===================================================================
RCS file: /cvsrepo/anoncvs/cvs/src/usr.bin/mandoc/mandocdb.c,v
retrieving revision 1.52
retrieving revision 1.53
diff -c -r1.52 -r1.53
*** src/usr.bin/mandoc/mandocdb.c	2014/01/02 22:19:38	1.52
--- src/usr.bin/mandoc/mandocdb.c	2014/01/02 22:44:07	1.53
***************
*** 1,4 ****
! /*	$Id: mandocdb.c,v 1.52 2014/01/02 22:19:38 schwarze Exp $ */
  /*
   * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
   * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
--- 1,4 ----
! /*	$Id: mandocdb.c,v 1.53 2014/01/02 22:44:07 schwarze Exp $ */
  /*
   * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
   * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
***************
*** 74,83 ****
  };
  
  struct	str {
! 	char		*utf8; /* key in UTF-8 form */
  	const struct mpage *mpage; /* if set, the owning parse */
  	uint64_t	 mask; /* bitmask in sequence */
! 	char		 key[]; /* the string itself */
  };
  
  struct	inodev {
--- 74,83 ----
  };
  
  struct	str {
! 	char		*rendered; /* key in UTF-8 or ASCII form */
  	const struct mpage *mpage; /* if set, the owning parse */
  	uint64_t	 mask; /* bitmask in sequence */
! 	char		 key[]; /* may contain escape sequences */
  };
  
  struct	inodev {
***************
*** 157,167 ****
  			const char *, size_t, uint64_t);
  static	void	 putmdockey(const struct mpage *,
  			const struct mdoc_node *, uint64_t);
  static	void	 say(const char *, const char *, ...);
  static	int	 set_basedir(const char *);
  static	int	 treescan(void);
  static	size_t	 utf8(unsigned int, char [7]);
- static	void	 utf8key(struct mchars *, struct str *);
  
  static	char		*progname;
  static	int	 	 use_all; /* use all found files */
--- 157,167 ----
  			const char *, size_t, uint64_t);
  static	void	 putmdockey(const struct mpage *,
  			const struct mdoc_node *, uint64_t);
+ static	void	 render_key(struct mchars *, struct str *);
  static	void	 say(const char *, const char *, ...);
  static	int	 set_basedir(const char *);
  static	int	 treescan(void);
  static	size_t	 utf8(unsigned int, char [7]);
  
  static	char		*progname;
  static	int	 	 use_all; /* use all found files */
***************
*** 1654,1664 ****
  }
  
  /*
!  * Store the UTF-8 version of a key, or alias the pointer if the key has
!  * no UTF-8 transcription marks in it.
   */
  static void
! utf8key(struct mchars *mc, struct str *key)
  {
  	size_t		 sz, bsz, pos;
  	char		 utfbuf[7], res[5];
--- 1654,1664 ----
  }
  
  /*
!  * Store the rendered version of a key, or alias the pointer
!  * if the key contains no escape sequences.
   */
  static void
! render_key(struct mchars *mc, struct str *key)
  {
  	size_t		 sz, bsz, pos;
  	char		 utfbuf[7], res[5];
***************
*** 1667,1673 ****
  	int		 len, u;
  	enum mandoc_esc	 esc;
  
! 	assert(NULL == key->utf8);
  
  	res[0] = '\\';
  	res[1] = '\t';
--- 1667,1673 ----
  	int		 len, u;
  	enum mandoc_esc	 esc;
  
! 	assert(NULL == key->rendered);
  
  	res[0] = '\\';
  	res[1] = '\t';
***************
*** 1683,1689 ****
  	 * pointer as ourselvse and get out of here.
  	 */
  	if (strcspn(val, res) == bsz) {
! 		key->utf8 = key->key;
  		return;
  	} 
  
--- 1683,1689 ----
  	 * pointer as ourselvse and get out of here.
  	 */
  	if (strcspn(val, res) == bsz) {
! 		key->rendered = key->key;
  		return;
  	} 
  
***************
*** 1762,1775 ****
  	}
  
  	buf[pos] = '\0';
! 	key->utf8 = buf;
  }
  
  /*
   * Flush the current page's terms (and their bits) into the database.
   * Wrap the entire set of additions in a transaction to make sqlite be a
   * little faster.
!  * Also, UTF-8-encode the description at the last possible moment.
   */
  static void
  dbindex(const struct mpage *mpage, struct mchars *mc)
--- 1762,1775 ----
  	}
  
  	buf[pos] = '\0';
! 	key->rendered = buf;
  }
  
  /*
   * Flush the current page's terms (and their bits) into the database.
   * Wrap the entire set of additions in a transaction to make sqlite be a
   * little faster.
!  * Also, handle escape sequences at the last possible moment.
   */
  static void
  dbindex(const struct mpage *mpage, struct mchars *mc)
***************
*** 1792,1800 ****
  		key = ohash_find(&strings,
  			ohash_qlookup(&strings, mpage->desc));
  		assert(NULL != key);
! 		if (NULL == key->utf8)
! 			utf8key(mc, key);
! 		desc = key->utf8;
  	}
  
  	SQL_EXEC("BEGIN TRANSACTION");
--- 1792,1800 ----
  		key = ohash_find(&strings,
  			ohash_qlookup(&strings, mpage->desc));
  		assert(NULL != key);
! 		if (NULL == key->rendered)
! 			render_key(mc, key);
! 		desc = key->rendered;
  	}
  
  	SQL_EXEC("BEGIN TRANSACTION");
***************
*** 1828,1843 ****
  	for (key = ohash_first(&strings, &slot); NULL != key;
  	     key = ohash_next(&strings, &slot)) {
  		assert(key->mpage == mpage);
! 		if (NULL == key->utf8)
! 			utf8key(mc, key);
  		i = 1;
  		SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
! 		SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->utf8);
  		SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno);
  		SQL_STEP(stmts[STMT_INSERT_KEY]);
  		sqlite3_reset(stmts[STMT_INSERT_KEY]);
! 		if (key->utf8 != key->key)
! 			free(key->utf8);
  		free(key);
  	}
  
--- 1828,1843 ----
  	for (key = ohash_first(&strings, &slot); NULL != key;
  	     key = ohash_next(&strings, &slot)) {
  		assert(key->mpage == mpage);
! 		if (NULL == key->rendered)
! 			render_key(mc, key);
  		i = 1;
  		SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
! 		SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
  		SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno);
  		SQL_STEP(stmts[STMT_INSERT_KEY]);
  		sqlite3_reset(stmts[STMT_INSERT_KEY]);
! 		if (key->rendered != key->key)
! 			free(key->rendered);
  		free(key);
  	}