X-Git-Url: http://gitweb.michael.orlitzky.com/?p=dead%2Fhalcyon.git;a=blobdiff_plain;f=src%2FTwitter%2FXml.hs;h=871d2164dee05737e9861d45c2e400dfde76e741;hp=0e23cb7e736905dcac1afc03bfd8c18f3cbd6341;hb=5454ac1d1f84984a7ce85cb1be340d18a86a5cfb;hpb=86efdb6afac4a7a165babba90a0a25e0e1185e5b

diff --git a/src/Twitter/Xml.hs b/src/Twitter/Xml.hs
index 0e23cb7..871d216 100644
--- a/src/Twitter/Xml.hs
+++ b/src/Twitter/Xml.hs
@@ -79,15 +79,20 @@ entity_from_codepoint codepoint =
       Just num -> [(chr num)]
 
 
--- |A list of tuples whose first entry is a regular expression
--- matching XML entities, and whose second entry is the ASCII
--- character represented by that entity.
+-- | A list of tuples whose first entry is a regular expression
+--   matching XML entities, and whose second entry is the ASCII
+--   character represented by that entity.
+--
+--   For some reason, ampersands are escaped twice in the status
+--   text. Rather than unescape everything twice, we just stick "amp"
+--   in the list again.
 xml_entities :: [(String, String)]
 xml_entities = [("[lr]dquo", "\""),
                 ("quot",     "\""),
                 ("[mn]dash", "-"),
                 ("nbsp",     " "),
                 ("amp",      "&"),
+                ("amp",      "&"),
                 ("lt",       "<"),
                 ("gt",       ">"),
                 ("hellip",   "…")]
@@ -130,7 +135,7 @@ unescape_recursive replacements target =
 
 
 xml_tests :: [Test]
-xml_tests = [ test_replace_entities ]
+xml_tests = [ test_replace_entities, test_double_unescape ]
 
 
 test_replace_entities :: Test
@@ -139,3 +144,11 @@ test_replace_entities =
     where
       actual_text = (replace_entities "&quot;The moon is gay&#8230;&hellip;&quot; said &lt;insert the current president of the United States of America&gt;. &ldquo;It&#8217;s OK&mdash;&ndash;he&#8217;s not a real doctor.&rdquo;")
       expected_text = "\"The moon is gay……\" said <insert the current president of the United States of America>. \"It’s OK--he’s not a real doctor.\""
+
+
+test_double_unescape :: Test
+test_double_unescape =
+    TestCase $ assertEqual "The status text is unescaped twice." expected_text actual_text
+    where
+      actual_text = (replace_entities "As a kid, I'd pull a girl's hair to let her know I liked her, but now that I'm older &amp;amp; wiser I simply hit her with my car.")
+      expected_text = "As a kid, I'd pull a girl's hair to let her know I liked her, but now that I'm older & wiser I simply hit her with my car."