56f94ec · typeable/html-parse

git.iko.soy/typeable/html-parse

Track self-closing tags

commit56f94ecef69fcba430969ff14998370d76839902 authorBen Gamari <ben@smart-cactus.org> dateMon 2 Jan 2017, 13:31 (10 years ago) parent801d861

2 files changed+10−7

unifiedsplit

Text/HTML/Parser.hs+3−1

@@ -53,6 +53,8 @@ type AttrValue = Text

5353data Token

5454 -- | An opening tag. Attribute ordering is arbitrary.

5555 = TagOpen !TagName [Attr]

56+ -- | A self-closing tag.

57+ | TagSelfClose !TagName [Attr]

5658 -- | A closing tag.

5759 | TagClose !TagName

5860 -- | The content between tags.

@@ -129,7 +131,7 @@ tagName' = do

129131-- | /§8.2.4.43/: Self-closing start tag state

130132selfClosingStartTag :: TagName -> [Attr] -> Parser Token

131133selfClosingStartTag tag attrs = do

132− (char '>' >> return (TagOpen tag attrs))

134+ (char '>' >> return (TagSelfClose tag attrs))

133135 <|> beforeAttrName tag attrs

134136

135137-- | /§8.2.4.34/: Before attribute name state

tests/Text/HTML/ParserSpec.hs+7−6

@@ -24,12 +24,13 @@ import Text.HTML.Parser

2424instance Arbitrary Token where

2525 arbitrary = oneof [validOpen, validClose, validFlat]

2626

27− shrink (TagOpen n as) = TagOpen n <$> shrink as

28− shrink (TagClose _) = []

29− shrink (ContentText _) = []

30− shrink (ContentChar _) = []

31− shrink (Comment b) = Comment . B.fromText <$> (shrink . TL.toStrict . B.toLazyText $ b)

32− shrink (Doctype t) = Doctype <$> shrink t

27+ shrink (TagOpen n as) = TagOpen n <$> shrink as

28+ shrink (TagSelfClose n as) = TagSelfClose n <$> shrink as

29+ shrink (TagClose _) = []

30+ shrink (ContentText _) = []

31+ shrink (ContentChar _) = []

32+ shrink (Comment b) = Comment . B.fromText <$> (shrink . TL.toStrict . B.toLazyText $ b)

33+ shrink (Doctype t) = Doctype <$> shrink t

3334

3435instance Arbitrary Attr where

3536 arbitrary = Attr <$> validXmlAttrName <*> validXmlAttrValue