Merge pull request #479 from gainan/issue_451

Extract charset encoding from HTTP headers or HTML headers.

Author: gustavo-iniguez-goya

cSploit/src/main/java/org/csploit/android/net/http/RequestParser.java

@@ -27,6 +27,8 @@
 import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.Iterator;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 public class RequestParser
 {
@@ -478,4 +480,48 @@ public static ArrayList<HttpCookie> getCookiesFromHeaders(ArrayList<String> head
 
     return null;
   }
+
+  /**
+   * extract the charset encoding from the HTTP response headers.
+   *
+   * @param contentType content-type header to be parsed
+   * @return returns the charset encoding if we've found it, or null.
+   */
+  public static String getCharsetFromHeaders(String contentType){
+    if (contentType != null && contentType.toLowerCase().trim().contains("charset=")){
+      String[] parts = contentType.toLowerCase().trim().split("=");
+      if (parts.length > 0)
+        return parts[1];
+    }
+
+    return null;
+  }
+
+  /**
+   * extract the charset encoding of a web site from the <meta> headers.
+   *
+   * @param body html body of the site to be parsed
+   * @return returns the charset encoding if we've found it, or null.
+   */
+  public static String getCharsetFromBody(String body) {
+    if (body != null) {
+      int headEnd = body.toLowerCase().trim().indexOf("</head>");
+
+      // return null if there's no head tags
+      if (headEnd == -1)
+        return null;
+
+      String body_head = body.toLowerCase().substring(0, headEnd);
+
+      Pattern p = Pattern.compile("charset=([\"\'a-z0-9A-Z-]+)");
+      Matcher m = p.matcher(body_head);
+      String str_match = "";
+      if (m.find()) {
+        str_match = m.toMatchResult().group(1);
+        return str_match.replaceAll("[\"']", "");
+      }
+    }
+
+    return null;
+  }
 }

cSploit/src/main/java/org/csploit/android/net/http/proxy/StreamThread.java

@@ -18,16 +18,17 @@
  */
 package org.csploit.android.net.http.proxy;
 
-import java.io.IOException;
-import java.io.InputStream;
-import java.io.OutputStream;
-
+import org.csploit.android.core.Logger;
 import org.csploit.android.core.Profiler;
 import org.csploit.android.core.System;
-import org.csploit.android.core.Logger;
 import org.csploit.android.net.ByteBuffer;
 import org.csploit.android.net.http.RequestParser;
 
+import java.io.IOException;
+import java.io.InputStream;
+import java.io.OutputStream;
+import java.io.UnsupportedEncodingException;
+
 public class StreamThread implements Runnable
 {
   private final static String[] FILTERED_CONTENT_TYPES = new String[]
@@ -159,7 +160,27 @@ public void run(){
 
         headers = patched;
 
-        mBuffer.setData((headers + HEAD_SEPARATOR + body).getBytes());
+        // try to get the charset encoding from the HTTP headers.
+        String charset = RequestParser.getCharsetFromHeaders(contentType);
+
+        // if we haven't found the charset encoding on the HTTP headers, try it out on the body.
+        if (charset == null) {
+          charset = RequestParser.getCharsetFromBody(body);
+        }
+
+        if (charset != null) {
+          try {
+            mBuffer.setData((headers + HEAD_SEPARATOR + body).getBytes(charset));
+          }
+          catch (UnsupportedEncodingException e){
+            Logger.error("UnsupportedEncoding: " + e.getLocalizedMessage());
+            mBuffer.setData((headers + HEAD_SEPARATOR + body).getBytes());
+          }
+        }
+        else {
+          // if we haven't found the charset encoding, just handle it on ByteBuffer()
+          mBuffer.setData((headers + HEAD_SEPARATOR + body).getBytes());
+        }
 
         mWriter.write(mBuffer.getData());
         mWriter.flush();