named groups

Author: andrii1812

tests/verbal_expressions_test.py

@@ -135,3 +135,11 @@ def test_should_match_email_address(self):
     def test_should_match_url(self):
         self.exp = self.v.start_of_line().then('http').maybe('s').then('://').maybe('www.').word().then('.').word().maybe('/').end_of_line().regex()
         self.assertRegexpMatches('https://www.google.com/', self.exp, 'Not a valid email')
+        
+    def test_should_find_named_groups(self):
+        name = "Linus Torvalds"
+        self.exp = self.v.start_of_line().word(name='first_name').then(' ').word(name='last_name').end_of_line().regex()
+        match = self.exp.match(name)
+        self.assertIsNotNone(match)
+        self.assertTrue(match.group('first_name') == 'Linus')
+        self.assertTrue(match.group('last_name') == 'Torvalds')

verbalexpressions/verbal_expressions.py

@@ -2,11 +2,16 @@
 
 
 def re_escape(fn):
-    def arg_escaped(this, *args):
+    def arg_escaped(this, *args, **kwargs):
         t = [isinstance(a, VerEx) and a.s or re.escape(str(a)) for a in args]
-        return fn(this, *t)
+        return fn(this, *t, **kwargs)
     return arg_escaped
 
+    
+def group(val, name=None):
+    prefix = '?P<{0}>'.format(name) if name else ''
+    return '(' + prefix + val + ')'
+    
 
 class VerEx(object):
     '''
@@ -50,49 +55,49 @@ def source(self):
 
     # ---------------------------------------------
 
-    def anything(self):
-        return self.add('(.*)')
+    def anything(self, name=None):
+        return self.add(group('.*', name))
 
     @re_escape
     def anything_but(self, value):
-        return self.add('([^' + value + ']*)')
+        return self.add(group('[^' + value + ']*'))
 
     def end_of_line(self):
         return self.add('$')
 
     @re_escape
     def maybe(self, value):
-        return self.add("(" + value + ")?")
+        return self.add(group(value) + "?")
 
     def start_of_line(self):
         return self.add('^')
 
     @re_escape
-    def find(self, value):
-        return self.add('(' + value + ')')
+    def find(self, value, name=None):        
+        return self.add(group(value, name))
     then = find
 
     # special characters and groups
 
     @re_escape
     def any(self, value):
-        return self.add("([" + value + "])")
+        return self.add(group("[" + value + "]"))
     any_of = any
 
     def line_break(self):
-        return self.add(r"(\n|(\r\n))")
+        return self.add(group(r"\n|(\r\n)"))
     br = line_break
 
     @re_escape
     def range(self, *args):
         from_tos = [args[i:i+2] for i in range(0, len(args), 2)]
-        return self.add("([" + ''.join(['-'.join(i) for i in from_tos]) + "])")
+        return self.add(group("[" + ''.join(['-'.join(i) for i in from_tos]) + "]"))
 
     def tab(self):
         return self.add(r'\t')
 
-    def word(self):
-        return self.add(r"(\w+)")
+    def word(self, name=None):
+        return self.add(group(r"\w+", name))
 
     def OR(self, value=None):
         ''' `or` is a python keyword so we use `OR` instead. '''