feat: comments, boolean and delimiters regex

Author: danielogen

src/PyReprism/languages/cpp.py

@@ -28,7 +28,7 @@ def keywords() -> list:
         return keyword
 
     @staticmethod
-    def comment_regex():
+    def comment_regex() -> re.Pattern:
         """
         Compile and return a regular expression pattern to identify different types of comments and non-comment code in C source files.
 
@@ -39,7 +39,7 @@ def comment_regex():
         return pattern
 
     @staticmethod
-    def number_regex():
+    def number_regex() -> re.Pattern:
         """
         Compile and return a regular expression pattern to identify numeric literals in C++ code.
 
@@ -50,7 +50,7 @@ def number_regex():
         return pattern
 
     @staticmethod
-    def operator_regex():
+    def operator_regex() -> re.Pattern:
         """
         Compile and return a regular expression pattern to identify C++ operators.
 
@@ -61,7 +61,7 @@ def operator_regex():
         return pattern
 
     @staticmethod
-    def keywords_regex():
+    def keywords_regex() -> re.Pattern:
         """
         Return a list of C++ keywords and built-in functions.
 
@@ -102,7 +102,7 @@ def remove_comments(source_code: str) -> str:
         return CPP.comment_regex().sub(lambda match: match.group('noncomment') if match.group('noncomment') else '', source_code).strip()
 
     @staticmethod
-    def remove_keywords(source: str):
+    def remove_keywords(source: str) -> str:
         """
         Remove all C++ keywords from the provided source code string.
 

src/PyReprism/languages/python.py

@@ -3,6 +3,9 @@
 
 
 class Python:
+    """
+    This is the class for processing Python source code
+    """
     def __init__(self):
         pass
 

src/PyReprism/languages/scala.py

@@ -16,26 +16,58 @@ def keywords() -> list:
         return keyword
 
     @staticmethod
-    def comment_regex():
+    def comment_regex() -> re.Pattern:
         pattern = re.compile(r'(?P<comment>//.*?$|/\*.*?\*/|/\*.*?$|^.*?\*/|[{}]+)|(?P<noncomment>\'(\\.|[^\\\'])*\'|"(\\.|[^\\"])*"|.[^/\'"{}]*)', re.DOTALL | re.MULTILINE)
         return pattern
 
     @staticmethod
-    def number_regex():
+    def number_regex() -> re.Pattern:
         pattern = re.compile(r'\b0x[\da-f]*\.?[\da-f]+|(?:\b\d+\.?\d*|\B\.\d+)(?:e\d+)?[dfl]?')
         return pattern
 
     @staticmethod
-    def operator_regex():
+    def operator_regex() -> re.Pattern:
         pattern = re.compile(r'(^|[^.])(?:\+[+=]?|-[-=]?|!=?|<<?=?|>>?>?=?|==?|&[&=]?|\|[|=]?|\*=?|\/=?|%=?|\^=?|[?:~])')
         return pattern
 
     @staticmethod
-    def keywords_regex():
+    def keywords_regex() -> re.Pattern:
         return re.compile(r'\b(' + '|'.join(Scala.keywords()) + r')\b')
 
+    @staticmethod
+    def boolean_regex() -> re.Pattern:
+        """
+        Compile and return a regular expression pattern to identify Scala boolean literals.
+
+        This function generates a regular expression that matches the Scala boolean literals `true`, `false`, and the special constant `null`.
+
+        :return: A compiled regex pattern to match Scala boolean literals and `null`.
+        :rtype: re.Pattern
+        """
+        return re.compile(r'\b(?:true|false|null)\b')
+
+    @staticmethod
+    def delimiters_regex() -> re.Pattern:
+        """
+        Compile and return a regular expression pattern to identify Scala language delimiters.
+
+        This function generates a regular expression that matches Scala language delimiters, which include parentheses `()`, brackets `[]`, braces `{}`, commas `,`, colons `:`, periods `.`, semicolons `;`, angle brackets `<`, `>`, the question mark `?`, and the underscore `_`.
+
+        :return: A compiled regex pattern to match Scala delimiters.
+        :rtype: re.Pattern
+        """
+        return re.compile(r'[()\[\]{}.,:;<>?_]')
+
     @staticmethod
     def remove_comments(source_code: str, isList: bool = False) -> str:
+        """
+        Remove comments from the provided Java source code string.
+
+        :param str source_code: The Java source code from which to remove comments.
+        :return: The source code with all comments removed.
+        :rtype: str
+        """
+        return Scala.comment_regex().sub(lambda match: match.group('noncomment') if match.group('noncomment') else '', source_code).strip()
         result = []
         for match in Scala.comment_regex().finditer(source_code):
             if match.group('noncomment'):
@@ -45,5 +77,5 @@ def remove_comments(source_code: str, isList: bool = False) -> str:
         return ''.join(result)
 
     @staticmethod
-    def remove_keywords(source: str):
+    def remove_keywords(source: str) -> str:
         return re.sub(re.compile(Scala.keywords_regex()), '', source)

src/PyReprism/utils/normalizer.py

@@ -6,10 +6,10 @@ def __init__():
         pass
 
     @staticmethod
-    def whitespaces_regex():
+    def whitespaces_regex() -> re.Pattern:
         return re.compile(r'[\t\x0b\x0c\r ]+|^\s*\n', re.MULTILINE)
 
     @staticmethod
-    def remove_whitespaces(source: str):
+    def remove_whitespaces(source: str) -> str:
         pattern = re.sub(Normalizer.whitespaces_regex(), '', source)
         return pattern.strip()