summaryrefslogtreecommitdiff
path: root/test/data/csdetect/tests2.dat
diff options
context:
space:
mode:
authorAndrew Sidwell <andy@entai.co.uk>2008-08-09 15:47:34 +0000
committerAndrew Sidwell <andy@entai.co.uk>2008-08-09 15:47:34 +0000
commit3d3e3b0dfda0fe808039b4b8e8b455bc0ea1b704 (patch)
treeacd2365a57b730844727138ba42ae407061e9b3f /test/data/csdetect/tests2.dat
parented627ac9bdd961cc98cba8fc455596d288f15212 (diff)
downloadlibhubbub-3d3e3b0dfda0fe808039b4b8e8b455bc0ea1b704.tar.gz
libhubbub-3d3e3b0dfda0fe808039b4b8e8b455bc0ea1b704.tar.bz2
Bring encoding detection tests up-to-date with html5lib, fix Hubbub-specific ones, and bring the csdetect algorithm up-to-date with that in the spec.
svn path=/trunk/hubbub/; revision=4974
Diffstat (limited to 'test/data/csdetect/tests2.dat')
-rw-r--r--test/data/csdetect/tests2.dat44
1 files changed, 38 insertions, 6 deletions
diff --git a/test/data/csdetect/tests2.dat b/test/data/csdetect/tests2.dat
index dd43f85..1908dec 100644
--- a/test/data/csdetect/tests2.dat
+++ b/test/data/csdetect/tests2.dat
@@ -19,20 +19,19 @@ windows-1252
windows-1252
#data
-<meta charset=EUC-jp
+<meta charset=euc_jp
#encoding
windows-1252
#data
-<meta <meta charset='EUC-jp'>
+<meta <meta charset='euc_jp'>
#encoding
-EUC-jp
+euc_jp
#data
-<meta charset = 'EUC-jp'>
+<meta charset = 'euc_jp'>
#encoding
-EUC-jp
-
+euc_jp
#data
<!-- -->
@@ -80,3 +79,36 @@ windows-1252
<meta content = "text/html;
#encoding
windows-1252
+
+#data
+<meta charset="UTF-16">
+#encoding
+utf-8
+
+#data
+<meta charset="UTF-16LE">
+#encoding
+utf-8
+
+#data
+<meta charset="UTF-16BE">
+#encoding
+utf-8
+
+#data
+<html a=ñ>
+<meta charset="utf-8">
+#encoding
+utf-8
+
+#data
+<html ñ>
+<meta charset="utf-8">
+#encoding
+utf-8
+
+#data
+<html>ñ
+<meta charset="utf-8">
+#encoding
+utf-8