MeUtils 2025.3.3.18.41.24__py3-none-any.whl → 2025.3.5.19.55.22__py3-none-any.whl

meutils/io/parsers/fileparser/filetype/__main__.py

@@ -0,0 +1,37 @@
+import sys
+
+import filetype
+
+
+def guess(path):
+    kind = filetype.guess(path)
+    if kind is None:
+        print('{}: File type determination failure.'.format(path))
+    else:
+        print('{}: {} ({})'.format(path, kind.extension, kind.mime))
+
+
+def main():
+    import argparse
+
+    parser = argparse.ArgumentParser(
+        prog='filetype', description='Determine type of FILEs.'
+    )
+    parser.add_argument('-f', '--file', nargs='+')
+    parser.add_argument(
+        '-v', '--version', action='version',
+        version='%(prog)s ' + filetype.version,
+        help='output version information and exit'
+    )
+
+    args = parser.parse_args()
+    if len(sys.argv) < 2:
+        parser.print_help()
+        sys.exit(1)
+
+    for i in args.file:
+        guess(i)
+
+
+if __name__ == '__main__':
+    main()

meutils/io/parsers/fileparser/filetype/filetype.py

@@ -0,0 +1,98 @@
+# -*- coding: utf-8 -*-
+
+from __future__ import absolute_import
+
+from .match import match
+from .types import TYPES, Type
+
+# Expose supported matchers types
+types = TYPES
+
+
+def guess(obj):
+    """
+    Infers the type of the given input.
+
+    Function is overloaded to accept multiple types in input
+    and peform the needed type inference based on it.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        The matched type instance. Otherwise None.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return match(obj) if obj else None
+
+
+def guess_mime(obj):
+    """
+    Infers the file type of the given input
+    and returns its MIME type.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        The matched MIME type as string. Otherwise None.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    kind = guess(obj)
+    return kind.mime if kind else kind
+
+
+def guess_extension(obj):
+    """
+    Infers the file type of the given input
+    and returns its RFC file extension.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        The matched file extension as string. Otherwise None.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    kind = guess(obj)
+    return kind.extension if kind else kind
+
+
+def get_type(mime=None, ext=None):
+    """
+    Returns the file type instance searching by
+    MIME type or file extension.
+
+    Args:
+        ext: file extension string. E.g: jpg, png, mp4, mp3
+        mime: MIME string. E.g: image/jpeg, video/mpeg
+
+    Returns:
+        The matched file type instance. Otherwise None.
+    """
+    for kind in types:
+        if kind.extension == ext or kind.mime == mime:
+            return kind
+    return None
+
+
+def add_type(instance):
+    """
+    Adds a new type matcher instance to the supported types.
+
+    Args:
+        instance: Type inherited instance.
+
+    Returns:
+        None
+    """
+    if not isinstance(instance, Type):
+        raise TypeError('instance must inherit from filetype.types.Type')
+
+    types.insert(0, instance)

meutils/io/parsers/fileparser/filetype/helpers.py

@@ -0,0 +1,140 @@
+# -*- coding: utf-8 -*-
+
+from __future__ import absolute_import
+from .types import TYPES
+from .match import (
+    image_match, font_match, document_match,
+    video_match, audio_match, archive_match
+)
+
+
+def is_extension_supported(ext):
+    """
+    Checks if the given extension string is
+    one of the supported by the file matchers.
+
+    Args:
+        ext (str): file extension string. E.g: jpg, png, mp4, mp3
+
+    Returns:
+        True if the file extension is supported.
+        Otherwise False.
+    """
+    for kind in TYPES:
+        if kind.extension == ext:
+            return True
+    return False
+
+
+def is_mime_supported(mime):
+    """
+    Checks if the given MIME type string is
+    one of the supported by the file matchers.
+
+    Args:
+        mime (str): MIME string. E.g: image/jpeg, video/mpeg
+
+    Returns:
+        True if the MIME type is supported.
+        Otherwise False.
+    """
+    for kind in TYPES:
+        if kind.mime == mime:
+            return True
+    return False
+
+
+def is_image(obj):
+    """
+    Checks if a given input is a supported type image.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        True if obj is a valid image. Otherwise False.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return image_match(obj) is not None
+
+
+def is_archive(obj):
+    """
+    Checks if a given input is a supported type archive.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        True if obj is a valid archive. Otherwise False.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return archive_match(obj) is not None
+
+
+def is_audio(obj):
+    """
+    Checks if a given input is a supported type audio.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        True if obj is a valid audio. Otherwise False.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return audio_match(obj) is not None
+
+
+def is_video(obj):
+    """
+    Checks if a given input is a supported type video.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        True if obj is a valid video. Otherwise False.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return video_match(obj) is not None
+
+
+def is_font(obj):
+    """
+    Checks if a given input is a supported type font.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        True if obj is a valid font. Otherwise False.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return font_match(obj) is not None
+
+
+def is_document(obj):
+    """
+    Checks if a given input is a supported type document.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        True if obj is a valid document. Otherwise False.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return document_match(obj) is not None

meutils/io/parsers/fileparser/filetype/match.py

@@ -0,0 +1,155 @@
+# -*- coding: utf-8 -*-
+
+from __future__ import absolute_import
+
+from .types import ARCHIVE as archive_matchers
+from .types import AUDIO as audio_matchers
+from .types import APPLICATION as application_matchers
+from .types import DOCUMENT as document_matchers
+from .types import FONT as font_matchers
+from .types import IMAGE as image_matchers
+from .types import VIDEO as video_matchers
+from .types import TYPES
+from .utils import get_bytes
+
+
+def match(obj, matchers=TYPES):
+    """
+    Matches the given input against the available
+    file type matchers.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        Type instance if type matches. Otherwise None.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    buf = get_bytes(obj)
+
+    for matcher in matchers:
+        if matcher.match(buf):
+            return matcher
+
+    return None
+
+
+def image_match(obj):
+    """
+    Matches the given input against the available
+    image type matchers.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        Type instance if matches. Otherwise None.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return match(obj, image_matchers)
+
+
+def font_match(obj):
+    """
+    Matches the given input against the available
+    font type matchers.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        Type instance if matches. Otherwise None.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return match(obj, font_matchers)
+
+
+def video_match(obj):
+    """
+    Matches the given input against the available
+    video type matchers.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        Type instance if matches. Otherwise None.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return match(obj, video_matchers)
+
+
+def audio_match(obj):
+    """
+    Matches the given input against the available
+    autio type matchers.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        Type instance if matches. Otherwise None.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return match(obj, audio_matchers)
+
+
+def archive_match(obj):
+    """
+    Matches the given input against the available
+    archive type matchers.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        Type instance if matches. Otherwise None.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return match(obj, archive_matchers)
+
+
+def application_match(obj):
+    """
+    Matches the given input against the available
+    application type matchers.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        Type instance if matches. Otherwise None.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return match(obj, application_matchers)
+
+
+def document_match(obj):
+    """
+    Matches the given input against the available
+    document type matchers.
+
+    Args:
+        obj: path to file, bytes or bytearray.
+
+    Returns:
+        Type instance if matches. Otherwise None.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return match(obj, document_matchers)

meutils/io/parsers/fileparser/filetype/types/__init__.py

@@ -0,0 +1,118 @@
+# -*- coding: utf-8 -*-
+
+from __future__ import absolute_import
+
+from . import archive
+from . import audio
+from . import application
+from . import document
+from . import font
+from . import image
+from . import video
+from .base import Type  # noqa
+
+# Supported image types
+IMAGE = (
+    image.Dwg(),
+    image.Xcf(),
+    image.Jpeg(),
+    image.Jpx(),
+    image.Apng(),
+    image.Png(),
+    image.Gif(),
+    image.Webp(),
+    image.Tiff(),
+    image.Cr2(),
+    image.Bmp(),
+    image.Jxr(),
+    image.Psd(),
+    image.Ico(),
+    image.Heic(),
+    image.Dcm(),
+    image.Avif(),
+)
+
+# Supported video types
+VIDEO = (
+    video.M3gp(),
+    video.Mp4(),
+    video.M4v(),
+    video.Mkv(),
+    video.Mov(),
+    video.Avi(),
+    video.Wmv(),
+    video.Mpeg(),
+    video.Webm(),
+    video.Flv(),
+)
+
+# Supported audio types
+AUDIO = (
+    audio.Aac(),
+    audio.Midi(),
+    audio.Mp3(),
+    audio.M4a(),
+    audio.Ogg(),
+    audio.Flac(),
+    audio.Wav(),
+    audio.Amr(),
+    audio.Aiff(),
+)
+
+# Supported font types
+FONT = (font.Woff(), font.Woff2(), font.Ttf(), font.Otf())
+
+# Supported archive container types
+ARCHIVE = (
+    archive.Br(),
+    archive.Rpm(),
+    archive.Dcm(),
+    archive.Epub(),
+    archive.Zip(),
+    archive.Tar(),
+    archive.Rar(),
+    archive.Gz(),
+    archive.Bz2(),
+    archive.SevenZ(),
+    archive.Pdf(),
+    archive.Exe(),
+    archive.Swf(),
+    archive.Rtf(),
+    archive.Nes(),
+    archive.Crx(),
+    archive.Cab(),
+    archive.Eot(),
+    archive.Ps(),
+    archive.Xz(),
+    archive.Sqlite(),
+    archive.Deb(),
+    archive.Ar(),
+    archive.Z(),
+    archive.Lzop(),
+    archive.Lz(),
+    archive.Elf(),
+    archive.Lz4(),
+    archive.Zstd(),
+)
+
+# Supported archive container types
+APPLICATION = (
+    application.Wasm(),
+)
+
+# Supported document types
+DOCUMENT = (
+    document.Doc(),
+    document.Docx(),
+    document.Odt(),
+    document.Xls(),
+    document.Xlsx(),
+    document.Ods(),
+    document.Ppt(),
+    document.Pptx(),
+    document.Odp(),
+)
+
+
+# Expose supported type matchers
+TYPES = list(IMAGE + AUDIO + VIDEO + FONT + DOCUMENT + ARCHIVE + APPLICATION)

meutils/io/parsers/fileparser/filetype/types/application.py

@@ -0,0 +1,22 @@
+# -*- coding: utf-8 -*-
+
+from __future__ import absolute_import
+
+from .base import Type
+
+
+class Wasm(Type):
+    """Implements the Wasm image type matcher."""
+
+    MIME = 'application/wasm'
+    EXTENSION = 'wasm'
+
+    def __init__(self):
+        super(Wasm, self).__init__(
+            mime=Wasm.MIME,
+            extension=Wasm.EXTENSION
+        )
+
+    def match(self, buf):
+        return buf[:8] == bytearray([0x00, 0x61, 0x73, 0x6d,
+                                     0x01, 0x00, 0x00, 0x00])