Bug 490916 - Trim space characters from encoding names in the HTML5 parser. r=jonas, a=jst.

2011-01-12 10:05:09 +02:00 · 2011-01-12 10:05:09 +02:00 · ae4be142ed
--- a/parser/html/nsHtml5MetaScannerCppSupplement.h
+++ b/parser/html/nsHtml5MetaScannerCppSupplement.h
@ -69,6 +69,7 @@ nsHtml5MetaScanner::tryCharset(nsString* charset)
  }
  nsCAutoString encoding;
  CopyUTF16toUTF8(*charset, encoding);
+  encoding.Trim(" \t\r\n\f");
  if (encoding.LowerCaseEqualsLiteral("utf-16") ||
      encoding.LowerCaseEqualsLiteral("utf-16be") ||
      encoding.LowerCaseEqualsLiteral("utf-16le")) {
--- a/parser/html/nsHtml5Parser.cpp
+++ b/parser/html/nsHtml5Parser.cpp
@ -144,8 +144,11 @@ nsHtml5Parser::SetDocumentCharset(const nsACString& aCharset,
  NS_PRECONDITION(!mExecutor->HasStarted(),
                  "Document charset set too late.");
  NS_PRECONDITION(mStreamParser, "Setting charset on a script-only parser.");
-  mStreamParser->SetDocumentCharset(aCharset, aCharsetSource);
-  mExecutor->SetDocumentCharsetAndSource((nsACString&)aCharset,
+  nsCAutoString trimmed;
+  trimmed.Assign(aCharset);
+  trimmed.Trim(" \t\r\n\f");
+  mStreamParser->SetDocumentCharset(trimmed, aCharsetSource);
+  mExecutor->SetDocumentCharsetAndSource(trimmed,
                                         aCharsetSource);
 }

--- a/parser/html/nsHtml5StreamParser.cpp
+++ b/parser/html/nsHtml5StreamParser.cpp
@ -768,7 +768,7 @@ nsHtml5StreamParser::internalEncodingDeclaration(nsString* aEncoding)

  nsCAutoString newEncoding;
  CopyUTF16toUTF8(*aEncoding, newEncoding);
-  // XXX spec says only UTF-16
+  newEncoding.Trim(" \t\r\n\f");
  if (newEncoding.LowerCaseEqualsLiteral("utf-16") ||
      newEncoding.LowerCaseEqualsLiteral("utf-16be") ||
      newEncoding.LowerCaseEqualsLiteral("utf-16le")) {
--- a/parser/htmlparser/tests/mochitest/Makefile.in
+++ b/parser/htmlparser/tests/mochitest/Makefile.in
@ -81,6 +81,7 @@ _TEST_FILES =	parser_datreader.js \
 		file_bug594730-6.html \
 		file_bug594730-7.html \
 		file_bug594730-8.html \
+		file_bug594730-9.html \
 		test_bug599584.html \
 		iframe_bug599584.html \
 		$(NULL)
--- a/parser/htmlparser/tests/mochitest/file_bug594730-9.html
+++ b/parser/htmlparser/tests/mochitest/file_bug594730-9.html
@ -0,0 +1,5 @@
+<!DOCTYPE html>
+<meta charset="
+  Windows-1250
+  ">
+<script>parent.is("ø", "\u0159", "Decoded bytes should have matched the Unicode escape.");</script>
--- a/parser/htmlparser/tests/mochitest/test_bug594730.html
+++ b/parser/htmlparser/tests/mochitest/test_bug594730.html
@ -22,6 +22,7 @@ https://bugzilla.mozilla.org/show_bug.cgi?id=594730
 <iframe src=file_bug594730-6.html></iframe>
 <iframe src=file_bug594730-7.html></iframe>
 <iframe src=file_bug594730-8.html></iframe>
+<iframe src=file_bug594730-9.html></iframe>
 <pre id="test">
 <script class="testbody" type="text/javascript">
 </script>