Fix: regex fails to capture certain langs (#30)

Qrbaker · web-flow · commit 2cc9a4d80d56 · 2022-07-14T07:55:13.000+02:00
Languages with non alphanumeric characters (i.e. c++)
would not be captured by the regex for syntax highlighting.
diff --git a/wikitexthtml/render/preprocess.py b/wikitexthtml/render/preprocess.py
@@ -18,8 +18,9 @@
 ONLYINCLUDE_PATTERN = regex.compile(r"<onlyinclude>([\s\S]*?)(?></onlyinclude>|\Z)")
 PRE_PATTERN = regex.compile(r"<pre>([\s\S]*?)(?></pre>|\Z)")
 SYNTAXHIGHLIGHT_PATTERN = regex.compile(
-    r"<syntaxhighlight( lang=\")?(?P<lang>[a-zA-z]+)?\"?"
-    r"(?P<line> line(?P<lineStart> \d*)?)?>(?P<code>[\s\S]*?)(?></syntaxhighlight>|\Z)"
+    r"<syntaxhighlight( lang=\"(?P<lang>[^\"]+)\")?"
+    r"(?P<line> line( start=\"(?P<lineStart>\d+)\")?)?>"
+    r"(?P<code>[\s\S]*?)(</syntaxhighlight>|\Z)"
 )
 
 
