Don't allow unescaped characters

openformats/formats/android_unescaped.py

@@ -10,22 +10,25 @@
 class AndroidUnescapedHandler(AndroidHandler):
     def _create_string(self, name, text, comment, product, child, pluralized=False):
         """Creates a string and returns it. If empty string it returns None.
+        Also checks if the provided text contains unescaped characters which are
+        invalid for the Android XML format.
 
         :param text: The strings text.
         :param name: The name of the string.
         :param comment: The developer's comment the string might have.
         :param product: Extra context for the string.
-        :param child: The child tag that the string is created from. Used to find
-                    line numbers when errors occur.
-        :returns: Returns an OpenString object if the text is not empty else None.
+        :param child: The child tag that the string is created from.
+                        Used to find line numbers when errors occur.
+        :returns: Returns an OpenString object if the text is not empty
+                    else None.
         """
+        AndroidUnescapedHandler._check_unescaped_characters(text)
         if XMLUtils.validate_not_empty_string(
             self.transcriber,
             text,
             child,
             error_context={"main_tag": "plural", "child_tag": "item"},
         ):
-            text = self._escape_quotes(text)
             if (name, product) in self.existing_hashes:
                 if child.tag in self.existing_hashes[(name, product)]:
                     format_dict = {"name": name, "child_tag": child.tag}
@@ -60,58 +63,64 @@ def _create_string(self, name, text, comment, product, child, pluralized=False):
             return string
         return None
 
-    def _escape_quotes(self, text):
-        """Allow single and double quotes to be uploaded unescaped but they must be
-        returned escaped
+    @staticmethod
+    def _check_unescaped_characters(text):
+        """Checks if the provided text contains unescaped characters which are
+        invalid for the Android XML format.
         """
         if type(text) == dict:
-            text = AndroidUnescapedHandler._escape_quotes_plural_string(text)
+            text = AndroidUnescapedHandler._check_unescaped_characters_in_plural_string(
+                text
+            )
         else:
-            text = AndroidUnescapedHandler._escape_quotes_simple_string(text)
+            text = AndroidUnescapedHandler._check_unescaped_characters_in_simple_string(
+                text
+            )
 
         return text
 
     @staticmethod
-    def _escape_quotes_simple_string(text):
-        text, protected_tags = AndroidUnescapedHandler._protect_inline_tags(text)
+    def _check_unescaped_characters_in_simple_string(text):
+        try:
+            protected_string, _ = AndroidUnescapedHandler._protect_inline_tags(text)
+        except Exception as e:
+            raise ParseError(
+                "Error escaping the string. Please check for any open tags or any "
+                "dangling < characters"
+            ) from e
 
-        text = re.sub(
+        not_allowed_unescaped = [
             r"(?<!\\)'",
-            "".join([DumbXml.BACKSLASH, DumbXml.SINGLE_QUOTE]),
-            text,
-        )
-        text = re.sub(
             r'(?<!\\)"',
-            "".join([DumbXml.BACKSLASH, DumbXml.DOUBLE_QUOTES]),
-            text,
-        )
-
-        text = AndroidUnescapedHandler._unprotect_inline_tags(text, protected_tags)
-
-        return text
+            r"(?<!\\)@",
+            r"(?<!\\)\?",
+            r"(?<!\\)\n",
+            r"(?<!\\)\t",
+        ]
+
+        for pattern in not_allowed_unescaped:
+            if re.search(pattern, protected_string):
+                raise ParseError(
+                    "You have one or more unescaped characters from the following "
+                    f"list ', \", @, ?, \\n, \\t in the string : {text}"
+                )
 
     @staticmethod
-    def _escape_quotes_plural_string(text):
-        escaped_dict = {}
-        for key, string in text.items():
-            escaped_string = AndroidUnescapedHandler._escape_quotes_simple_string(
-                string
-            )
-            escaped_dict[key] = escaped_string
-
-        return escaped_dict
+    def _check_unescaped_characters_in_plural_string(text):
+        for _, string in text.items():
+            AndroidUnescapedHandler._check_unescaped_characters_in_simple_string(string)
 
     @staticmethod
     def _protect_inline_tags(text):
-        """Protect INLINE_TAGS from escaping single and double quotes"""
+        """Protect INLINE_TAGS from escaping special characters"""
         protected_tags = {}
         wrapped_text = f"<x>{text}</x>"
         parsed = DumbXml(wrapped_text)
         children_iterator = parsed.find_children()
 
         for child in children_iterator:
             if child.tag in AndroidHandler.INLINE_TAGS:
-                child_content = child.source[child.start : child.end]
+                child_content = child.source[child.position : child.tail_position]
                 string_hash = md5(child_content.encode("utf-8")).hexdigest()
                 text = text.replace(child_content, string_hash)
                 protected_tags[string_hash] = child_content
@@ -131,11 +140,8 @@ def escape(string):
             string, protected_tags = AndroidUnescapedHandler._protect_inline_tags(
                 string
             )
-        except Exception as e:
-            raise ParseError(
-                "Error escaping the string. Please check for any open tags or any "
-                "dangling < characters"
-            )
+        except Exception as _:
+            return AndroidHandler.escape(string)
 
         string = AndroidHandler.escape(string)
         string = (