Script tags can contain "less-than" symbol

html-conduit/ChangeLog.md

@@ -1,3 +1,7 @@
+## 1.3.2
+
+* Fix a bug that was removing `<` symbols in script tags.
+
 ## 1.3.1
 
 * Inline tagstream-conduit for entity decoding in attribute value bug

html-conduit/html-conduit.cabal

@@ -1,5 +1,5 @@
 Name:                html-conduit
-Version:             1.3.1
+Version:             1.3.2
 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. Note that, since version 1.3.1, it uses an inlined copy of tagstream-conduit with entity decoding bugfixes applied.
 Homepage:            https://github.com/snoyberg/xml

html-conduit/src/Text/HTML/TagStream.hs

@@ -167,7 +167,7 @@ tillScriptEnd open =
       chunk <- takeTill (== '<')
       let acc' = acc <> B.fromText chunk
           finish = pure [open, Text $ L.toStrict $ B.toLazyText acc', TagClose "script"]
-          hasContent = (string "/script>" *> finish) <|> loop acc'
+          hasContent = (string "/script>" *> finish) <|> loop (acc' <> "<")
       (char '<' *> hasContent) <|> finish
 
 tokens :: Parser [Token]

html-conduit/test/main.hs

@@ -120,9 +120,9 @@ main = hspec $ do
 
     describe "script tags" $ do
       it "ignores funny characters" $
-        let html = "<script>hello > world</script>"
+        let html = "<script>hello <> world</script>"
             doc = X.Document (X.Prologue [] Nothing []) root []
-            root = X.Element "script" Map.empty [X.NodeContent "hello > world"]
+            root = X.Element "script" Map.empty [X.NodeContent "hello <> world"]
          in H.parseLBS html @?= doc
 
       {-