Menu
▾
▴
[HtmlUnit] SVN: [7892] trunk/htmlunit/src
|
From: <asa...@us...> - 2012-12-20 03:12:30
|
Revision: 7892
http://sourceforge.net/p/htmlunit/code/7892
Author: asashour
Date: 2012-12-20 03:12:25 +0000 (Thu, 20 Dec 2012)
Log Message:
-----------
Upper-case the charset detected from "meta" tag.
Modified Paths:
--------------
trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/html/HtmlForm.java
trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/util/EncodingSniffer.java
trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/WebResponse2Test.java
trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/html/HtmlFormTest.java
trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/html/HtmlPage3Test.java
trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/html/HtmlPageTest.java
trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/javascript/host/html/HTMLDocumentTest.java
trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/util/EncodingSnifferTest.java
Modified: trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/html/HtmlForm.java
===================================================================
--- trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/html/HtmlForm.java 2012-12-19 17:55:04 UTC (rev 7891)
+++ trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/html/HtmlForm.java 2012-12-20 03:12:25 UTC (rev 7892)
@@ -230,7 +230,7 @@
*/
private String getSubmitCharset() {
if (getAcceptCharsetAttribute().length() > 0) {
- return SUBMIT_CHARSET_PATTERN.matcher(getAcceptCharsetAttribute().trim()).replaceAll("");
+ return SUBMIT_CHARSET_PATTERN.matcher(getAcceptCharsetAttribute().trim()).replaceAll("").toUpperCase();
}
return getPage().getPageEncoding();
}
Modified: trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/util/EncodingSniffer.java
===================================================================
--- trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/util/EncodingSniffer.java 2012-12-19 17:55:04 UTC (rev 7891)
+++ trunk/htmlunit/src/main/java/com/gargoylesoftware/htmlunit/util/EncodingSniffer.java 2012-12-20 03:12:25 UTC (rev 7892)
@@ -295,6 +295,7 @@
final String value = pair.getValue();
encoding = extractEncodingFromContentType(value);
if (encoding != null) {
+ encoding = encoding.toUpperCase();
break;
}
}
Modified: trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/WebResponse2Test.java
===================================================================
--- trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/WebResponse2Test.java 2012-12-19 17:55:04 UTC (rev 7891)
+++ trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/WebResponse2Test.java 2012-12-20 03:12:25 UTC (rev 7892)
@@ -40,7 +40,7 @@
+ "<body>foo</body>\n"
+ "</html>";
final HtmlPage page = loadPage(html);
- assertEquals("utf-8", page.getWebResponse().getContentCharsetOrNull());
+ assertEquals("UTF-8", page.getWebResponse().getContentCharsetOrNull());
assertEquals(html, page.getWebResponse().getContentAsString());
}
Modified: trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/html/HtmlFormTest.java
===================================================================
--- trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/html/HtmlFormTest.java 2012-12-19 17:55:04 UTC (rev 7891)
+++ trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/html/HtmlFormTest.java 2012-12-20 03:12:25 UTC (rev 7892)
@@ -1149,12 +1149,12 @@
*/
@Test
public void submitRequestCharset() throws Exception {
- submitRequestCharset("utf-8", null, null, "utf-8");
- submitRequestCharset(null, "utf-8", null, "utf-8");
- submitRequestCharset("iso-8859-1", null, "utf-8", "utf-8");
- submitRequestCharset("iso-8859-1", null, "utf-8, iso-8859-1", "utf-8");
- submitRequestCharset("utf-8", null, "iso-8859-1 utf-8", "iso-8859-1");
- submitRequestCharset("iso-8859-1", null, "utf-8, iso-8859-1", "utf-8");
+ submitRequestCharset("utf-8", null, null, "UTF-8");
+ submitRequestCharset(null, "utf-8", null, "UTF-8");
+ submitRequestCharset("iso-8859-1", null, "utf-8", "UTF-8");
+ submitRequestCharset("iso-8859-1", null, "utf-8, iso-8859-1", "UTF-8");
+ submitRequestCharset("utf-8", null, "iso-8859-1 utf-8", "ISO-8859-1");
+ submitRequestCharset("iso-8859-1", null, "utf-8, iso-8859-1", "UTF-8");
}
/**
@@ -1207,7 +1207,7 @@
webConnection.setDefaultResponse(html, 200, "ok", contentType);
final HtmlPage page = client.getPage(getDefaultUrl());
- final String firstPageEncoding = StringUtils.defaultString(metaCharset, headerCharset);
+ final String firstPageEncoding = StringUtils.defaultString(metaCharset, headerCharset).toUpperCase();
assertEquals(firstPageEncoding, page.getPageEncoding());
final HtmlForm form = page.getFormByName("form1");
Modified: trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/html/HtmlPage3Test.java
===================================================================
--- trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/html/HtmlPage3Test.java 2012-12-19 17:55:04 UTC (rev 7891)
+++ trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/html/HtmlPage3Test.java 2012-12-20 03:12:25 UTC (rev 7892)
@@ -24,6 +24,7 @@
import org.openqa.selenium.WebElement;
import com.gargoylesoftware.htmlunit.BrowserRunner;
+import com.gargoylesoftware.htmlunit.BrowserRunner.Alerts;
import com.gargoylesoftware.htmlunit.WebDriverTestCase;
/**
@@ -93,4 +94,36 @@
Assert.assertEquals("js", input.getAttribute("addedBy"));
Assert.assertEquals("js", input.getAttribute("addedby"));
}
+
+ /**
+ * @throws Exception if the test fails
+ */
+ @Test
+ @Alerts(IE = { "undefined", "undefined", "iso-8859-1", "windows-1252" },
+ FF = { "ISO-8859-1", "ISO-8859-1", "undefined", "undefined" },
+ CHROME = { "ISO-8859-1", "ISO-8859-1", "ISO-8859-1", "ISO-8859-1" })
+ public void getPageEncoding() throws Exception {
+ final String htmlContent = "<html><head>\n"
+ + " <title>foo</title>\n"
+ + " <meta http-equiv='Content-Type' content='text/html; charset=Shift_JIS'>\n"
+ + " <script>\n"
+ + " function test() {\n"
+ + " alert(document.inputEncoding);\n"
+ + " alert(document.characterSet);\n"
+ + " alert(document.charset);\n"
+ + " alert(document.defaultCharset);\n"
+ + " }\n"
+ + " </script>\n"
+ + "</head><body onload='test()'>\n"
+ + "<table><tr><td>\n"
+ + "<meta name=vs_targetSchema content=\"http://schemas.microsoft.com/intellisense/ie5\">\n"
+ + "<form name='form1'>\n"
+ + " <input type='text' name='textfield1' id='textfield1' value='foo' />\n"
+ + " <input type='text' name='textfield2' id='textfield2'/>\n"
+ + "</form>\n"
+ + "</td></tr></table>\n"
+ + "</body></html>";
+ loadPageWithAlerts2(htmlContent);
+ }
+
}
Modified: trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/html/HtmlPageTest.java
===================================================================
--- trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/html/HtmlPageTest.java 2012-12-19 17:55:04 UTC (rev 7891)
+++ trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/html/HtmlPageTest.java 2012-12-20 03:12:25 UTC (rev 7892)
@@ -607,28 +607,6 @@
}
/**
- * @throws Exception if the test fails
- */
- @Test
- public void testGetPageEncoding() throws Exception {
- final String htmlContent = "<html><head>\n"
- + "<title>foo</title>\n"
- + "<meta http-equiv='Content-Type' content='text/html ;charset=Shift_JIS'>\n"
- + "</head><body>\n"
- + "<table><tr><td>\n"
- + "<meta name=vs_targetSchema content=\"http://schemas.microsoft.com/intellisense/ie5\">\n"
- + "<form name='form1'>\n"
- + " <input type='text' name='textfield1' id='textfield1' value='foo' />\n"
- + " <input type='text' name='textfield2' id='textfield2'/>\n"
- + "</form>\n"
- + "</td></tr></table>\n"
- + "</body></html>";
- final HtmlPage page = loadPage(htmlContent);
-
- assertEquals("shift_jis", page.getPageEncoding());
- }
-
- /**
* Verifies that an empty charset in a content-type meta tag is ignored. See bug 2484753.
* @throws Exception if an error occurs
*/
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 2012-12-19 17:55:04 UTC (rev 7891)
+++ trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/javascript/host/html/HTMLDocumentTest.java 2012-12-20 03:12:25 UTC (rev 7892)
@@ -1562,7 +1562,7 @@
public void encoding5() throws Exception {
final String html = "<html>\n"
+ "<head>\n"
- + "<meta charset='UTF-8'>\n"
+ + " <meta http-equiv='Content-Type' content='text/html; charset=iso-8859-1'>\n"
+ " <script>\n"
+ " function test() {\n"
+ " alert(document.inputEncoding);\n"
@@ -1572,6 +1572,33 @@
+ " }\n"
+ " </script>\n"
+ "</head><body onload='test()'>\n"
+ + "</body></html>";
+
+ final String[] expectedAlerts = getExpectedAlerts();
+ final WebDriver driver = loadPage2(html, URL_FIRST, "text/html;charset=utf-8", "ISO-8859-1");
+ verifyAlerts(DEFAULT_WAIT_TIME, expectedAlerts, driver);
+ }
+
+ /**
+ * @throws Exception if the test fails
+ */
+ @Test
+ @Alerts(IE = { "undefined", "undefined", "utf-8", "windows-1252" },
+ FF = { "UTF-8", "UTF-8", "undefined", "undefined" },
+ CHROME = { "UTF-8", "UTF-8", "UTF-8", "ISO-8859-1" })
+ public void encoding6() throws Exception {
+ final String html = "<html>\n"
+ + "<head>\n"
+ + " <meta charset='UTF-8'>\n"
+ + " <script>\n"
+ + " function test() {\n"
+ + " alert(document.inputEncoding);\n"
+ + " alert(document.characterSet);\n"
+ + " alert(document.charset);\n"
+ + " alert(document.defaultCharset);\n"
+ + " }\n"
+ + " </script>\n"
+ + "</head><body onload='test()'>\n"
+ " <a id='myId' href='test?è=è'>test</a>"
+ "</body></html>";
Modified: trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/util/EncodingSnifferTest.java
===================================================================
--- trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/util/EncodingSnifferTest.java 2012-12-19 17:55:04 UTC (rev 7891)
+++ trunk/htmlunit/src/test/java/com/gargoylesoftware/htmlunit/util/EncodingSnifferTest.java 2012-12-20 03:12:25 UTC (rev 7892)
@@ -41,8 +41,8 @@
testHeader(null, "foo", "bar");
testHeader(null, "Content-Type", "blah");
testHeader(null, "Content-Type", "text/html;charset=blah");
- testHeader("utf-8", "Content-Type", "text/html;charset=utf-8");
- testHeader("utf-8", "Content-Type", "text/html;charset=utf-8;");
+ testHeader("UTF-8", "Content-Type", "text/html;charset=utf-8");
+ testHeader("UTF-8", "Content-Type", "text/html;charset=utf-8;");
}
private void testHeader(final String expectedEncoding, final String headerName, final String headerValue) {
@@ -72,10 +72,10 @@
testMeta(null, "<meta a='b'");
testMeta(null, "<meta a='b' c=d e=\"f\"/>");
testMeta(null, "<meta a='b' c=d e=\"f\" content='text/html; charset=blah' />");
- testMeta("utf-8", "<meta a='b' c=d e=\"f\" content='text/html; charset=utf-8' />");
- testMeta("utf-8", "abc <meta http-equiv='Content-Type' content='text/html; charset=utf-8'/>");
- testMeta("utf-8", "abc <meta http-equiv='Content-Type' content='text/html; CHARSET=UTF-8'/>");
- testMeta("utf-8", "abc <meta http-equiv='Content-Type' content='text/html; chArsEt=UtF-8'/>");
+ testMeta("UTF-8", "<meta a='b' c=d e=\"f\" content='text/html; charset=utf-8' />");
+ testMeta("UTF-8", "abc <meta http-equiv='Content-Type' content='text/html; charset=utf-8'/>");
+ testMeta("UTF-8", "abc <meta http-equiv='Content-Type' content='text/html; CHARSET=UTF-8'/>");
+ testMeta("UTF-8", "abc <meta http-equiv='Content-Type' content='text/html; chArsEt=UtF-8'/>");
}
private void testMeta(final String expectedEncoding, final String content) {
|
