Changeset 118

Timestamp:

04/11/08 17:06:18 (4 years ago)

Author:

phil

Message:

pygments SPARQL Lexer addition

Files:

• tools/trunk/pygments/lexers/sw.py (modified) (2 diffs)

tools/trunk/pygments/lexers/sw.py

@@ -71 +71 @@
 class SparqlLexer(RegexLexer):
     """
-    Lexer for SPARQL This is a stubb It don't work yet
-    I haven't started on it.  
+    Lexer for SPARQL Not Complete
     """
     name = 'SPARQL'
@@ -80 +79 @@
     flags = re.IGNORECASE
     tokens = {
+        'comments': [
+            (r'(\s*#.*)', Comment)
+        ],
         'root': [
+            include('comments'),        
             (r'(\s*PREFIX\s)(\s*\w*:\s+)(<[^> ]*>\s*)',bygroups(Keyword,Name.Variable,Name.Namespace)),
             (r'(\s*#.*)', Comment),
-            (r'\s+', Text),
-            (r'--.*?\n', Comment.Single),
-            (r'/\*', Comment.Multiline, 'multiline-comments'),
-            (r'(SELECT|FILTER|LIMIT|WHERE|'
-             r'WITH|WITHOUT|WORK|WRITE|YEAR|ZONE)\b', Keyword),
-            (r'(ARRAY|BIGINT|BINARY|BIT|BLOB|BOOLEAN|CHAR|CHARACTER|DATE|'
-             r'DEC|DECIMAL|FLOAT|INT|INTEGER|INTERVAL|NUMBER|NUMERIC|REAL|'
-             r'SERIAL|SMALLINT|VARCHAR|VARYING|INT8|SERIAL8|TEXT)\b',
-             Name.Builtin),
-            (r'[+*/<>=~!@#%^&|`?^-]', Operator),
-            (r'[0-9]+', Number.Integer),
-            (r"'(''|[^'])*'", String.Single),
-            (r'"(""|[^"])*"', String.Symbol), 
-            (r"`(``|[^`])*`", String.Backtick),
-            (r'[a-zA-Z_][a-zA-Z0-9_]*', Name),
-            (r'[;:()\[\],\.]', Punctuation)
+            (r'((?:SELECT|ASK|CONSTRUCT)\s*(?:DISTINCT|REDUCED)?\s*)((?:\?[a-zA-Z0-9_-]+\s*)+|\*)(\s*)',bygroups(Keyword,Name.Variable,Text)),
+            (r'(FROM\s*(?:NAMED)?)(\s*.*)', bygroups(Keyword,Text)),
+            (r'(WHERE)?\s*({)',bygroups(Keyword,Text),'graph'),
+            (r'(LIMIT|OFFSET)(\s*[+-]?[0-9]+)',bygroups(Keyword,Literal.String)),
         ],
-        'multiline-comments': [
-            (r'/\*', Comment.Multiline, 'multiline-comments'),
-            (r'\*/', Comment.Multiline, '#pop'),
-            (r'[^/\*]+', Comment.Multiline),
-            (r'[/*]', Comment.Multiline)
-        ]
+        'graph':[
+            (r'\s*(<[^>]*\>)', Name.Class, ('triple','predObj')),
+            (r'(\s*[a-zA-Z_0-9\-]*:[a-zA-Z0-9\-_]*\s)', Name.Class, ('triple','predObj')),
+            (r'(\s*\?[a-zA-Z0-9_-]*)', Name.Variable, ('triple','predObj')),            
+            (r'\s*\[\]\s*', Name.Class, ('triple','predObj')),
+            (r'\s*(FILTER\s*)((?:regex)?\()',bygroups(Keyword,Text),'filterExp'),
+            (r'\s*}', Text, '#pop'),
+        ],
+        'triple' : [
+            (r'(?=\s*})', Text, '#pop'),                    
+            (r'\s*\.\s*', Text, '#pop'),
+        ],
+        'predObj': [
+            include('comments'),
+            (r'(\s*\?[a-zA-Z0-9_-]*\b\s*)', Name.Variable,'object'),            
+            (r'(\s*[a-zA-Z_:][a-zA-Z0-9\-_:]*\b\s*)', Operator, 'object'),
+            (r'\s*(<[^>]*\>)', Operator, 'object'),
+            (r'\s*\]\s*', Text, '#pop'),
+            (r'(?=\s*\.\s*)', Keyword, '#pop'), 
+        ],
+        'objList': [
+            (r'\s*\)', Text, '#pop'),
+            include('object'),
+        ],
+        'object': [
+            include('variable'),
+            (r'\s*\[', Text, 'predObj'),
+            (r'\s*<[^> ]*>', Name.Attribute),
+            (r'\s*("""(?:.|\n)*?""")(\@[a-z]{2-4}|\^\^<?[a-zA-Z0-9\-\:_#/\.]*>?)?\s*', bygroups(Literal.String,Text)),
+            (r'\s*".*?[^\\]"(?:\@[a-z]{2-4}|\^\^<?[a-zA-Z0-9\-\:_#/\.]*>?)?\s*', Literal.String),
+            (r'\s*[a-zA-Z0-9\-_\:]\s*', Name.Attribute),
+            (r'\s*\(', Text, 'objList'),
+            (r'\s*;\s*', Text, '#pop'),
+            (r'(?=\])', Text, '#pop'),            
+            (r'(?=\.)', Text, '#pop'),           
+        ],
+        'variable':[
+            (r'(\?[a-zA-Z0-9\-_]+\s*)', Name.Variable),            
+        ],
+        'filterExp':[
+            include('variable'),
+            include('object'),
+            (r'[+*/<>=~!%&|-]+', Operator),
+            (r'\s*\)', Text, '#pop'),            
+        ],
+
     }