✨ Add the negation [^...] in PythonRegex

pyformlang/regular_expression/python_regex.py

@@ -23,7 +23,8 @@
     "$": "\\$",
     "\n": "",
     " ": "\\ ",
-    '\\': '\\\\'
+    '\\': '\\\\',
+    "?": "\\?"
 }
 
 RECOMBINE = {
@@ -218,13 +219,14 @@ def _preprocess_brackets_content(self, bracket_content):
         bracket_content_temp = []
         previous_is_valid_for_range = False
         for i, symbol in enumerate(bracket_content):
-            if (symbol == "-" and not self._should_escape_next_symbol(
-                    bracket_content_temp)):
-                if (not previous_is_valid_for_range
-                        or i == len(bracket_content) - 1):
+            # We have a range
+            if symbol == "-" and not self._should_escape_next_symbol(bracket_content_temp):
+                if not previous_is_valid_for_range or i == len(bracket_content) - 1:
+                    # False alarm, no range
                     bracket_content_temp.append("-")
                     previous_is_valid_for_range = True
                 else:
+                    # We insert all the characters in the range
                     bracket_content[i - 1] = self._recombine(bracket_content[i - 1])
                     for j in range(ord(bracket_content[i - 1][-1]) + 1,
                                    ord(bracket_content[i + 1][-1])):
@@ -244,10 +246,18 @@ def _preprocess_brackets_content(self, bracket_content):
                     previous_is_valid_for_range = False
                 else:
                     previous_is_valid_for_range = True
+        bracket_content_temp = self._preprocess_negation(bracket_content_temp)
         bracket_content_temp = self._insert_or(bracket_content_temp)
         bracket_content_temp = self._recombine(bracket_content_temp)
         return bracket_content_temp
 
+    @staticmethod
+    def _preprocess_negation(bracket_content):
+        if not bracket_content or bracket_content[0] != "^":
+            return bracket_content
+        # We inverse everything
+        return [x for x in ESCAPED_PRINTABLES if x not in bracket_content]
+
     @staticmethod
     def _insert_or(l_to_modify):
         res = []

pyformlang/regular_expression/tests/test_python_regex.py

@@ -313,3 +313,18 @@ def test_error_backslash(self):
         self._test_compare(r"[a\\\\\\]]", "\\]")
         self._test_compare(r"\"([d\"\\\\]|\\\\.)*\"", '"d\\"')
         self._test_compare(r"[a\\\\]", "a")
+
+    def test_negation_brackets(self):
+        self._test_compare(r"[^abc]*", "")
+        self._test_compare(r"[^abc]*", "a")
+        self._test_compare(r"[^abc]*", "b")
+        self._test_compare(r"[^abc]*", "c")
+        self._test_compare(r"[^abc]*", "d")
+        self._test_compare(r"[^abc]*", "dga")
+        self._test_compare(r"[^abc]*", "dgh")
+        self._test_compare(r"[^?]*", "dgh")
+
+    def test_question_mark(self):
+        self._test_compare(r".", "?")
+        self._test_compare(r"a(a|b)?", "a")
+        self._test_compare(r"a(a|b)\?", "ab?")

setup.py

@@ -11,7 +11,7 @@
 
 setuptools.setup(
     name='pyformlang',
-    version='1.0.8',
+    version='1.0.9',
     #scripts=['pyformlang'] ,
     author="Julien Romero",
     author_email="[email protected]",