Subclass the standard Groovy lexer

elevans · elevans · commit bf3e39e98dbe · 2024-03-27T11:23:52.000-05:00
Instead of duplicating the Groovy lexer to
modify it, we can simply subclass it
and inherit it's tokens.

This commit also renames the lexer to
"SciJavaGroovyLexer".
diff --git a/docs/lexers.py b/docs/lexers.py
@@ -1,82 +1,11 @@
-import re
-from pygments.lexer import RegexLexer, bygroups, using, this, default
-from pygments.token import Comment, Operator, Keyword, Name, String, Number, Whitespace
-from pygments.util import shebang_matches
-
-class FijiGroovyLexer(RegexLexer):
-    """
-    For Groovy source code.
-
-    .. versionadded:: 1.5
-    """
-
-    name = 'Groovy'
-    url = 'https://groovy-lang.org/'
-    aliases = ['fijigroovy']
-    filenames = ['*.groovy','*.gradle']
-    mimetypes = ['text/x-groovy']
-
-    flags = re.MULTILINE | re.DOTALL
+from pygments.lexer import inherit
+from pygments.lexers.jvm import GroovyLexer
+from pygments.token import Comment
 
+class SciJavaGroovyLexer(GroovyLexer):
     tokens = {
-        'root': [
-            # Groovy allows a file to start with a shebang
-            (r'#!(.*?)$', Comment.Preproc, 'base'),
-            default('base'),
-        ],
         'base': [
-            (r'[^\S\n]+', Whitespace),
             (r'#@.*?$', Comment.Single),
-            (r'(//.*?)(\n)', bygroups(Comment.Single, Whitespace)),
-            (r'/\*.*?\*/', Comment.Multiline),
-            # keywords: go before method names to avoid lexing "throw new XYZ"
-            # as a method signature
-            (r'(assert|break|case|catch|continue|default|do|else|finally|for|'
-             r'if|goto|instanceof|new|return|switch|this|throw|try|while|in|as)\b',
-             Keyword),
-            # method names
-            (r'^(\s*(?:[a-zA-Z_][\w.\[\]]*\s+)+?)'  # return arguments
-             r'('
-             r'[a-zA-Z_]\w*'                        # method name
-             r'|"(?:\\\\|\\[^\\]|[^"\\])*"'         # or double-quoted method name
-             r"|'(?:\\\\|\\[^\\]|[^'\\])*'"         # or single-quoted method name
-             r')'
-             r'(\s*)(\()',                          # signature start
-             bygroups(using(this), Name.Function, Whitespace, Operator)),
-            (r'@[a-zA-Z_][\w.]*', Name.Decorator),
-            (r'(abstract|const|enum|extends|final|implements|native|private|'
-             r'protected|public|static|strictfp|super|synchronized|throws|'
-             r'transient|volatile)\b', Keyword.Declaration),
-            (r'(def|boolean|byte|char|double|float|int|long|short|void)\b',
-             Keyword.Type),
-            (r'(package)(\s+)', bygroups(Keyword.Namespace, Whitespace)),
-            (r'(true|false|null)\b', Keyword.Constant),
-            (r'(class|interface)(\s+)', bygroups(Keyword.Declaration, Whitespace),
-             'class'),
-            (r'(import)(\s+)', bygroups(Keyword.Namespace, Whitespace), 'import'),
-            (r'""".*?"""', String.Double),
-            (r"'''.*?'''", String.Single),
-            (r'"(\\\\|\\[^\\]|[^"\\])*"', String.Double),
-            (r"'(\\\\|\\[^\\]|[^'\\])*'", String.Single),
-            (r'\$/((?!/\$).)*/\$', String),
-            (r'/(\\\\|\\[^\\]|[^/\\])*/', String),
-            (r"'\\.'|'[^\\]'|'\\u[0-9a-fA-F]{4}'", String.Char),
-            (r'(\.)([a-zA-Z_]\w*)', bygroups(Operator, Name.Attribute)),
-            (r'[a-zA-Z_]\w*:', Name.Label),
-            (r'[a-zA-Z_$]\w*', Name),
-            (r'[~^*!%&\[\](){}<>|+=:;,./?-]', Operator),
-            (r'[0-9][0-9]*\.[0-9]+([eE][0-9]+)?[fd]?', Number.Float),
-            (r'0x[0-9a-fA-F]+', Number.Hex),
-            (r'[0-9]+L?', Number.Integer),
-            (r'\n', Whitespace)
-        ],
-        'class': [
-            (r'[a-zA-Z_]\w*', Name.Class, '#pop')
-        ],
-        'import': [
-            (r'[\w.]+\*?', Name.Namespace, '#pop')
-        ],
-    }
-
-    def analyse_text(text):
-        return shebang_matches(text, r'groovy')
+            inherit,
+        ]
+    }
