Fix for begin-but-no-end rules (xml)

babi/highlight.py

@@ -123,6 +123,10 @@ class Rule(NamedTuple):
         else:
             while_captures = ()
 
+        # some grammars (at least xml) have begin rules with no end
+        if begin is not None and end is None and while_ is None:
+            end = '$impossible^'
+
         # Using the captures key for a begin/end/while rule is short-hand for
         # giving both beginCaptures and endCaptures with same values
         if begin and end and captures:

bin/download-syntax

@@ -58,6 +58,8 @@ SYNTAXES = (
     # TODO: https://github.com/zargony/atom-language-rust/pull/149
     Syntax('rust', Ext.CSON, 'https://raw.githubusercontent.com/asottile/atom-language-rust/e113ca67/grammars/rust.cson'),  # noqa: E501
     Syntax('shell', Ext.CSON, 'https://raw.githubusercontent.com/atom/language-shellscript/7008ea926867d8a231003e78094091471c4fccf8/grammars/shell-unix-bash.cson'),  # noqa: E501
+    # TODO: https://github.com/atom/language-xml/pull/99
+    Syntax('xml', Ext.CSON, 'https://raw.githubusercontent.com/asottile/language-xml/2d76bc1f/grammars/xml.cson'),  # noqa: E501
     Syntax('yaml', Ext.PLIST, 'https://raw.githubusercontent.com/textmate/yaml.tmbundle/e54ceae3/Syntaxes/YAML.tmLanguage'),  # noqa: E501
 )
 

tests/highlight_test.py

@@ -534,3 +534,27 @@ def test_include_base():
         Region(2, 3, ('test', 'tick')),
         Region(3, 4, ('test',)),
     )
+
+
+def test_rule_with_begin_and_no_end():
+    compiler, state = _compiler_state({
+        'scopeName': 'test',
+        'patterns': [
+            {
+                'begin': '!', 'end': '!', 'name': 'bang',
+                'patterns': [{'begin': '--', 'name': 'invalid'}],
+            },
+        ],
+    })
+
+    state, regions = highlight_line(compiler, state, '!x! !--!', True)
+
+    assert regions == (
+        Region(0, 1, ('test', 'bang')),
+        Region(1, 2, ('test', 'bang')),
+        Region(2, 3, ('test', 'bang')),
+        Region(3, 4, ('test',)),
+        Region(4, 5, ('test', 'bang')),
+        Region(5, 7, ('test', 'bang', 'invalid')),
+        Region(7, 8, ('test', 'bang', 'invalid')),
+    )