+ support extracting archives (zip, rar, 7z, ...)

* added 7-Zip-JBinding libs and native dependencies
* added CLI option -extract and make it available in scripting environment
* allow --output to be used as output folder in -rename CLI call

build.xml

@@ -6,7 +6,9 @@
 	<property name="title" value="${application.name}" />
 	<property name="version" value="${application.version}" />
 
-	<tstamp><format property="today" pattern="yyyy-MM-dd" /></tstamp>
+	<tstamp>
+		<format property="today" pattern="yyyy-MM-dd" />
+	</tstamp>
 
 	<!-- define source dirs -->
 	<property name="dir.source" location="${basedir}/source" />
@@ -66,64 +68,64 @@
 				<include name="org/apache/**" />
 				<include name="org/w3c/dom/html/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/nekohtml.jar">
 				<include name="org/cyberneko/html/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/json-simple.jar">
 				<include name="org/json/simple/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/simmetrics.jar">
 				<include name="uk/ac/shef/wit/simmetrics/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/glazedlists.jar">
 				<include name="ca/odell/glazedlists/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/miglayout.jar">
 				<include name="net/miginfocom/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/xmlrpc.jar">
 				<include name="redstone/xmlrpc/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/args4j.jar">
 				<include name="org/kohsuke/args4j/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/ehcache.jar">
 				<include name="net/sf/ehcache/**" />
 				<include name="ehcache-failsafe.xml" />
 				<include name="ehcache-version.properties" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/slf4j.jar">
 				<include name="org/slf4j/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/slf4j-jdk.jar">
 				<include name="org/slf4j/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/commons-io.jar">
 				<include name="org/apache/commons/io/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/jna.jar">
 				<!-- include classes and native libraries -->
 				<include name="com/sun/jna/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/groovy.jar">
 				<include name="groovy*/**" />
 				<include name="org/codehaus/groovy/**" />
 				<include name="META-INF/dgminfo" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/icu4j.jar">
 				<include name="com/ibm/icu/**" />
 			</zipfileset>
@@ -131,11 +133,11 @@
 			<zipfileset src="${dir.lib}/sublight-ws.jar">
 				<include name="net/sublight/webservice/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/junrar-custom.jar">
 				<include name="de/innosystec/unrar/**" />
 			</zipfileset>
-			
+
 			<zipfileset src="${dir.lib}/jgat-custom.jar">
 				<include name="com/dmurph/tracking/**" />
 			</zipfileset>
@@ -146,16 +148,16 @@
 	<target name="appbundle" depends="fatjar" description="Build an OSX application bundle">
 		<taskdef name="jarbundler" classname="net.sourceforge.jarbundler.JarBundler" classpath="${dir.lib}/build/jarbundler.jar" />
 		<copy tofile="${dir.dist}/appbundle/FileBot.jar" file="${path.fatjar}" />
-		
+
 		<!-- build app bundle folder and add native libs -->
 		<jarbundler dir="${dir.dist}" name="${title}" version="${version}" build="${svn.revision}" icon="${dir.installer}/appbundle/icon.icns" bundleid="net.sourceforge.filebot" jar="${dir.dist}/appbundle/FileBot.jar" stubfile="${dir.installer}/appbundle/JavaApplicationStub" workingdirectory="$APP_PACKAGE/Contents/Resources/Java" mainclass="net.sourceforge.filebot.Main" jvmversion="1.6+" vmoptions="-Xmx256m">
-			<javaproperty name="application.deployment" value="app"/>
+			<javaproperty name="application.deployment" value="app" />
 		</jarbundler>
-		
+
 		<!-- shell scripts -->
 		<copy todir="${dir.dist}/${title}.app/Contents/MacOS" file="${dir.installer}/appbundle/filebot" />
 		<copy todir="${dir.dist}/${title}.app/Contents/MacOS" file="${dir.installer}/appbundle/install.sh" />
-		
+
 		<copy todir="${dir.dist}/${title}.app/Contents/Resources/Java">
 			<fileset dir="${dir.lib}/native/mac-x86_64" includes="*.dylib" />
 		</copy>
@@ -163,14 +165,15 @@
 		<!-- application bundle folder as .tar.gz -->
 		<tar destfile="${path.appbundle.tar.gz}" compression="gzip">
 			<tarfileset dir="${dir.dist}" includes="${title}.app/**" excludes="**/MacOS/**" />
-			<tarfileset dir="${dir.dist}" includes="${title}.app/**/MacOS/**" filemode="755" /> <!-- IMPORTANT application stub must be executable!! -->
+			<tarfileset dir="${dir.dist}" includes="${title}.app/**/MacOS/**" filemode="755" />
+			<!-- IMPORTANT application stub must be executable!! -->
 		</tar>
 	</target>
 
 
 	<target name="deb" depends="fatjar" description="Build debian package for i386 and amd64">
 		<taskdef resource="ant_deb_task.properties" classpath="${dir.lib}/build/ant-deb.jar" />
-		
+
 		<antcall target="deb-arch">
 			<param name="arch" value="i386" />
 		</antcall>
@@ -204,31 +207,33 @@
 
 	<target name="msi-arch">
 		<property name="mediainfo" location="${dir.lib}/native/win32-${arch}/MediaInfo.dll" />
+		<property name="lib7z_binding" location="${dir.lib}/native/win32-${arch}/lib7z-JBinding.dll" />
+		<property name="lib7z_gcc" location="${dir.lib}/native/win32-${arch}/lib7z-gcc.dll" />
 		<property name="installer" location="${dir.dist}/FileBot_${version}_${arch}.msi" />
 
 		<exec executable="candle.exe" dir="${dir.installer}/msi" failonerror="true">
-			<arg line="filebot-wix.xml -out ${dir.dist}/msi.wixobj -arch ${arch} -dreleaseversion=${version} -dfatjar=${path.fatjar} -dmediainfo=${mediainfo}" />
+			<arg line="filebot-wix.xml -out ${dir.dist}/msi.wixobj -arch ${arch} -dreleaseversion=${version} -dfatjar=${path.fatjar} -dmediainfo=${mediainfo} -dlib7z_binding=${lib7z_binding} -dlib7z_gcc=${lib7z_gcc}" />
 		</exec>
 		<exec executable="light.exe" dir="${dir.installer}/msi" failonerror="true">
 			<arg line="${dir.dist}/msi.wixobj -sval -ext WixUIExtension -out ${installer}" />
 		</exec>
 	</target>
-	
-	
+
+
 	<target name="portable" depends="fatjar" description="Portable application package">
-		<mkdir dir="${dir.dist}/portable"/>
+		<mkdir dir="${dir.dist}/portable" />
 		<copy file="${path.fatjar}" tofile="${dir.dist}/portable/FileBot.jar" />
 		<copy todir="${dir.dist}/portable">
 			<fileset dir="${dir.installer}/portable" includes="*.exe, *.ini, *.sh" />
 		</copy>
-		
+
 		<zip destfile="${dir.dist}/FileBot_${version}-portable.zip">
 			<zipfileset dir="${dir.dist}/portable" includes="*.jar, *.exe, *.ini" />
 			<zipfileset dir="${dir.dist}/portable" includes="*.sh" filemode="755" />
 		</zip>
 	</target>
-	
-	
+
+
 	<target name="webstart" depends="jar" description="Build and compress jars used for webstart deployment">
 		<!-- create dirs -->
 		<mkdir dir="${dir.dist}/webstart" />
@@ -248,7 +253,7 @@
 		<jar destfile="${dir.dist}/webstart/jna.jar">
 			<zipfileset src="${dir.lib}/jna.jar" includes="**/*.class" />
 		</jar>
-		
+
 		<!-- create mediainfo jar as seperate jar and use as trigger for lazy loading the native libs -->
 		<jar destfile="${dir.dist}/webstart/mediainfo.jar">
 			<fileset dir="${dir.build}" includes="net/sourceforge/filebot/mediainfo/**" />
@@ -292,7 +297,7 @@
 		<signjar alias="filebot" keystore="filebot.keystore" storepass="secret">
 			<fileset id="signjar" dir="${dir.dist}/webstart" includes="**/*.jar" />
 		</signjar>
-		
+
 		<!-- pack200 all jars -->
 		<apply executable="pack200" dest="${dir.dist}/webstart">
 			<!-- workaround for bug 6575373, see http://bugs.sun.com/view_bug.do?bug_id=6575373 -->
@@ -383,11 +388,17 @@
 
 	<target name="svn-update">
 		<typedef resource="org/tigris/subversion/svnant/svnantlib.xml">
-			<classpath><fileset dir="${dir.lib}/build" includes="*.jar" /></classpath>
+			<classpath>
+				<fileset dir="${dir.lib}/build" includes="*.jar" />
+			</classpath>
 		</typedef>
 		<svnSetting id="svn.settings" svnkit="true" javahl="false" />
-		<svn refid="svn.settings"><update dir="${dir.source}" revision="HEAD" recurse="false" /></svn>
-		<svn refid="svn.settings"><status path="${dir.source}" revisionProperty="svn.revision" /></svn>
+		<svn refid="svn.settings">
+			<update dir="${dir.source}" revision="HEAD" recurse="false" />
+		</svn>
+		<svn refid="svn.settings">
+			<status path="${dir.source}" revisionProperty="svn.revision" />
+		</svn>
 		<echo>Revision: ${svn.revision}</echo>
 	</target>
 
@@ -419,7 +430,7 @@
 		<scp todir="${sf.user}:${sf.password}@${deploy.release}" trust="yes" verbose="true">
 			<fileset dir="${dir.dist}/release" includes="**/*.jar" />
 		</scp>
-		
+
 		<!-- deploy portable application package -->
 		<sleep seconds="5" />
 		<scp todir="${sf.user}:${sf.password}@${deploy.release}" trust="yes" verbose="true">

installer/deb/filebot.sh

@@ -1,2 +1,2 @@
 #!/bin/bash
-java -Dapplication.deployment=deb -Djna.library.path=/usr/share/filebot -Xmx256m -jar /usr/share/filebot/FileBot.jar "$@"
+java -Dapplication.deployment=deb -Djna.library.path=/usr/share/filebot -Djava.library.path=/usr/share/filebot -Xmx256m -jar /usr/share/filebot/FileBot.jar "$@"

installer/msi/filebot-wix.xml

@@ -32,6 +32,8 @@
 					<Component Id='ApplicationBase' Guid='9E365344-A00C-45DE-A2A4-266412C3D06E'>
 						<File Id='FileBot.jar' Name='FileBot.jar' Source='$(var.fatjar)' KeyPath='yes' />
 						<File Id='MediaInfo.dll' Name='MediaInfo.dll' Source='$(var.mediainfo)' />
+						<File Id='lib7z_JBinding.dll' Name='7z-JBinding.dll' Source='$(var.lib7z_binding)' />
+						<File Id='lib7z_gcc.dll' Name='7z-gcc.dll' Source='$(var.lib7z_gcc)' />
 					</Component>
 				</Directory>
 			</Directory>

installer/portable/filebot.sh

@@ -4,4 +4,4 @@ while [ -h "$SOURCE" ] ; do SOURCE="$(readlink "$SOURCE")"; done
 dir_bin="$( cd -P "$( dirname "$SOURCE" )" && pwd )"
 
 # WARNING: NOT TESTED / HERE THERE BE DRAGONS
-java -Dapplication.deployment=portable "-Dapplication.dir=$dir_bin" "-Duser.home=$dir_bin" "-Djna.library.path=$dir_bin" -Djava.util.prefs.PreferencesFactory=net.sourceforge.tuned.prefs.FilePreferencesFactory -Dnet.sourceforge.tuned.prefs.file=prefs.properties -Xmx256m -jar "$dir_app/FileBot.jar" "$@"
+java -Dapplication.deployment=portable "-Dapplication.dir=$dir_bin" "-Duser.home=$dir_bin" "-Djna.library.path=$dir_bin" "-Djava.library.path=$dir_bin" -Djava.util.prefs.PreferencesFactory=net.sourceforge.tuned.prefs.FilePreferencesFactory -Dnet.sourceforge.tuned.prefs.file=prefs.properties -Xmx256m -jar "$dir_app/FileBot.jar" "$@"

source/net/sourceforge/filebot/archive/Archive.java

@@ -0,0 +1,135 @@
+
+package net.sourceforge.filebot.archive;
+
+
+import static org.apache.commons.io.FilenameUtils.*;
+
+import java.io.Closeable;
+import java.io.File;
+import java.io.FileFilter;
+import java.io.FileNotFoundException;
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.EnumMap;
+import java.util.List;
+import java.util.Map;
+import java.util.logging.Logger;
+import java.util.regex.Pattern;
+
+import com.sun.jna.Platform;
+
+import net.sf.sevenzipjbinding.ArchiveFormat;
+import net.sf.sevenzipjbinding.IInStream;
+import net.sf.sevenzipjbinding.ISevenZipInArchive;
+import net.sf.sevenzipjbinding.PropID;
+import net.sf.sevenzipjbinding.SevenZip;
+import net.sf.sevenzipjbinding.SevenZipException;
+
+
+public class Archive implements Closeable {
+	
+	static {
+		// initialize 7z-JBinding native libs
+		try {
+			if (Platform.isWindows()) {
+				System.loadLibrary("lib7z-gcc");
+			}
+			
+			System.loadLibrary("lib7z-JBinding");
+			SevenZip.initLoadedLibraries();
+		} catch (Throwable e) {
+			Logger.getLogger(Archive.class.getName()).warning("Failed to load 7z-JBinding");
+		}
+	}
+	
+	private ISevenZipInArchive inArchive;
+	private Closeable openVolume;
+	
+	
+	public Archive(File file) throws SevenZipException, IOException {
+		if (!file.exists()) {
+			throw new FileNotFoundException(file.getAbsolutePath());
+		}
+		
+		ArchiveOpenVolumeCallback openVolumeCallback = new ArchiveOpenVolumeCallback();
+		IInStream inStream = openVolumeCallback.getStream(file.getAbsolutePath());
+		
+		inArchive = SevenZip.openInArchive(null, inStream, openVolumeCallback);
+		openVolume = openVolumeCallback;
+	}
+	
+	
+	public int itemCount() throws SevenZipException {
+		return inArchive.getNumberOfItems();
+	}
+	
+	
+	public Map<PropID, Object> getItem(int index) throws SevenZipException {
+		Map<PropID, Object> item = new EnumMap<PropID, Object>(PropID.class);
+		
+		for (PropID prop : PropID.values()) {
+			Object value = inArchive.getProperty(index, prop);
+			if (value != null) {
+				item.put(prop, value);
+			}
+		}
+		
+		return item;
+	}
+	
+	
+	public List<File> listFiles() throws SevenZipException {
+		List<File> paths = new ArrayList<File>();
+		
+		for (int i = 0; i < inArchive.getNumberOfItems(); i++) {
+			boolean isFolder = (Boolean) inArchive.getProperty(i, PropID.IS_FOLDER);
+			if (!isFolder) {
+				String path = (String) inArchive.getProperty(i, PropID.PATH);
+				paths.add(new File(path));
+			}
+		}
+		
+		return paths;
+	}
+	
+	
+	public void extract(ExtractOutProvider outputMapper) throws SevenZipException {
+		inArchive.extract(null, false, new ExtractCallback(inArchive, outputMapper));
+	}
+	
+	
+	@Override
+	public void close() throws IOException {
+		try {
+			inArchive.close();
+		} catch (SevenZipException e) {
+			throw new IOException(e);
+		} finally {
+			openVolume.close();
+		}
+	}
+	
+	
+	public static final FileFilter VOLUME_ONE_FILTER = new FileFilter() {
+		
+		private Pattern exclude = Pattern.compile("[.]r[0-9]+$|[.]part[0-9]*[2-9][.]rar$|[.][0-9]*[2-9]$", Pattern.CASE_INSENSITIVE);
+		
+		
+		@Override
+		public boolean accept(File path) {
+			if (exclude.matcher(path.getName()).find()) {
+				return false;
+			}
+			
+			String ext = getExtension(path.getName());
+			for (ArchiveFormat it : ArchiveFormat.values()) {
+				if (it.getMethodName().equalsIgnoreCase(ext)) {
+					return true;
+				}
+			}
+			
+			return false;
+		}
+	};
+	
+}

source/net/sourceforge/filebot/archive/ArchiveOpenVolumeCallback.java

@@ -0,0 +1,109 @@
+
+package net.sourceforge.filebot.archive;
+
+
+import java.io.*;
+import java.util.*;
+
+import net.sf.sevenzipjbinding.*;
+import net.sf.sevenzipjbinding.impl.*;
+
+
+class ArchiveOpenVolumeCallback implements IArchiveOpenVolumeCallback, IArchiveOpenCallback, Closeable {
+	
+	/**
+	 * Cache for opened file streams
+	 */
+	private Map<String, RandomAccessFile> openedRandomAccessFileList = new HashMap<String, RandomAccessFile>();
+	
+	/**
+	 * Name of the last volume returned by {@link #getStream(String)}
+	 */
+	private String name;
+	
+	
+	/**
+	 * This method should at least provide the name of the last
+	 * opened volume (propID=PropID.NAME).
+	 * 
+	 * @see IArchiveOpenVolumeCallback#getProperty(PropID)
+	 */
+	public Object getProperty(PropID propID) throws SevenZipException {
+		switch (propID) {
+			case NAME:
+				return name;
+		}
+		return null;
+	}
+	
+	
+	/**
+	 * The name of the required volume will be calculated out of the
+	 * name of the first volume and a volume index. In case of RAR file,
+	 * the substring ".partNN." in the name of the volume file will
+	 * indicate a volume with id NN. For example:
+	 * <ul>
+	 * <li>test.rar - single part archive or multi-part archive with a single volume</li>
+	 * <li>test.part23.rar - 23-th part of a multi-part archive</li>
+	 * <li>test.part001.rar - first part of a multi-part archive. "00" indicates, that at least 100 volumes must exist.</li>
+	 * </ul>
+	 */
+	public IInStream getStream(String filename) throws SevenZipException {
+		try {
+			// We use caching of opened streams, so check cache first
+			RandomAccessFile randomAccessFile = openedRandomAccessFileList.get(filename);
+			if (randomAccessFile != null) { // Cache hit.
+				// Move the file pointer back to the beginning
+				// in order to emulating new stream
+				randomAccessFile.seek(0);
+				
+				// Save current volume name in case getProperty() will be called
+				name = filename;
+				
+				return new RandomAccessFileInStream(randomAccessFile);
+			}
+			
+			// Nothing useful in cache. Open required volume.
+			randomAccessFile = new RandomAccessFile(filename, "r");
+			
+			// Put new stream in the cache
+			openedRandomAccessFileList.put(filename, randomAccessFile);
+			
+			// Save current volume name in case getProperty() will be called
+			name = filename;
+			return new RandomAccessFileInStream(randomAccessFile);
+		} catch (FileNotFoundException fileNotFoundException) {
+			// Required volume doesn't exist. This happens if the volume:
+			// 1. never exists. 7-Zip doesn't know how many volumes should
+			//    exist, so it have to try each volume.
+			// 2. should be there, but doesn't. This is an error case.
+			
+			// Since normal and error cases are possible,
+			// we can't throw an error message
+			return null; // We return always null in this case
+		} catch (Exception e) {
+			throw new RuntimeException(e);
+		}
+	}
+	
+	
+	/**
+	 * Close all opened streams
+	 */
+	public void close() throws IOException {
+		for (RandomAccessFile file : openedRandomAccessFileList.values()) {
+			file.close();
+		}
+	}
+	
+	
+	@Override
+	public void setCompleted(Long files, Long bytes) throws SevenZipException {
+	}
+	
+	
+	@Override
+	public void setTotal(Long files, Long bytes) throws SevenZipException {
+	}
+	
+}

source/net/sourceforge/filebot/archive/ExtractCallback.java

@@ -0,0 +1,85 @@
+
+package net.sourceforge.filebot.archive;
+
+
+import java.io.File;
+import java.io.IOException;
+import java.io.OutputStream;
+
+import net.sf.sevenzipjbinding.ExtractAskMode;
+import net.sf.sevenzipjbinding.ExtractOperationResult;
+import net.sf.sevenzipjbinding.IArchiveExtractCallback;
+import net.sf.sevenzipjbinding.ISequentialOutStream;
+import net.sf.sevenzipjbinding.ISevenZipInArchive;
+import net.sf.sevenzipjbinding.PropID;
+import net.sf.sevenzipjbinding.SevenZipException;
+
+
+class ExtractCallback implements IArchiveExtractCallback {
+	
+	private ISevenZipInArchive inArchive;
+	private ExtractOutProvider extractOut;
+	
+	private ExtractOutStream output = null;
+	
+	
+	public ExtractCallback(ISevenZipInArchive inArchive, ExtractOutProvider extractOut) {
+		this.inArchive = inArchive;
+		this.extractOut = extractOut;
+	}
+	
+	
+	public ISequentialOutStream getStream(int index, ExtractAskMode extractAskMode) throws SevenZipException {
+		if (extractAskMode != ExtractAskMode.EXTRACT) {
+			return null;
+		}
+		
+		boolean isFolder = (Boolean) inArchive.getProperty(index, PropID.IS_FOLDER);
+		if (isFolder) {
+			return null;
+		}
+		
+		String path = (String) inArchive.getProperty(index, PropID.PATH);
+		try {
+			OutputStream target = extractOut.getStream(new File(path));
+			if (target == null) {
+				return null;
+			}
+			
+			output = new ExtractOutStream(target);
+			return output;
+		} catch (IOException e) {
+			throw new SevenZipException(e);
+		}
+	}
+	
+	
+	public void prepareOperation(ExtractAskMode extractAskMode) throws SevenZipException {
+	}
+	
+	
+	public void setOperationResult(ExtractOperationResult extractOperationResult) throws SevenZipException {
+		if (output != null) {
+			try {
+				output.close();
+			} catch (IOException e) {
+				throw new SevenZipException(e);
+			} finally {
+				output = null;
+			}
+		}
+		
+		if (extractOperationResult != ExtractOperationResult.OK) {
+			throw new SevenZipException("Extraction Error: " + extractOperationResult);
+		}
+	}
+	
+	
+	public void setCompleted(long completeValue) throws SevenZipException {
+	}
+	
+	
+	public void setTotal(long total) throws SevenZipException {
+	}
+	
+}

source/net/sourceforge/filebot/archive/ExtractOutProvider.java

@@ -0,0 +1,12 @@
+
+package net.sourceforge.filebot.archive;
+
+
+import java.io.*;
+
+
+public interface ExtractOutProvider {
+	
+	OutputStream getStream(File archivePath) throws IOException;
+	
+}

source/net/sourceforge/filebot/archive/ExtractOutStream.java

@@ -0,0 +1,36 @@
+
+package net.sourceforge.filebot.archive;
+
+
+import java.io.*;
+
+import net.sf.sevenzipjbinding.*;
+
+
+class ExtractOutStream implements ISequentialOutStream, Closeable {
+	
+	private OutputStream out;
+	
+	
+	public ExtractOutStream(OutputStream out) {
+		this.out = out;
+	}
+	
+	
+	@Override
+	public int write(byte[] data) throws SevenZipException {
+		try {
+			out.write(data);
+		} catch (IOException e) {
+			throw new SevenZipException(e);
+		}
+		return data.length; // return amount of proceed data
+	}
+	
+	
+	@Override
+	public void close() throws IOException {
+		out.close();
+	}
+	
+}

source/net/sourceforge/filebot/archive/FileMapper.java

@@ -0,0 +1,40 @@
+
+package net.sourceforge.filebot.archive;
+
+
+import java.io.File;
+import java.io.FileOutputStream;
+import java.io.IOException;
+import java.io.OutputStream;
+
+
+public class FileMapper implements ExtractOutProvider {
+	
+	private File outputDir;
+	private boolean flatten;
+	
+	
+	public FileMapper(File outputDir, boolean flatten) {
+		this.outputDir = outputDir;
+		this.flatten = flatten;
+	};
+	
+	
+	public File getOutputFile(File entry) {
+		return new File(outputDir, flatten ? entry.getName() : entry.getPath());
+	}
+	
+	
+	@Override
+	public OutputStream getStream(File entry) throws IOException {
+		File outputFile = getOutputFile(entry);
+		File outputFolder = outputFile.getParentFile();
+		
+		// create parent folder if necessary
+		if (!outputFolder.isDirectory() && !outputFolder.mkdirs()) {
+			throw new IOException("Failed to create folder: " + outputFolder);
+		}
+		
+		return new FileOutputStream(outputFile);
+	}
+}

source/net/sourceforge/filebot/cli/ArgumentBean.java

@@ -52,10 +52,10 @@ public class ArgumentBean {
 	@Option(name = "-check", usage = "Create/Check verification file", metaVar = "fileset")
 	public boolean check;
 	
-	@Option(name = "--output", usage = "Output options", metaVar = "[sfv, md5, sha1] or [srt]")
+	@Option(name = "--output", usage = "Output path / format", metaVar = "Output options")
 	public String output;
 	
-	@Option(name = "--encoding", usage = "Character encoding", metaVar = "[UTF-8, windows-1252, GB18030, etc]")
+	@Option(name = "--encoding", usage = "Output character encoding", metaVar = "[UTF-8, windows-1252, GB18030, etc]")
 	public String encoding;
 	
 	@Option(name = "-list", usage = "Fetch episode list")
@@ -64,6 +64,9 @@ public class ArgumentBean {
 	@Option(name = "-mediainfo", usage = "Get media info")
 	public boolean mediaInfo = false;
 	
+	@Option(name = "-extract", usage = "Extract archives")
+	public boolean extract = false;
+	
 	@Option(name = "-script", usage = "Run Groovy script", metaVar = "robot.groovy")
 	public String script = null;
 	
@@ -93,7 +96,7 @@ public class ArgumentBean {
 	
 	
 	public boolean runCLI() {
-		return rename || getSubtitles || getMissingSubtitles || check || list || mediaInfo || script != null;
+		return rename || getSubtitles || getMissingSubtitles || check || list || mediaInfo || extract || script != null;
 	}
 	
 	

source/net/sourceforge/filebot/cli/ArgumentProcessor.java

@@ -8,8 +8,8 @@ import static net.sourceforge.tuned.FileUtilities.*;
 
 import java.io.File;
 import java.security.AccessController;
+import java.util.Collection;
 import java.util.LinkedHashSet;
-import java.util.Set;
 import java.util.logging.Level;
 
 import javax.script.Bindings;
@@ -60,7 +60,11 @@ public class ArgumentProcessor {
 			// execute CLI operations
 			if (args.script == null) {
 				// file operations
-				Set<File> files = new LinkedHashSet<File>(args.getFiles(true));
+				Collection<File> files = new LinkedHashSet<File>(args.getFiles(true));
+				
+				if (args.extract) {
+					files.addAll(cli.extract(files, args.output));
+				}
 				
 				if (args.getSubtitles) {
 					files.addAll(cli.getSubtitles(files, args.query, args.lang, args.output, args.encoding, !args.nonStrict));
@@ -69,7 +73,7 @@ public class ArgumentProcessor {
 				}
 				
 				if (args.rename) {
-					cli.rename(files, args.query, args.format, args.db, args.order, args.lang, !args.nonStrict);
+					cli.rename(files, args.query, args.output, args.format, args.db, args.order, args.lang, !args.nonStrict);
 				}
 				
 				if (args.check) {

source/net/sourceforge/filebot/cli/CmdlineInterface.java

@@ -9,7 +9,7 @@ import java.util.List;
 
 public interface CmdlineInterface {
 	
-	List<File> rename(Collection<File> files, String query, String format, String db, String sortOrder, String lang, boolean strict) throws Exception;
+	List<File> rename(Collection<File> files, String query, String output, String format, String db, String sortOrder, String lang, boolean strict) throws Exception;
 	
 	
 	List<File> getSubtitles(Collection<File> files, String query, String lang, String output, String encoding, boolean strict) throws Exception;
@@ -29,4 +29,7 @@ public interface CmdlineInterface {
 	
 	String getMediaInfo(File file, String format) throws Exception;
 	
+	
+	List<File> extract(Collection<File> files, String output) throws Exception;
+	
 }

source/net/sourceforge/filebot/cli/CmdlineOperations.java

@@ -42,6 +42,8 @@ import java.util.concurrent.Future;
 import net.sourceforge.filebot.Analytics;
 import net.sourceforge.filebot.MediaTypes;
 import net.sourceforge.filebot.WebServices;
+import net.sourceforge.filebot.archive.Archive;
+import net.sourceforge.filebot.archive.FileMapper;
 import net.sourceforge.filebot.format.ExpressionFormat;
 import net.sourceforge.filebot.format.MediaBindingBean;
 import net.sourceforge.filebot.hash.HashType;
@@ -78,9 +80,10 @@ import net.sourceforge.tuned.FileUtilities.FolderFilter;
 public class CmdlineOperations implements CmdlineInterface {
 	
 	@Override
-	public List<File> rename(Collection<File> files, String query, String expression, String db, String sortOrder, String languageName, boolean strict) throws Exception {
+	public List<File> rename(Collection<File> files, String query, String output, String expression, String db, String sortOrder, String languageName, boolean strict) throws Exception {
 		ExpressionFormat format = (expression != null) ? new ExpressionFormat(expression) : null;
 		Locale locale = getLanguage(languageName).toLocale();
+		File outputDir = (output != null && output.length() > 0) ? new File(output) : null;
 		
 		List<File> mediaFiles = filter(files, VIDEO_FILES, SUBTITLE_FILES);
 		if (mediaFiles.isEmpty()) {
@@ -89,12 +92,12 @@ public class CmdlineOperations implements CmdlineInterface {
 		
 		if (getEpisodeListProvider(db) != null) {
 			// tv series mode
-			return renameSeries(files, query, format, getEpisodeListProvider(db), SortOrder.forName(sortOrder), locale, strict);
+			return renameSeries(files, query, outputDir, format, getEpisodeListProvider(db), SortOrder.forName(sortOrder), locale, strict);
 		}
 		
 		if (getMovieIdentificationService(db) != null) {
 			// movie mode
-			return renameMovie(files, query, format, getMovieIdentificationService(db), locale, strict);
+			return renameMovie(files, query, outputDir, format, getMovieIdentificationService(db), locale, strict);
 		}
 		
 		// auto-determine mode
@@ -125,14 +128,14 @@ public class CmdlineOperations implements CmdlineInterface {
 		
 		CLILogger.finest(format("Filename pattern: [%.02f] SxE, [%.02f] CWS", sxe / max, cws / max));
 		if (sxe >= (max * 0.65) || cws >= (max * 0.65)) {
-			return renameSeries(files, query, format, getEpisodeListProviders()[0], SortOrder.forName(sortOrder), locale, strict); // use default episode db
+			return renameSeries(files, query, outputDir, format, getEpisodeListProviders()[0], SortOrder.forName(sortOrder), locale, strict); // use default episode db
 		} else {
-			return renameMovie(files, query, format, getMovieIdentificationServices()[0], locale, strict); // use default movie db
+			return renameMovie(files, query, outputDir, format, getMovieIdentificationServices()[0], locale, strict); // use default movie db
 		}
 	}
 	
 	
-	public List<File> renameSeries(Collection<File> files, String query, ExpressionFormat format, EpisodeListProvider db, SortOrder sortOrder, Locale locale, boolean strict) throws Exception {
+	public List<File> renameSeries(Collection<File> files, String query, File outputDir, ExpressionFormat format, EpisodeListProvider db, SortOrder sortOrder, Locale locale, boolean strict) throws Exception {
 		CLILogger.config(format("Rename episodes using [%s]", db.getName()));
 		List<File> mediaFiles = filter(files, VIDEO_FILES, SUBTITLE_FILES);
 		
@@ -179,7 +182,7 @@ public class CmdlineOperations implements CmdlineInterface {
 			File file = match.getValue();
 			Episode episode = match.getCandidate();
 			String newName = (format != null) ? format.format(new MediaBindingBean(episode, file)) : validateFileName(EpisodeFormat.SeasonEpisode.format(episode));
-			File newFile = new File(newName + "." + getExtension(file));
+			File newFile = new File(outputDir, newName + "." + getExtension(file));
 			
 			if (isInvalidFilePath(newFile)) {
 				CLILogger.config("Stripping invalid characters from new name: " + newName);
@@ -264,7 +267,7 @@ public class CmdlineOperations implements CmdlineInterface {
 	}
 	
 	
-	public List<File> renameMovie(Collection<File> files, String query, ExpressionFormat format, MovieIdentificationService service, Locale locale, boolean strict) throws Exception {
+	public List<File> renameMovie(Collection<File> files, String query, File outputDir, ExpressionFormat format, MovieIdentificationService service, Locale locale, boolean strict) throws Exception {
 		CLILogger.config(format("Rename movies using [%s]", service.getName()));
 		
 		// handle movie files
@@ -910,4 +913,30 @@ public class CmdlineOperations implements CmdlineInterface {
 		return format.format(new MediaBindingBean(file, file));
 	}
 	
+	
+	@Override
+	public List<File> extract(Collection<File> files, String output) throws Exception {
+		// only keep single-volume archives or first part of multi-volume archives
+		List<File> archiveFiles = filter(files, Archive.VOLUME_ONE_FILTER);
+		List<File> extractedFiles = new ArrayList<File>();
+		
+		for (File file : archiveFiles) {
+			Archive archive = new Archive(file);
+			File outputFolder = (output != null) ? new File(output).getAbsoluteFile() : new File(file.getParentFile(), getNameWithoutExtension(file.getName()));
+			
+			CLILogger.info(String.format("Extract archive [%s] to [%s]", file.getName(), outputFolder));
+			FileMapper outputMapper = new FileMapper(outputFolder, false);
+			
+			List<File> entries = archive.listFiles();
+			for (File entry : entries) {
+				extractedFiles.add(outputMapper.getOutputFile(entry));
+			}
+			
+			CLILogger.finest("Extract files " + entries);
+			archive.extract(outputMapper);
+		}
+		
+		return extractedFiles;
+	}
+	
 }

source/net/sourceforge/filebot/cli/ScriptShell.lib.groovy

@@ -61,11 +61,13 @@ import java.nio.ByteBuffer
 import java.nio.charset.Charset
 import static net.sourceforge.filebot.web.WebRequest.*
 
-URL.metaClass.get = { readAll(getReader(delegate.openConnection())) }
+URL.metaClass.getText = { readAll(getReader(delegate.openConnection())) }
+URL.metaClass.getHtml = { new XmlParser(new org.cyberneko.html.parsers.SAXParser()).parseText(delegate.getText()) }
 URL.metaClass.fetch = { fetch(delegate) }
-URL.metaClass.getHtml = { new XmlParser(new org.cyberneko.html.parsers.SAXParser()).parseText(readAll(getReader(delegate.openConnection()))) }
-ByteBuffer.metaClass.getHtml = { csn = "utf-8" -> new XmlParser(new org.cyberneko.html.parsers.SAXParser()).parseText(Charset.forName(csn).decode(delegate.duplicate()).toString()) }
+ByteBuffer.metaClass.getText = { csn = "utf-8" -> Charset.forName(csn).decode(delegate.duplicate()).toString() }
+ByteBuffer.metaClass.getHtml = { csn = "utf-8" -> new XmlParser(new org.cyberneko.html.parsers.SAXParser()).parseText(delegate.getText(csn)) }
 
+URL.metaClass.get = { delegate.getText() }
 URL.metaClass.post = { Map parameters -> post(delegate.openConnection(), parameters) }
 URL.metaClass.post = { byte[] data, contentType = 'application/octet-stream' -> post(delegate.openConnection(), data, contentType) }
 URL.metaClass.post = { String text, csn = 'utf-8' -> delegate.post(text.getBytes(csn), 'text/plain') }
@@ -212,6 +214,12 @@ def compute(args) { args = _defaults(args)
 	}
 }
 
+def extract(args) { args = _defaults(args)
+	synchronized (_cli) {
+		_guarded { _cli.extract(_files(args), args.output) }
+	}
+}
+
 def fetchEpisodeList(args) { args = _defaults(args)
 	synchronized (_cli) {
 		_guarded { _cli.fetchEpisodeList(args.query, args.format, args.db, args.order, args.lang) }

source/net/sourceforge/filebot/media.types

@@ -30,7 +30,10 @@
 		<extension>zip</extension>
 	</type>
 	<type name="archive/rar">
-		<extension>rar</extension>	
+		<extension>rar</extension>
+	</type>
+	<type name="archive/7z">
+		<extension>7z</extension>
 	</type>
 	<type name="archive/gzip">
 		<extension>gzip</extension>
@@ -39,6 +42,9 @@
 	<type name="archive/tar">
 		<extension>tar</extension>
 	</type>
+	<type name="archive/bzip2">
+		<extension>bzip2</extension>
+	</type>
 
 	<!-- Audio -->
 	<type name="audio/mp3">
@@ -119,7 +125,7 @@
 		<extension>sami</extension>
 	</type>
 	<type name="subtitle/VobSub">
-		<extension>sub</extension>	
+		<extension>sub</extension>
 		<extension>idx</extension>
 	</type>
 </media-types>

website/cli.html

@@ -133,6 +133,10 @@
 					</p>
 					<h3 id="scripting">Scripting</h3>
 					<p>
+					<code><span class="cmd">filebot</span> <span class="option">-extract</span> <span class="argument">"300.part01.rar"</span> <span class="option">--output</span> <span class="argument">path/to/folder</span></code></code>
+					Extract files from single-volume or multi-volume archives (e.g. rar).
+					</p>
+					<p>
 					<code><span class="cmd">filebot</span> <span class="option">-list</span> <span class="option">--db</span> <span class="argument">thetvdb</span> <span class="option">--q</span> <span class="argument">Dexter</span> <span class="option">--format</span> <span class="argument">"{s}x{e.pad(2)} {t}"</span></code>
 					Fetch episode list and print to console.
 					</p>
@@ -223,6 +227,11 @@
 							<td>create/check verification file</td>
 							<td>folder or sfv file</td>
 						</tr>
+						<tr>
+							<td>-extract</td>
+							<td>extract archives</td>
+							<td>folder or archive file</td>
+						</tr>
 						<tr>
 							<td>--output</td>
 							<td>output format and/or path</td>

website/script.html

@@ -129,13 +129,14 @@
 					
 					<div class="documentation">
 					<h4>Rename media files</h4>
-					<pre><span class="return">File[]</span> <span class="method">rename</span>(<span class="property">folder</span>|<span class="property">file</span>, <span class="property">query</span>, <span class="property">format</span>, <span class="property">db</span>, <span class="property">lang</span>, <span class="property">strict</span>)</pre>
+					<pre><span class="return">File[]</span> <span class="method">rename</span>(<span class="property">folder</span>|<span class="property">file</span>, <span class="property">query</span>, <span class="property">output</span>, <span class="property">format</span>, <span class="property">db</span>, <span class="property">lang</span>, <span class="property">strict</span>)</pre>
 					<div class="text">Match files with episode/movie data and rename according to given naming scheme.</div>
 					<dl>
 						<dt>Parameters:</dt>
 						<dd><span class="property">folder</span> - process media files in this folder</dd>
 						<dd><span class="property">file</span> - process these media files</dd>
 						<dd><span class="property">query</span> - force series/movie name, auto-detect if not set</dd>
+						<dd><span class="property">output</span> - output folder, if format expression is not absolute</dd>
 						<dd><span class="property">format</span> - episode/movie naming scheme</dd>
 						<dd><span class="property">db</span> - episode/movie database</dd>
 						<dd><span class="property">lang</span> - preferred language for episode/movie titles</dd>
@@ -186,6 +187,15 @@
 					
 					<div class="documentation">
 					<h4>Other</h4>
+					<pre><span class="return">File[]</span> <span class="method">extract</span>(<span class="property">folder</span>|<span class="property">file</span>, <span class="property">output</span>)</pre>
+					<div class="text">Extract files from single-volume or multi-volume archives (.zip, .rar, .7z, etc).</div>
+					<dl>
+						<dt>Parameters:</dt>
+						<dd><span class="property">folder</span> - extract all archives that are in this folder</dd>
+						<dd><span class="property">file</span> - extract this archive</dd>
+						<dd><span class="property">output</span> - output folder, defaults to archive path</dd>
+					</dl>
+					<hr/>
 					<pre><span class="return">String[]</span> <span class="method">fetchEpisodeList</span>(<span class="property">query</span>, <span class="property">format</span>, <span class="property">db</span>, <span class="property">lang</span>)</pre>
 					<div class="text">Fetch episode data for the given tv show and format episode names.</div>
 					<dl>