From 70b7c16010afef16a75eceb5ba1aa51143ba4057 Mon Sep 17 00:00:00 2001
From: Swenyu Duan <dsy@sina.com>
Date: Thu, 5 Jul 2012 09:25:48 -0400
Subject: [PATCH] A optimized version of normalize function in C. In preivous
 version of normalize function. The norm2_getInstance function is being called
 each time calls the normalize function. This function will read in some
 really large file to memory which will be a huge consumption in  time. In the
 new implementation, this function is moved to a PG_fini  function which will
 be called when the extension's shared library is loading into the memory. In
 this way, the initialized normalizer will could be shared as a global
 variable instead of loading it in each normalize operation. Besides it, some
 of unneccessary memory allocation function is also removed. Signed-off-by:
 Swenyu Duan <dsy@sina.com>

---
 .../sql/Pg/extensions/normalize.functions_in_c.c   | 72 +++++++++++++---------
 1 file changed, 43 insertions(+), 29 deletions(-)

diff --git a/Open-ILS/src/sql/Pg/extensions/normalize.functions_in_c.c b/Open-ILS/src/sql/Pg/extensions/normalize.functions_in_c.c
index a2128bba63..82422be685 100755
--- a/Open-ILS/src/sql/Pg/extensions/normalize.functions_in_c.c
+++ b/Open-ILS/src/sql/Pg/extensions/normalize.functions_in_c.c
@@ -23,13 +23,15 @@
 PG_MODULE_MAGIC;
 #endif
 
+UNormalizer2 *normalizer;
+
 static int32_t regexp_expand_string(const UChar *src,
                                     int32_t src_len,
                                     UChar *des,
                                     int32_t des_capacity)
 {
     int des_len;
-    UChar s, e;
+    UChar s;
     const UChar *cur_pos;
     const UChar *pre_pos;
 
@@ -88,7 +90,7 @@ static int32_t regexp_transliterate(const UChar *search_list,
 									UChar *des,
 									int32_t des_capacity)
 {
-	int i, j;
+	int i;
 	int32_t des_len;
 	UChar *cur_pos;
 
@@ -108,7 +110,7 @@ static int32_t regexp_transliterate(const UChar *search_list,
 		}
 		des_len++;
 
-		if(cur_pos = u_strchr(search_list, src[i]))
+		if((cur_pos = u_strchr(search_list, src[i])))
 		{
 			if(cur_pos - search_list > replacement_list_len)
 				des_len--;
@@ -290,26 +292,32 @@ static UChar *u_strreplace(UChar *regexp,
 							 replacement_len,
 							 str,
 							 str_len,
-							 NULL,
-							 0,
+							 str,
+							 str_len,
 							 is_global);
 
-	des = palloc((des_len + 1)* sizeof(UChar));
-
-	des_len = regexp_replace(regexp,
-							regexp_len,
-							replacement,
-							replacement_len,
-							str,
-							str_len,
-							des,
-							des_len,
-							is_global);
-	if(des != NULL)
-		des[des_len] = '\0';
-
-	pfree(str);
-	return des;
+	if (des_len > str_capacity)
+	{
+		pfree(str);
+
+		des = palloc((des_len + 1)* sizeof(UChar));
+
+		des_len = regexp_replace(regexp,
+			regexp_len,
+			replacement,
+			replacement_len,
+			str,
+			str_len,
+			des,
+			des_len,
+			is_global);
+		if(des != NULL)
+			des[des_len] = '\0';
+		
+		return des;
+	}
+	
+	return str;
 }
 
 static UChar *additional_substitutions(UChar *nustr, int is_search)
@@ -522,26 +530,32 @@ static UChar *leading_trailing_spaces(UChar * nustr)
 
 }
 
+void _PG_fini()
+{
+	UErrorCode err = 0;
+	normalizer = unorm2_getInstance(NULL, "nfkc", UNORM2_DECOMPOSE, &err);
+	if (U_FAILURE(err))
+	{
+		normalizer = NULL;
+	}
+	return;
+}
+
 text *normalize(text *str, text *sf, int is_search)
 {
 	UChar *ustr = NULL, *nustr = NULL, *temp = NULL, *usf = NULL;
 	int32_t nustr_len, temp_len, str_len, sf_len, ustr_len;
-	UNormalizer2 *normalizer;
-	char *regexp, *result;
+	char *result;
 	char *s;
 	UChar uregexp[200], replacement[200];
 	UErrorCode err = 0;
-
+	
 	if (str == NULL || sf == NULL)
 	{
 		return NULL;
 	}
 
-	normalizer = unorm2_getNFKDInstance(&err);
-	if (U_FAILURE(err))
-	{
-		return NULL;
-	}
+	
 
 	s = VARDATA(str);
 
-- 
2.11.0