fix(TAG_FULLMATCH): handle invalid /

5j9 · 5j9 · commit 5b84792b24cf · 2022-07-04T22:17:47.000+04:30
Also update changelog for the previous commit. closes #108
diff --git a/CHANGELOG.rst b/CHANGELOG.rst
@@ -1,3 +1,8 @@
+Unreleased
+----------
+- Fixed an issue in handling of / in tags. (#108)
+- Fixed a false-positive detection of invalid external links. (#109)
+
 v0.49.2
 -------
 - Fixed an issue in ``Template.normal_name()`` causing IndexError on empty/invalid template names, e.g. ``{{Template:}}``. (#105)
diff --git a/tests/test_tag.py b/tests/test_tag.py
@@ -76,14 +76,16 @@ def test_contents():
     assert t.contents == '\nc\n'
     t.contents = 'n'
     assert t.string == '<t>n</t>'
+
     t = Tag('<t></t>')
     assert t.contents == ''
     t.contents = 'n'
     assert t.string == '<t>n</t>'
+
     t = Tag('<t/>')
     assert t.contents == ''
     t.contents = 'n'
-    assert t.string == '<t>n</t>'
+    assert t.string == '<t/>n</t>'
 
 
 def test_get_attr_value():
@@ -165,3 +167,11 @@ def test_ref_with_invalid_attr():  # 47,48
     assert Tag('<ref name="a"3></ref>').attrs == {'name': 'a'}
     assert Tag('<ref name=""></ref>').attrs == {'name': ''}
     assert Tag('<ref "16/32"></ref>').attrs == {}
+
+
+def test_ref_tag_name():  # 108
+    # This is actually an invalid syntax on Mediawiki, but the same syntax
+    # is valid for `pre`, `noinclude`, and `includeonly`.
+    # I think it should be valid and the / should be treated as an invalid
+    # attribute and be ignored like in normal HTML tags.
+    assert Tag('<ref/ ></ref>').name == 'ref'
diff --git a/wikitextparser/_spans.py b/wikitextparser/_spans.py
@@ -149,9 +149,9 @@
 ATTRS_PATTERN = ( # noqa
     rb'(?<attr>'
         rb'[' + SPACE_CHARS + rb']++(?>' + ATTR_NAME + ATTR_VAL + rb')'
-        # Invalid attribute. Todo: could the / be removed? see
-        # https://stackoverflow.com/a/3558200/2705757
-        + rb'|(?>[^>/]++|/(?!\s*+>))++'
+        # See https://stackoverflow.com/a/3558200/2705757 for how HTML5
+        # treats self-closing marks.
+        + rb'|[^>]++'
     rb')*+(?<attr_insert>)')
 ATTRS_MATCH = regex_compile(
     # Leading space is not required at the start of the attribute string.
diff --git a/wikitextparser/_tag.py b/wikitextparser/_tag.py
@@ -32,10 +32,9 @@
     <(?<name>[A-Za-z0-9]++)''' + ATTRS_PATTERN + rb'''
     [''' + SPACE_CHARS + rb''']*+
     (?>
-        (?<self_closing>/\s*>)
-        |>(?<contents>.*)''' + END_TAG_PATTERN.replace(
+        >(?<contents>.*)''' + END_TAG_PATTERN.replace(
             rb'{name}', rb'(?<end_name>[A-Za-z0-9]++)') +  # noqa
-        rb'''|>  # only start; no end tag
+        rb'''|>  # only start; no end tag; could be self-closing
     )''', DOTALL | VERBOSE).fullmatch
 
 
@@ -182,8 +181,7 @@ def contents(self, contents: str) -> None:
             self[start:end] = contents
         else:
             # This is a self-closing tag.
-            s, e = match.span('self_closing')
-            self[s:e] = '>{0}</{1}>'.format(contents, match['name'].decode())
+            self[-1:] = f'>{contents}</{match["name"].decode()}>'
 
     @property
     def parsed_contents(self) -> SubWikiText:
