src/Twitter/Xml.hs

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