Fix auto_identifiers_ascii extension.

Omit non-ASCII characters that don't have an (unaccented) ASCII equivalent. Add tests to test suite.
jgm · Jun 17, 2024 · a3baddd · a3baddd
1 parent b7f0ca3
commit a3baddd
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 6 deletions.
diff --git a/commonmark-extensions/src/Commonmark/Extensions/AutoIdentifiers.hs b/commonmark-extensions/src/Commonmark/Extensions/AutoIdentifiers.hs
@@ -82,9 +82,13 @@ makeIdentifier ascii = toIdent . T.toLower
                     = fromchar c
         | otherwise = mempty
     fromchar c
-      | ascii
-      , not (isAscii c) = maybe mempty T.singleton $ M.lookup c asciiMap
-      | otherwise       = T.singleton c
+      | ascii =
+        if isAscii c
+           then T.singleton c
+           else case M.lookup c asciiMap of
+                       Nothing -> mempty
+                       Just d -> T.singleton d
+      | otherwise = T.singleton c
 
 asciiMap :: M.Map Char Char
 asciiMap = M.fromList

diff --git a/commonmark-extensions/test/auto_identifiers_ascii.md b/commonmark-extensions/test/auto_identifiers_ascii.md
@@ -3,10 +3,17 @@
 The `auto_identifiers_ascii` extension is like
 `auto_identifiers` but limits identifiers to ASCII.
 Accented Latin characters are converted into the
-closest ASCII equivalent.
+closest ASCII equivalent. Other non-ASCII characters
+are simply omitted.
 
 ```````````````````````````````` example
-# Heading with &auml;
+# Heading with &auml; and &alpha;
 .
-<h1 id="heading-with-a">Heading with &auml;</h1>
+<h1 id="heading-with-a-and-">Heading with ä and α</h1>
+````````````````````````````````
+
+```````````````````````````````` example
+# Heading with ❤️
+.
+<h1 id="heading-with-heart">Heading with ❤️</h1>
 ````````````````````````````````
diff --git a/commonmark-extensions/test/test-commonmark-extensions.hs b/commonmark-extensions/test/test-commonmark-extensions.hs
@@ -45,6 +45,7 @@ main = do
              , ("test/bracketed_spans.md", bracketedSpanSpec)
              , ("test/fenced_divs.md", fencedDivSpec)
              , ("test/auto_identifiers.md", autoIdentifiersSpec <> attributesSpec)
+             , ("test/auto_identifiers_ascii.md", autoIdentifiersAsciiSpec <> attributesSpec)
              , ("test/implicit_heading_references.md",
                  autoIdentifiersSpec <> attributesSpec <> implicitHeadingReferencesSpec)
              , ("test/wikilinks_title_before_pipe.md", wikilinksSpec TitleBeforePipe)