[XML] Optional charset encoding

   - Via System property com.reandroid.xml.KeepCharsetEncoding, turns on by any non-null value.
   - If this property is ON (true), the decoded XML string string will have same encoding as string pool.
   - By default only AndroidManifest.xml will be encoded with utf-16
   - If utf-16 activated, "encoding" attribute under xml declaration will be removed, to let pull parser detect charset by itself.

Author: REAndroid

src/main/java/com/reandroid/apk/xmlencoder/XMLEncodeSource.java

@@ -1,8 +1,10 @@
 package com.reandroid.apk.xmlencoder;
 
 import com.reandroid.apk.APKLogger;
+import com.reandroid.app.AndroidManifest;
 import com.reandroid.archive.ByteInputSource;
 import com.reandroid.arsc.chunk.PackageBlock;
+import com.reandroid.arsc.chunk.xml.AndroidManifestBlock;
 import com.reandroid.arsc.chunk.xml.ResXmlDocument;
 import com.reandroid.utils.Crc32;
 import com.reandroid.utils.io.IOUtil;
@@ -68,9 +70,15 @@ private byte[] getArray() throws IOException{
     }
     private ResXmlDocument encode() throws XmlPullParserException, IOException {
         XMLParserSource parserSource = this.parserSource;
-        logVerbose("Encoding: " + parserSource.getPath());
+        String path = parserSource.getPath();
+        logVerbose("Encoding: " + path);
         XmlPullParser parser = parserSource.getParser();
-        ResXmlDocument resXmlDocument = new ResXmlDocument();
+        ResXmlDocument resXmlDocument;
+        if (AndroidManifest.FILE_NAME.equals(path)) {
+            resXmlDocument = new AndroidManifestBlock();
+        } else {
+            resXmlDocument = new ResXmlDocument();
+        }
         resXmlDocument.setPackageBlock(this.packageBlock);
         resXmlDocument.parse(parser);
         IOUtil.close(parser);

src/main/java/com/reandroid/arsc/chunk/xml/ResXmlDocument.java

@@ -15,6 +15,7 @@
 import com.reandroid.utils.io.FileUtil;
 import com.reandroid.xml.XMLDocument;
 import com.reandroid.xml.XMLFactory;
+import com.reandroid.xml.XMLUtil;
 import com.reandroid.xml.base.Document;
 import org.xmlpull.v1.XmlPullParser;
 import org.xmlpull.v1.XmlPullParserException;
@@ -104,7 +105,7 @@ ResXmlDocumentChunk getChunk() {
     @Override
     public XMLDocument toXml(boolean decode) {
         XMLDocument xmlDocument = new XMLDocument();
-        xmlDocument.setEncoding("utf-8");
+        xmlDocument.setEncoding(getEncoding());
         Iterator<ResXmlNode> iterator = iterator();
         while (iterator.hasNext()) {
             ResXmlNode node = iterator.next();
@@ -152,23 +153,29 @@ String nodeTypeName() {
     @Override
     public JSONObject toJson() {
         JSONObject jsonObject = new JSONObject();
-        jsonObject.put(JSON_encoding, getEncoding());
+        jsonObject.put(JSON_encoding, getStringPool().getEncoding());
         jsonObject.put(JSON_node_type, nodeTypeName());
         jsonObject.put(JSON_nodes, nodesToJson());
         return jsonObject;
     }
     @Override
     public void fromJson(JSONObject json) {
-        setEncoding(json.optString(JSON_encoding));
+        getStringPool().setEncoding(json.optString(JSON_encoding));
         nodesFromJson(json);
         refresh();
     }
 
     public String getEncoding() {
-        return getStringPool().getEncoding();
+        if (XMLUtil.KEEP_CHARSET_ENCODING) {
+            return getStringPool().getEncoding();
+        } else {
+            return "utf-8";
+        }
     }
     public void setEncoding(String encoding) {
-        getStringPool().setEncoding(encoding);
+        if (XMLUtil.KEEP_CHARSET_ENCODING) {
+            getStringPool().setEncoding(encoding);
+        }
     }
     @Override
     public int getLineNumber() {
@@ -237,13 +244,15 @@ public ResXmlIDMap getResXmlIDMap() {
     public void serialize(XmlSerializer serializer) throws IOException {
         serialize(serializer, true);
     }
+    @Override
     public void serialize(XmlSerializer serializer, boolean decode) throws IOException {
         PackageBlock packageBlock = getPackageBlock();
-        if(decode && packageBlock == null) {
+        if (decode && packageBlock == null) {
             throw new IOException("Can not decode without package");
         }
         setIndent(serializer, true);
-        serializer.startDocument(getEncoding(), null);
+        String encoding = getEncoding();
+        serializer.startDocument(encoding, encoding == null ? Boolean.FALSE : null);
         fixNamespaces();
         removeUnusedNamespaces();
 

src/main/java/com/reandroid/arsc/pool/StringPool.java

@@ -344,16 +344,10 @@ public String getEncoding() {
         if (isUtf8()) {
             return "utf-8";
         }
-        return "utf-16";
+        return null;
     }
     public void setEncoding(String encoding) {
-        boolean utf8;
-        if (encoding != null) {
-            utf8 = !StringsUtil.toLowercase(encoding).startsWith("utf-16");
-        } else {
-            utf8 = true;
-        }
-        setUtf8(utf8);
+        setUtf8(encoding != null && !StringsUtil.toLowercase(encoding).startsWith("utf-16"));
     }
 
     abstract StringArray<T> newInstance(OffsetArray offsets, IntegerItem itemCount, IntegerItem itemStart, boolean is_utf8);

src/main/java/com/reandroid/xml/XMLFactory.java

@@ -23,7 +23,6 @@
 import org.xmlpull.v1.XmlSerializer;
 
 import java.io.*;
-import java.nio.charset.StandardCharsets;
 
 public class XMLFactory {
 
@@ -37,6 +36,7 @@ public static XmlPullParser newPullParser(File file) throws XmlPullParserExcepti
         XmlPullParser parser = newPullParser();
         try {
             parser.setInput(FileUtil.inputStream(file), null);
+            XMLUtil.setLocation(parser, file);
         } catch (IOException ex) {
             throw new XmlPullParserException(ex.getMessage());
         }
@@ -74,7 +74,7 @@ public static XmlSerializer newSerializer(File file, String encoding) throws IOE
     }
     public static XmlSerializer newSerializer(OutputStream outputStream) throws IOException{
         XmlSerializer serializer = newSerializer();
-        serializer.setOutput(outputStream, StandardCharsets.UTF_8.name());
+        serializer.setOutput(outputStream, "utf-8");
         return serializer;
     }
     public static XmlSerializer newSerializer(OutputStream outputStream, String encoding) throws IOException{

src/main/java/com/reandroid/xml/XMLUtil.java

@@ -26,6 +26,9 @@
 
 public class XMLUtil {
 
+    public static final boolean KEEP_CHARSET_ENCODING = System.getProperty("com.reandroid.xml.KeepCharsetEncoding") != null;
+
+
     public static void expectEvent(XmlPullParser parser, int expect) throws XmlPullParserException {
         int event = parser.getEventType();
         if (event != expect) {