html-conduit 1.1.1.1 → 1.1.1.2

raw patch · 5 files changed

+20/−25 lines, 5 filesdep ~tagstream-conduitPVP ok

version bump matches the API change (PVP)

Dependency ranges changed: tagstream-conduit

API changes (from Hackage documentation)

Files

• ChangeLog.md +3/−0
• README.md +4/−0
• Text/HTML/DOM.hs +2/−22
• html-conduit.cabal +3/−3
• test/main.hs +8/−0

@@ -0,0 +1,3 @@+## 1.1.1.2++* Fix a bug with double-unescaping of entities

@@ -0,0 +1,4 @@+This package uses tagstream-conduit for its parser. It automatically balances+mismatched tags, so that there shouldn't be any parse failures. It does not+handle a full HTML document rendering, such as adding missing html and head+tags.

@@ -1,4 +1,4 @@-{-# LANGUAGE OverloadedStrings, CPP #-}+{-# LANGUAGE OverloadedStrings #-} module Text.HTML.DOM     ( eventConduit     , sinkDoc@@ -10,9 +10,7 @@ import Control.Monad.Trans.Resource import Prelude hiding (readFile) import qualified Data.ByteString as S-#if MIN_VERSION_tagstream_conduit(0,5,0) import qualified Text.HTML.TagStream.Text as TS-#endif import qualified Text.HTML.TagStream as TS import qualified Data.XML.Types as XT import Data.Conduit@@ -41,7 +39,7 @@   where     go stack = do         mx <- await-        case fmap entities mx of+        case mx of             Nothing -> closeStack stack              -- Ignore processing instructions (or pseudo-instructions)@@ -74,24 +72,6 @@             Just TS.Incomplete{} -> go stack     toName l = XT.Name l Nothing Nothing     closeStack = mapM_ (yield . XT.EventEndElement)--    entities :: TS.Token' Text -> TS.Token' Text-    entities (TS.TagOpen x pairs b) = TS.TagOpen x (map (second entities') pairs) b-    entities (TS.Text x) = TS.Text $ entities' x-    entities ts = ts--    entities' :: Text -> Text-    entities' t =-        case T.break (== '&') t of-            (_, "") -> t-            (before, t') ->-                case T.break (== ';') $ T.drop 1 t' of-                    (_, "") -> t-                    (entity, rest') ->-                        let rest = T.drop 1 rest'-                         in case decodeHtmlEntities entity of-                                XT.ContentText entity' -> T.concat [before, entity', entities' rest]-                                XT.ContentEntity _ -> T.concat [before, "&", entity, entities' rest']      isVoid = flip Set.member $ Set.fromList         [ "area"

@@ -1,5 +1,5 @@ Name:                html-conduit-Version:             1.1.1.1+Version:             1.1.1.2 Synopsis:            Parse HTML documents using xml-conduit datatypes. Description:         This package uses tagstream-conduit for its parser. It automatically balances mismatched tags, so that there shouldn't be any parse failures. It does not handle a full HTML document rendering, such as adding missing html and head tags. Homepage:            https://github.com/snoyberg/xml@@ -9,7 +9,7 @@ Maintainer:          michael@snoyman.com Category:            Web, Text, Conduit Build-type:          Simple-Extra-source-files:  test/main.hs+Extra-source-files:  test/main.hs README.md ChangeLog.md Cabal-version:       >=1.8  Library@@ -24,7 +24,7 @@                      , conduit-extra                    >= 1.1.1                      , system-filepath                  >= 0.4            && < 0.5                      , xml-conduit                      >= 1.1            && < 1.3-                     , tagstream-conduit                >= 0.4            && < 0.6+                     , tagstream-conduit                >= 0.5.5.3        && < 0.6                      , xml-types                        >= 0.3            && < 0.4  test-suite test

@@ -86,3 +86,11 @@                         ]                     ]              in H.parseLBS html @?= doc++    it "doesn't double unescape" $+        let html = "<p>Hello &amp;gt; World</p>"+            doc = X.Document (X.Prologue [] Nothing []) root []+            root = X.Element "p" Map.empty+                [ X.NodeContent "Hello &gt; World"+                ]+         in H.parseLBS html @?= doc