src/TSN/Xml.hs

   1 -- | Minimal XML functionality needed to parse each document's
   2 --   XML_File_ID.
   3 --
   4 module TSN.Xml (
   5   parse_xmlfid,
   6   xml_prologue )
   7 where
   8
   9 import Data.Maybe (listToMaybe, mapMaybe)
  10 import Text.Read (readMaybe)
  11 import Text.XML.HXT.Core (
  12   (>>>),
  13   (/>),
  14   getChildren,
  15   getText,
  16   hasName,
  17   runLA,
  18   xreadDoc )
  19
  20 -- | A tiny parser written in HXT to extract the "XML_File_ID" element
  21 --   from a document.
  22 parse_xmlfid :: String -> Maybe Integer
  23 parse_xmlfid =
  24   listToMaybe . mapMaybe readMaybe . parse
  25   where
  26     parse :: String -> [String]
  27     parse =
  28       runLA (xreadDoc
  29                >>> hasName "message"
  30                />  hasName "XML_File_ID"
  31                >>> getChildren
  32                >>> getText)
  33
  34
  35 -- | The opening "tag" for the XML prologue.
  36 xml_prologue :: String
  37 xml_prologue = "<?xml "