Added CM_AUTO, which automatically selects CM_STORE or CM_DEFLATE based on file extension. Used to avoid double-compression of already compressed file formats such as images.

obiltschnig · obiltschnig · commit c9590ff7d4c1 · 2014-09-08T11:06:57.000+02:00
diff --git a/Zip/include/Poco/Zip/Compress.h b/Zip/include/Poco/Zip/Compress.h
@@ -25,6 +25,7 @@
 #include "Poco/FIFOEvent.h"
 #include <istream>
 #include <ostream>
+#include <set>
 
 
 namespace Poco {
@@ -69,6 +70,25 @@ class Zip_API Compress
 	ZipArchive close();
 		/// Finalizes the ZipArchive, closes it.
 
+	void setStoreExtensions(const std::set<std::string>& extensions);
+		/// Sets the file extensions for which the CM_STORE compression method
+		/// is used if CM_AUTO is specified in addFile() or addRecursive().
+		/// For all other extensions, CM_DEFLATE is used. This is used to avoid
+		/// double compression of already compressed file formats, which usually
+		/// leads to worse results. Extensions will be converted to lower case.
+		///
+		/// The default extensions are:
+		///   - gif
+		///   - jpg
+		///   - jpeg
+		///   - png
+		
+	const std::set<std::string>& getStoreExtensions() const;
+		/// Returns the file extensions for which the CM_STORE compression method
+		/// is used if CM_AUTO is specified in addFile() or addRecursive().
+		///
+		/// See setStoreExtensions() for more information.
+
 private:
 	enum
 	{
@@ -86,6 +106,7 @@ class Zip_API Compress
 		/// copys an already compressed ZipEntry from in
 
 private:
+	std::set<std::string>      _storeExtensions;
 	std::ostream&              _out;
 	bool                       _seekableOut;
 	ZipArchive::FileHeaders    _files;
@@ -114,6 +135,12 @@ inline const std::string& Compress::getZipComment() const
 }
 
 
+inline const std::set<std::string>& Compress::getStoreExtensions() const
+{
+	return _storeExtensions;
+}
+
+
 } } // namespace Poco::Zip
 
 
diff --git a/Zip/include/Poco/Zip/ZipCommon.h b/Zip/include/Poco/Zip/ZipCommon.h
@@ -49,7 +49,8 @@ class Zip_API ZipCommon
 		CM_DEFLATE = 8,
 		CM_ENHANCEDDEFLATE = 9,
 		CM_DATECOMPRIMPLODING = 10,
-		CM_UNUSED = 11
+		CM_UNUSED = 11,
+		CM_AUTO = 255 /// automatically select DM_DEFLATE or CM_STORE based on file type (extension)
 	};
 
 	enum CompressionLevel
diff --git a/Zip/src/Compress.cpp b/Zip/src/Compress.cpp
@@ -23,6 +23,7 @@
 #include "Poco/StreamCopier.h"
 #include "Poco/File.h"
 #include "Poco/FileStream.h"
+#include "Poco/String.h"
 
 
 namespace Poco {
@@ -37,6 +38,10 @@ Compress::Compress(std::ostream& out, bool seekableOut):
 	_dirs(),
 	_offset(0)
 {
+	_storeExtensions.insert("gif");
+	_storeExtensions.insert("png");
+	_storeExtensions.insert("jpg");
+	_storeExtensions.insert("jpeg");
 }
 
 
@@ -47,6 +52,15 @@ Compress::~Compress()
 
 void Compress::addEntry(std::istream& in, const Poco::DateTime& lastModifiedAt, const Poco::Path& fileName, ZipCommon::CompressionMethod cm, ZipCommon::CompressionLevel cl)
 {
+	if (cm == ZipCommon::CM_AUTO)
+	{
+		std::string ext = Poco::toLower(fileName.getExtension());
+		if (_storeExtensions.find(ext) != _storeExtensions.end())
+			cm = ZipCommon::CM_STORE;
+		else
+			cm = ZipCommon::CM_DEFLATE;
+	}
+
 	std::string fn = ZipUtil::validZipEntryFileName(fileName);
 
 	if (_files.size() >= 65535)
@@ -300,4 +314,14 @@ ZipArchive Compress::close()
 }
 
 
+void Compress::setStoreExtensions(const std::set<std::string>& extensions)
+{
+	_storeExtensions.clear();
+	for (std::set<std::string>::const_iterator it = extensions.begin(); it != extensions.end(); ++it)
+	{
+		_storeExtensions.insert(Poco::toLower(*it));
+	}
+}
+
+
 } } // namespace Poco::Zip
