From 9d371f295ce42e1f87aff5b061f9652a5d897be6 Mon Sep 17 00:00:00 2001
From: kelson42 <kelson42@users.sourceforge.net>
Date: Mon, 31 Oct 2011 09:55:34 +0000
Subject: [PATCH] + use libicu in regexTools

---
 src/common/regexTools.cpp | 27 +++++++++++++++------------
 src/common/regexTools.h   |  4 ++--
 2 files changed, 17 insertions(+), 14 deletions(-)

diff --git a/src/common/regexTools.cpp b/src/common/regexTools.cpp
index 202c55821..f5257f812 100644
--- a/src/common/regexTools.cpp
+++ b/src/common/regexTools.cpp
@@ -19,33 +19,36 @@
 
 #include "regexTools.h"
 
-std::map<std::string, regex_t> regexCache;
+std::map<std::string, RegexMatcher*> regexCache;
 
-regex_t buildRegex(const std::string &regex) {
-  regex_t regexStruct;
-  std::map<std::string, regex_t>::iterator itr = regexCache.find(regex);
+RegexMatcher *buildRegex(const std::string &regex) {
+  RegexMatcher *matcher;
+  std::map<std::string, RegexMatcher*>::iterator itr = regexCache.find(regex);
   
   /* Regex is in cache */
   if (itr != regexCache.end()) {
-    regexStruct = itr->second;
+    matcher = itr->second;
   }
 
   /* Regex needs to be parsed (and cached) */
   else {
-    regcomp(&regexStruct, regex.data(), REG_ICASE);
-    regexCache[regex] = regexStruct;
+    UErrorCode status = U_ZERO_ERROR;
+    UnicodeString uregex = UnicodeString(regex.c_str());
+    matcher = new RegexMatcher(uregex, UREGEX_CASE_INSENSITIVE, status);
+    regexCache[regex] = matcher;
   }
 
-  return regexStruct;
+  return matcher;
 }
 
 /* todo */
 void freeRegexCache() {
-  //regfree(&regexStructure);
 }
 
 bool matchRegex(const std::string &content, const std::string &regex) {
-  regex_t regexStructure = buildRegex(regex);
-  bool result = !regexec(&regexStructure, content.data(), 0, 0, 0);
-  return result;
+  ucnv_setDefaultName("UTF-8");
+  UnicodeString ucontent = UnicodeString(content.c_str());
+  RegexMatcher *matcher = buildRegex(regex);
+  matcher->reset(ucontent);
+  return matcher->find();
 }
diff --git a/src/common/regexTools.h b/src/common/regexTools.h
index 3e4443d10..7037c7789 100644
--- a/src/common/regexTools.h
+++ b/src/common/regexTools.h
@@ -20,9 +20,9 @@
 #ifndef KIWIX_REGEXTOOLS_H
 #define KIWIX_REGEXTOOLS_H
 
-#include <regex.h>
+#include <unicode/regex.h>
+#include <unicode/ucnv.h>
 #include <string>
-#include <iostream>
 #include <map>
 
 bool matchRegex(const std::string &content, const std::string &regex);