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 top-level <status> elements.
  19 -- The name is due to the fact that if we retrieve more than
  20 -- one status, they will be wrapped in a <statuses> tag, and
  21 -- thus not be top-level.
  22 single_status :: CFilter
  23 single_status = (tag "status")
  24
  25 -- |A 'CFilter' returning all <status> tags within <statuses>.
  26 all_statuses :: CFilter
  27 all_statuses = (tag "statuses" /> tag "status")
  28
  29 -- |Finds the text of the <id> element contained within some other
  30 -- content. Called unique_id here because status_id is used elsewhere.
  31 unique_id :: CFilter
  32 unique_id = keep /> (tag "id") /> txt
  33
  34 -- |Finds the text of the <created_at> element contained within some
  35 -- other element.
  36 status_created_at :: CFilter
  37 status_created_at = keep /> (tag "created_at") /> txt
  38
  39 -- |Finds the text of the <text> element contained within some
  40 -- other element.
  41 status_text :: CFilter
  42 status_text = keep /> (tag "text") /> txt
  43
  44 -- |Finds the XML of the <user> element contained within some other
  45 -- element.
  46 status_user :: CFilter
  47 status_user = keep /> (tag "user")
  48
  49 -- |Finds the text of the <screen_name> element contained within some
  50 -- other element.
  51 user_screen_name :: CFilter
  52 user_screen_name = keep /> (tag "screen_name") /> txt
  53
  54
  55 -- |A list of tuples whose first entry is a regular expression
  56 -- matching XML entities, and whose second entry is the ASCII
  57 -- character represented by that entity.
  58 xml_entities :: [(String, String)]
  59 xml_entities = [("[lr]dquo", "\""),
  60                 ("quot",     "\""),
  61                 ("[mn]dash", "-"),
  62                 ("nbsp",     " "),
  63                 ("#8217",    "'"),
  64                 ("amp",      "&"),
  65                 ("lt",       "<"),
  66                 ("gt",       ">"),
  67                 ("#8230",    "..."),
  68                 ("hellip",   "...")]
  69
  70 -- |Replace all of the XML entities in target.
  71 replace_entities :: String -> String
  72 replace_entities target = unescape_recursive xml_entities target
  73
  74 -- |The recursive function which does the real work for
  75 -- 'replace_entities'.
  76 unescape_recursive :: [(String, String)] -> String -> String
  77 unescape_recursive [] target = target
  78 unescape_recursive replacements target =
  79     unescape_recursive (tail replacements) (subRegex (mkRegex from) target to)
  80     where
  81       replacement = (replacements !! 0)
  82       from = "&" ++ (fst replacement) ++ ";"
  83       to = (snd replacement)
  84
  85
  86
  87 xml_tests :: [Test]
  88 xml_tests = [ test_replace_entities ]
  89
  90
  91 test_replace_entities :: Test
  92 test_replace_entities =
  93     TestCase $ assertEqual "All entities are replaced correctly." expected_text actual_text
  94     where
  95       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;")
  96       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.\""