Add the ability to twat a single status (as a debugging tool).

[dead/halcyon.git] / src / Twitter / Xml.hs

diff --git a/src/Twitter/Xml.hs b/src/Twitter/Xml.hs
index 20015d37abcd58a43167b413a3552e45fb062d85..1b1ea48e9015ce1732e6898d4dcbfdf7b98b3176 100644 (file)
--- a/src/Twitter/Xml.hs
+++ b/src/Twitter/Xml.hs
@@ -2,7 +2,7 @@
 module Twitter.Xml
 where
 
-import Data.Maybe
+import Test.HUnit
 import Text.Regex (mkRegex, subRegex)
 import Text.XML.HaXml
 
@@ -15,6 +15,13 @@ get_char_data (CRef ref) = Just (verbatim ref) -- Entities.
 get_char_data _ = Nothing
 
 
+-- |A 'CFilter' returning all top-level <status> elements.
+-- The name is due to the fact that if we retrieve more than
+-- one status, they will be wrapped in a <statuses> tag, and
+-- thus not be top-level.
+single_status :: CFilter
+single_status = (tag "status")
+
 -- |A 'CFilter' returning all <status> tags within <statuses>.
 all_statuses :: CFilter
 all_statuses = (tag "statuses" /> tag "status")
@@ -50,12 +57,15 @@ user_screen_name = keep /> (tag "screen_name") /> txt
 -- character represented by that entity.
 xml_entities :: [(String, String)]
 xml_entities = [("[lr]dquo", "\""),
+                ("quot",     "\""),
                 ("[mn]dash", "-"),
                 ("nbsp",     " "),
                 ("#8217",    "'"),
                 ("amp",      "&"),
                 ("lt",       "<"),
-                ("gt",       ">")]
+                ("gt",       ">"),
+                ("#8230",    "..."),
+                ("hellip",   "...")]
 
 -- |Replace all of the XML entities in target.
 replace_entities :: String -> String
@@ -71,3 +81,16 @@ unescape_recursive replacements target =
       replacement = (replacements !! 0)
       from = "&" ++ (fst replacement) ++ ";"
       to = (snd replacement)
+
+
+
+xml_tests :: [Test]
+xml_tests = [ test_replace_entities ]
+
+
+test_replace_entities :: Test
+test_replace_entities =
+    TestCase $ assertEqual "All entities are replaced correctly." expected_text actual_text
+    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.\""