Menu
▾
▴
[nekohtml-dev] SF.net SVN: nekohtml:[347] trunk
|
From: <mgu...@us...> - 2015-04-16 15:53:18
|
Revision: 347
http://sourceforge.net/p/nekohtml/code/347
Author: mguillem
Date: 2015-04-16 15:53:16 +0000 (Thu, 16 Apr 2015)
Log Message:
-----------
improved detection of compatible encodings from meta charset when only decode is supported (patch from Steve McKay)
Issue #20
Modified Paths:
--------------
trunk/doc/changes.html
trunk/src/org/cyberneko/html/HTMLScanner.java
Added Paths:
-----------
trunk/data/meta/test-meta-encoding3.html
trunk/data/meta/test-meta-encoding3.html.canonical
trunk/data/meta/test-meta-encoding3.html.settings
Added: trunk/data/meta/test-meta-encoding3.html
===================================================================
--- trunk/data/meta/test-meta-encoding3.html (rev 0)
+++ trunk/data/meta/test-meta-encoding3.html 2015-04-16 15:53:16 UTC (rev 347)
@@ -0,0 +1,14 @@
+<head>
+<meta http-equiv="Content-Type" content="text/html;charset=iso-2022-cn">
+</head>
+$)AKNLe
+
+PB
+
+$)G\XM||U
+
+IzN"~
+
+d;\XM||U
+
+$)A#?
Added: trunk/data/meta/test-meta-encoding3.html.canonical
===================================================================
--- trunk/data/meta/test-meta-encoding3.html.canonical (rev 0)
+++ trunk/data/meta/test-meta-encoding3.html.canonical 2015-04-16 15:53:16 UTC (rev 347)
@@ -0,0 +1,13 @@
+(HTML
+(HEAD
+"\n
+(META
+Acontent text/html;charset=iso-2022-cn
+Ahttp-equiv Content-Type
+)META
+"\n
+)HEAD
+(BODY
+"\n宋体\n\n新\n\n細明體\n\n宋体\n\n浠茇忘�\n\n？\n\n
+)BODY
+)HTML
Added: trunk/data/meta/test-meta-encoding3.html.settings
===================================================================
--- trunk/data/meta/test-meta-encoding3.html.settings (rev 0)
+++ trunk/data/meta/test-meta-encoding3.html.settings 2015-04-16 15:53:16 UTC (rev 347)
@@ -0,0 +1 @@
+property http://cyberneko.org/html/properties/default-encoding UTF-8
Modified: trunk/doc/changes.html
===================================================================
--- trunk/doc/changes.html 2015-04-16 15:42:57 UTC (rev 346)
+++ trunk/doc/changes.html 2015-04-16 15:53:16 UTC (rev 347)
@@ -31,7 +31,8 @@
<dt>Version 1.9.22 (to be released)</dt>
<dd>Element <code>NOBR</code> closes <code>NOBR</code>, <code>BUTTON</code> closes <code>BUTTON</code> (patch from Ronald Brill),
element <code>EMBED</code> has no body (patch from Ronald Brill),
- element <code>A</code> shouldn't be inline (patch from Ahmed Ashour).
+ element <code>A</code> shouldn't be inline (patch from Ahmed Ashour),
+ improved detection of compatible encodings from meta charset when only decode is supported (patch from Steve McKay).
</dd>
<dt>Version 1.9.21 (2 Jun 2014)</dt>
Modified: trunk/src/org/cyberneko/html/HTMLScanner.java
===================================================================
--- trunk/src/org/cyberneko/html/HTMLScanner.java 2015-04-16 15:42:57 UTC (rev 346)
+++ trunk/src/org/cyberneko/html/HTMLScanner.java 2015-04-16 15:53:16 UTC (rev 347)
@@ -3709,17 +3709,33 @@
* be the same in both encodings
*/
boolean isEncodingCompatible(final String encoding1, final String encoding2) {
- final String reference = "<html><head><meta http-equiv=\"Content-Type\" content=\"text/html;charset=";
try {
- final byte[] bytesEncoding1 = reference.getBytes(encoding1);
- final String referenceWithEncoding2 = new String(bytesEncoding1, encoding2);
- return reference.equals(referenceWithEncoding2);
+ try {
+ return canRoundtrip(encoding1, encoding2);
+ }
+ catch (final UnsupportedOperationException e) {
+ // if encoding1 only supports decode, we can test it the other way to only decode with it
+ try {
+ return canRoundtrip(encoding2, encoding1);
+ }
+ catch (final UnsupportedOperationException e1) {
+ // encoding2 only supports decode too. Time to give up.
+ return false;
+ }
+ }
}
catch (final UnsupportedEncodingException e) {
return false;
}
}
+ private boolean canRoundtrip(final String encodeCharset, final String decodeCharset) throws UnsupportedEncodingException {
+ final String reference = "<html><head><meta http-equiv=\"Content-Type\" content=\"text/html;charset=";
+ final byte[] bytesEncoding1 = reference.getBytes(encodeCharset);
+ final String referenceWithEncoding2 = new String(bytesEncoding1, decodeCharset);
+ return reference.equals(referenceWithEncoding2);
+ }
+
private boolean endsWith(final XMLStringBuffer buffer, final String string) {
final int l = string.length();
if (buffer.length < l) {
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
