Initial Commit

dml/__init__.py

@@ -0,0 +1,16 @@
+"""
+Dotted Markup Language
+
+Translate text to and from DML with ease.
+"""
+
+__title__ = "DML"
+__author__ = "RGBCube"
+__license__ = "MIT"
+__copyright__ = "Copyright (c) 2022-present RGBCube"
+__version__ = "1.0.0"
+
+from .encoder import *
+from .errors import *
+from .fileutils import *
+from .run import *

dml/encoder.py

@@ -0,0 +1,61 @@
+import typing as t
+
+from .errors import DecodeError
+from .symbols import symbols as s
+
+__all__ = ("encode", "decode")
+
+
+def encode(text: t.Union[t.Generator, t.List[str], str], *, out: str = None) -> t.Optional[t.Generator[str, None, None]]:
+    """Encodes text into Dotted Markup Language (DML)
+
+    Arguments:
+        text (Union[Generator, List[str], str]): The text to encode.
+        out (str): The filepath to write the encoded file to. If not specified, the encoded text will be returned as a generator.
+
+    Returns:
+        Optional[Generator[str, None, None]]: The encoded text as a generator.
+    """
+
+    def inner() -> t.Generator[str, None, None]:
+        for char_ in text:
+            yield format(ord(char_), "b").replace("0", s["0"]).replace("1", s["1"]) + s["stop"]
+
+    if out:
+        out = out + ".dml" if not out.endswith(".dml") else out
+        with open(out, "w") as f:
+            for char in inner():
+                f.write(char)
+    else:
+        return inner()
+
+
+def decode(text: t.Union[t.Generator, t.List[str], str], *, out: str = None) -> t.Optional[t.Generator[str, None, None]]:
+    """Decodes text from Dotted Markup Language (DML)
+
+    Arguments:
+        text (Union[Generator, List[str], str]): The text to decode.
+        out (str): The filepath to write the decoded file to. If not specified, the decoded text will be returned as a generator.
+
+    Returns:
+        Optional[Generator[str, None, None]]: The decoded text as a generator.
+
+    Raises:
+        DecodeError: If the text is not valid DML.
+    """
+
+    def inner() -> t.Generator[str, None, None]:
+        nonlocal text
+        if isinstance(text, str):
+            text = [e + s["stop"] for e in text.split(s["stop"]) if e]
+        for char_ in text:
+            if any(e not in s.values() for e in char_):
+                raise DecodeError(f"Invalid character: '{char_}', expected {s['1']}, {s['0']} or {s['stop']}")
+            yield chr(int(char_[:-1].replace(s['0'], "0").replace(s['1'], "1"), 2))
+
+    if out:
+        with open(out, "w") as f:
+            for char in inner():
+                f.write(char)
+    else:
+        return inner()

dml/errors.py

@@ -0,0 +1,11 @@
+__all__ = ("DottedMarkupLanguageException", "DecodeError")
+
+
+class DottedMarkupLanguageException(Exception):
+    """Base class for all exceptions in this module."""
+    pass
+
+
+class DecodeError(DottedMarkupLanguageException):
+    """Raised when there is an error decoding a string."""
+    pass

dml/fileutils.py

@@ -0,0 +1,67 @@
+import typing as t
+
+from .encoder import encode, decode
+from .errors import DecodeError
+from .symbols import symbols as s
+
+__all__ = ("decode_file", "encode_file")
+
+
+def encode_file(fp: str, *, out: str = None) -> t.Optional[t.Generator[str, None, None]]:
+    """Encodes a file to DML.
+
+    Arguments:
+        fp (str): The filepath to the file to encode.
+        out (str): The filepath to write the encoded file to. If not specified, the encoded text will be returned as a generator.
+
+    Returns:
+        Optional[Generator[str, None, None]]: The encoded text as a generator.
+    """
+
+    def read_file() -> t.Generator[str, None, None]:
+        with open(fp) as f_:
+            while char_ := f_.read(1):
+                yield char_
+
+    if out:
+        out = out + ".dml" if not out.endswith(".dml") else out
+        with open(out, "w") as f:
+            for char in encode(read_file()):
+                f.write(char)
+    else:
+        return encode(read_file())
+
+
+def decode_file(fp: str, *, out: str = None) -> t.Optional[t.Generator[str, None, None]]:
+    """Decodes a file that has DML encoded in it.
+
+    Arguments:
+         fp (str): The filepath to the file to decode.
+         out (str): The filepath to write the decoded file to. If not specified, the decoded text will be returned as a generator.
+
+    Returns:
+        Optional[Generator[str, None, None]]: The decoded text as a generator.
+
+    Raises:
+        DecodeError: If the file is not valid DML.
+    """
+
+    def read_file() -> t.Generator[str, None, None]:
+        buffer = ""
+        with open(fp) as f_:
+            while char_ := f_.read(1):
+                if char_ not in s.values():
+                    raise DecodeError(f"Invalid character: '{char_}', expected {s['1']}, {s['0']} or {s['stop']}")
+                elif char_ != s["stop"]:
+                    buffer += char_
+                else:
+                    yield buffer + s["stop"]
+                    buffer = ""
+
+    if out:
+        out = out + ".dml" if not out.endswith(".dml") else out
+        with open(out, "w") as f:
+            for char in decode(read_file()):
+                f.write(char)
+    else:
+        return decode(read_file())

dml/run.py

@@ -0,0 +1,12 @@
+from .encoder import decode
+from .fileutils import decode_file
+
+__all__ = ("run_file", "run")
+
+
+def run(dml: str) -> None:
+    exec("".join(decode(dml)))
+
+
+def run_file(fp: str) -> None:
+    exec("".join(decode_file(fp)))

dml/symbols.py

@@ -0,0 +1,5 @@
+symbols = {
+    "1": "\u0701",  # Syriac Supralinear Full Stop
+    "0": "\u0660",  # Arabic-Indic Digit Zero
+    "stop": "\u06F0",  # Extended Arabic-Indic Digit Zero
+}