chore(START_TAG_PATTERN): remove self_closing group

5j9 · 5j9 · commit 611f97be2f8a · 2022-07-04T22:52:40.000+04:30
diff --git a/tests/test_tag.py b/tests/test_tag.py
@@ -21,30 +21,23 @@ def test_start_tag_patterns():
     assert start_tag_finder(b'<b>').groupdict() == {
         'name': b'b', 'attr': None, 'quote': None,
         'start_tag': b'<b>', 'attr_name': None,
-        'self_closing': None, 'attr_value': None, 'attr_insert': b'',}
+        'attr_value': None, 'attr_insert': b'',}
     assert start_tag_finder(b'<b t>').groupdict() == {
         'name': b'b', 'attr': b' t', 'quote': None,
         'start_tag': b'<b t>', 'attr_name': b't', 'attr_value': b'',
-        'self_closing': None, 'attr_insert': b'',}
+        'attr_insert': b'',}
     assert start_tag_finder(b'<div value=yes>').groupdict() == {
         'name': b'div', 'attr': b' value=yes', 'quote': None,
         'start_tag': b'<div value=yes>', 'attr_name': b'value',
-        'attr_value': b'yes', 'self_closing': None, 'attr_insert': b'',}
+        'attr_value': b'yes', 'attr_insert': b'',}
     assert start_tag_finder(b"<div class='body'>").groupdict() == {
         'name': b'div', 'attr': b" class='body'", 'quote': b"'",
         'start_tag': b"<div class='body'>", 'attr_name': b'class',
-        'attr_value': b'body', 'self_closing': None, 'attr_insert': b'',}
-    # This is not standard HTML5, but could be useful to have.
-    # ae(
-    #     START_TAG_MATCH('<s style=>').groupdict(),
-    #     {'name': 's', 'attr': 'style=', 'quote': None,
-    #      'start': '<s style=>', 'attr_name': 'style', 'attr_value': ''
-    #      'self_closing': None}
-    # )
+        'attr_value': b'body', 'attr_insert': b'',}
     assert start_tag_finder(b"<table a1=v1 a2=v2>").capturesdict() == {
         'attr_name': [b'a1', b'a2'], 'start_tag': [b'<table a1=v1 a2=v2>'],
         'attr': [b' a1=v1', b' a2=v2'], 'quote': [],
-        'attr_value': [b'v1', b'v2'], 'self_closing': [],
+        'attr_value': [b'v1', b'v2'],
         'name': [b'table'],  'attr_insert': [b'']}
 
 
diff --git a/wikitextparser/_spans.py b/wikitextparser/_spans.py
@@ -173,8 +173,7 @@
 START_TAG_PATTERN = ( # noqa
     rb'(?<start_tag>'
         rb'<{name}' + ATTRS_PATTERN +
-        rb'[' + SPACE_CHARS + rb']*+'
-        rb'(?:(?<self_closing>/[' + SPACE_CHARS + b']*+>)|>)'
+        rb'[' + SPACE_CHARS + rb']*+>'
     rb')')
 HTML_START_TAG_FINDITER = regex_compile(
     START_TAG_PATTERN.replace(b'{name}', _HTML_TAG_NAME, 1)).finditer
diff --git a/wikitextparser/_wikitext.py b/wikitextparser/_wikitext.py
@@ -606,7 +606,7 @@ def remove(b: int, e: int):
         if callable(replace_parser_functions):
             for pf in self.parser_functions:
                 b, e = pf._span_data[:2]
-                if lst[b] is None: # overwrriten
+                if lst[b] is None:  # already overwritten
                     continue
                 lst[b] = replace_parser_functions(pf)
                 remove(b + 1, e)
@@ -1350,8 +1350,8 @@ def get_tags(self, name=None) -> List['Tag']:
         span_tuple_to_span_get = {(s[0], s[1]): s for s in spans}.get
         spans_append = spans.append
         for start_match in reversed_start_matches:
-            if start_match['self_closing']:
-                # Don't look for the end tag
+            if start_match[0].rstrip(b' \t\n>')[-1] == 47:  # ord('/') == 47
+                # Self-closing tag. Don't look for the end tag.
                 ms, me = start_match.span()
                 span = [ss + ms, ss + me, None, shadow_copy[ms:me]]
             else:
