Glob now uses UTF-8 strings and support caseless comparison

Foundation/include/Poco/Glob.h

@@ -1,7 +1,7 @@
 //
 // Glob.h
 //
-// $Id: //poco/svn/Foundation/include/Poco/Glob.h#2 $
+// $Id: //poco/1.3/Foundation/include/Poco/Glob.h#3 $
 //
 // Library: Foundation
 // Package: Filesystem
@@ -9,7 +9,7 @@
 //
 // Definition of the Glob class.
 //
-// Copyright (c) 2004-2006, Applied Informatics Software Engineering GmbH.
+// Copyright (c) 2004-2009, Applied Informatics Software Engineering GmbH.
 // and Contributors.
 //
 // Permission is hereby granted, free of charge, to any person or organization
@@ -41,6 +41,7 @@
 
 
 #include "Poco/Foundation.h"
+#include "Poco/TextIterator.h"
 #include <set>
 
 
@@ -67,7 +68,7 @@ class Foundation_API Glob
 	/// To suppress the special syntactic significance of any of '[]*?!-\',
 	/// and match the character exactly, precede it with a backslash.
 	///
-	/// UTF-8 encoded strings are not supported.
+	/// All strings are assumed to be UTF-8 encoded.
 {
 public:
 	enum Options
@@ -76,6 +77,7 @@ public:
 		GLOB_DEFAULT         = 0x00, /// default behavior
 		GLOB_DOT_SPECIAL     = 0x01, /// '*' and '?' do not match '.' at beginning of subject
 		GLOB_FOLLOW_SYMLINKS = 0x02, /// follow symbolic links
+		GLOB_CASELESS        = 0x04, /// ignore case when comparing characters
 		GLOB_DIRS_ONLY       = 0x80  /// only glob for directories (for internal use only)
 	};
 	
@@ -138,9 +140,9 @@ public:
 		/// ignored.
 
 protected:
-	bool match(std::string::const_iterator& itp, const std::string::const_iterator& endp, std::string::const_iterator& its, const std::string::const_iterator& ends);
-	bool matchAfterAsterisk(std::string::const_iterator itp, const std::string::const_iterator& endp, std::string::const_iterator its, const std::string::const_iterator& ends);
-	bool matchSet(std::string::const_iterator& itp, const std::string::const_iterator& endp, char c);
+	bool match(TextIterator& itp, const TextIterator& endp, TextIterator& its, const TextIterator& ends);
+	bool matchAfterAsterisk(TextIterator itp, const TextIterator& endp, TextIterator its, const TextIterator& ends);
+	bool matchSet(TextIterator& itp, const TextIterator& endp, int c);
 	static void collect(const Path& pathPattern, const Path& base, const Path& current, const std::string& pattern, std::set<std::string>& files, int options);
 	static bool isDirectory(const Path& path, bool followSymlink);
 	

Foundation/src/Glob.cpp

@@ -1,7 +1,7 @@
 //
 // Glob.cpp
 //
-// $Id: //poco/svn/Foundation/src/Glob.cpp#2 $
+// $Id: //poco/1.3/Foundation/src/Glob.cpp#3 $
 //
 // Library: Foundation
 // Package: Filesystem
@@ -39,6 +39,8 @@
 #include "Poco/Exception.h"
 #include "Poco/DirectoryIterator.h"
 #include "Poco/File.h"
+#include "Poco/UTF8Encoding.h"
+#include "Poco/Unicode.h"
 
 
 namespace Poco {
@@ -59,10 +61,11 @@ Glob::~Glob()
 
 bool Glob::match(const std::string& subject)
 {
-	std::string::const_iterator itp  = _pattern.begin();
-	std::string::const_iterator endp = _pattern.end();
-	std::string::const_iterator its  = subject.begin();
-	std::string::const_iterator ends = subject.end();
+	UTF8Encoding utf8;
+	TextIterator itp(_pattern, utf8);
+	TextIterator endp(_pattern);
+	TextIterator its(subject, utf8);
+	TextIterator ends(subject);
 	
 	if ((_options & GLOB_DOT_SPECIAL) && its != ends && *its == '.' && (*itp == '?' || *itp == '*'))
 		return false;
@@ -100,7 +103,7 @@ void Glob::glob(const Path& pathPattern, std::set<std::string>& files, int optio
 }
 
 
-bool Glob::match(std::string::const_iterator& itp, const std::string::const_iterator& endp, std::string::const_iterator& its, const std::string::const_iterator& ends)
+bool Glob::match(TextIterator& itp, const TextIterator& endp, TextIterator& its, const TextIterator& ends)
 {
 	while (itp != endp)
 	{
@@ -138,7 +141,14 @@ bool Glob::match(std::string::const_iterator& itp, const std::string::const_iter
 			if (++itp == endp) throw SyntaxException("backslash must be followed by character in glob pattern");
 			// fallthrough
 		default:
-			if (*itp != *its) return false;
+			if (_options & GLOB_CASELESS)
+			{
+				if (Unicode::toLower(*itp) != Unicode::toLower(*its)) return false;
+			}
+			else
+			{
+				if (*itp != *its) return false;
+			}
 			++itp; ++its;
 		}
 	}
@@ -146,14 +156,17 @@ bool Glob::match(std::string::const_iterator& itp, const std::string::const_iter
 }
 
 
-bool Glob::matchAfterAsterisk(std::string::const_iterator itp, const std::string::const_iterator& endp, std::string::const_iterator its, const std::string::const_iterator& ends)
+bool Glob::matchAfterAsterisk(TextIterator itp, const TextIterator& endp, TextIterator its, const TextIterator& ends)
 {
 	return match(itp, endp, its, ends);
 }
 
 
-bool Glob::matchSet(std::string::const_iterator& itp, const std::string::const_iterator& endp, char c)
+bool Glob::matchSet(TextIterator& itp, const TextIterator& endp, int c)
 {
+	if (_options & GLOB_CASELESS)
+		c = Unicode::toLower(c);
+
 	while (itp != endp)
 	{
 		switch (*itp)
@@ -164,8 +177,8 @@ bool Glob::matchSet(std::string::const_iterator& itp, const std::string::const_i
 		case '\\':
 			if (++itp == endp) throw SyntaxException("backslash must be followed by character in glob pattern");
 		}
-		char first = *itp;
-		char last  = first;
+		int first = *itp;
+		int last  = first;
 		if (++itp != endp && *itp == '-')
 		{
 			if (++itp != endp)
@@ -173,6 +186,11 @@ bool Glob::matchSet(std::string::const_iterator& itp, const std::string::const_i
 			else
 				throw SyntaxException("bad range syntax in glob pattern");
 		}
+		if (_options & GLOB_CASELESS)
+		{
+			first = Unicode::toLower(first);
+			last  = Unicode::toLower(last);
+		}
 		if (first <= c && c <= last)
 		{
 			while (itp != endp)

Foundation/testsuite/src/GlobTest.cpp

@@ -1,7 +1,7 @@
 //
 // GlobTest.cpp
 //
-// $Id: //poco/svn/Foundation/testsuite/src/GlobTest.cpp#2 $
+// $Id: //poco/1.3/Foundation/testsuite/src/GlobTest.cpp#2 $
 //
 // Copyright (c) 2004-2006, Applied Informatics Software Engineering GmbH.
 // and Contributors.
@@ -331,6 +331,72 @@ void GlobTest::testMisc()
 }
 
 
+void GlobTest::testCaseless()
+{
+	Glob g1("*.cpp", Glob::GLOB_CASELESS);
+	assert (g1.match("Glob.cpp"));
+	assert (!g1.match("Glob.h"));
+	assert (g1.match("Glob.CPP"));
+	assert (!g1.match("Glob.H"));
+	
+	Glob g2("*.[hc]", Glob::GLOB_CASELESS);
+	assert (g2.match("foo.c"));
+	assert (g2.match("foo.h"));
+	assert (!g2.match("foo.i"));
+	assert (g2.match("foo.C"));
+	assert (g2.match("foo.H"));
+	assert (!g2.match("foo.I"));
+		
+	Glob g4("File*.?pp", Glob::GLOB_CASELESS);
+	assert (g4.match("file.hpp"));
+	assert (g4.match("FILE.CPP"));
+	assert (g4.match("filesystem.hpp"));
+	assert (g4.match("FILESYSTEM.HPP"));
+	assert (!g4.match("FILE.H"));
+	assert (!g4.match("file.h"));
+	
+	Glob g5("File*.[ch]*", Glob::GLOB_CASELESS);
+	assert (g5.match("file.hpp"));
+	assert (g5.match("FILE.HPP"));
+	assert (g5.match("file.cpp"));
+	assert (g5.match("FILE.CPP"));
+	assert (g5.match("filesystem.hpp"));
+	assert (g5.match("FILESYSTEM.HPP"));
+	assert (g5.match("file.h"));
+	assert (g5.match("FILE.H"));
+	assert (g5.match("filesystem.cp"));
+	assert (g5.match("FILESYSTEM.CP"));
+
+	Glob g6("[abc]", Glob::GLOB_CASELESS);
+	assert (g6.match("a"));
+	assert (g6.match("b"));
+	assert (g6.match("c"));
+	assert (g6.match("A"));
+	assert (g6.match("B"));
+	assert (g6.match("C"));
+
+	Glob g7("[a-f]", Glob::GLOB_CASELESS);
+	assert (g7.match("a"));
+	assert (g7.match("b"));
+	assert (g7.match("f"));
+	assert (!g7.match("g"));
+	assert (g7.match("A"));
+	assert (g7.match("B"));
+	assert (g7.match("F"));
+	assert (!g7.match("G"));
+
+	Glob g8("[A-F]", Glob::GLOB_CASELESS);
+	assert (g8.match("a"));
+	assert (g8.match("b"));
+	assert (g8.match("f"));
+	assert (!g8.match("g"));
+	assert (g8.match("A"));
+	assert (g8.match("B"));
+	assert (g8.match("F"));
+	assert (!g8.match("G"));
+}
+
+
 void GlobTest::testGlob()
 {
 	createFile("globtest/Makefile");
@@ -353,6 +419,16 @@ void GlobTest::testGlob()
 	assert (files.find("globtest/include/") != files.end());
 	assert (files.find("globtest/src/") != files.end());
 	assert (files.find("globtest/testsuite/") != files.end());
+
+	files.clear();
+	Glob::glob("GlobTest/*", files, Glob::GLOB_CASELESS);
+	translatePaths(files);
+	assert (files.size() == 5);
+	assert (files.find("globtest/Makefile") != files.end());
+	assert (files.find("globtest/.hidden") != files.end());
+	assert (files.find("globtest/include/") != files.end());
+	assert (files.find("globtest/src/") != files.end());
+	assert (files.find("globtest/testsuite/") != files.end());
 	
 	files.clear();
 	Glob::glob("globtest/*/*.[hc]", files);
@@ -370,6 +446,13 @@ void GlobTest::testGlob()
 	assert (files.size() == 2);
 	assert (files.find("globtest/testsuite/src/test.c") != files.end());
 	assert (files.find("globtest/testsuite/src/main.c") != files.end());
+
+	files.clear();
+	Glob::glob("Gl?bT?st/*/*/*.C", files, Glob::GLOB_CASELESS);
+	translatePaths(files);
+	assert (files.size() == 2);
+	assert (files.find("globtest/testsuite/src/test.c") != files.end());
+	assert (files.find("globtest/testsuite/src/main.c") != files.end());
 	
 	files.clear();
 	Glob::glob("globtest/*/src/*", files);
@@ -439,6 +522,7 @@ CppUnit::Test* GlobTest::suite()
 	CppUnit_addTest(pSuite, GlobTest, testMatchAsterisk);
 	CppUnit_addTest(pSuite, GlobTest, testMatchRange);
 	CppUnit_addTest(pSuite, GlobTest, testMisc);
+	CppUnit_addTest(pSuite, GlobTest, testCaseless);
 	CppUnit_addTest(pSuite, GlobTest, testGlob);
 
 	return pSuite;

Foundation/testsuite/src/GlobTest.h

@@ -1,7 +1,7 @@
 //
 // GlobTest.h
 //
-// $Id: //poco/svn/Foundation/testsuite/src/GlobTest.h#2 $
+// $Id: //poco/1.3/Foundation/testsuite/src/GlobTest.h#2 $
 //
 // Definition of the GlobTest class.
 //
@@ -53,6 +53,7 @@ public:
 	void testMatchRange();
 	void testMisc();
 	void testGlob();
+	void testCaseless();
 
 	void setUp();
 	void tearDown();