[Android] Add API for specifying a charset for data Urls.

content/public/android/java/src/org/chromium/content/browser/LoadUrlParams.java

 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 package org.chromium.content.browser;
 
 import org.chromium.base.CalledByNative;
 import org.chromium.base.JNINamespace;
 
 import java.util.Map;
@@ skipping 11 matching lines @@
     public static int LOAD_TYPE_BROWSER_INITIATED_HTTP_POST;
     public static int LOAD_TYPE_DATA;
 
     // Should match NavigationController::UserAgentOverrideOption exactly.
     // See comments there for proper usage. Values are initialized in
     // initializeConstants.
     public static int UA_OVERRIDE_INHERIT;
     public static int UA_OVERRIDE_FALSE;
     public static int UA_OVERRIDE_TRUE;
 
+    // The default character encoding for data URLs as defined in
+    // RFC 2397.
+    private static final String DATA_URL_DEFAULT_CHARSET = "US-ASCII";
+
     // Fields with counterparts in NavigationController::LoadURLParams.
     // Package private so that ContentViewCore.loadUrl can pass them down to
     // native code. Should not be accessed directly anywhere else outside of
     // this class.
     final String mUrl;
     int mLoadUrlType;
     int mTransitionType;
     int mUaOverrideOption;
     private Map<String, String> mExtraHeaders;
     byte[] mPostData;
     String mBaseUrlForDataUrl;
     String mVirtualUrlForDataUrl;
     boolean mCanLoadLocalResources;
 
     public LoadUrlParams(String url) {
         // Check initializeConstants was called.
         assert LOAD_TYPE_DEFAULT != LOAD_TYPE_BROWSER_INITIATED_HTTP_POST;
 
         mUrl = url;
         mLoadUrlType = LOAD_TYPE_DEFAULT;
         mTransitionType = ContentViewCore.PAGE_TRANSITION_LINK;
         mUaOverrideOption = UA_OVERRIDE_INHERIT;
         mPostData = null;
         mBaseUrlForDataUrl = null;
         mVirtualUrlForDataUrl = null;
     }
 
     /**
+     * Helper method to create a LoadUrlParams object for data url. Defaults
+     * to US-ASCII charset, as per RFC2397.
+     * @param data Data to be loaded.
+     * @param mimeType Mime type of the data.
+     * @param isBase64Encoded True if the data is encoded in Base 64 format.
+     */
+    public static LoadUrlParams createLoadDataParams(
+        String data, String mimeType, boolean isBase64Encoded) {
+        return createLoadDataParams(data, mimeType, isBase64Encoded, DATA_URL_DEFAULT_CHARSET);
+    }
+
+    /**
      * Helper method to create a LoadUrlParams object for data url.
      * @param data Data to be loaded.
      * @param mimeType Mime type of the data.
      * @param isBase64Encoded True if the data is encoded in Base 64 format.
+     * @param charset The character set for the data.
      */
     public static LoadUrlParams createLoadDataParams(
-            String data, String mimeType, boolean isBase64Encoded) {
+            String data, String mimeType, boolean isBase64Encoded, String charset) {
         StringBuilder dataUrl = new StringBuilder("data:");
         dataUrl.append(mimeType);
+        dataUrl.append(";charset=" + charset);

[joth, 2012/12/14 18:02:51]

-AIUI the charset is spurious if the mimetype is not text/* -- what would it
mean to have a shift-jis charset JPEG or UCS-32 .exe ? (if the payload is base64
encoded, the charset applies to the decoded form, I believe. the base64  encoded
form is unavoidably us-ascii.)
-the receiver defaults to US-ASCII so we can always omit it in that case too,
but it doesn't make much odds if we include it anyway there.

[benm (inactive), 2012/12/14 18:13:58]

Good point. I think the charset is considered a parameter to the MIME type, so
in all likelihood the underlying implementation will ignore it if it doesn't
make sense given the MIME type (seems chromium does this).

But I also agree it does seem a bit odd to append it when the embedder didn't
specify it and to do so given the MIME type wouldn't make much sense. I suppose
text/* is the only case that a charset does make sense so probably best to only
add it in that case.

         if (isBase64Encoded) {
             dataUrl.append(";base64");
         }
         dataUrl.append(",");
         dataUrl.append(data);
 
         LoadUrlParams params = new LoadUrlParams(dataUrl.toString());
         params.setLoadType(LoadUrlParams.LOAD_TYPE_DATA);
         params.setTransitionType(ContentViewCore.PAGE_TRANSITION_TYPED);
         return params;
     }
 
     /**
      * Helper method to create a LoadUrlParams object for data url with base
-     * and virtual url.
+     * and virtual url. Defaults to US-ASCII charset, as per RFC2397.
      * @param data Data to be loaded.
      * @param mimeType Mime type of the data.
      * @param isBase64Encoded True if the data is encoded in Base 64 format.
      * @param baseUrl Base url of this data load. Note that for WebView compatibility,
      *                baseUrl and historyUrl are ignored if this is a data: url.
      *                Defaults to about:blank if null.
      * @param historyUrl History url for this data load. Note that for WebView compatibility,
      *                   this is ignored if baseUrl is a data: url. Defaults to about:blank
      *                   if null.
      */
     public static LoadUrlParams createLoadDataParamsWithBaseUrl(
             String data, String mimeType, boolean isBase64Encoded,
             String baseUrl, String historyUrl) {
-        LoadUrlParams params = createLoadDataParams(data, mimeType, isBase64Encoded);
+        return createLoadDataParamsWithBaseUrl(data, mimeType, isBase64Encoded,
+                baseUrl, historyUrl, DATA_URL_DEFAULT_CHARSET);
+    }
+
+    /**
+     * Helper method to create a LoadUrlParams object for data url with base
+     * and virtual url.
+     * @param data Data to be loaded.
+     * @param mimeType Mime type of the data.
+     * @param isBase64Encoded True if the data is encoded in Base 64 format.
+     * @param baseUrl Base url of this data load. Note that for WebView compatibility,
+     *                baseUrl and historyUrl are ignored if this is a data: url.
+     *                Defaults to about:blank if null.
+     * @param historyUrl History url for this data load. Note that for WebView compatibility,
+     *                   this is ignored if baseUrl is a data: url. Defaults to about:blank
+     *                   if null.
+     * @param charset The character set for the data.
+     */
+    public static LoadUrlParams createLoadDataParamsWithBaseUrl(
+            String data, String mimeType, boolean isBase64Encoded,
+            String baseUrl, String historyUrl, String charset) {
+        LoadUrlParams params = createLoadDataParams(data, mimeType, isBase64Encoded, charset);
         // For WebView compatibility, when the base URL has the 'data:'
         // scheme, we treat it as a regular data URL load and skip setting
         // baseUrl and historyUrl.
         if (baseUrl == null || !baseUrl.toLowerCase().startsWith("data:")) {
             params.setBaseUrlForDataUrl(baseUrl != null ? baseUrl : "about:blank");
             params.setVirtualUrlForDataUrl(historyUrl != null ? historyUrl : "about:blank");
         }
         return params;
     }
 
@@ skipping 144 matching lines @@
         UA_OVERRIDE_FALSE = ua_override_false;
         UA_OVERRIDE_TRUE = ua_override_true;
     }
 
     /**
      * Parses |url| as a GURL on the native side, and
      * returns true if it's scheme is data:.
      */
     private static native boolean nativeIsDataScheme(String url);
 }