X-Git-Url: http://gitweb.michael.orlitzky.com/?a=blobdiff_plain;f=src%2FXml.hs;h=e8b7c4de311e0e90203dbda322e4bae2d2767676;hb=88b80555c3df3649799c8caa4de8d9c87c50be45;hp=21a806847ca1761845260f1a2d4f92344d07916d;hpb=53d408ee8691bc123dd55eba64e2c00cd17c8f57;p=dead%2Fhtsn-import.git

diff --git a/src/Xml.hs b/src/Xml.hs
index 21a8068..e8b7c4d 100644
--- a/src/Xml.hs
+++ b/src/Xml.hs
@@ -3,76 +3,76 @@
 -- | General XML stuff.
 --
 module Xml (
+  DtdName(..),
   FromXml(..),
   parse_opts,
-  pickle_unpickle )
+  pickle_unpickle,
+  unpickleable )
 where
 
+import Control.Exception ( SomeException(..), catch )
 import Text.XML.HXT.Core (
   (>>>),
   (/>),
+  PU,
   SysConfigList,
   XmlPickler(..),
   hasName,
-  no,
   readDocument,
   runX,
-  withPreserveComment,
   withRemoveWS,
-  withSubstDTDEntities,
-  withValidate,
   xpickleVal,
   xunpickleVal,
   yes )
 
 
--- | A typeclass for types which can be converted into an associated
---   XML type. The story behind this is long, but basically, we need
---   to different types for each XML thingie we're going to import: a
---   database type and an XML type. Both Groundhog and HXT are very
---   particular about the types that they can use, and there's no way
---   to reuse e.g. a type that HXT can pickle in Groundhog. So this
---   typeclass gives us a way to get the XML type from the Groundhog
---   type.
+-- | A typeclass for XML types that can be converted into an
+--   associated database type. The story behind this is long, but
+--   basically, we need to different types most XML thingies we're
+--   going to import: a database type and an XML type.
 --
---   At first there appears to be an equally-valid approach, getting the
---   Groundhog type from the XML one. But Groundhog won't use type family
---   instances, so here we are.
+--   Both Groundhog and HXT are very particular about the types that
+--   they can use, and there's no way to reuse e.g. a type that HXT
+--   can pickle in Groundhog. This typeclass gives us a standard way
+--   to get the database type from the XML type that we have to define
+--   for HXT.
 --
 class FromXml a where
-  -- | Each instance a must declare its associated XML type (Xml a)
-  type Xml a :: *
+  -- | Each instance @a@ must declare its associated database type @Db a@.
+  type Db a :: *
 
-  -- | And provide a function for getting an (Xml a) out of an "a."
-  to_xml :: a -> Xml a
+  -- | And provide a function for getting a @Db a@ out of an @a@.
+  from_xml :: a -> Db a
 
 
+-- | Represents the DTD filename (\"SYSTEM\") part of the DOCTYPE
+--   definition.
+newtype DtdName = DtdName String
+
 -- | A list of options passed to 'readDocument' when we parse an XML
---   document. We don't validate because the DTDs from TSN are
---   wrong. As a result, we don't want to keep useless DTDs
---   areound. Thus we disable 'withSubstDTDEntities' which, when
---   combined with "withValidate no", prevents HXT from trying to read
---   the DTD at all.
+--   document. All cosmetic whitespace should be removed, otherwise we
+--   would have to parse whitespace in each (un)pickler.
 --
 parse_opts :: SysConfigList
-parse_opts =
-  [ withPreserveComment no,
-    withRemoveWS yes,
-    withSubstDTDEntities no,
-    withValidate no ]
+parse_opts = [ withRemoveWS yes ]
 
 
 -- | Given a root element name and a file path, return both the
---   original unpickled root "object" and the one that was constructed
---   by pickled and unpickling the original. This is used in a number
---   of XML tests which pickle/unpickle and then make sure that the
---   output is the same as the input.
+--   original unpickled root \"object\" and the one that was
+--   constructed by pickled and unpickling the original. This is used
+--   in a number of XML tests which pickle/unpickle and then make sure
+--   that the output is the same as the input.
 --
 --   We return the object instead of an XmlTree (which would save us
 --   an unpickle call) because otherwise the type of @a@ in the call
 --   to 'xpickle' would be ambiguous. By returning some @a@s, we allow
 --   the caller to annotate its type.
 --
+--   Note that this will happily pickle nothing to nothing and then
+--   unpickle it back to more nothing. So the fact that the
+--   before/after results from this function agree does not mean that
+--   the document was successfully unpickled!
+--
 pickle_unpickle :: XmlPickler a
                 => String
                 -> FilePath
@@ -80,7 +80,7 @@ pickle_unpickle :: XmlPickler a
 pickle_unpickle root_element filepath = do
   -- We need to check only the root message element since
   -- readDocument produces a bunch of other junk.
-    expected <- runX $ arr_getobj
+    expected <- runX arr_getobj
     actual <- runX $ arr_getobj
                      >>>
                      xpickleVal xpickle
@@ -94,3 +94,23 @@ pickle_unpickle root_element filepath = do
                    hasName root_element
                    >>>
                    xunpickleVal xpickle
+
+
+
+-- | Is the given XML file unpickleable? Unpickling will be attempted
+--   using the @unpickler@ argument. If we unilaterally used the
+--   generic 'xpickle' function for our unpickler, a type ambiguity
+--   would result. By taking the unpickler as an argument, we allow
+--   the caller to indirectly specify a concrete type.
+--
+--   Apologies the the name; unpickleable means \"we can unpickle
+--   it\", not \"not pickleable.\"
+--
+unpickleable :: XmlPickler a => FilePath -> PU a -> IO Bool
+unpickleable filepath unpickler = do
+  xmldoc <- try_unpickle `catch` (\(SomeException _) -> return [])
+  return $ (not . null) xmldoc
+  where
+    try_unpickle = runX $ readDocument parse_opts filepath
+                          >>>
+                          xunpickleVal unpickler