Home | History | Annotate | Line # | Download | only in makemandb
makemandb.c revision 1.67
      1 /*	$NetBSD: makemandb.c,v 1.67 2023/01/01 21:27:14 gutteridge Exp $	*/
      2 /*
      3  * Copyright (c) 2011 Abhinav Upadhyay <er.abhinav.upadhyay (at) gmail.com>
      4  * Copyright (c) 2011 Kristaps Dzonsons <kristaps (at) bsd.lv>
      5  *
      6  * Permission to use, copy, modify, and distribute this software for any
      7  * purpose with or without fee is hereby granted, provided that the above
      8  * copyright notice and this permission notice appear in all copies.
      9  *
     10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
     11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
     12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
     13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
     14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
     15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
     16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
     17  */
     18 
     19 #include <sys/cdefs.h>
     20 __RCSID("$NetBSD: makemandb.c,v 1.67 2023/01/01 21:27:14 gutteridge Exp $");
     21 
     22 #include <sys/stat.h>
     23 #include <sys/types.h>
     24 
     25 #include <assert.h>
     26 #include <dirent.h>
     27 #include <err.h>
     28 #include <archive.h>
     29 #include <libgen.h>
     30 #include <md5.h>
     31 #include <stdio.h>
     32 #include <stdlib.h>
     33 #include <string.h>
     34 #include <unistd.h>
     35 #include <util.h>
     36 
     37 #include "apropos-utils.h"
     38 #include "dist/man.h"
     39 #include "dist/mandoc.h"
     40 #include "dist/mdoc.h"
     41 #include "dist/roff.h"
     42 #include "dist/mandoc_parse.h"
     43 
     44 #define BUFLEN 1024
     45 #define MDOC 0	//If the page is of mdoc(7) type
     46 #define MAN 1	//If the page  is of man(7) type
     47 
     48 /*
     49  * A data structure for holding section specific data.
     50  */
     51 typedef struct secbuff {
     52 	char *data;
     53 	size_t buflen;	//Total length of buffer allocated initially
     54 	size_t offset;	// Current offset in the buffer.
     55 } secbuff;
     56 
     57 typedef struct makemandb_flags {
     58 	int optimize;
     59 	int limit;	// limit the indexing to only NAME section
     60 	int recreate;	// Database was created from scratch
     61 	int verbosity;	// 0: quiet, 1: default, 2: verbose
     62 } makemandb_flags;
     63 
     64 typedef struct roff_mandb_rec {
     65 	/* Fields for mandb table */
     66 	char *name;	// for storing the name of the man page
     67 	char *name_desc; // for storing the one line description (.Nd)
     68 	secbuff desc; // for storing the DESCRIPTION section
     69 	secbuff lib; // for the LIBRARY section
     70 	secbuff return_vals; // RETURN VALUES
     71 	secbuff env; // ENVIRONMENT
     72 	secbuff files; // FILES
     73 	secbuff exit_status; // EXIT STATUS
     74 	secbuff diagnostics; // DIAGNOSTICS
     75 	secbuff errors; // ERRORS
     76 	char *section;
     77 
     78 	int xr_found; // To track whether a .Xr was seen when parsing a section
     79 
     80 	/* Fields for mandb_meta table */
     81 	char *md5_hash;
     82 	dev_t device;
     83 	ino_t inode;
     84 	time_t mtime;
     85 
     86 	/* Fields for mandb_links table */
     87 	char *machine;
     88 	char *links; //all the links to a page in a space separated form
     89 	char *file_path;
     90 
     91 	/* Non-db fields */
     92 	int page_type; //Indicates the type of page: mdoc or man
     93 } mandb_rec;
     94 
     95 typedef	void (*proff_nf)(const struct roff_node *n, mandb_rec *);
     96 
     97 static void append(secbuff *sbuff, const char *src);
     98 static void init_secbuffs(mandb_rec *);
     99 static void free_secbuffs(mandb_rec *);
    100 static int check_md5(const char *, sqlite3 *, char **, void *, size_t);
    101 static void cleanup(mandb_rec *);
    102 static void set_section(const struct roff_meta *, mandb_rec *);
    103 static void set_machine(const struct roff_meta *, mandb_rec *);
    104 static int insert_into_db(sqlite3 *, mandb_rec *);
    105 static	void begin_parse(const char *, struct mparse *, mandb_rec *, int);
    106 static void proff_node(const struct roff_node *, mandb_rec *,
    107 		       struct roff_meta *, const proff_nf *);
    108 static void pmdoc_Nm(const struct roff_node *, mandb_rec *);
    109 static void pmdoc_Nd(const struct roff_node *, mandb_rec *);
    110 static void pmdoc_Sh(const struct roff_node *, mandb_rec *);
    111 static void mdoc_parse_Sh(const struct roff_node *, mandb_rec *);
    112 static void pmdoc_Xr(const struct roff_node *, mandb_rec *);
    113 static void pmdoc_Pp(const struct roff_node *, mandb_rec *);
    114 static void pmdoc_macro_handler(const struct roff_node *, mandb_rec *, int);
    115 static void pman_parse_node(const struct roff_node *, secbuff *);
    116 static void pman_parse_name(const struct roff_node *, mandb_rec *);
    117 static void pman_sh(const struct roff_node *, mandb_rec *);
    118 static void pman_block(const struct roff_node *, mandb_rec *);
    119 static void traversedir(const char *, const char *, sqlite3 *, struct mparse *);
    120 static void mdoc_parse_section(enum roff_sec, const char *, mandb_rec *);
    121 static void man_parse_section(enum man_sec, const struct roff_node *, mandb_rec *);
    122 static void build_file_cache(sqlite3 *, const char *, const char *,
    123 			     struct stat *);
    124 static void update_db(sqlite3 *, struct mparse *, mandb_rec *);
    125 __dead static void usage(void);
    126 static void optimize(sqlite3 *);
    127 static char *parse_escape(const char *);
    128 static void replace_hyph(char *);
    129 static makemandb_flags mflags = { .verbosity = 1 };
    130 
    131 static	const proff_nf mdocs[MDOC_MAX - MDOC_Dd] = {
    132 	NULL, /* Dd */
    133 	NULL, /* Dt */
    134 	NULL, /* Os */
    135 	pmdoc_Sh, /* Sh */
    136 	NULL, /* Ss */
    137 	pmdoc_Pp, /* Pp */
    138 	NULL, /* D1 */
    139 	NULL, /* Dl */
    140 	NULL, /* Bd */
    141 	NULL, /* Ed */
    142 	NULL, /* Bl */
    143 	NULL, /* El */
    144 	NULL, /* It */
    145 	NULL, /* Ad */
    146 	NULL, /* An */
    147 	NULL, /* Ap */
    148 	NULL, /* Ar */
    149 	NULL, /* Cd */
    150 	NULL, /* Cm */
    151 	NULL, /* Dv */
    152 	NULL, /* Er */
    153 	NULL, /* Ev */
    154 	NULL, /* Ex */
    155 	NULL, /* Fa */
    156 	NULL, /* Fd */
    157 	NULL, /* Fl */
    158 	NULL, /* Fn */
    159 	NULL, /* Ft */
    160 	NULL, /* Ic */
    161 	NULL, /* In */
    162 	NULL, /* Li */
    163 	pmdoc_Nd, /* Nd */
    164 	pmdoc_Nm, /* Nm */
    165 	NULL, /* Op */
    166 	NULL, /* Ot */
    167 	NULL, /* Pa */
    168 	NULL, /* Rv */
    169 	NULL, /* St */
    170 	NULL, /* Va */
    171 	NULL, /* Vt */
    172 	pmdoc_Xr, /* Xr */
    173 	NULL, /* %A */
    174 	NULL, /* %B */
    175 	NULL, /* %D */
    176 	NULL, /* %I */
    177 	NULL, /* %J */
    178 	NULL, /* %N */
    179 	NULL, /* %O */
    180 	NULL, /* %P */
    181 	NULL, /* %R */
    182 	NULL, /* %T */
    183 	NULL, /* %V */
    184 	NULL, /* Ac */
    185 	NULL, /* Ao */
    186 	NULL, /* Aq */
    187 	NULL, /* At */
    188 	NULL, /* Bc */
    189 	NULL, /* Bf */
    190 	NULL, /* Bo */
    191 	NULL, /* Bq */
    192 	NULL, /* Bsx */
    193 	NULL, /* Bx */
    194 	NULL, /* Db */
    195 	NULL, /* Dc */
    196 	NULL, /* Do */
    197 	NULL, /* Dq */
    198 	NULL, /* Ec */
    199 	NULL, /* Ef */
    200 	NULL, /* Em */
    201 	NULL, /* Eo */
    202 	NULL, /* Fx */
    203 	NULL, /* Ms */
    204 	NULL, /* No */
    205 	NULL, /* Ns */
    206 	NULL, /* Nx */
    207 	NULL, /* Ox */
    208 	NULL, /* Pc */
    209 	NULL, /* Pf */
    210 	NULL, /* Po */
    211 	NULL, /* Pq */
    212 	NULL, /* Qc */
    213 	NULL, /* Ql */
    214 	NULL, /* Qo */
    215 	NULL, /* Qq */
    216 	NULL, /* Re */
    217 	NULL, /* Rs */
    218 	NULL, /* Sc */
    219 	NULL, /* So */
    220 	NULL, /* Sq */
    221 	NULL, /* Sm */
    222 	NULL, /* Sx */
    223 	NULL, /* Sy */
    224 	NULL, /* Tn */
    225 	NULL, /* Ux */
    226 	NULL, /* Xc */
    227 	NULL, /* Xo */
    228 	NULL, /* Fo */
    229 	NULL, /* Fc */
    230 	NULL, /* Oo */
    231 	NULL, /* Oc */
    232 	NULL, /* Bk */
    233 	NULL, /* Ek */
    234 	NULL, /* Bt */
    235 	NULL, /* Hf */
    236 	NULL, /* Fr */
    237 	NULL, /* Ud */
    238 	NULL, /* Lb */
    239 	NULL, /* Lp */
    240 	NULL, /* Lk */
    241 	NULL, /* Mt */
    242 	NULL, /* Brq */
    243 	NULL, /* Bro */
    244 	NULL, /* Brc */
    245 	NULL, /* %C */
    246 	NULL, /* Es */
    247 	NULL, /* En */
    248 	NULL, /* Dx */
    249 	NULL, /* %Q */
    250 	NULL, /* %U */
    251 	NULL /* Ta */
    252 };
    253 
    254 static	const proff_nf mans[MAN_MAX - MAN_TH] = {
    255 	NULL,	//TH
    256 	pman_sh, //SH
    257 	NULL,	//SS
    258 	NULL,	//TP
    259 	NULL,	//LP
    260 	NULL,	//PP
    261 	NULL,	//P
    262 	NULL,	//IP
    263 	NULL,	//HP
    264 	NULL,	//SM
    265 	NULL,	//SB
    266 	NULL,	//BI
    267 	NULL,	//IB
    268 	NULL,	//BR
    269 	NULL,	//RB
    270 	NULL,	//R
    271 	pman_block,	//B
    272 	NULL,	//I
    273 	NULL,	//IR
    274 	NULL,	//RI
    275 	NULL,	//nf
    276 	NULL,	//fi
    277 	NULL,	//RE
    278 	NULL,	//RS
    279 	NULL,	//DT
    280 	NULL,	//UC
    281 	NULL,	//PD
    282 	NULL,	//AT
    283 	NULL,	//in
    284 	NULL,	//OP
    285 	NULL,	//EX
    286 	NULL,	//EE
    287 	NULL,	//UR
    288 	NULL,	//UE
    289 	NULL,	//MT
    290 	NULL   //ME
    291 };
    292 
    293 int
    294 main(int argc, char *argv[])
    295 {
    296 	FILE *file;
    297 	const char *sqlstr, *manconf = NULL;
    298 	char *line, *command;
    299 	char *errmsg;
    300 	int ch;
    301 	struct mparse *mp;
    302 	sqlite3 *db;
    303 	ssize_t len;
    304 	size_t linesize;
    305 	struct roff_mandb_rec rec;
    306 
    307 	while ((ch = getopt(argc, argv, "C:floQqv")) != -1) {
    308 		switch (ch) {
    309 		case 'C':
    310 			manconf = optarg;
    311 			break;
    312 		case 'f':
    313 			mflags.recreate = 1;
    314 			break;
    315 		case 'l':
    316 			mflags.limit = 1;
    317 			break;
    318 		case 'o':
    319 			mflags.optimize = 1;
    320 			break;
    321 		case 'Q':
    322 			mflags.verbosity = 0;
    323 			break;
    324 		case 'q':
    325 			mflags.verbosity = 1;
    326 			break;
    327 		case 'v':
    328 			mflags.verbosity = 2;
    329 			break;
    330 		default:
    331 			usage();
    332 		}
    333 	}
    334 
    335 	memset(&rec, 0, sizeof(rec));
    336 
    337 	init_secbuffs(&rec);
    338 	mchars_alloc();
    339 	mp = mparse_alloc(MPARSE_SO | MPARSE_UTF8 | MPARSE_LATIN1 |
    340 	    MPARSE_VALIDATE, MANDOC_OS_OTHER, NULL);
    341 
    342 	if (manconf) {
    343 		char *arg;
    344 		size_t command_len = shquote(manconf, NULL, 0) + 1;
    345 		arg = emalloc(command_len);
    346 		shquote(manconf, arg, command_len);
    347 		easprintf(&command, "man -p -C %s", arg);
    348 		free(arg);
    349 	} else {
    350 		command = estrdup("man -p");
    351 		manconf = MANCONF;
    352 	}
    353 
    354 	/* Call man -p to get the list of man page dirs */
    355 	if ((file = popen(command, "r")) == NULL) {
    356 		free(command);
    357 		err(EXIT_FAILURE, "popen failed");
    358 	}
    359 	free(command);
    360 
    361 	if (mflags.recreate) {
    362 		char *dbp = get_dbpath(manconf);
    363 		/* No error here, it will fail in init_db in the same call */
    364 		if (dbp != NULL)
    365 			remove(dbp);
    366 	}
    367 
    368 	if ((db = init_db(MANDB_CREATE, manconf)) == NULL)
    369 		exit(EXIT_FAILURE);
    370 
    371 	sqlite3_exec(db, "PRAGMA synchronous = 0", NULL, NULL, 	&errmsg);
    372 	if (errmsg != NULL) {
    373 		warnx("%s", errmsg);
    374 		free(errmsg);
    375 		close_db(db);
    376 		exit(EXIT_FAILURE);
    377 	}
    378 
    379 	sqlite3_exec(db, "ATTACH DATABASE \':memory:\' AS metadb", NULL, NULL,
    380 	    &errmsg);
    381 	if (errmsg != NULL) {
    382 		warnx("%s", errmsg);
    383 		free(errmsg);
    384 		close_db(db);
    385 		exit(EXIT_FAILURE);
    386 	}
    387 
    388 	/* Begin the transaction for indexing the pages	*/
    389 	sqlite3_exec(db, "BEGIN", NULL, NULL, &errmsg);
    390 	if (errmsg != NULL) {
    391 		warnx("%s", errmsg);
    392 		free(errmsg);
    393 		close_db(db);
    394 		exit(EXIT_FAILURE);
    395 	}
    396 
    397 	sqlstr = "CREATE TABLE metadb.file_cache(device, inode, mtime, parent,"
    398 		 " file PRIMARY KEY);"
    399 		 "CREATE UNIQUE INDEX metadb.index_file_cache_dev"
    400 		 " ON file_cache (device, inode)";
    401 
    402 	sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg);
    403 	if (errmsg != NULL) {
    404 		warnx("%s", errmsg);
    405 		free(errmsg);
    406 		close_db(db);
    407 		exit(EXIT_FAILURE);
    408 	}
    409 
    410 	if (mflags.verbosity)
    411 		printf("Building temporary file cache\n");
    412 	line = NULL;
    413 	linesize = 0;
    414 	while ((len = getline(&line, &linesize, file)) != -1) {
    415 		/* Replace the new line character at the end of string with '\0' */
    416 		line[len - 1] = '\0';
    417 		char *pdir = estrdup(dirname(line));
    418 		/* Traverse the man page directories and parse the pages */
    419 		traversedir(pdir, line, db, mp);
    420 		free(pdir);
    421 	}
    422 	free(line);
    423 
    424 	if (pclose(file) == -1) {
    425 		close_db(db);
    426 		cleanup(&rec);
    427 		free_secbuffs(&rec);
    428 		err(EXIT_FAILURE, "pclose error");
    429 	}
    430 
    431 	if (mflags.verbosity)
    432 		printf("Performing index update\n");
    433 	update_db(db, mp, &rec);
    434 	mparse_free(mp);
    435 	mchars_free();
    436 	free_secbuffs(&rec);
    437 
    438 	/* Commit the transaction */
    439 	sqlite3_exec(db, "COMMIT", NULL, NULL, &errmsg);
    440 	if (errmsg != NULL) {
    441 		warnx("%s", errmsg);
    442 		free(errmsg);
    443 		close_db(db);
    444 		exit(EXIT_FAILURE);
    445 	}
    446 
    447 	if (mflags.optimize)
    448 		optimize(db);
    449 
    450 	close_db(db);
    451 	return 0;
    452 }
    453 
    454 /*
    455  * traversedir --
    456  *  Traverses the given directory recursively and passes all the man page files
    457  *  in the way to build_file_cache()
    458  */
    459 static void
    460 traversedir(const char *parent, const char *file, sqlite3 *db,
    461             struct mparse *mp)
    462 {
    463 	struct stat sb;
    464 	struct dirent *dirp;
    465 	DIR *dp;
    466 	char *buf;
    467 
    468 	if (stat(file, &sb) < 0) {
    469 		if (mflags.verbosity)
    470 			warn("stat failed: %s", file);
    471 		return;
    472 	}
    473 
    474 	/* If it is a directory, traverse it recursively */
    475 	if (S_ISDIR(sb.st_mode)) {
    476 		if ((dp = opendir(file)) == NULL) {
    477 			if (mflags.verbosity)
    478 				warn("opendir error: %s", file);
    479 			return;
    480 		}
    481 
    482 		while ((dirp = readdir(dp)) != NULL) {
    483 			/* Avoid . and .. entries in a directory */
    484 			if (dirp->d_name[0] != '.') {
    485 				easprintf(&buf, "%s/%s", file, dirp->d_name);
    486 				traversedir(parent, buf, db, mp);
    487 				free(buf);
    488 			}
    489 		}
    490 		closedir(dp);
    491 		return;
    492 	}
    493 
    494 	if (!S_ISREG(sb.st_mode))
    495 		return;
    496 
    497 	if (sb.st_size == 0) {
    498 		if (mflags.verbosity)
    499 			warnx("Empty file: %s", file);
    500 		return;
    501 	}
    502 	build_file_cache(db, parent, file, &sb);
    503 }
    504 
    505 /* build_file_cache --
    506  *   This function generates an md5 hash of the file passed as its 2nd parameter
    507  *   and stores it in a temporary table file_cache along with the full file path.
    508  *   This is done to support incremental update of the database.
    509  *   The temporary table file_cache is dropped thereafter in the function
    510  *   update_db(), once the database has been updated.
    511  */
    512 static void
    513 build_file_cache(sqlite3 *db, const char *parent, const char *file,
    514 		 struct stat *sb)
    515 {
    516 	const char *sqlstr;
    517 	sqlite3_stmt *stmt = NULL;
    518 	int rc, idx;
    519 	assert(file != NULL);
    520 	dev_t device_cache = sb->st_dev;
    521 	ino_t inode_cache = sb->st_ino;
    522 	time_t mtime_cache = sb->st_mtime;
    523 
    524 	sqlstr = "INSERT INTO metadb.file_cache VALUES (:device, :inode,"
    525 		 " :mtime, :parent, :file)";
    526 	rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL);
    527 	if (rc != SQLITE_OK) {
    528 		if (mflags.verbosity)
    529 			warnx("%s", sqlite3_errmsg(db));
    530 		return;
    531 	}
    532 
    533 	idx = sqlite3_bind_parameter_index(stmt, ":device");
    534 	rc = sqlite3_bind_int64(stmt, idx, device_cache);
    535 	if (rc != SQLITE_OK) {
    536 		if (mflags.verbosity)
    537 			warnx("%s", sqlite3_errmsg(db));
    538 		sqlite3_finalize(stmt);
    539 		return;
    540 	}
    541 
    542 	idx = sqlite3_bind_parameter_index(stmt, ":inode");
    543 	rc = sqlite3_bind_int64(stmt, idx, inode_cache);
    544 	if (rc != SQLITE_OK) {
    545 		if (mflags.verbosity)
    546 			warnx("%s", sqlite3_errmsg(db));
    547 		sqlite3_finalize(stmt);
    548 		return;
    549 	}
    550 
    551 	idx = sqlite3_bind_parameter_index(stmt, ":mtime");
    552 	rc = sqlite3_bind_int64(stmt, idx, mtime_cache);
    553 	if (rc != SQLITE_OK) {
    554 		if (mflags.verbosity)
    555 			warnx("%s", sqlite3_errmsg(db));
    556 		sqlite3_finalize(stmt);
    557 		return;
    558 	}
    559 
    560 	idx = sqlite3_bind_parameter_index(stmt, ":parent");
    561 	rc = sqlite3_bind_text(stmt, idx, parent, -1, NULL);
    562 	if (rc != SQLITE_OK) {
    563 		if (mflags.verbosity)
    564 			warnx("%s", sqlite3_errmsg(db));
    565 		sqlite3_finalize(stmt);
    566 		return;
    567 	}
    568 
    569 	idx = sqlite3_bind_parameter_index(stmt, ":file");
    570 	rc = sqlite3_bind_text(stmt, idx, file, -1, NULL);
    571 	if (rc != SQLITE_OK) {
    572 		if (mflags.verbosity)
    573 			warnx("%s", sqlite3_errmsg(db));
    574 		sqlite3_finalize(stmt);
    575 		return;
    576 	}
    577 
    578 	sqlite3_step(stmt);
    579 	sqlite3_finalize(stmt);
    580 }
    581 
    582 /* read_and_decompress --
    583  *	Reads the given file into memory. If it is compressed, decompress
    584  *	it before returning to the caller.
    585  */
    586 static int
    587 read_and_decompress(const char *file, void **bufp, size_t *len)
    588 {
    589 	size_t off;
    590 	ssize_t r;
    591 	struct archive *a;
    592 	struct archive_entry *ae;
    593 	char *buf;
    594 
    595 	if ((a = archive_read_new()) == NULL)
    596 		errx(EXIT_FAILURE, "memory allocation failed");
    597 
    598 	*bufp = NULL;
    599 	if (archive_read_support_filter_all(a) != ARCHIVE_OK ||
    600 	    archive_read_support_format_raw(a) != ARCHIVE_OK ||
    601 	    archive_read_open_filename(a, file, 65536) != ARCHIVE_OK ||
    602 	    archive_read_next_header(a, &ae) != ARCHIVE_OK)
    603 		goto archive_error;
    604 	*len = 65536;
    605 	buf = emalloc(*len);
    606 	off = 0;
    607 	for (;;) {
    608 		r = archive_read_data(a, buf + off, *len - off);
    609 		if (r == ARCHIVE_OK) {
    610 			archive_read_free(a);
    611 			*bufp = buf;
    612 			*len = off;
    613 			return 0;
    614 		}
    615 		if (r <= 0) {
    616 			free(buf);
    617 			break;
    618 		}
    619 		off += r;
    620 		if (off == *len) {
    621 			*len *= 2;
    622 			if (*len < off) {
    623 				if (mflags.verbosity)
    624 					warnx("File too large: %s", file);
    625 				free(buf);
    626 				archive_read_free(a);
    627 				return -1;
    628 			}
    629 			buf = erealloc(buf, *len);
    630 		}
    631 	}
    632 
    633 archive_error:
    634 	warnx("Error while reading `%s': %s", file, archive_error_string(a));
    635 	archive_read_free(a);
    636 	return -1;
    637 }
    638 
    639 static void
    640 update_existing_entry(sqlite3 *db, const char *file, const char *hash,
    641     mandb_rec *rec, int *new_count, int *link_count, int *err_count)
    642 {
    643 	int update_count, rc, idx;
    644 	const char *inner_sqlstr;
    645 	sqlite3_stmt *inner_stmt;
    646 
    647 	update_count = sqlite3_total_changes(db);
    648 	inner_sqlstr = "UPDATE mandb_meta SET device = :device,"
    649 		       " inode = :inode, mtime = :mtime WHERE"
    650 		       " md5_hash = :md5 AND file = :file AND"
    651 		       " (device <> :device2 OR inode <> "
    652 		       "  :inode2 OR mtime <> :mtime2)";
    653 	rc = sqlite3_prepare_v2(db, inner_sqlstr, -1, &inner_stmt, NULL);
    654 	if (rc != SQLITE_OK) {
    655 		if (mflags.verbosity)
    656 			warnx("%s", sqlite3_errmsg(db));
    657 		return;
    658 	}
    659 	idx = sqlite3_bind_parameter_index(inner_stmt, ":device");
    660 	sqlite3_bind_int64(inner_stmt, idx, rec->device);
    661 	idx = sqlite3_bind_parameter_index(inner_stmt, ":inode");
    662 	sqlite3_bind_int64(inner_stmt, idx, rec->inode);
    663 	idx = sqlite3_bind_parameter_index(inner_stmt, ":mtime");
    664 	sqlite3_bind_int64(inner_stmt, idx, rec->mtime);
    665 	idx = sqlite3_bind_parameter_index(inner_stmt, ":md5");
    666 	sqlite3_bind_text(inner_stmt, idx, hash, -1, NULL);
    667 	idx = sqlite3_bind_parameter_index(inner_stmt, ":file");
    668 	sqlite3_bind_text(inner_stmt, idx, file, -1, NULL);
    669 	idx = sqlite3_bind_parameter_index(inner_stmt, ":device2");
    670 	sqlite3_bind_int64(inner_stmt, idx, rec->device);
    671 	idx = sqlite3_bind_parameter_index(inner_stmt, ":inode2");
    672 	sqlite3_bind_int64(inner_stmt, idx, rec->inode);
    673 	idx = sqlite3_bind_parameter_index(inner_stmt, ":mtime2");
    674 	sqlite3_bind_int64(inner_stmt, idx, rec->mtime);
    675 
    676 	rc = sqlite3_step(inner_stmt);
    677 	if (rc == SQLITE_DONE) {
    678 		/* Check if an update has been performed. */
    679 		if (update_count != sqlite3_total_changes(db)) {
    680 			if (mflags.verbosity == 2)
    681 				printf("Updated %s\n", file);
    682 			(*new_count)++;
    683 		} else {
    684 			/* Otherwise it was a hardlink. */
    685 			(*link_count)++;
    686 		}
    687 	} else {
    688 		if (mflags.verbosity == 2)
    689 			warnx("Could not update the metadata for %s", file);
    690 		(*err_count)++;
    691 	}
    692 	sqlite3_finalize(inner_stmt);
    693 }
    694 
    695 /* update_db --
    696  *	Does an incremental update of the database by checking the file_cache.
    697  *	It parses and adds the pages which are present in file_cache,
    698  *	but not in the database.
    699  *	It also removes the pages which are present in the database,
    700  *	but not in the file_cache.
    701  */
    702 static void
    703 update_db(sqlite3 *db, struct mparse *mp, mandb_rec *rec)
    704 {
    705 	const char *sqlstr;
    706 	sqlite3_stmt *stmt = NULL;
    707 	char *file;
    708 	char *parent;
    709 	char *errmsg = NULL;
    710 	char *md5sum;
    711 	void *buf;
    712 	size_t buflen;
    713 	struct sql_row {
    714 		struct sql_row *next;
    715 		dev_t device;
    716 		ino_t inode;
    717 		time_t mtime;
    718 		char *parent;
    719 		char *file;
    720 	} *rows, *row;
    721 	int new_count = 0;	/* Counter for newly indexed/updated pages */
    722 	int total_count = 0;	/* Counter for total number of pages */
    723 	int err_count = 0;	/* Counter for number of failed pages */
    724 	int link_count = 0;	/* Counter for number of hard/sym links */
    725 	int md5_status;
    726 	int rc;
    727 
    728 	sqlstr = "SELECT device, inode, mtime, parent, file"
    729 	         " FROM metadb.file_cache fc"
    730 	         " WHERE NOT EXISTS(SELECT 1 FROM mandb_meta WHERE"
    731 	         "  device = fc.device AND inode = fc.inode AND "
    732 	         "  mtime = fc.mtime AND file = fc.file)";
    733 
    734 	rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL);
    735 	if (rc != SQLITE_OK) {
    736 		if (mflags.verbosity)
    737 			warnx("%s", sqlite3_errmsg(db));
    738 		close_db(db);
    739 		errx(EXIT_FAILURE, "Could not query file cache");
    740 	}
    741 
    742 	buf = NULL;
    743 	rows = NULL;
    744 	while (sqlite3_step(stmt) == SQLITE_ROW) {
    745 		row = emalloc(sizeof(struct sql_row));
    746 		row->device = sqlite3_column_int64(stmt, 0);
    747 		row->inode = sqlite3_column_int64(stmt, 1);
    748 		row->mtime = sqlite3_column_int64(stmt, 2);
    749 		row->parent = estrdup((const char *) sqlite3_column_text(stmt, 3));
    750 		row->file = estrdup((const char *) sqlite3_column_text(stmt, 4));
    751 		row->next = rows;
    752 		rows = row;
    753 		total_count++;
    754 	}
    755 	sqlite3_finalize(stmt);
    756 
    757 	for ( ; rows != NULL; free(parent), free(file), free(buf)) {
    758 		int fd;
    759 
    760 		row = rows;
    761 		rows = rows->next;
    762 
    763 		rec->device = row->device;
    764 		rec->inode = row->inode;
    765 		rec->mtime = row->mtime;
    766 		parent = row->parent;
    767 		file = row->file;
    768 		free(row);
    769 
    770 		// XXX: reading twice!
    771 		if (read_and_decompress(file, &buf, &buflen)) {
    772 			err_count++;
    773 			continue;
    774 		}
    775 		if ((fd = mparse_open(mp, file)) == -1) {
    776 			err_count++;
    777 			continue;
    778 		}
    779 
    780 		md5_status = check_md5(file, db, &md5sum, buf, buflen);
    781 		assert(md5sum != NULL);
    782 		if (md5_status == -1) {
    783 			if (mflags.verbosity)
    784 				warnx("An error occurred in checking md5 value"
    785 			      " for file %s", file);
    786 			err_count++;
    787 			close(fd);
    788 			continue;
    789 		}
    790 
    791 		if (md5_status == 0) {
    792 			/*
    793 			 * The MD5 hash is already present in the database,
    794 			 * so simply update the metadata.
    795 			 */
    796 			update_existing_entry(db, file, md5sum, rec,
    797 			    &new_count, &link_count, &err_count);
    798 			free(md5sum);
    799 			close(fd);
    800 			continue;
    801 		}
    802 
    803 		if (md5_status == 1) {
    804 			/*
    805 			 * The MD5 hash was not present in the database.
    806 			 * This means is either a new file or an updated file.
    807 			 * We should go ahead with parsing.
    808 			 */
    809 			if (chdir(parent) == -1) {
    810 				if (mflags.verbosity)
    811 					warn("chdir failed for `%s', could "
    812 					    "not index `%s'", parent, file);
    813 				err_count++;
    814 				free(md5sum);
    815 				close(fd);
    816 				continue;
    817 			}
    818 
    819 			if (mflags.verbosity == 2)
    820 				printf("Parsing: %s\n", file);
    821 			rec->md5_hash = md5sum;
    822 			rec->file_path = estrdup(file);
    823 			// file_path is freed by insert_into_db itself.
    824 			begin_parse(file, mp, rec, fd);
    825 			if (insert_into_db(db, rec) < 0) {
    826 				if (mflags.verbosity)
    827 					warnx("Error in indexing `%s'", file);
    828 				err_count++;
    829 			} else {
    830 				new_count++;
    831 			}
    832 		}
    833 		close(fd);
    834 	}
    835 
    836 	if (mflags.verbosity == 2) {
    837 		printf("Number of new or updated pages encountered: %d\n"
    838 		    "Number of hard links found: %d\n"
    839 		    "Number of pages that were successfully"
    840 		    " indexed or updated: %d\n"
    841 		    "Number of pages that could not be indexed"
    842 		    " due to errors: %d\n",
    843 		    total_count - link_count, link_count, new_count, err_count);
    844 	}
    845 
    846 	if (mflags.recreate)
    847 		return;
    848 
    849 	if (mflags.verbosity == 2)
    850 		printf("Deleting stale index entries\n");
    851 
    852 	sqlstr = "DELETE FROM mandb_meta WHERE file NOT IN"
    853 		 " (SELECT file FROM metadb.file_cache);"
    854 		 "DELETE FROM mandb_links WHERE md5_hash NOT IN"
    855 		 " (SELECT md5_hash from mandb_meta);"
    856 		 "DROP TABLE metadb.file_cache;"
    857 		 "DELETE FROM mandb WHERE rowid NOT IN"
    858 		 " (SELECT id FROM mandb_meta);";
    859 
    860 	sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg);
    861 	if (errmsg != NULL) {
    862 		warnx("Removing old entries failed: %s", errmsg);
    863 		warnx("Please rebuild database from scratch with -f.");
    864 		free(errmsg);
    865 		return;
    866 	}
    867 }
    868 
    869 /*
    870  * begin_parse --
    871  *  parses the man page using libmandoc
    872  */
    873 static void
    874 begin_parse(const char *file, struct mparse *mp, mandb_rec *rec, int fd)
    875 {
    876 	struct roff_meta *roff;
    877 	mparse_reset(mp);
    878 
    879 	rec->xr_found = 0;
    880 
    881 	mparse_readfd(mp, fd, file);
    882 	roff = mparse_result(mp);
    883 	if (roff == NULL) {
    884 		if (mflags.verbosity == 2)
    885 			warnx("Not a roff(7) page");
    886 		return;
    887 	}
    888 
    889 	if (roff->macroset == MACROSET_MDOC) {
    890 		rec->page_type = MDOC;
    891 		proff_node(roff->first->child, rec, roff, mdocs);
    892 	} else if (roff->macroset == MACROSET_MAN) {
    893 		rec->page_type = MAN;
    894 		proff_node(roff->first->child, rec, roff, mans);
    895 	} else
    896 		warnx("Unknown macroset %d", roff->macroset);
    897 	set_machine(roff, rec);
    898 	set_section(roff, rec);
    899 }
    900 
    901 /*
    902  * set_section --
    903  *  Extracts the section number and normalizes it to only the numeric part
    904  *  (Which should be the first character of the string).
    905  */
    906 static void
    907 set_section(const struct roff_meta *rm, mandb_rec *rec)
    908 {
    909 	if (!rm)
    910 		return;
    911 	const char *s = rm->msec == NULL ? "?" : rm->msec;
    912 	easprintf(&rec->section, "%s", s);
    913 	if (rec->section[0] == '?' && mflags.verbosity == 2)
    914 		warnx("%s: Missing section number", rec->file_path);
    915 }
    916 
    917 /*
    918  * get_machine --
    919  *  Extracts the machine architecture information if available.
    920  */
    921 static void
    922 set_machine(const struct roff_meta *rm, mandb_rec *rec)
    923 {
    924 	if (rm == NULL)
    925 		return;
    926 	if (rm->arch)
    927 		rec->machine = estrdup(rm->arch);
    928 }
    929 
    930 /*
    931  * pmdoc_Nm --
    932  *  Extracts the Name of the manual page from the .Nm macro
    933  */
    934 static void
    935 pmdoc_Nm(const struct roff_node *n, mandb_rec *rec)
    936 {
    937 	if (n->sec != SEC_NAME)
    938 		return;
    939 
    940 	for (n = n->child; n; n = n->next) {
    941 		if (n->type == ROFFT_TEXT) {
    942 			char *escaped_name = parse_escape(n->string);
    943 			concat(&rec->name, escaped_name);
    944 			free(escaped_name);
    945 		}
    946 	}
    947 }
    948 
    949 /*
    950  * pmdoc_Nd --
    951  *  Extracts the one line description of the man page from the .Nd macro
    952  */
    953 static void
    954 pmdoc_Nd(const struct roff_node *n, mandb_rec *rec)
    955 {
    956 	if (n->type == ROFFT_BODY)
    957 		deroff(&rec->name_desc, n);
    958 	if (rec->name_desc)
    959 		replace_hyph(rec->name_desc);
    960 
    961 }
    962 
    963 /*
    964  * pmdoc_macro_handler--
    965  *  This function is a single point of handling all the special macros that we
    966  *  want to handle especially. For example the .Xr macro for properly parsing
    967  *  the referenced page name along with the section number, or the .Pp macro
    968  *  for adding a new line whenever we encounter it.
    969  */
    970 static void
    971 pmdoc_macro_handler(const struct roff_node *n, mandb_rec *rec, int doct)
    972 {
    973 	const struct roff_node *sn;
    974 	assert(n);
    975 
    976 	switch (doct) {
    977 	/*  Parse the man page references.
    978 	 * Basically the .Xr macros are used like:
    979 	 *  .Xr ls 1
    980  	 *  and formatted like this:
    981 	 *  ls(1)
    982 	 *  Prepare a buffer to format the data like the above example and call
    983 	 *  pmdoc_parse_section to append it.
    984 	 */
    985 	case MDOC_Xr:
    986 		n = n->child;
    987 		while (n->type != ROFFT_TEXT && n->next)
    988 			n = n->next;
    989 
    990 		if (n && n->type != ROFFT_TEXT)
    991 			return;
    992 		sn = n;
    993 		if (n->next)
    994 			n = n->next;
    995 
    996 		while (n->type != ROFFT_TEXT && n->next)
    997 			n = n->next;
    998 
    999 		if (n && n->type == ROFFT_TEXT) {
   1000 			char *buf;
   1001 			easprintf(&buf, "%s(%s)", sn->string, n->string);
   1002 			mdoc_parse_section(n->sec, buf, rec);
   1003 			free(buf);
   1004 		}
   1005 
   1006 		break;
   1007 
   1008 	/* Parse the .Pp macro to add a new line */
   1009 	case MDOC_Pp:
   1010 		if (n->type == ROFFT_TEXT)
   1011 			mdoc_parse_section(n->sec, "\n", rec);
   1012 		break;
   1013 	default:
   1014 		break;
   1015 	}
   1016 
   1017 }
   1018 
   1019 /*
   1020  * pmdoc_Xr, pmdoc_Pp--
   1021  *  Empty stubs.
   1022  *  The parser calls these functions each time it encounters
   1023  *  a .Xr or .Pp macro. We are parsing all the data from
   1024  *  the pmdoc_Sh function, so don't do anything here.
   1025  *  (See if else blocks in pmdoc_Sh.)
   1026  */
   1027 static void
   1028 pmdoc_Xr(const struct roff_node *n, mandb_rec *rec)
   1029 {
   1030 }
   1031 
   1032 static void
   1033 pmdoc_Pp(const struct roff_node *n, mandb_rec *rec)
   1034 {
   1035 }
   1036 
   1037 /*
   1038  * pmdoc_Sh --
   1039  *  Called when a .Sh macro is encountered and tries to parse its body
   1040  */
   1041 static void
   1042 pmdoc_Sh(const struct roff_node *n, mandb_rec *rec)
   1043 {
   1044 	if (n == NULL)
   1045 		return;
   1046 
   1047 	switch (n->sec) {
   1048 	case SEC_NAME:
   1049 	case SEC_SYNOPSIS:
   1050 	case SEC_EXAMPLES:
   1051 	case SEC_STANDARDS:
   1052 	case SEC_HISTORY:
   1053 	case SEC_AUTHORS:
   1054 	case SEC_BUGS:
   1055 		/*
   1056 		 * We don't care about text from these sections
   1057 		 */
   1058 		return;
   1059 	default:
   1060 		break;
   1061 	}
   1062 
   1063 	if (n->type == ROFFT_BLOCK)
   1064 		mdoc_parse_Sh(n->body, rec);
   1065 }
   1066 
   1067 /*
   1068  *  Called from pmdoc_Sh to parse body of a .Sh macro. It calls
   1069  *  mdoc_parse_section to append the data to the section specific buffer.
   1070  *  The .Xr macro needs special handling, thus the separate if branch for it.
   1071  */
   1072 static void
   1073 mdoc_parse_Sh(const struct roff_node *n, mandb_rec *rec)
   1074 {
   1075 	if (n == NULL || (n->type != ROFFT_TEXT && n->tok == MDOC_MAX))
   1076 		return;
   1077 	int xr_found = 0;
   1078 
   1079 	if (n->type == ROFFT_TEXT) {
   1080 		mdoc_parse_section(n->sec, n->string, rec);
   1081 	} else if (n->tok >= MDOC_Dd && n->tok < MDOC_MAX) {
   1082 		const int tok_idx = n->tok - MDOC_Dd;
   1083 		if (mdocs[tok_idx] == pmdoc_Xr) {
   1084 			/*
   1085 			* When encountering other inline macros,
   1086 			* call pmdoc_macro_handler.
   1087 			*/
   1088 			pmdoc_macro_handler(n, rec, MDOC_Xr);
   1089 			xr_found = 1;
   1090 		} else if (mdocs[tok_idx] == pmdoc_Pp) {
   1091 			pmdoc_macro_handler(n, rec, MDOC_Pp);
   1092 		}
   1093 	}
   1094 
   1095 	/*
   1096 	 * If an Xr macro was encountered then the child node has
   1097 	 * already been explored by pmdoc_macro_handler.
   1098 	 */
   1099 	if (xr_found == 0)
   1100 		mdoc_parse_Sh(n->child, rec);
   1101 	mdoc_parse_Sh(n->next, rec);
   1102 }
   1103 
   1104 /*
   1105  * mdoc_parse_section--
   1106  *  Utility function for parsing sections of the mdoc type pages.
   1107  *  Takes two params:
   1108  *   1. sec is an enum which indicates the section in which we are present
   1109  *   2. string is the string which we need to append to the secbuff for this
   1110  *      particular section.
   1111  *  The function appends string to the global section buffer and returns.
   1112  */
   1113 static void
   1114 mdoc_parse_section(enum roff_sec sec, const char *string, mandb_rec *rec)
   1115 {
   1116 	/*
   1117 	 * If the user specified the 'l' flag, then parse and store only the
   1118 	 * NAME section. Ignore the rest.
   1119 	 */
   1120 	if (mflags.limit)
   1121 		return;
   1122 
   1123 	switch (sec) {
   1124 	case SEC_LIBRARY:
   1125 		append(&rec->lib, string);
   1126 		break;
   1127 	case SEC_RETURN_VALUES:
   1128 		append(&rec->return_vals, string);
   1129 		break;
   1130 	case SEC_ENVIRONMENT:
   1131 		append(&rec->env, string);
   1132 		break;
   1133 	case SEC_FILES:
   1134 		append(&rec->files, string);
   1135 		break;
   1136 	case SEC_EXIT_STATUS:
   1137 		append(&rec->exit_status, string);
   1138 		break;
   1139 	case SEC_DIAGNOSTICS:
   1140 		append(&rec->diagnostics, string);
   1141 		break;
   1142 	case SEC_ERRORS:
   1143 		append(&rec->errors, string);
   1144 		break;
   1145 	default:
   1146 		append(&rec->desc, string);
   1147 		break;
   1148 	}
   1149 }
   1150 
   1151 static void
   1152 proff_node(const struct roff_node * n, mandb_rec * rec,
   1153     struct roff_meta * roff, const proff_nf * func)
   1154 {
   1155 	if (n == NULL)
   1156 		return;
   1157 
   1158 	int tok_idx;
   1159 
   1160 	switch (n->type) {
   1161 	case (ROFFT_BODY):
   1162 		/* FALLTHROUGH */
   1163 	case (ROFFT_BLOCK):
   1164 		/* FALLTHROUGH */
   1165 	case (ROFFT_ELEM):
   1166 		if (roff->macroset == MACROSET_MAN)
   1167 			tok_idx = n->tok - MAN_TH;
   1168 		else if (roff->macroset == MACROSET_MDOC)
   1169 			tok_idx = n->tok - MDOC_Dd;
   1170 		else
   1171 			tok_idx = -1;
   1172 		if (tok_idx >= 0 && func[tok_idx] != NULL)
   1173 			(*func[tok_idx]) (n, rec);
   1174 		break;
   1175 	default:
   1176 		break;
   1177 	}
   1178 
   1179 	proff_node(n->child, rec, roff, func);
   1180 	proff_node(n->next, rec, roff, func);
   1181 }
   1182 
   1183 /*
   1184  * pman_parse_name --
   1185  *  Parses the NAME section and puts the complete content in the name_desc
   1186  *  variable.
   1187  */
   1188 static void
   1189 pman_parse_name(const struct roff_node *n, mandb_rec *rec)
   1190 {
   1191 	if (n == NULL)
   1192 		return;
   1193 
   1194 	if (n->type == ROFFT_TEXT) {
   1195 		char *tmp = parse_escape(n->string);
   1196 		concat(&rec->name_desc, tmp);
   1197 		free(tmp);
   1198 	}
   1199 
   1200 	if (n->child)
   1201 		pman_parse_name(n->child, rec);
   1202 
   1203 	if(n->next)
   1204 		pman_parse_name(n->next, rec);
   1205 }
   1206 
   1207 /*
   1208  * A stub function to be able to parse the macros like .B embedded inside
   1209  * a section.
   1210  */
   1211 static void
   1212 pman_block(const struct roff_node *n, mandb_rec *rec)
   1213 {
   1214 }
   1215 
   1216 /*
   1217  * pman_sh --
   1218  * This function does one of the two things:
   1219  *  1. If the present section is NAME, then it will:
   1220  *    (a) Extract the name of the page (in case of multiple comma separated
   1221  *        names, it will pick up the first one).
   1222  *    (b) Build a space separated list of all the symlinks/hardlinks to
   1223  *        this page and store in the buffer 'links'. These are extracted from
   1224  *        the comma separated list of names in the NAME section as well.
   1225  *    (c) Move on to the one line description section, which is after the list
   1226  *        of names in the NAME section.
   1227  *  2. Otherwise, it will check the section name and call the man_parse_section
   1228  *     function, passing the enum corresponding to that section.
   1229  */
   1230 static void
   1231 pman_sh(const struct roff_node *n, mandb_rec *rec)
   1232 {
   1233 	static const struct {
   1234 		enum man_sec section;
   1235 		const char *header;
   1236 	} mapping[] = {
   1237 	    { MANSEC_DESCRIPTION, "DESCRIPTION" },
   1238 	    { MANSEC_SYNOPSIS, "SYNOPSIS" },
   1239 	    { MANSEC_LIBRARY, "LIBRARY" },
   1240 	    { MANSEC_ERRORS, "ERRORS" },
   1241 	    { MANSEC_FILES, "FILES" },
   1242 	    { MANSEC_RETURN_VALUES, "RETURN VALUE" },
   1243 	    { MANSEC_RETURN_VALUES, "RETURN VALUES" },
   1244 	    { MANSEC_EXIT_STATUS, "EXIT STATUS" },
   1245 	    { MANSEC_EXAMPLES, "EXAMPLES" },
   1246 	    { MANSEC_EXAMPLES, "EXAMPLE" },
   1247 	    { MANSEC_STANDARDS, "STANDARDS" },
   1248 	    { MANSEC_HISTORY, "HISTORY" },
   1249 	    { MANSEC_BUGS, "BUGS" },
   1250 	    { MANSEC_AUTHORS, "AUTHORS" },
   1251 	    { MANSEC_COPYRIGHT, "COPYRIGHT" },
   1252 	};
   1253 	const struct roff_node *head;
   1254 	char *name_desc;
   1255 	size_t sz;
   1256 	size_t i;
   1257 
   1258 	if ((head = n->parent->head) == NULL || (head = head->child) == NULL ||
   1259 	    head->type != ROFFT_TEXT)
   1260 		return;
   1261 
   1262 	/*
   1263 	 * Check if this section should be extracted and
   1264 	 * where it should be stored. Handled the trivial cases first.
   1265 	 */
   1266 	for (i = 0; i < sizeof(mapping) / sizeof(mapping[0]); ++i) {
   1267 		if (strcmp(head->string, mapping[i].header) == 0) {
   1268 			man_parse_section(mapping[i].section, n, rec);
   1269 			return;
   1270 		}
   1271 	}
   1272 
   1273 	if (strcmp(head->string, "NAME") == 0) {
   1274 		/*
   1275 		 * We are in the NAME section.
   1276 		 * pman_parse_name will put the complete content in name_desc.
   1277 		 */
   1278 		pman_parse_name(n, rec);
   1279 
   1280 		name_desc = rec->name_desc;
   1281 		if (name_desc == NULL)
   1282 			return;
   1283 
   1284 		/* Remove any leading spaces. */
   1285 		while (name_desc[0] == ' ')
   1286 			name_desc++;
   1287 
   1288 		/* If the line begins with a "\&", avoid those */
   1289 		if (name_desc[0] == '\\' && name_desc[1] == '&')
   1290 			name_desc += 2;
   1291 
   1292 		/* Now name_desc should be left with a comma-space
   1293 		 * separated list of names and the one line description
   1294 		 * of the page:
   1295 		 *     "a, b, c \- sample description"
   1296 		 * Take out the first name, before the first comma
   1297 		 * (or space) and store it in rec->name.
   1298 		 * If the page has aliases then they should be
   1299 		 * in the form of a comma separated list.
   1300 		 * Keep looping while there is a comma in name_desc,
   1301 		 * extract the alias name and store in rec->links.
   1302 		 * When there are no more commas left, break out.
   1303 		 */
   1304 		int has_alias = 0;	// Any more aliases left?
   1305 		while (*name_desc) {
   1306 			/* Remove any leading spaces or hyphens. */
   1307 			if (name_desc[0] == ' ' || name_desc[0] == '-') {
   1308 				name_desc++;
   1309 				continue;
   1310 			}
   1311 			sz = strcspn(name_desc, ", ");
   1312 
   1313 			/* Extract the first term and store it in rec->name. */
   1314 			if (rec->name == NULL) {
   1315 				if (name_desc[sz] == ',')
   1316 					has_alias = 1;
   1317 				rec->name = estrndup(name_desc, sz);
   1318 				/* XXX This would only happen with a poorly
   1319 				 * written man page, maybe warn? */
   1320 				if (name_desc[sz] == '\0')
   1321 					break;
   1322 				name_desc += sz + 1;
   1323 				continue;
   1324 			}
   1325 
   1326 			/*
   1327 			 * Once rec->name is set, rest of the names
   1328 			 * are to be treated as links or aliases.
   1329 			 */
   1330 			if (rec->name && has_alias) {
   1331 				if (name_desc[sz] != ',') {
   1332 					/* No more commas left --> no more
   1333 					 * aliases to take out */
   1334 					has_alias = 0;
   1335 				}
   1336 				concat2(&rec->links, name_desc, sz);
   1337 				/* XXX This would only happen with a poorly
   1338 				 * written man page, maybe warn? */
   1339 				if (name_desc[sz] == '\0')
   1340 					break;
   1341 				name_desc += sz + 1;
   1342 				continue;
   1343 			}
   1344 			break;
   1345 		}
   1346 
   1347 		/* Parse any escape sequences that might be there */
   1348 		char *temp = parse_escape(name_desc);
   1349 		free(rec->name_desc);
   1350 		rec->name_desc = temp;
   1351 		temp = parse_escape(rec->name);
   1352 		free(rec->name);
   1353 		rec->name = temp;
   1354 		return;
   1355 	}
   1356 
   1357 	/* The RETURN VALUE section might be specified in multiple ways */
   1358 	if (strcmp(head->string, "RETURN") == 0 &&
   1359 	    head->next != NULL && head->next->type == ROFFT_TEXT &&
   1360 	    (strcmp(head->next->string, "VALUE") == 0 ||
   1361 	    strcmp(head->next->string, "VALUES") == 0)) {
   1362 		man_parse_section(MANSEC_RETURN_VALUES, n, rec);
   1363 		return;
   1364 	}
   1365 
   1366 	/*
   1367 	 * EXIT STATUS section can also be specified all on one line or on two
   1368 	 * separate lines.
   1369 	 */
   1370 	if (strcmp(head->string, "EXIT") == 0 &&
   1371 	    head->next != NULL && head->next->type == ROFFT_TEXT &&
   1372 	    strcmp(head->next->string, "STATUS") == 0) {
   1373 		man_parse_section(MANSEC_EXIT_STATUS, n, rec);
   1374 		return;
   1375 	}
   1376 
   1377 	/* Store the rest of the content in desc. */
   1378 	man_parse_section(MANSEC_NONE, n, rec);
   1379 }
   1380 
   1381 /*
   1382  * pman_parse_node --
   1383  *  Generic function to iterate through a node. Usually called from
   1384  *  man_parse_section to parse a particular section of the man page.
   1385  */
   1386 static void
   1387 pman_parse_node(const struct roff_node *n, secbuff *s)
   1388 {
   1389 	if (n == NULL)
   1390 		return;
   1391 
   1392 	if (n->type == ROFFT_TEXT)
   1393 		append(s, n->string);
   1394 
   1395 	pman_parse_node(n->child, s);
   1396 	pman_parse_node(n->next, s);
   1397 }
   1398 
   1399 /*
   1400  * man_parse_section --
   1401  *  Takes two parameters:
   1402  *   sec: Tells which section we are present in
   1403  *   n: Is the present node of the AST.
   1404  * Depending on the section, we call pman_parse_node to parse that section and
   1405  * concatenate the content from that section into the buffer for that section.
   1406  */
   1407 static void
   1408 man_parse_section(enum man_sec sec, const struct roff_node *n, mandb_rec *rec)
   1409 {
   1410 	/*
   1411 	 * If the user sepecified the 'l' flag then just parse
   1412 	 * the NAME section, ignore the rest.
   1413 	 */
   1414 	if (mflags.limit)
   1415 		return;
   1416 
   1417 	switch (sec) {
   1418 	case MANSEC_LIBRARY:
   1419 		pman_parse_node(n, &rec->lib);
   1420 		break;
   1421 	case MANSEC_RETURN_VALUES:
   1422 		pman_parse_node(n, &rec->return_vals);
   1423 		break;
   1424 	case MANSEC_ENVIRONMENT:
   1425 		pman_parse_node(n, &rec->env);
   1426 		break;
   1427 	case MANSEC_FILES:
   1428 		pman_parse_node(n, &rec->files);
   1429 		break;
   1430 	case MANSEC_EXIT_STATUS:
   1431 		pman_parse_node(n, &rec->exit_status);
   1432 		break;
   1433 	case MANSEC_DIAGNOSTICS:
   1434 		pman_parse_node(n, &rec->diagnostics);
   1435 		break;
   1436 	case MANSEC_ERRORS:
   1437 		pman_parse_node(n, &rec->errors);
   1438 		break;
   1439 	case MANSEC_NAME:
   1440 	case MANSEC_SYNOPSIS:
   1441 	case MANSEC_EXAMPLES:
   1442 	case MANSEC_STANDARDS:
   1443 	case MANSEC_HISTORY:
   1444 	case MANSEC_BUGS:
   1445 	case MANSEC_AUTHORS:
   1446 	case MANSEC_COPYRIGHT:
   1447 		break;
   1448 	default:
   1449 		pman_parse_node(n, &rec->desc);
   1450 		break;
   1451 	}
   1452 
   1453 }
   1454 
   1455 /*
   1456  * insert_into_db --
   1457  *  Inserts the parsed data of the man page in the SQLite database.
   1458  *  If any of the values is NULL, then we cleanup and return -1 indicating
   1459  *  an error.
   1460  *  Otherwise, store the data in the database and return 0.
   1461  */
   1462 static int
   1463 insert_into_db(sqlite3 *db, mandb_rec *rec)
   1464 {
   1465 	int rc = 0;
   1466 	int idx = -1;
   1467 	const char *sqlstr = NULL;
   1468 	sqlite3_stmt *stmt = NULL;
   1469 	char *ln = NULL;
   1470 	char *errmsg = NULL;
   1471 	long int mandb_rowid;
   1472 
   1473 	/*
   1474 	 * At the very minimum we want to make sure that we store
   1475 	 * the following data:
   1476 	 *   Name, one line description, and the MD5 hash
   1477 	 */
   1478 	if (rec->name == NULL || rec->name_desc == NULL ||
   1479 	    rec->md5_hash == NULL) {
   1480 		cleanup(rec);
   1481 		return -1;
   1482 	}
   1483 
   1484 	/* Write null byte at the end of all the sec_buffs */
   1485 	rec->desc.data[rec->desc.offset] = 0;
   1486 	rec->lib.data[rec->lib.offset] = 0;
   1487 	rec->env.data[rec->env.offset] = 0;
   1488 	rec->return_vals.data[rec->return_vals.offset] = 0;
   1489 	rec->exit_status.data[rec->exit_status.offset] = 0;
   1490 	rec->files.data[rec->files.offset] = 0;
   1491 	rec->diagnostics.data[rec->diagnostics.offset] = 0;
   1492 	rec->errors.data[rec->errors.offset] = 0;
   1493 
   1494 	/*
   1495 	 * In case of a mdoc page: (sorry, no better place to put this code)
   1496 	 * parse the comma separated list of names of man pages,
   1497 	 * the first name will be stored in the mandb table, rest will be
   1498 	 * treated as links and put in the mandb_links table.
   1499 	 */
   1500 	if (rec->page_type == MDOC) {
   1501 		char *tmp;
   1502 		rec->links = estrdup(rec->name);
   1503 		free(rec->name);
   1504 		size_t sz = strcspn(rec->links, " \0");
   1505 		rec->name = emalloc(sz + 1);
   1506 		memcpy(rec->name, rec->links, sz);
   1507 		if(rec->name[sz - 1] == ',')
   1508 			rec->name[sz - 1] = 0;
   1509 		else
   1510 			rec->name[sz] = 0;
   1511 		while (rec->links[sz] == ' ')
   1512 			++sz;
   1513 		tmp = estrdup(rec->links + sz);
   1514 		free(rec->links);
   1515 		rec->links = tmp;
   1516 	}
   1517 
   1518 /*------------------------ Populate the mandb table---------------------------*/
   1519 	sqlstr = "INSERT INTO mandb VALUES (:section, :name, :name_desc, :desc,"
   1520 		 " :lib, :return_vals, :env, :files, :exit_status,"
   1521 		 " :diagnostics, :errors, :md5_hash, :machine)";
   1522 
   1523 	rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL);
   1524 	if (rc != SQLITE_OK)
   1525 		goto Out;
   1526 
   1527 	idx = sqlite3_bind_parameter_index(stmt, ":name");
   1528 	rc = sqlite3_bind_text(stmt, idx, rec->name, -1, NULL);
   1529 	if (rc != SQLITE_OK) {
   1530 		sqlite3_finalize(stmt);
   1531 		goto Out;
   1532 	}
   1533 
   1534 	idx = sqlite3_bind_parameter_index(stmt, ":section");
   1535 	rc = sqlite3_bind_text(stmt, idx, rec->section, -1, NULL);
   1536 	if (rc != SQLITE_OK) {
   1537 		sqlite3_finalize(stmt);
   1538 		goto Out;
   1539 	}
   1540 
   1541 	idx = sqlite3_bind_parameter_index(stmt, ":name_desc");
   1542 	rc = sqlite3_bind_text(stmt, idx, rec->name_desc, -1, NULL);
   1543 	if (rc != SQLITE_OK) {
   1544 		sqlite3_finalize(stmt);
   1545 		goto Out;
   1546 	}
   1547 
   1548 	idx = sqlite3_bind_parameter_index(stmt, ":desc");
   1549 	rc = sqlite3_bind_text(stmt, idx, rec->desc.data,
   1550 	                       rec->desc.offset + 1, NULL);
   1551 	if (rc != SQLITE_OK) {
   1552 		sqlite3_finalize(stmt);
   1553 		goto Out;
   1554 	}
   1555 
   1556 	idx = sqlite3_bind_parameter_index(stmt, ":lib");
   1557 	rc = sqlite3_bind_text(stmt, idx, rec->lib.data,
   1558 	    rec->lib.offset + 1, NULL);
   1559 	if (rc != SQLITE_OK) {
   1560 		sqlite3_finalize(stmt);
   1561 		goto Out;
   1562 	}
   1563 
   1564 	idx = sqlite3_bind_parameter_index(stmt, ":return_vals");
   1565 	rc = sqlite3_bind_text(stmt, idx, rec->return_vals.data,
   1566 	                      rec->return_vals.offset + 1, NULL);
   1567 	if (rc != SQLITE_OK) {
   1568 		sqlite3_finalize(stmt);
   1569 		goto Out;
   1570 	}
   1571 
   1572 	idx = sqlite3_bind_parameter_index(stmt, ":env");
   1573 	rc = sqlite3_bind_text(stmt, idx, rec->env.data,
   1574 	    rec->env.offset + 1, NULL);
   1575 	if (rc != SQLITE_OK) {
   1576 		sqlite3_finalize(stmt);
   1577 		goto Out;
   1578 	}
   1579 
   1580 	idx = sqlite3_bind_parameter_index(stmt, ":files");
   1581 	rc = sqlite3_bind_text(stmt, idx, rec->files.data,
   1582 	                       rec->files.offset + 1, NULL);
   1583 	if (rc != SQLITE_OK) {
   1584 		sqlite3_finalize(stmt);
   1585 		goto Out;
   1586 	}
   1587 
   1588 	idx = sqlite3_bind_parameter_index(stmt, ":exit_status");
   1589 	rc = sqlite3_bind_text(stmt, idx, rec->exit_status.data,
   1590 	                       rec->exit_status.offset + 1, NULL);
   1591 	if (rc != SQLITE_OK) {
   1592 		sqlite3_finalize(stmt);
   1593 		goto Out;
   1594 	}
   1595 
   1596 	idx = sqlite3_bind_parameter_index(stmt, ":diagnostics");
   1597 	rc = sqlite3_bind_text(stmt, idx, rec->diagnostics.data,
   1598 	                       rec->diagnostics.offset + 1, NULL);
   1599 	if (rc != SQLITE_OK) {
   1600 		sqlite3_finalize(stmt);
   1601 		goto Out;
   1602 	}
   1603 
   1604 	idx = sqlite3_bind_parameter_index(stmt, ":errors");
   1605 	rc = sqlite3_bind_text(stmt, idx, rec->errors.data,
   1606 	                       rec->errors.offset + 1, NULL);
   1607 	if (rc != SQLITE_OK) {
   1608 		sqlite3_finalize(stmt);
   1609 		goto Out;
   1610 	}
   1611 
   1612 	idx = sqlite3_bind_parameter_index(stmt, ":md5_hash");
   1613 	rc = sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL);
   1614 	if (rc != SQLITE_OK) {
   1615 		sqlite3_finalize(stmt);
   1616 		goto Out;
   1617 	}
   1618 
   1619 	idx = sqlite3_bind_parameter_index(stmt, ":machine");
   1620 	if (rec->machine)
   1621 		rc = sqlite3_bind_text(stmt, idx, rec->machine, -1, NULL);
   1622 	else
   1623 		rc = sqlite3_bind_null(stmt, idx);
   1624 	if (rc != SQLITE_OK) {
   1625 		sqlite3_finalize(stmt);
   1626 		goto Out;
   1627 	}
   1628 
   1629 	rc = sqlite3_step(stmt);
   1630 	if (rc != SQLITE_DONE) {
   1631 		sqlite3_finalize(stmt);
   1632 		goto Out;
   1633 	}
   1634 
   1635 	sqlite3_finalize(stmt);
   1636 
   1637 	/* Get the row id of the last inserted row */
   1638 	mandb_rowid = sqlite3_last_insert_rowid(db);
   1639 
   1640 /*------------------------Populate the mandb_meta table-----------------------*/
   1641 	sqlstr = "INSERT INTO mandb_meta VALUES (:device, :inode, :mtime,"
   1642 		 " :file, :md5_hash, :id)";
   1643 	rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL);
   1644 	if (rc != SQLITE_OK)
   1645 		goto Out;
   1646 
   1647 	idx = sqlite3_bind_parameter_index(stmt, ":device");
   1648 	rc = sqlite3_bind_int64(stmt, idx, rec->device);
   1649 	if (rc != SQLITE_OK) {
   1650 		sqlite3_finalize(stmt);
   1651 		goto Out;
   1652 	}
   1653 
   1654 	idx = sqlite3_bind_parameter_index(stmt, ":inode");
   1655 	rc = sqlite3_bind_int64(stmt, idx, rec->inode);
   1656 	if (rc != SQLITE_OK) {
   1657 		sqlite3_finalize(stmt);
   1658 		goto Out;
   1659 	}
   1660 
   1661 	idx = sqlite3_bind_parameter_index(stmt, ":mtime");
   1662 	rc = sqlite3_bind_int64(stmt, idx, rec->mtime);
   1663 	if (rc != SQLITE_OK) {
   1664 		sqlite3_finalize(stmt);
   1665 		goto Out;
   1666 	}
   1667 
   1668 	idx = sqlite3_bind_parameter_index(stmt, ":file");
   1669 	rc = sqlite3_bind_text(stmt, idx, rec->file_path, -1, NULL);
   1670 	if (rc != SQLITE_OK) {
   1671 		sqlite3_finalize(stmt);
   1672 		goto Out;
   1673 	}
   1674 
   1675 	idx = sqlite3_bind_parameter_index(stmt, ":md5_hash");
   1676 	rc = sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL);
   1677 	if (rc != SQLITE_OK) {
   1678 		sqlite3_finalize(stmt);
   1679 		goto Out;
   1680 	}
   1681 
   1682 	idx = sqlite3_bind_parameter_index(stmt, ":id");
   1683 	rc = sqlite3_bind_int64(stmt, idx, mandb_rowid);
   1684 	if (rc != SQLITE_OK) {
   1685 		sqlite3_finalize(stmt);
   1686 		goto Out;
   1687 	}
   1688 
   1689 	rc = sqlite3_step(stmt);
   1690 	sqlite3_finalize(stmt);
   1691 	if (rc == SQLITE_CONSTRAINT_UNIQUE) {
   1692 		/* The *most* probable reason for reaching here is that
   1693 		 * the UNIQUE constraint on the file column of the mandb_meta
   1694 		 * table was violated.
   1695 		 * This can happen when a file was updated/modified.
   1696 		 * To fix this we need to do two things:
   1697 		 * 1. Delete the row for the older version of this file
   1698 		 *    from mandb table.
   1699 		 * 2. Run an UPDATE query to update the row for this file
   1700 		 *    in the mandb_meta table.
   1701 		 */
   1702 		warnx("Trying to update index for %s", rec->file_path);
   1703 		char *sql = sqlite3_mprintf("DELETE FROM mandb "
   1704 					    "WHERE rowid = (SELECT id"
   1705 					    "  FROM mandb_meta"
   1706 					    "  WHERE file = %Q)",
   1707 					    rec->file_path);
   1708 		sqlite3_exec(db, sql, NULL, NULL, &errmsg);
   1709 		sqlite3_free(sql);
   1710 		if (errmsg != NULL) {
   1711 			if (mflags.verbosity)
   1712 				warnx("%s", errmsg);
   1713 			free(errmsg);
   1714 		}
   1715 		sqlstr = "UPDATE mandb_meta SET device = :device,"
   1716 			 " inode = :inode, mtime = :mtime, id = :id,"
   1717 			 " md5_hash = :md5 WHERE file = :file";
   1718 		rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL);
   1719 		if (rc != SQLITE_OK) {
   1720 			if (mflags.verbosity)
   1721 				warnx("Update failed with error: %s",
   1722 			    sqlite3_errmsg(db));
   1723 			close_db(db);
   1724 			cleanup(rec);
   1725 			errx(EXIT_FAILURE,
   1726 			    "Consider running makemandb with -f option");
   1727 		}
   1728 
   1729 		idx = sqlite3_bind_parameter_index(stmt, ":device");
   1730 		sqlite3_bind_int64(stmt, idx, rec->device);
   1731 		idx = sqlite3_bind_parameter_index(stmt, ":inode");
   1732 		sqlite3_bind_int64(stmt, idx, rec->inode);
   1733 		idx = sqlite3_bind_parameter_index(stmt, ":mtime");
   1734 		sqlite3_bind_int64(stmt, idx, rec->mtime);
   1735 		idx = sqlite3_bind_parameter_index(stmt, ":id");
   1736 		sqlite3_bind_int64(stmt, idx, mandb_rowid);
   1737 		idx = sqlite3_bind_parameter_index(stmt, ":md5");
   1738 		sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL);
   1739 		idx = sqlite3_bind_parameter_index(stmt, ":file");
   1740 		sqlite3_bind_text(stmt, idx, rec->file_path, -1, NULL);
   1741 		rc = sqlite3_step(stmt);
   1742 		sqlite3_finalize(stmt);
   1743 
   1744 		if (rc != SQLITE_DONE) {
   1745 			if (mflags.verbosity)
   1746 				warnx("%s", sqlite3_errmsg(db));
   1747 			close_db(db);
   1748 			cleanup(rec);
   1749 			errx(EXIT_FAILURE,
   1750 			    "Consider running makemandb with -f option");
   1751 		}
   1752 	} else if (rc != SQLITE_DONE) {
   1753 		/* Otherwise make this error fatal */
   1754 		warnx("Failed at %s\n%s", rec->file_path, sqlite3_errmsg(db));
   1755 		cleanup(rec);
   1756 		close_db(db);
   1757 		exit(EXIT_FAILURE);
   1758 	}
   1759 
   1760 /*------------------------ Populate the mandb_links table---------------------*/
   1761 	char *str = NULL;
   1762 	char *links;
   1763 	if (rec->links && strlen(rec->links)) {
   1764 		links = rec->links;
   1765 		for(ln = strtok(links, " "); ln; ln = strtok(NULL, " ")) {
   1766 			if (ln[0] == ',')
   1767 				ln++;
   1768 			if(ln[strlen(ln) - 1] == ',')
   1769 				ln[strlen(ln) - 1] = 0;
   1770 
   1771 			str = sqlite3_mprintf("INSERT INTO mandb_links"
   1772 					      " VALUES (%Q, %Q, %Q, %Q, %Q, %Q)",
   1773 					      ln, rec->name, rec->section,
   1774 					      rec->machine, rec->md5_hash, rec->name_desc);
   1775 			sqlite3_exec(db, str, NULL, NULL, &errmsg);
   1776 			sqlite3_free(str);
   1777 			if (errmsg != NULL) {
   1778 				warnx("%s", errmsg);
   1779 				cleanup(rec);
   1780 				free(errmsg);
   1781 				return -1;
   1782 			}
   1783 		}
   1784 	}
   1785 
   1786 	cleanup(rec);
   1787 	return 0;
   1788 
   1789   Out:
   1790 	if (mflags.verbosity)
   1791 		warnx("%s", sqlite3_errmsg(db));
   1792 	cleanup(rec);
   1793 	return -1;
   1794 }
   1795 
   1796 /*
   1797  * check_md5--
   1798  *  Generates the md5 hash of the file and checks if it already doesn't exist
   1799  *  in the table.
   1800  *  This function is being used to avoid hardlinks.
   1801  *  On successful completion it will also set the value of the fourth parameter
   1802  *  to the md5 hash of the file (computed previously). It is the responsibility
   1803  *  of the caller to free this buffer.
   1804  *  Return values:
   1805  *  -1: If an error occurs somewhere and sets the md5 return buffer to NULL.
   1806  *  0: If the md5 hash does not exist in the table.
   1807  *  1: If the hash exists in the database.
   1808  */
   1809 static int
   1810 check_md5(const char *file, sqlite3 *db, char **md5, void *buf, size_t buflen)
   1811 {
   1812 	int rc = 0;
   1813 	int idx = -1;
   1814 	char *sqlstr = NULL;
   1815 	char *mymd5;
   1816 	sqlite3_stmt *stmt = NULL;
   1817 	*md5 = NULL;
   1818 
   1819 	assert(file != NULL);
   1820 	if ((mymd5 = MD5Data(buf, buflen, NULL)) == NULL) {
   1821 		if (mflags.verbosity)
   1822 			warn("md5 failed: %s", file);
   1823 		return -1;
   1824 	}
   1825 
   1826 	easprintf(&sqlstr, "SELECT * FROM mandb_meta WHERE md5_hash = :md5_hash");
   1827 	rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL);
   1828 	if (rc != SQLITE_OK) {
   1829 		free(sqlstr);
   1830 		free(mymd5);
   1831 		return -1;
   1832 	}
   1833 
   1834 	idx = sqlite3_bind_parameter_index(stmt, ":md5_hash");
   1835 	rc = sqlite3_bind_text(stmt, idx, mymd5, -1, NULL);
   1836 	if (rc != SQLITE_OK) {
   1837 		if (mflags.verbosity)
   1838 			warnx("%s", sqlite3_errmsg(db));
   1839 		sqlite3_finalize(stmt);
   1840 		free(sqlstr);
   1841 		free(mymd5);
   1842 		return -1;
   1843 	}
   1844 
   1845 	*md5 = mymd5;
   1846 	if (sqlite3_step(stmt) == SQLITE_ROW) {
   1847 		sqlite3_finalize(stmt);
   1848 		free(sqlstr);
   1849 		return 0;
   1850 	}
   1851 
   1852 	sqlite3_finalize(stmt);
   1853 	free(sqlstr);
   1854 	return 1;
   1855 }
   1856 
   1857 /* Optimize the index for faster search */
   1858 static void
   1859 optimize(sqlite3 *db)
   1860 {
   1861 	const char *sqlstr;
   1862 	char *errmsg = NULL;
   1863 
   1864 	if (mflags.verbosity == 2)
   1865 		printf("Optimizing the database index\n");
   1866 	sqlstr = "INSERT INTO mandb(mandb) VALUES (\'optimize\');"
   1867 		 "VACUUM";
   1868 	sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg);
   1869 	if (errmsg != NULL) {
   1870 		if (mflags.verbosity)
   1871 			warnx("%s", errmsg);
   1872 		free(errmsg);
   1873 		return;
   1874 	}
   1875 }
   1876 
   1877 /*
   1878  * cleanup --
   1879  *  cleans up the global buffers
   1880  */
   1881 static void
   1882 cleanup(mandb_rec *rec)
   1883 {
   1884 	rec->desc.offset = 0;
   1885 	rec->lib.offset = 0;
   1886 	rec->return_vals.offset = 0;
   1887 	rec->env.offset = 0;
   1888 	rec->exit_status.offset = 0;
   1889 	rec->diagnostics.offset = 0;
   1890 	rec->errors.offset = 0;
   1891 	rec->files.offset = 0;
   1892 
   1893 	free(rec->machine);
   1894 	rec->machine = NULL;
   1895 
   1896 	free(rec->links);
   1897 	rec->links = NULL;
   1898 
   1899 	free(rec->file_path);
   1900 	rec->file_path = NULL;
   1901 
   1902 	free(rec->name);
   1903 	rec->name = NULL;
   1904 
   1905 	free(rec->name_desc);
   1906 	rec->name_desc = NULL;
   1907 
   1908 	free(rec->md5_hash);
   1909 	rec->md5_hash = NULL;
   1910 
   1911 	free(rec->section);
   1912 	rec->section = NULL;
   1913 }
   1914 
   1915 /*
   1916  * init_secbuffs--
   1917  *  Sets the value of buflen for all the sec_buff field of rec. And then
   1918  *  allocate memory to each sec_buff member of rec.
   1919  */
   1920 static void
   1921 init_secbuffs(mandb_rec *rec)
   1922 {
   1923 	/*
   1924 	 * Some sec_buff might need more memory, for example desc,
   1925 	 * which stores the data of the DESCRIPTION section,
   1926 	 * while some might need very small amount of memory.
   1927 	 * Therefore explicitly setting the value of buflen field for
   1928 	 * each sec_buff.
   1929 	 */
   1930 	rec->desc.buflen = 10 * BUFLEN;
   1931 	rec->desc.data = emalloc(rec->desc.buflen);
   1932 	rec->desc.offset = 0;
   1933 
   1934 	rec->lib.buflen = BUFLEN / 2;
   1935 	rec->lib.data = emalloc(rec->lib.buflen);
   1936 	rec->lib.offset = 0;
   1937 
   1938 	rec->return_vals.buflen = BUFLEN;
   1939 	rec->return_vals.data = emalloc(rec->return_vals.buflen);
   1940 	rec->return_vals.offset = 0;
   1941 
   1942 	rec->exit_status.buflen = BUFLEN;
   1943 	rec->exit_status.data = emalloc(rec->exit_status.buflen);
   1944 	rec->exit_status.offset = 0;
   1945 
   1946 	rec->env.buflen = BUFLEN;
   1947 	rec->env.data = emalloc(rec->env.buflen);
   1948 	rec->env.offset = 0;
   1949 
   1950 	rec->files.buflen = BUFLEN;
   1951 	rec->files.data = emalloc(rec->files.buflen);
   1952 	rec->files.offset = 0;
   1953 
   1954 	rec->diagnostics.buflen = BUFLEN;
   1955 	rec->diagnostics.data = emalloc(rec->diagnostics.buflen);
   1956 	rec->diagnostics.offset = 0;
   1957 
   1958 	rec->errors.buflen = BUFLEN;
   1959 	rec->errors.data = emalloc(rec->errors.buflen);
   1960 	rec->errors.offset = 0;
   1961 }
   1962 
   1963 /*
   1964  * free_secbuffs--
   1965  *  This function should be called at the end, when all the pages have been
   1966  *  parsed.
   1967  *  It frees the memory allocated to sec_buffs by init_secbuffs in the starting.
   1968  */
   1969 static void
   1970 free_secbuffs(mandb_rec *rec)
   1971 {
   1972 	free(rec->desc.data);
   1973 	free(rec->lib.data);
   1974 	free(rec->return_vals.data);
   1975 	free(rec->exit_status.data);
   1976 	free(rec->env.data);
   1977 	free(rec->files.data);
   1978 	free(rec->diagnostics.data);
   1979 	free(rec->errors.data);
   1980 }
   1981 
   1982 static void
   1983 replace_hyph(char *str)
   1984 {
   1985 	char *iter = str;
   1986 	while ((iter = strchr(iter, ASCII_HYPH)) != NULL)
   1987 		*iter = '-';
   1988 
   1989 	iter = str;
   1990 	while ((iter = strchr(iter, ASCII_NBRSP)) != NULL)
   1991 		*iter = '-';
   1992 }
   1993 
   1994 static char *
   1995 parse_escape(const char *str)
   1996 {
   1997 	const char *backslash, *last_backslash;
   1998 	char *result, *iter;
   1999 	size_t len;
   2000 
   2001 	assert(str);
   2002 
   2003 	last_backslash = str;
   2004 	backslash = strchr(str, '\\');
   2005 	if (backslash == NULL) {
   2006 		result = estrdup(str);
   2007 		replace_hyph(result);
   2008 		return result;
   2009 	}
   2010 
   2011 	result = emalloc(strlen(str) + 1);
   2012 	iter = result;
   2013 
   2014 	do {
   2015 		len = backslash - last_backslash;
   2016 		memcpy(iter, last_backslash, len);
   2017 		iter += len;
   2018 		if (backslash[1] == '-' || backslash[1] == ' ') {
   2019 			*iter++ = backslash[1];
   2020 			last_backslash = backslash + 2;
   2021 			backslash = strchr(last_backslash, '\\');
   2022 		} else {
   2023 			++backslash;
   2024 			mandoc_escape(&backslash, NULL, NULL);
   2025 			last_backslash = backslash;
   2026 			if (backslash == NULL)
   2027 				break;
   2028 			backslash = strchr(last_backslash, '\\');
   2029 		}
   2030 	} while (backslash != NULL);
   2031 	if (last_backslash != NULL)
   2032 		strcpy(iter, last_backslash);
   2033 
   2034 	replace_hyph(result);
   2035 	return result;
   2036 }
   2037 
   2038 /*
   2039  * append--
   2040  *  Concatenates a space and src at the end of sbuff->data (much like concat in
   2041  *  apropos-utils.c).
   2042  *  Rather than reallocating space for writing data, it uses the value of the
   2043  *  offset field of sec_buff to write new data at the free space left in the
   2044  *  buffer.
   2045  *  In case the size of the data to be appended exceeds the number of bytes left
   2046  *  in the buffer, it reallocates buflen number of bytes and then continues.
   2047  *  Value of offset field should be adjusted as new data is written.
   2048  *
   2049  *  NOTE: This function does not write the null byte at the end of the buffers,
   2050  *  write a null byte at the position pointed to by offset before inserting data
   2051  *  in the db.
   2052  */
   2053 static void
   2054 append(secbuff *sbuff, const char *src)
   2055 {
   2056 	short flag = 0;
   2057 	size_t srclen, newlen;
   2058 	char *temp;
   2059 
   2060 	assert(src != NULL);
   2061 	temp = parse_escape(src);
   2062 	srclen = strlen(temp);
   2063 
   2064 	if (sbuff->data == NULL) {
   2065 		sbuff->data = emalloc(sbuff->buflen);
   2066 		sbuff->offset = 0;
   2067 	}
   2068 
   2069 	newlen = sbuff->offset + srclen + 2;
   2070 	if (newlen >= sbuff->buflen) {
   2071 		while (sbuff->buflen < newlen)
   2072 			sbuff->buflen += sbuff->buflen;
   2073 		sbuff->data = erealloc(sbuff->data, sbuff->buflen);
   2074 		flag = 1;
   2075 	}
   2076 
   2077 	/* Append a space at the end of the buffer. */
   2078 	if (sbuff->offset || flag)
   2079 		sbuff->data[sbuff->offset++] = ' ';
   2080 	/* Now, copy src at the end of the buffer. */
   2081 	memcpy(sbuff->data + sbuff->offset, temp, srclen);
   2082 	sbuff->offset += srclen;
   2083 	free(temp);
   2084 }
   2085 
   2086 static void
   2087 usage(void)
   2088 {
   2089 	fprintf(stderr, "Usage: %s [-floQqv] [-C path]\n", getprogname());
   2090 	exit(1);
   2091 }
   2092