src/Twitter/Xml.hs

   1 -- |Application-specific XML functions.
   2 module Twitter.Xml
   3 where
   4
   5 import Test.HUnit
   6 import Text.Regex (mkRegex, subRegex)
   7 import Text.XML.HaXml
   8
   9 -- |Returns the 'CharData' contained within the given 'Content', or
  10 -- 'Nothing' if no acceptable CharData was found. It will parse either
  11 -- a 'CString' ('String') or 'CRef' (XML entity reference).
  12 get_char_data :: Content -> (Maybe CharData)
  13 get_char_data (CString _ cd) = Just cd
  14 get_char_data (CRef ref) = Just (verbatim ref) -- Entities.
  15 get_char_data _ = Nothing
  16
  17
  18 -- |A 'CFilter' returning all <status> tags within <statuses>.
  19 all_statuses :: CFilter
  20 all_statuses = (tag "statuses" /> tag "status")
  21
  22 -- |Finds the text of the <id> element contained within some other
  23 -- content. Called unique_id here because status_id is used elsewhere.
  24 unique_id :: CFilter
  25 unique_id = keep /> (tag "id") /> txt
  26
  27 -- |Finds the text of the <created_at> element contained within some
  28 -- other element.
  29 status_created_at :: CFilter
  30 status_created_at = keep /> (tag "created_at") /> txt
  31
  32 -- |Finds the text of the <text> element contained within some
  33 -- other element.
  34 status_text :: CFilter
  35 status_text = keep /> (tag "text") /> txt
  36
  37 -- |Finds the XML of the <user> element contained within some other
  38 -- element.
  39 status_user :: CFilter
  40 status_user = keep /> (tag "user")
  41
  42 -- |Finds the text of the <screen_name> element contained within some
  43 -- other element.
  44 user_screen_name :: CFilter
  45 user_screen_name = keep /> (tag "screen_name") /> txt
  46
  47
  48 -- |A list of tuples whose first entry is a regular expression
  49 -- matching XML entities, and whose second entry is the ASCII
  50 -- character represented by that entity.
  51 xml_entities :: [(String, String)]
  52 xml_entities = [("[lr]dquo", "\""),
  53                 ("quot",     "\""),
  54                 ("[mn]dash", "-"),
  55                 ("nbsp",     " "),
  56                 ("#8217",    "'"),
  57                 ("amp",      "&"),
  58                 ("lt",       "<"),
  59                 ("gt",       ">"),
  60                 ("#8230",    "..."),
  61                 ("hellip",   "...")]
  62
  63 -- |Replace all of the XML entities in target.
  64 replace_entities :: String -> String
  65 replace_entities target = unescape_recursive xml_entities target
  66
  67 -- |The recursive function which does the real work for
  68 -- 'replace_entities'.
  69 unescape_recursive :: [(String, String)] -> String -> String
  70 unescape_recursive [] target = target
  71 unescape_recursive replacements target =
  72     unescape_recursive (tail replacements) (subRegex (mkRegex from) target to)
  73     where
  74       replacement = (replacements !! 0)
  75       from = "&" ++ (fst replacement) ++ ";"
  76       to = (snd replacement)
  77
  78
  79
  80 xml_tests :: [Test]
  81 xml_tests = [ test_replace_entities ]
  82
  83
  84 test_replace_entities :: Test
  85 test_replace_entities =
  86     TestCase $ assertEqual "All entities are replaced correctly." expected_text actual_text
  87     where
  88       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;")
  89       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.\""