From 17c92cf5b2b0dfdfdbbdb5c41354634ca98ae1b4 Mon Sep 17 00:00:00 2001
From: Mike Buland <eichlan@xagasoft.com>
Date: Thu, 7 Jan 2010 00:24:08 +0000
Subject: Added a new class, RegEx, it does extended regular expressions for
 now, more to come.

---
 src/regex.cpp       | 88 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 src/regex.h         | 37 ++++++++++++++++++++++
 src/tests/regex.cpp | 31 +++++++++++++++++++
 3 files changed, 156 insertions(+)
 create mode 100644 src/regex.cpp
 create mode 100644 src/regex.h
 create mode 100644 src/tests/regex.cpp

diff --git a/src/regex.cpp b/src/regex.cpp
new file mode 100644
index 0000000..a9e09bb
--- /dev/null
+++ b/src/regex.cpp
@@ -0,0 +1,88 @@
+#include "bu/regex.h"
+
+#include <regex.h> // Please, please include the system regex.h file
+
+#define re ((regex_t *)pRegEx)
+#define aSubStr ((regmatch_t *)paSubStr)
+
+Bu::RegEx::RegEx() :
+	pRegEx( NULL ),
+	bCompiled( false ),
+	paSubStr( NULL )
+{
+}
+
+Bu::RegEx::RegEx( const Bu::FString &sSrc ) :
+	pRegEx( NULL ),
+	bCompiled( false ),
+	paSubStr( NULL )
+{
+	compile( sSrc );
+}
+
+Bu::RegEx::~RegEx()
+{
+	if( bCompiled )
+	{
+		regfree( re );
+		delete re;
+		delete[] aSubStr;
+	}
+}
+
+void Bu::RegEx::compile( const Bu::FString &sSrc )
+{
+	if( bCompiled )
+	{
+		regfree( re );
+		delete re;
+		delete[] aSubStr;
+		bCompiled = false;
+	}
+	pRegEx = (void *)(new regex_t);
+
+	int nErr = regcomp( re, sSrc.getStr(), REG_EXTENDED|REG_NEWLINE );
+	if( nErr )
+	{
+		size_t length = regerror( nErr, re, NULL, 0 );
+		char *buffer = new char[length];
+		(void) regerror( nErr, re, buffer, length );
+		Bu::FString s( buffer );
+		delete[] buffer;
+		throw "???"; // BuildException( s.getStr() );
+	}
+	bCompiled = true;
+	this->sSrc = sSrc;
+
+	nSubStr = re->re_nsub+1;
+	paSubStr = (void *)(new regmatch_t[nSubStr]);
+}
+
+int Bu::RegEx::getNumSubStrings()
+{
+	return nSubStr;
+}
+
+bool Bu::RegEx::execute( const Bu::FString &sSrc )
+{
+	sTest = sSrc;
+	if( regexec( re, sSrc.getStr(), nSubStr, aSubStr, 0 ) )
+		return false;
+	return true;
+}
+
+void Bu::RegEx::getSubStringRange( int nIndex, int &iStart, int &iEnd )
+{
+	iStart = aSubStr[nIndex].rm_so;
+	iEnd = aSubStr[nIndex].rm_eo;
+}
+
+Bu::FString Bu::RegEx::getSubString( int nIndex )
+{
+	regmatch_t *Subs = aSubStr;
+	return Bu::FString(
+		sTest.getStr()+aSubStr[nIndex].rm_so,
+		aSubStr[nIndex].rm_eo - aSubStr[nIndex].rm_so
+		);
+}
+
diff --git a/src/regex.h b/src/regex.h
new file mode 100644
index 0000000..11ff9f0
--- /dev/null
+++ b/src/regex.h
@@ -0,0 +1,37 @@
+#ifndef BU_REG_EX_H
+#define BU_REG_EX_H
+
+#include "bu/fstring.h"
+
+#include <stdint.h>
+
+namespace Bu
+{
+	class RegEx
+	{
+	public:
+		RegEx();
+		RegEx( const Bu::FString &sSrc );
+		virtual ~RegEx();
+
+		void compile( const Bu::FString &sSrc );
+		int getNumSubStrings();
+		bool execute( const Bu::FString &sSrc );
+		void getSubStringRange( int nIndex, int &iStart, int &iEnd );
+		Bu::FString getSubString( int nIndex );
+		const Bu::FString &getSource()
+		{
+			return sSrc;
+		}
+
+	private:
+		Bu::FString sSrc;
+		Bu::FString sTest;
+		void *pRegEx;
+		bool bCompiled;
+		int nSubStr;
+		void *paSubStr;
+	};
+};
+
+#endif
diff --git a/src/tests/regex.cpp b/src/tests/regex.cpp
new file mode 100644
index 0000000..eb3aff6
--- /dev/null
+++ b/src/tests/regex.cpp
@@ -0,0 +1,31 @@
+#include <bu/regex.h>
+#include <stdio.h>
+
+int main( int argc, char *argv[] )
+{
+	if( argc < 3 )
+	{
+		printf("No...  %s <regex> <string>\n\n", argv[0] );
+		return 0;
+	}
+
+	Bu::RegEx re( argv[1] );
+
+	printf("Regex:     %s\n", argv[1] );
+	printf("Match:     %s\n", argv[2] );
+	
+	if( re.execute( argv[2] ) )
+	{
+		for( int j = 0; j < re.getNumSubStrings(); j++ )
+		{
+			printf("SubStr %d: %s\n", j, re.getSubString( j ).getStr() );
+		}
+	}
+	else
+	{
+		printf("Regex did not match.\n");
+	}
+
+	return 0;
+}
+
-- 
cgit v1.2.3