Regularized the scheme for unique header identifiers in HTML writer:

- punctuation is now all removed (except -)
 - spaces are turned into -
 - all lowercase
This scheme should be fairly predictable.
Updated tests accordingly.


git-svn-id: https://pandoc.googlecode.com/svn/trunk@655 788f1e2b-df1e-0410-8736-df70ead52e1b

src/Text/Pandoc/Writers/HTML.hs

@@ -34,7 +34,7 @@ import Text.Pandoc.Entities (decodeEntities)
 import Text.Regex ( mkRegex, matchRegex )
 import Numeric ( showHex )
 import Data.Char ( ord, toLower )
-import Data.List ( isPrefixOf, partition )
+import Data.List ( isPrefixOf, partition, intersperse )
 import Control.Monad.State
 import Text.XHtml.Strict
 
@@ -169,22 +169,32 @@ obfuscateChar char =
 obfuscateString :: String -> String
 obfuscateString = (concatMap obfuscateChar) . decodeEntities
 
+-- | True if character is a punctuation character (unicode).
+isPunctuation :: Char -> Bool
+isPunctuation c =
+  let c' = ord c in
+  if (c `elem` "!\"'()*,-./:;<>?[\\]`{|}~") || (c' >= 0x2000 && c' <= 0x206F) ||
+     (c' >= 0xE000 && c' <= 0xE0FF)
+     then True
+     else False
+
 -- | Convert Pandoc inline list to plain text identifier.
 inlineListToIdentifier :: [Inline] -> String
 inlineListToIdentifier [] = ""
 inlineListToIdentifier (x:xs) = 
   xAsText ++ inlineListToIdentifier xs
   where xAsText = case x of
-                       Str s        -> map toLower s
+                       Str s        -> filter (\c -> (c == '-') || not (isPunctuation c)) $
+                                       concat $ intersperse "-" $ words $ map toLower s
                        Emph lst     -> inlineListToIdentifier lst
                        Strong lst   -> inlineListToIdentifier lst
                        Quoted _ lst -> inlineListToIdentifier lst
                        Code s       -> s
                        Space        -> "-"
-                       EmDash       -> "--"
+                       EmDash       -> "-"
                        EnDash       -> "-"
                        Apostrophe   -> ""
-                       Ellipses     -> "..."
+                       Ellipses     -> ""
                        LineBreak    -> "-"
                        TeX _        -> ""
                        HtmlInline _ -> ""

tests/s5.basic.html

@@ -749,7 +749,7 @@ window.onresize = function(){setTimeout('fontScale()', 50);}</script>
 <div id="currentSlide"></div>
 <div id="header"></div>
 <div id="footer">
-<h1 id="july 15, 2006"
+<h1 id="july-15-2006"
     >July 15, 2006</h1
     ><h2 id="my-s5-document"
     >My S5 Document</h2
@@ -760,9 +760,9 @@ window.onresize = function(){setTimeout('fontScale()', 50);}</script>
 <div class="slide">
 <h1 id="my-s5-document-1"
     >My S5 Document</h1
-    ><h3 id="sam smith, jen jones"
+    ><h3 id="sam-smith-jen-jones"
     >Sam Smith, Jen Jones</h3
-    ><h4 id="july 15, 2006-1"
+    ><h4 id="july-15-2006-1"
     >July 15, 2006</h4
     ></div>
 <div class="slide">

tests/s5.fancy.html

@@ -1661,7 +1661,7 @@ else
 <div id="currentSlide"></div>
 <div id="header"></div>
 <div id="footer">
-<h1 id="july 15, 2006"
+<h1 id="july-15-2006"
     >July 15, 2006</h1
     ><h2 id="my-s5-document"
     >My S5 Document</h2
@@ -1672,9 +1672,9 @@ else
 <div class="slide">
 <h1 id="my-s5-document-1"
     >My S5 Document</h1
-    ><h3 id="sam smith, jen jones"
+    ><h3 id="sam-smith-jen-jones"
     >Sam Smith, Jen Jones</h3
-    ><h4 id="july 15, 2006-1"
+    ><h4 id="july-15-2006-1"
     >July 15, 2006</h4
     ></div>
 <div class="slide">

tests/writer.html

@@ -636,7 +636,7 @@ Blah
       >&lt;html&gt;</code
       >.</p
     ><hr
-     /><h1 id="smart-quotes,-ellipses,-dashes"
+     /><h1 id="smart-quotes-ellipses-dashes"
     >Smart quotes, ellipses, dashes</h1
     ><p
     >&ldquo;Hello,&rdquo; said the spider. &ldquo;&lsquo;Shelob&rsquo; is my name.&rdquo;</p