Menu
▾
▴
[HtmlUnit] SVN: [7899] trunk/htmlunit/src
|
From: <asa...@us...> - 2012-12-22 11:32:03
|
Revision: 7899
http://sourceforge.net/p/htmlunit/code/7899
Author: asashour
Date: 2012-12-22 11:31:58 +0000 (Sat, 22 Dec 2012)
Log Message:
-----------
- Fix the encoding of clicked links to match the enclosing page (FF).
- UrlUtils: deprecate encodeUrl(URL, boolean).
Issue 1402
Modified Paths:
--------------
trunk/htmlunit/src/changes/changes.xml
trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/HttpWebConnection.java
trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/WebClient.java
trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/html/HtmlAnchor.java
trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/util/UrlUtils.java
trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/javascript/host/html/HTMLDocumentTest.java
trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/util/UrlUtilsTest.java
Modified: trunk/htmlunit/src/changes/changes.xml
===================================================================
--- trunk/htmlunit/src/changes/changes.xml 2012-12-22 06:13:04 UTC (rev 7898)
+++ trunk/htmlunit/src/changes/changes.xml 2012-12-22 11:31:58 UTC (rev 7899)
@@ -9,6 +9,12 @@
<body>
<release version="2.12" date="???" description="Bugfixes">
<action type="update" dev="asashour">
+ UrlUtils: deprecate encodeUrl(URL, boolean).
+ </action>
+ <action type="fix" dev="asashour" issue="1402">
+ Fix the encoding of clicked links to match the enclosing page (FF).
+ </action>
+ <action type="update" dev="asashour">
Upper-case the charset detected from "meta" tag.
</action>
<action type="update" dev="mguillem">
Modified: trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/HttpWebConnection.java
===================================================================
--- trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/HttpWebConnection.java 2012-12-22 06:13:04 UTC (rev 7898)
+++ trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/HttpWebConnection.java 2012-12-22 11:31:58 UTC (rev 7899)
@@ -241,7 +241,7 @@
// URLs; because of this we allow some Unicode chars in URLs. However, at this point we're
// handing things over the HttpClient, and HttpClient will blow up if we leave these Unicode
// chars in the URL.
- final URL url = UrlUtils.encodeUrl(webRequest.getUrl(), false);
+ final URL url = UrlUtils.encodeUrl(webRequest.getUrl(), false, webRequest.getCharset());
final String charset = webRequest.getCharset();
// URIUtils.createURI is deprecated but as of httpclient-4.2.1, URIBuilder doesn't work here as it encodes path
Modified: trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/WebClient.java
===================================================================
--- trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/WebClient.java 2012-12-22 06:13:04 UTC (rev 7898)
+++ trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/WebClient.java 2012-12-22 11:31:58 UTC (rev 7899)
@@ -1422,7 +1422,8 @@
WebAssert.notNull("method", method);
WebAssert.notNull("parameters", parameters);
- url = UrlUtils.encodeUrl(url, getBrowserVersion().hasFeature(URL_MINIMAL_QUERY_ENCODING));
+ url = UrlUtils.encodeUrl(url, getBrowserVersion().hasFeature(URL_MINIMAL_QUERY_ENCODING),
+ webRequest.getCharset());
webRequest.setUrl(url);
if (LOG.isDebugEnabled()) {
Modified: trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/html/HtmlAnchor.java
===================================================================
--- trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/html/HtmlAnchor.java 2012-12-22 06:13:04 UTC (rev 7898)
+++ trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/html/HtmlAnchor.java 2012-12-22 11:31:58 UTC (rev 7899)
@@ -116,6 +116,7 @@
}
final WebRequest webRequest = new WebRequest(url);
+ webRequest.setCharset(page.getPageEncoding());
webRequest.setAdditionalHeader("Referer", page.getUrl().toExternalForm());
if (LOG.isDebugEnabled()) {
LOG.debug(
Modified: trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/util/UrlUtils.java
===================================================================
--- trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/util/UrlUtils.java 2012-12-22 06:13:04 UTC (rev 7898)
+++ trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/util/UrlUtils.java 2012-12-22 11:31:58 UTC (rev 7899)
@@ -22,6 +22,7 @@
import org.apache.commons.codec.DecoderException;
import org.apache.commons.codec.net.URLCodec;
+import com.gargoylesoftware.htmlunit.TextUtil;
import com.gargoylesoftware.htmlunit.WebAssert;
/**
@@ -203,8 +204,25 @@
* @param url the URL to encode
* @param minimalQueryEncoding whether or not to perform minimal query encoding, like IE does
* @return the encoded URL
+ * @deprecated as of 2.12, please use {@link #encodeUrl(URL, boolean, String)} instead
*/
+ @Deprecated
public static URL encodeUrl(final URL url, final boolean minimalQueryEncoding) {
+ return encodeUrl(url, minimalQueryEncoding, TextUtil.DEFAULT_CHARSET);
+ }
+
+ /**
+ * <p>Encodes illegal characters in the specified URL's path, query string and anchor according to the URL
+ * encoding rules observed in real browsers.</p>
+ *
+ * <p>For example, this method changes <tt>"http://first/?a=b c"</tt> to <tt>"http://first/?a=b%20c"</tt>.</p>
+ *
+ * @param url the URL to encode
+ * @param minimalQueryEncoding whether or not to perform minimal query encoding, like IE does
+ * @param charset the charset
+ * @return the encoded URL
+ */
+ public static URL encodeUrl(final URL url, final boolean minimalQueryEncoding, final String charset) {
final String p = url.getProtocol();
if ("javascript".equalsIgnoreCase(p) || "about".equalsIgnoreCase(p) || "data".equalsIgnoreCase(p)) {
// Special exception.
@@ -213,7 +231,7 @@
try {
String path = url.getPath();
if (path != null) {
- path = encode(path, PATH_ALLOWED_CHARS, "utf-8");
+ path = encode(path, PATH_ALLOWED_CHARS, "UTF-8");
}
String query = url.getQuery();
if (query != null) {
@@ -221,12 +239,12 @@
query = org.apache.commons.lang3.StringUtils.replace(query, " ", "%20");
}
else {
- query = encode(query, QUERY_ALLOWED_CHARS, "windows-1252");
+ query = encode(query, QUERY_ALLOWED_CHARS, charset);
}
}
String anchor = url.getRef();
if (anchor != null) {
- anchor = encode(anchor, ANCHOR_ALLOWED_CHARS, "utf-8");
+ anchor = encode(anchor, ANCHOR_ALLOWED_CHARS, "UTF-8");
}
return createNewUrl(url.getProtocol(), url.getHost(), url.getPort(), path, anchor, query);
}
@@ -244,7 +262,7 @@
*/
public static String encodeAnchor(String anchor) {
if (anchor != null) {
- anchor = encode(anchor, ANCHOR_ALLOWED_CHARS, "utf-8");
+ anchor = encode(anchor, ANCHOR_ALLOWED_CHARS, "UTF-8");
}
return anchor;
}
Modified: trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/javascript/host/html/HTMLDocumentTest.java
===================================================================
--- trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/javascript/host/html/HTMLDocumentTest.java 2012-12-22 06:13:04 UTC (rev 7898)
+++ trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/javascript/host/html/HTMLDocumentTest.java 2012-12-22 11:31:58 UTC (rev 7899)
@@ -1606,4 +1606,30 @@
final WebDriver driver = loadPage2(html, URL_FIRST, "text/html", "UTF-8");
verifyAlerts(DEFAULT_WAIT_TIME, expectedAlerts, driver);
}
+
+ /**
+ * @throws Exception if the test fails
+ */
+ @Test
+ public void encoding7() throws Exception {
+ final String html = "<html>\n"
+ + "<head>\n"
+ + "<meta charset='UTF-8'>\n"
+ + "</head><body>\n"
+ + " <a id='myId' href='test?\u00E8=\u00E8'>test</a>"
+ + "</body></html>";
+
+ final WebDriver driver = loadPage2(html, URL_FIRST, "text/html", "UTF-8");
+ driver.findElement(By.id("myId")).click();
+ String actualQuery = driver.getCurrentUrl();
+ actualQuery = actualQuery.substring(actualQuery.indexOf('?'));
+ final String expectedQuery;
+ if (getBrowserVersion().isIE()) {
+ expectedQuery = "?\u00E8=\u00E8";
+ }
+ else {
+ expectedQuery = "?%C3%A8=%C3%A8";
+ }
+ assertTrue(actualQuery.endsWith(expectedQuery));
+ }
}
Modified: trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/util/UrlUtilsTest.java
===================================================================
--- trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/util/UrlUtilsTest.java 2012-12-22 06:13:04 UTC (rev 7898)
+++ trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/util/UrlUtilsTest.java 2012-12-22 11:31:58 UTC (rev 7899)
@@ -19,6 +19,7 @@
import org.junit.Test;
import com.gargoylesoftware.htmlunit.SimpleWebTestCase;
+import com.gargoylesoftware.htmlunit.TextUtil;
/**
* Tests for {@link UrlUtils}.
@@ -230,7 +231,8 @@
@Test
public void percent() throws Exception {
final URL url = new URL("http://localhost/bug%21.html");
- assertEquals("http://localhost/bug%21.html", UrlUtils.encodeUrl(url, false).toExternalForm());
+ assertEquals("http://localhost/bug%21.html",
+ UrlUtils.encodeUrl(url, false, TextUtil.DEFAULT_CHARSET).toExternalForm());
}
/**
|
