hxt-tagsoup 9.0.0 → 9.1.0

raw patch · 3 files changed

+49/−21 lines, 3 filesdep +hxt-unicodedep ~hxtdep ~tagsoupPVP ok

version bump matches the API change (PVP)

Dependencies added: hxt-unicode

Dependency ranges changed: hxt, tagsoup

API changes (from Hackage documentation)

Files

• hxt-tagsoup.cabal +5/−3
• src/Text/XML/HXT/Arrow/TagSoupInterface.hs +31/−5
• src/Text/XML/HXT/Parser/TagSoup.hs +13/−13

@@ -1,6 +1,6 @@ -- arch-tag: Haskell XML Toolbox main description file Name:           hxt-tagsoup-Version:        9.0.0+Version:        9.1.0 Synopsis:       TagSoup parser for HXT Description:    The Tagsoup interface for the HXT lazy HTML parser. License:        OtherLicense@@ -28,11 +28,13 @@  hs-source-dirs: src   ghc-options: -Wall+ ghc-prof-options: -auto-all -caf-all   extensions: MultiParamTypeClasses DeriveDataTypeable FunctionalDependencies FlexibleInstances   build-depends: base               >= 4    && < 5,                 haskell98          >= 1    && < 2,-                tagsoup            >= 0.10 && < 0.11,+                tagsoup            >= 0.10 && < 0.13,                 hxt-charproperties >= 9    && < 10,-                hxt                >= 9    && < 10+                hxt-unicode        >= 9    && < 10,+                hxt                >= 9.1  && < 10

@@ -2,7 +2,7 @@  {- |    Module     : Text.XML.HXT.Arrow.TagSoupInterface-   Copyright  : Copyright (C) 2005 Uwe Schmidt+   Copyright  : Copyright (C) 2010 Uwe Schmidt    License    : MIT     Maintainer : Uwe Schmidt (uwe@fh-wedel.de)@@ -22,6 +22,8 @@ import Control.Arrow.ArrowList import Control.Arrow.ArrowTree +import Data.String.Unicode            ( normalizeNL )+ import Text.XML.HXT.DOM.Interface  import Text.XML.HXT.Arrow.XmlArrow@@ -32,11 +34,35 @@  -- ------------------------------------------------------------ --- | enable TagSoup parsing+{- |+   The system config option to enable the tagsoup parser +Here is an example, how to use it:++> ...+> import Text.HXT.XML.Core+> import Text.HXT.XML.TagSoup+> ...+>+> readDocument [ withExpat ] "some-file.xml"+> ...++reads the given document and parses it with the lazy tagsoup parser.+There is no validation enabled.+-}++ withTagSoup                     :: SysConfig-withTagSoup                     = setS (theTagSoup .&&&. theTagSoupParser) (True, parseHtmlTagSoup)+withTagSoup                     = setS (theTagSoup       .&&&.+                                        theExpat         .&&&.+                                        theTagSoupParser+                                       ) (True, (False, parseHtmlTagSoup)) +-- | Turns off tagsoup parsing. The build in HXT parser will be used.++withoutTagSoup                  :: SysConfig+withoutTagSoup                  = setS theTagSoup False+ -- ------------------------------------------------------------  -- | The Tagsoup parser arrow@@ -59,8 +85,8 @@                                   replaceChildren                                   ( ( getAttrValue a_source               -- get source name                                       &&&-                                      xshow getChildren-                                    )                                     -- get string to be parsed+                                      (xshow getChildren >>^ normalizeNL) -- get string to be parsed and normalize newline char+                                    )                                     >>>                                     arr2L (TS.parseHtmlTagSoup withNamespaces' withWarnings' preserveCmt' removeWS' lowerCaseNames')                                   )

@@ -42,8 +42,8 @@                                         , NsEnv                                         , toNsEnv                                         , newXName+                                        , newQName                                         , nullXName-                                        , mkQName'                                         , mkName                                         , isWellformedQualifiedName                                         , c_warn@@ -53,11 +53,11 @@                                         , xmlnsNamespace                                         ) import Text.XML.HXT.DOM.XmlNode         ( isElem-                                        , mkError-                                        , mkCmt-                                        , mkText+                                        , mkError'+                                        , mkCmt'+                                        , mkText'                                         , mkElement-                                        , mkAttr+                                        , mkAttr'                                         )  -- ----------------------------------------@@ -142,7 +142,7 @@ -- primitive symbol parsers  getTag          :: Parser STag-getTag          = P $ \ (t1:ts1) -> (t1, ts1)+getTag          = P $ \ t -> (head t, tail t)  getSym          :: (STag -> a) -> Parser a getSym f        = do@@ -208,8 +208,8 @@     where     qn1         | isAttr && isSimpleName        = s'-        | isSimpleName                  = mkQName' nullXName (newXName s) (nsUri nullXName)-        | isWellformedQualifiedName s   = mkQName' px'        lp'         (nsUri px')+        | isSimpleName                  = newQName (newXName s) nullXName (nsUri nullXName)+        | isWellformedQualifiedName s   = newQName lp'          px'       (nsUri px')         | otherwise                     = s'     qn0                                 = s' @@ -359,18 +359,18 @@     wrap                = (:[])      warn-        | withWarnings  = wrap . mkError c_warn . show . (doc ++) . (" " ++)+        | withWarnings  = wrap . mkError' c_warn . show . (doc ++) . (" " ++)         | otherwise     = const []     cmt-        | withComment   = wrap . mkCmt+        | withComment   = wrap . mkCmt'         | otherwise     = const []     txt         | removeWhiteSpace                         = \ t ->                           if all isXmlSpaceChar t                           then []-                          else wrap . mkText $ t-        | otherwise     = wrap . mkText+                          else wrap . mkText' $ t+        | otherwise     = wrap . mkText'      isEmptyElem         | asHtml        = isEmptyHtmlTag@@ -486,7 +486,7 @@             mkAttrs             = mapM (uncurry mkA)             mkA an av           = do                                   qan <- mkAttrQN nenv an-                                  return (mkAttr qan (wrap . mkText $ av))+                                  return (mkAttr' qan (wrap . mkText' $ av))          closeAll                :: ([String], NsEnv) -> Parser XmlTrees         closeAll (ns',_)        = return (concatMap wrn ns')