🗓 Dec 19, 2023 8:19:47 PM

🤖 types added/updated
🐙 improve automatic delimiter detector
🎨 minor changes

chepy/modules/dataformat.py

@@ -568,10 +568,14 @@ def from_hex(self, delimiter: str = None, join_by: str = " ") -> DataFormatT:
             b"AAA"
         """
         data = self._convert_to_bytes()
-        delimiter = detect_delimiter(data, default_delimiter=None)
+        if delimiter is None:
+            delimiter = detect_delimiter(data, default_delimiter=None)
         if delimiter is not None:
             self.state = join_by.encode().join(
-                list(binascii.unhexlify(x) for x in data.split(delimiter))
+                list(
+                    binascii.unhexlify(x)
+                    for x in data.split(self._str_to_bytes(delimiter))
+                )
             )
         else:
             self.state = binascii.unhexlify(data)
@@ -885,7 +889,9 @@ def from_charcode(
         """
         data = self._convert_to_str()
         out = []
-        delimiter = detect_delimiter(data)
+        if not delimiter:
+            delimiter = detect_delimiter(data)
+        print("🟢 ", delimiter)
         for c in data.split(delimiter):
             out.append(chr(int(c, base)))
         self.state = join_by.join(out)
@@ -926,7 +932,8 @@ def from_decimal(self, delimiter: str = None, join_by: str = "") -> DataFormatT:
             "ㅎ"
         """
         data = self._convert_to_str()
-        delimiter = detect_delimiter(data)
+        if not delimiter:
+            delimiter = detect_delimiter(data)
         self.state = join_by.join(
             list(chr(int(s)) for s in data.strip().split(delimiter))
         )
@@ -973,7 +980,8 @@ def from_binary(self, delimiter: str = None, byte_length: int = 8) -> DataFormat
             "abc"
         """
         data = self._convert_to_str()
-        delimiter = detect_delimiter(data)
+        if not delimiter:
+            delimiter = detect_delimiter(data)
         n = int(
             "".join([x[byte_length - 8 :] for x in data.split(delimiter)]),
             2,
@@ -1016,7 +1024,8 @@ def from_octal(self, delimiter: str = None, join_by: str = "") -> DataFormatT:
             "ab"
         """
         data = self._convert_to_str()
-        delimiter = detect_delimiter(data, default_delimiter=delimiter)
+        if not delimiter:
+            delimiter = detect_delimiter(data, default_delimiter=delimiter)
         self.state = join_by.join(
             list(chr(int(str(x), 8)) for x in data.split(delimiter))
         )
@@ -1372,7 +1381,7 @@ def substitute(self, x: str, y: str) -> DataFormatT:
         return self
 
     @ChepyDecorators.call_stack
-    def remove_nonprintable(self, replace_with: bytes = b"") -> DataFormatT:
+    def remove_nonprintable(self, replace_with: Union[str, bytes] = b"") -> DataFormatT:
         """Remove non-printable characters from string.
 
         Args:
@@ -1381,8 +1390,7 @@ def remove_nonprintable(self, replace_with: bytes = b"") -> DataFormatT:
         Returns:
             Chepy: The Chepy object.
         """
-        if isinstance(replace_with, str):
-            replace_with = replace_with.encode()
+        replace_with = self._str_to_bytes(replace_with)
         data = self._convert_to_bytes()
         self.state = re.sub(b"[^[:print:]]", replace_with, data)
         return self

chepy/modules/dataformat.pyi

@@ -71,11 +71,11 @@ class DataFormat(ChepyCore):
     def swap_strings(self: DataFormatT, by:int) -> DataFormatT: ...
     def to_string(self: DataFormatT) -> DataFormatT: ...
     def stringify(self: DataFormatT, compact:bool=...) -> DataFormatT: ...
-    def select(self: DataFormatT, start: int, end: int) -> DataFormatT: ...
+    def select(self: DataFormatT, start: int, end: int=None) -> DataFormatT: ...
     def length(self: DataFormatT) -> DataFormatT: ...
     def to_leetcode(self: DataFormatT, replace_space: str=...) -> DataFormatT: ...
     def substitute(self: DataFormatT, x: str=..., y: str=...) -> DataFormatT: ...
-    def remove_nonprintable(self: DataFormatT, replace_with: bytes = ...): ...
+    def remove_nonprintable(self: DataFormatT, replace_with: Union[str, bytes] = ...): ...
     def to_base91(self: DataFormatT) -> DataFormatT: ...
     def from_base91(self: DataFormatT) -> DataFormatT: ...
     def swap_endianness(self: DataFormatT, word_length: int=...) -> DataFormatT: ...

chepy/modules/encryptionencoding.py

@@ -1505,7 +1505,7 @@ def to_letter_number_code(
 
     @ChepyDecorators.call_stack
     def from_letter_number_code(
-        self, delimiter: Union[str, bytes] = None, join_by: Union[str, bytes] = ""
+        self, delimiter: Union[str, bytes] = ' ', join_by: Union[str, bytes] = ""
     ) -> EncryptionEncodingT:
         """Decode A1Z26
 
@@ -1517,7 +1517,7 @@ def from_letter_number_code(
             Chepy: The Chepy object.
         """
         data = self._convert_to_str()
-        delimiter = detect_delimiter(data)
+        delimiter = detect_delimiter(data, default_delimiter=delimiter)
         data = data.split(delimiter)
         hold = ["§" for _ in range(len(data))]
         for d in data:

chepy/modules/internal/helpers.py

@@ -137,19 +137,24 @@ def set(self, index, value):
 def detect_delimiter(
     data: Union[str, bytes],
     delimiters: List[Union[str, bytes]] = [
-        " ",
+        "; ",
         ";",
+        ". ",
         ".",
+        "- ",
         "-",
         "\\",
+        ": ",
         ":",
         "/",
+        ", ",
         ",",
+        " ",
         "\n",
         "\\x",
         "\\0x",
     ],
-    default_delimiter: str = " ",
+    default_delimiter: str = "",
 ) -> Union[str, bytes, None]:
     """Detect delimiter
 
@@ -161,6 +166,9 @@ def detect_delimiter(
     Returns:
         Union[str, bytes, None]: Delimiter or None if one is not found
     """
+    if default_delimiter:
+        return default_delimiter
+
     is_bytes = False
     if isinstance(data, bytes):  # pragma: no cover
         delimiters = [d.encode() for d in delimiters]