eb6fb62 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

HTML reader: Use sets instead of lists for block tag lookup.

eb6fb62John MacFarlane9 years ago

src/Text/Pandoc/Readers/HTML.hs+43−50

compare with master

unifiedsplit

@@ -869,62 +869,54 @@ pSpace = many1 (satisfy isSpace) >>= \xs ->

869-- Constants

870--

871

872−eitherBlockOrInline :: [String]

872+eitherBlockOrInline :: Set.Set String

873+eitherBlockOrInline = Set.fromList

873−eitherBlockOrInline = ["audio", "applet", "button", "iframe", "embed",

874+ ["audio", "applet", "button", "iframe", "embed",

874− "del", "ins",

875− "progress", "map", "area", "noscript", "script",

875+ "del", "ins", "progress", "map", "area", "noscript", "script",

876− "object", "svg", "video", "source"]

876+ "object", "svg", "video", "source"]

877−

877+

878−{-

879−inlineHtmlTags :: [[Char]]

880−inlineHtmlTags = ["a", "abbr", "acronym", "b", "basefont", "bdo", "big",

881− "br", "cite", "code", "dfn", "em", "font", "i", "img",

882− "input", "kbd", "label", "q", "s", "samp", "select",

883− "small", "span", "strike", "strong", "sub", "sup",

884− "textarea", "tt", "u", "var"]

885−-}

886−

887−blockHtmlTags :: [String]

878+blockHtmlTags :: Set.Set String

879+blockHtmlTags = Set.fromList

888−blockHtmlTags = ["?xml", "!DOCTYPE", "address", "article", "aside",

880+ ["?xml", "!DOCTYPE", "address", "article", "aside",

889− "blockquote", "body", "canvas",

881+ "blockquote", "body", "canvas",

890− "caption", "center", "col", "colgroup", "dd", "details",

882+ "caption", "center", "col", "colgroup", "dd", "details",

891− "dir", "div",

892− "dl", "dt", "fieldset", "figcaption", "figure",

883+ "dir", "div", "dl", "dt", "fieldset", "figcaption", "figure",

893− "footer", "form", "h1", "h2", "h3", "h4",

884+ "footer", "form", "h1", "h2", "h3", "h4",

894− "h5", "h6", "head", "header", "hgroup", "hr", "html",

885+ "h5", "h6", "head", "header", "hgroup", "hr", "html",

895− "isindex", "menu", "noframes", "ol", "output", "p", "pre",

886+ "isindex", "menu", "noframes", "ol", "output", "p", "pre",

896− "section", "table", "tbody", "textarea",

887+ "section", "table", "tbody", "textarea",

897− "thead", "tfoot", "ul", "dd",

888+ "thead", "tfoot", "ul", "dd",

898− "dt", "frameset", "li", "tbody", "td", "tfoot",

889+ "dt", "frameset", "li", "tbody", "td", "tfoot",

899− "th", "thead", "tr", "script", "style"]

890+ "th", "thead", "tr", "script", "style"]

900

891

901-- We want to allow raw docbook in markdown documents, so we

892-- We want to allow raw docbook in markdown documents, so we

902-- include docbook block tags here too.

893-- include docbook block tags here too.

903−blockDocBookTags :: [String]

894+blockDocBookTags :: Set.Set String

895+blockDocBookTags = Set.fromList

904−blockDocBookTags = ["calloutlist", "bibliolist", "glosslist", "itemizedlist",

896+ ["calloutlist", "bibliolist", "glosslist", "itemizedlist",

905− "orderedlist", "segmentedlist", "simplelist",

897+ "orderedlist", "segmentedlist", "simplelist",

906− "variablelist", "caution", "important", "note", "tip",

898+ "variablelist", "caution", "important", "note", "tip",

907− "warning", "address", "literallayout", "programlisting",

899+ "warning", "address", "literallayout", "programlisting",

908− "programlistingco", "screen", "screenco", "screenshot",

900+ "programlistingco", "screen", "screenco", "screenshot",

909− "synopsis", "example", "informalexample", "figure",

901+ "synopsis", "example", "informalexample", "figure",

910− "informalfigure", "table", "informaltable", "para",

902+ "informalfigure", "table", "informaltable", "para",

911− "simpara", "formalpara", "equation", "informalequation",

903+ "simpara", "formalpara", "equation", "informalequation",

912− "figure", "screenshot", "mediaobject", "qandaset",

904+ "figure", "screenshot", "mediaobject", "qandaset",

913− "procedure", "task", "cmdsynopsis", "funcsynopsis",

905+ "procedure", "task", "cmdsynopsis", "funcsynopsis",

914− "classsynopsis", "blockquote", "epigraph", "msgset",

906+ "classsynopsis", "blockquote", "epigraph", "msgset",

915− "sidebar", "title"]

907+ "sidebar", "title"]

916−

908+

917−epubTags :: [String]

909+epubTags :: Set.Set String

918−epubTags = ["case", "switch", "default"]

910+epubTags = Set.fromList ["case", "switch", "default"]

919−

911+

920−blockTags :: [String]

912+blockTags :: Set.Set String

921−blockTags = blockHtmlTags ++ blockDocBookTags ++ epubTags

913+blockTags = Set.unions [blockHtmlTags, blockDocBookTags, epubTags]

922

914

923isInlineTag :: Tag String -> Bool

915isInlineTag :: Tag String -> Bool

924isInlineTag t = tagOpen isInlineTagName (const True) t ||

916isInlineTag t = tagOpen isInlineTagName (const True) t ||

925 tagClose isInlineTagName t ||

917 tagClose isInlineTagName t ||

926 tagComment (const True) t

918 tagComment (const True) t

927− where isInlineTagName x = x `notElem` blockTags

919+ where isInlineTagName x = x `Set.notMember` blockTags

928

920

929isBlockTag :: Tag String -> Bool

921isBlockTag :: Tag String -> Bool

930isBlockTag t = tagOpen isBlockTagName (const True) t ||

922isBlockTag t = tagOpen isBlockTagName (const True) t ||

@@ -976,8 +968,9 @@ t `closes` t2 |

976 t `elem` ["h1","h2","h3","h4","h5","h6","dl","ol","ul","table","div","p"] &&

968 t `elem` ["h1","h2","h3","h4","h5","h6","dl","ol","ul","table","div","p"] &&

977 t2 `elem` ["h1","h2","h3","h4","h5","h6","p" ] = True -- not "div"

969 t2 `elem` ["h1","h2","h3","h4","h5","h6","p" ] = True -- not "div"

978t1 `closes` t2 |

970t1 `closes` t2 |

979− t1 `elem` blockTags &&

971+ t1 `Set.member` blockTags &&

972+ t2 `Set.notMember` blockTags &&

980− t2 `notElem` (blockTags ++ eitherBlockOrInline) = True

973+ t2 `Set.notMember` eitherBlockOrInline = True

981_ `closes` _ = False

974_ `closes` _ = False

982

975

983--- parsers for use in markdown, textile readers

976--- parsers for use in markdown, textile readers