From 5ba0affdd7b18ed970c039961b75da2cb925d6cd Mon Sep 17 00:00:00 2001
From: axel <axel@liljencrantz.se>
Date: Wed, 26 Oct 2005 20:51:02 +1000
Subject: [PATCH] Minor performance tweaks

darcs-hash:20051026105102-ac50b-ffa35c43fd9e1aad47229260e5d7da4249cacdcf.gz
---
 expand.c    | 49 +++++++++++++++++++++++++++++++++++++++++++++++--
 proc.c      |  1 -
 reader.c    |  2 +-
 tokenizer.c | 21 +++++++++++++++++++--
 wgetopt.h   |  4 ++--
 5 files changed, 69 insertions(+), 8 deletions(-)

diff --git a/expand.c b/expand.c
index 4236d2ab1..fce58d839 100644
--- a/expand.c
+++ b/expand.c
@@ -77,6 +77,42 @@ parameter expansion.
 */
 #define LAST_STR L"last"
 
+/**
+   Characters which make a string unclean if they are the first
+   character of the string. See \c is_clean().
+*/
+#define UNCLEAN_FIRST L"~%"
+/**
+   Unclean characters. See \c is_clean().
+*/
+#define UNCLEAN L"$*?\\\"'({})"
+
+/**
+   Test if the specified argument is clean, i.e. it does not contin
+   any tokens which are expanded. Clean strings can be passed through
+   expand_string and expand_one without changing them. About 90% of
+   all strings are clean, so skipping expantion on them actually does
+   save a small amount of time.
+*/
+static int is_clean( const wchar_t *in )
+{
+	
+	const wchar_t * str = in;
+
+	if( wcschr( UNCLEAN_FIRST, *str ) )
+		return 0;
+	while( *str )
+	{
+		if( wcschr( UNCLEAN, *str ) )
+			return 0;
+		str++;
+	}
+
+//	debug( 1, L"%ls", in );
+	
+	return 1;
+}
+
 /**
    Return the environment variable value for the string starting at in 
 */
@@ -1311,7 +1347,13 @@ int expand_string( wchar_t *str,
 	
 	int i;
 	int subshell_ok = 1;
-	
+
+	if( (!(flags & ACCEPT_INCOMPLETE)) && is_clean( str ) )
+	{
+		al_push( end_out, str );
+		return 1;
+	}
+
 	al_init( &list1 );
 	al_init( &list2 );
 
@@ -1501,6 +1543,9 @@ wchar_t *expand_one( wchar_t *string, int flags )
 	array_list_t l;
 	int res;
 	wchar_t *one;
+
+	if( (!(flags & ACCEPT_INCOMPLETE)) &&  is_clean( string ) )
+		return string;	
 	
 	al_init( &l );
 	res = expand_string( string, &l, flags );
@@ -1520,7 +1565,7 @@ wchar_t *expand_one( wchar_t *string, int flags )
 			al_set( &l, 0, 0 );
 		}
 	}
-	
+
 	al_foreach( &l, (void(*)(const void *))&free );
 	al_destroy( &l );
 	return one;
diff --git a/proc.c b/proc.c
index bc27a2147..9bae4774d 100644
--- a/proc.c
+++ b/proc.c
@@ -486,7 +486,6 @@ static void fire_process_event( const wchar_t *msg, int type, pid_t pid, int sta
 {
 	static event_t ev;
 	event_t e;
-
 	
 	e.function_name=0;
 	
diff --git a/reader.c b/reader.c
index 3057a35c1..98c9c2c66 100644
--- a/reader.c
+++ b/reader.c
@@ -93,7 +93,7 @@ commence.
 
 /**
    The maximum number of characters to read from the keyboard without
-   repainting. Note that this readahead wil only occur if new
+   repainting. Note that this readahead will only occur if new
    characters are avaialble for reading, fish will never block for
    more input without repainting.
 */
diff --git a/tokenizer.c b/tokenizer.c
index 42b2781a4..5e0fa42c7 100644
--- a/tokenizer.c
+++ b/tokenizer.c
@@ -46,7 +46,7 @@
 /**
   Characters that separate tokens. They are ordered by frequency of occurrence to increase parsing speed.
 */
-#define SEP L" \n;|#\t\r<>^&"
+#define SEP L" \n|\t;#\r<>^&"
 /**
    Tests if the tokenizer buffer is large enough to hold contents of
    the specified length, and if not, reallocates the tokenizer buffer.
@@ -192,8 +192,18 @@ static int is_string_char( wchar_t c )
 	{
 		return 0;
 	}
+	return 1;
+}
 
-  return 1;
+/**
+   Quick test to catch the most common 'non-magical' characters, makes
+   read_string slightly faster by adding a fast path for the most
+   common characters. This is obviously not a suitable replacement for
+   iswalpha.
+*/
+static int myal( wchar_t c )
+{
+	return (c>=L'a' && c<=L'z') || (c>=L'A'&&c<=L'Z');
 }
 
 /**
@@ -212,6 +222,11 @@ static void read_string( tokenizer *tok )
 
 	while( 1 )
 	{
+		
+		if( !myal( *tok->buff ) )
+		{
+//			debug(1, L"%lc", *tok->buff );
+
 		if( *tok->buff == L'\\' )
 		{
 			tok->buff++;
@@ -338,6 +353,8 @@ static void read_string( tokenizer *tok )
 				}
 				break;
 		}
+		}
+		
 
 		if( !do_loop )
 			break;
diff --git a/wgetopt.h b/wgetopt.h
index 7856add5d..9c5c9947a 100644
--- a/wgetopt.h
+++ b/wgetopt.h
@@ -64,7 +64,7 @@ extern wchar_t *woptarg;
    When `getopt' returns EOF, this is the index of the first of the
    non-option elements that the caller should itself scan.
 
-   Otherwise, `optind' communicates from one call to the next
+   Otherwise, `woptind' communicates from one call to the next
    how much of ARGV has been scanned so far.  */
 
 extern int woptind;
@@ -149,7 +149,7 @@ struct woption
 #ifdef __GNU_LIBRARY__
 /** 
 	Get options from argument list. See the glibc manual for information on how to use this function.
- */
+*/
 extern int wgetopt (int argc, wchar_t *const *argv, const wchar_t *shortopts);
 #else /* not __GNU_LIBRARY__ */