TEXT NOT NULL,\n"
2406: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2407: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2408: "ON DELETE CASCADE\n"
1.81 schwarze 2409: ");\n"
1.136 schwarze 2410: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2411: "\n"
1.133 schwarze 2412: "CREATE TABLE \"names\" (\n"
2413: " \"bits\" INTEGER NOT NULL,\n"
2414: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2415: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.175 schwarze 2416: "ON DELETE CASCADE,\n"
2417: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.133 schwarze 2418: ");\n"
2419: "\n"
1.50 kristaps 2420: "CREATE TABLE \"keys\" (\n"
2421: " \"bits\" INTEGER NOT NULL,\n"
2422: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2423: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2424: "ON DELETE CASCADE\n"
1.136 schwarze 2425: ");\n"
2426: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2427:
1.50 kristaps 2428: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2429: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2430: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2431: sqlite3_close(db);
1.197 schwarze 2432: return 0;
1.50 kristaps 2433: }
1.4 kristaps 2434:
1.57 schwarze 2435: prepare_statements:
1.146 schwarze 2436: if (SQLITE_OK != sqlite3_exec(db,
2437: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2438: exitcode = (int)MANDOCLEVEL_SYSERR;
2439: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2440: sqlite3_errmsg(db));
2441: sqlite3_close(db);
1.197 schwarze 2442: return 0;
1.146 schwarze 2443: }
2444:
1.137 schwarze 2445: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2446: "(SELECT pageid FROM mlinks WHERE "
2447: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2448: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2449: sql = "INSERT INTO mpages "
1.132 schwarze 2450: "(desc,form) VALUES (?,?)";
1.81 schwarze 2451: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2452: sql = "INSERT INTO mlinks "
1.104 schwarze 2453: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2454: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.175 schwarze 2455: sql = "SELECT bits FROM names where pageid = ?";
2456: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.133 schwarze 2457: sql = "INSERT INTO names "
2458: "(bits,name,pageid) VALUES (?,?,?)";
2459: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2460: sql = "INSERT INTO keys "
1.81 schwarze 2461: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2462: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2463:
2464: #ifndef __APPLE__
2465: /*
2466: * When opening a new database, we can turn off
2467: * synchronous mode for much better performance.
2468: */
2469:
1.146 schwarze 2470: if (real && SQLITE_OK != sqlite3_exec(db,
2471: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2472: exitcode = (int)MANDOCLEVEL_SYSERR;
2473: say(MANDOC_DB, "PRAGMA synchronous: %s",
1.162 schwarze 2474: sqlite3_errmsg(db));
1.146 schwarze 2475: sqlite3_close(db);
1.197 schwarze 2476: return 0;
1.146 schwarze 2477: }
1.70 schwarze 2478: #endif
2479:
1.197 schwarze 2480: return 1;
1.4 kristaps 2481: }
2482:
1.50 kristaps 2483: static int
1.165 schwarze 2484: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2485: {
1.59 schwarze 2486: static char startdir[PATH_MAX];
1.151 schwarze 2487: static int getcwd_status; /* 1 = ok, 2 = failure */
2488: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2489: char *cp;
1.4 kristaps 2490:
1.59 schwarze 2491: /*
1.151 schwarze 2492: * Remember the original working directory, if possible.
2493: * This will be needed if the second or a later directory
2494: * on the command line is given as a relative path.
2495: * Do not error out if the current directory is not
2496: * searchable: Maybe it won't be needed after all.
2497: */
2498: if (0 == getcwd_status) {
2499: if (NULL == getcwd(startdir, sizeof(startdir))) {
2500: getcwd_status = 2;
2501: (void)strlcpy(startdir, strerror(errno),
2502: sizeof(startdir));
2503: } else
2504: getcwd_status = 1;
2505: }
2506:
2507: /*
2508: * We are leaving the old base directory.
2509: * Do not use it any longer, not even for messages.
2510: */
2511: *basedir = '\0';
2512:
2513: /*
2514: * If and only if the directory was changed earlier and
2515: * the next directory to process is given as a relative path,
2516: * first go back, or bail out if that is impossible.
1.59 schwarze 2517: */
1.151 schwarze 2518: if (chdir_status && '/' != *targetdir) {
2519: if (2 == getcwd_status) {
1.59 schwarze 2520: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2521: say("", "getcwd: %s", startdir);
1.197 schwarze 2522: return 0;
1.59 schwarze 2523: }
1.151 schwarze 2524: if (-1 == chdir(startdir)) {
1.59 schwarze 2525: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2526: say("", "&chdir %s", startdir);
1.197 schwarze 2527: return 0;
1.59 schwarze 2528: }
2529: }
1.151 schwarze 2530:
2531: /*
2532: * Always resolve basedir to the canonicalized absolute
2533: * pathname and append a trailing slash, such that
2534: * we can reliably check whether files are inside.
2535: */
1.59 schwarze 2536: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2537: if (report_baddir || errno != ENOENT) {
2538: exitcode = (int)MANDOCLEVEL_BADARG;
2539: say("", "&%s: realpath", targetdir);
2540: }
1.197 schwarze 2541: return 0;
1.59 schwarze 2542: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2543: if (report_baddir || errno != ENOENT) {
2544: exitcode = (int)MANDOCLEVEL_BADARG;
2545: say("", "&chdir");
2546: }
1.197 schwarze 2547: return 0;
1.150 schwarze 2548: }
1.151 schwarze 2549: chdir_status = 1;
1.150 schwarze 2550: cp = strchr(basedir, '\0');
2551: if ('/' != cp[-1]) {
2552: if (cp - basedir >= PATH_MAX - 1) {
2553: exitcode = (int)MANDOCLEVEL_SYSERR;
2554: say("", "Filename too long");
1.197 schwarze 2555: return 0;
1.150 schwarze 2556: }
2557: *cp++ = '/';
2558: *cp = '\0';
1.4 kristaps 2559: }
1.197 schwarze 2560: return 1;
1.56 schwarze 2561: }
2562:
2563: static void
1.59 schwarze 2564: say(const char *file, const char *format, ...)
1.56 schwarze 2565: {
2566: va_list ap;
1.123 schwarze 2567: int use_errno;
1.56 schwarze 2568:
1.59 schwarze 2569: if ('\0' != *basedir)
2570: fprintf(stderr, "%s", basedir);
2571: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2572: fputc('/', stderr);
1.56 schwarze 2573: if ('\0' != *file)
1.59 schwarze 2574: fprintf(stderr, "%s", file);
2575:
1.123 schwarze 2576: use_errno = 1;
2577: if (NULL != format) {
2578: switch (*format) {
1.141 schwarze 2579: case '&':
1.123 schwarze 2580: format++;
2581: break;
1.141 schwarze 2582: case '\0':
1.123 schwarze 2583: format = NULL;
2584: break;
2585: default:
2586: use_errno = 0;
2587: break;
2588: }
2589: }
2590: if (NULL != format) {
2591: if ('\0' != *basedir || '\0' != *file)
2592: fputs(": ", stderr);
2593: va_start(ap, format);
2594: vfprintf(stderr, format, ap);
2595: va_end(ap);
2596: }
2597: if (use_errno) {
2598: if ('\0' != *basedir || '\0' != *file || NULL != format)
2599: fputs(": ", stderr);
1.59 schwarze 2600: perror(NULL);
1.123 schwarze 2601: } else
2602: fputc('\n', stderr);
1.1 kristaps 2603: }
CVSweb