Resolve issue #19

pyformlang/regular_expression/python_regex.py

@@ -111,10 +111,14 @@ def _preprocess_brackets(self):
         in_brackets = 0
         in_brackets_temp = []
         for symbol in self._python_regex:
-            if symbol == "[" and (not regex_temp or regex_temp[-1] != "\\"):
+            if symbol == "[" and (not regex_temp or regex_temp[-1] != "\\") and \
+                    (in_brackets == 0 or not in_brackets_temp[-1] or in_brackets_temp[-1][-1] != "\\"):
                 in_brackets += 1
                 in_brackets_temp.append([])
-            elif symbol == "]" and (not regex_temp or regex_temp[-1] != "\\"):
+            elif symbol == "]" and (not regex_temp or regex_temp[-1] != "\\") and \
+                    (in_brackets == 0 or not in_brackets_temp[-1] or
+                     (in_brackets_temp[-1][-1] != "\\" or
+                      (len(in_brackets_temp[-1]) > 1 and in_brackets_temp[-1][-2] == "\\"))):
                 if len(in_brackets_temp) == 1:
                     regex_temp.append("(")
                     regex_temp += self._preprocess_brackets_content(
@@ -151,7 +155,11 @@ def _preprocess_brackets_content(self, bracket_content):
                 else:
                     for j in range(ord(bracket_content[i - 1]) + 1,
                                    ord(bracket_content[i + 1])):
-                        bracket_content_temp.append(chr(j))
+                        next_char = chr(j)
+                        if next_char in TRANSFORMATIONS:
+                            bracket_content_temp.append(TRANSFORMATIONS[next_char])
+                        else:
+                            bracket_content_temp.append(next_char)
                     previous_is_valid_for_range = False
             else:
                 if self._should_escape_next_symbol(bracket_content_temp):

pyformlang/regular_expression/tests/test_python_regex.py

@@ -122,12 +122,12 @@ def test_escape_backslash_opening_bracket(self):
         regex = PythonRegex(r"a\\[ab]")
         self.assertTrue(regex.accepts(["a", "\\", "a"]))
         self.assertTrue(regex.accepts(["a", "\\", "b"]))
+        self._test_compare(r"a\\[ab]", "a\\a")
 
     def test_escape_backslash_closing_bracket(self):
-        regex = PythonRegex(r"a[ab\\]")
-        self.assertTrue(regex.accepts(["a", "a"]))
-        self.assertTrue(regex.accepts(["a", "b"]))
-        self.assertTrue(regex.accepts(["a", "\\"]))
+        self._test_compare(r"a[ab\\]", "aa")
+        self._test_compare(r"a[ab\\]", "ab")
+        self._test_compare(r"a[ab\\]", "a\\")
 
     def test_escape_backslash_question_mark(self):
         regex = PythonRegex(r"a\\?")
@@ -196,13 +196,33 @@ def test_shortcut_word(self):
         self.assertTrue(regex.accepts(["a", "A"]))
         self.assertTrue(regex.accepts(["a", "f"]))
 
+    def _test_compare(self, regex, s_test):
+        r_pyformlang = PythonRegex(regex)
+        r_python = re.compile(regex)
+        self.assertEqual(r_pyformlang.accepts(s_test), r_python.match(s_test) is not None)
+
     def test_backslash(self):
-        r_all = PythonRegex('.*')
-        r_python = re.compile(".*")
-        self.assertEqual(r_all.accepts(']'), r_python.match("]") is not None)
-        self.assertEqual(r_all.accepts('\\'), r_python.match("\\") is not None)
+        self._test_compare(".*", "]")
+        self._test_compare(".*", "\\")
 
     def test_escape_dot(self):
-        r_pyformlang = PythonRegex('\\.')
-        r_re = re.compile("\\.")
-        self.assertEqual(r_pyformlang.accepts("."), r_re.match(".") is not None)
+        self._test_compare("\\.", ".")
+
+    def test_brackets(self):
+        self._test_compare(r"[{}]", "{")
+        self._test_compare(r"[{-}]", "{")
+        self._test_compare(r"[{-}]", "-")
+        self._test_compare(r"[{-}]", "}")
+
+    def test_brackets_escape(self):
+        self._test_compare(r"[\[]", "[")
+        self._test_compare(r"[Z-\[]", "Z")
+        self._test_compare(r"[Z-\[]", "[")
+        self._test_compare(r"[\[-a]", "[")
+        self._test_compare(r"[\[-a]", "a")
+        self._test_compare(r"[\[-\]]", "[")
+        self._test_compare(r"[\[-\]]", "]")
+        self._test_compare(r"[Z-\]]", "Z")
+        self._test_compare(r"[Z-\]]", "]")
+        self._test_compare(r"[\]-a]", "]")
+        self._test_compare(r"[\]-a]", "a")

setup.py

@@ -11,7 +11,7 @@
 
 setuptools.setup(
     name='pyformlang',
-    version='1.0.4',
+    version='1.0.5',
     #scripts=['pyformlang'] ,
     author="Julien Romero",
     author_email="[email protected]",