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/types/audio.py

@@ -0,0 +1,212 @@
+# -*- coding: utf-8 -*-
+
+from __future__ import absolute_import
+
+from .base import Type
+
+
+class Midi(Type):
+    """
+    Implements the Midi audio type matcher.
+    """
+    MIME = 'audio/midi'
+    EXTENSION = 'midi'
+
+    def __init__(self):
+        super(Midi, self).__init__(
+            mime=Midi.MIME,
+            extension=Midi.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 3 and
+                buf[0] == 0x4D and
+                buf[1] == 0x54 and
+                buf[2] == 0x68 and
+                buf[3] == 0x64)
+
+
+class Mp3(Type):
+    """
+    Implements the MP3 audio type matcher.
+    """
+    MIME = 'audio/mpeg'
+    EXTENSION = 'mp3'
+
+    def __init__(self):
+        super(Mp3, self).__init__(
+            mime=Mp3.MIME,
+            extension=Mp3.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 2 and
+                ((buf[0] == 0x49 and
+                  buf[1] == 0x44 and
+                  buf[2] == 0x33) or
+                 (buf[0] == 0xFF and
+                  buf[1] == 0xF2) or
+                 (buf[0] == 0xFF and
+                  buf[1] == 0xF3) or
+                 (buf[0] == 0xFF and
+                  buf[1] == 0xFB)))
+
+
+class M4a(Type):
+    """
+    Implements the M4A audio type matcher.
+    """
+    MIME = 'audio/mp4'
+    EXTENSION = 'm4a'
+
+    def __init__(self):
+        super(M4a, self).__init__(
+            mime=M4a.MIME,
+            extension=M4a.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 10 and
+                ((buf[4] == 0x66 and
+                    buf[5] == 0x74 and
+                    buf[6] == 0x79 and
+                    buf[7] == 0x70 and
+                    buf[8] == 0x4D and
+                    buf[9] == 0x34 and
+                    buf[10] == 0x41) or
+                (buf[0] == 0x4D and
+                    buf[1] == 0x34 and
+                    buf[2] == 0x41 and
+                    buf[3] == 0x20)))
+
+
+class Ogg(Type):
+    """
+    Implements the OGG audio type matcher.
+    """
+    MIME = 'audio/ogg'
+    EXTENSION = 'ogg'
+
+    def __init__(self):
+        super(Ogg, self).__init__(
+            mime=Ogg.MIME,
+            extension=Ogg.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 3 and
+                buf[0] == 0x4F and
+                buf[1] == 0x67 and
+                buf[2] == 0x67 and
+                buf[3] == 0x53)
+
+
+class Flac(Type):
+    """
+    Implements the FLAC audio type matcher.
+    """
+    MIME = 'audio/x-flac'
+    EXTENSION = 'flac'
+
+    def __init__(self):
+        super(Flac, self).__init__(
+            mime=Flac.MIME,
+            extension=Flac.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 3 and
+                buf[0] == 0x66 and
+                buf[1] == 0x4C and
+                buf[2] == 0x61 and
+                buf[3] == 0x43)
+
+
+class Wav(Type):
+    """
+    Implements the WAV audio type matcher.
+    """
+    MIME = 'audio/x-wav'
+    EXTENSION = 'wav'
+
+    def __init__(self):
+        super(Wav, self).__init__(
+            mime=Wav.MIME,
+            extension=Wav.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 11 and
+                buf[0] == 0x52 and
+                buf[1] == 0x49 and
+                buf[2] == 0x46 and
+                buf[3] == 0x46 and
+                buf[8] == 0x57 and
+                buf[9] == 0x41 and
+                buf[10] == 0x56 and
+                buf[11] == 0x45)
+
+
+class Amr(Type):
+    """
+    Implements the AMR audio type matcher.
+    """
+    MIME = 'audio/amr'
+    EXTENSION = 'amr'
+
+    def __init__(self):
+        super(Amr, self).__init__(
+            mime=Amr.MIME,
+            extension=Amr.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 11 and
+                buf[0] == 0x23 and
+                buf[1] == 0x21 and
+                buf[2] == 0x41 and
+                buf[3] == 0x4D and
+                buf[4] == 0x52 and
+                buf[5] == 0x0A)
+
+
+class Aac(Type):
+    """Implements the Aac audio type matcher."""
+
+    MIME = 'audio/aac'
+    EXTENSION = 'aac'
+
+    def __init__(self):
+        super(Aac, self).__init__(
+            mime=Aac.MIME,
+            extension=Aac.EXTENSION
+        )
+
+    def match(self, buf):
+        return (buf[:2] == bytearray([0xff, 0xf1]) or
+                buf[:2] == bytearray([0xff, 0xf9]))
+
+
+class Aiff(Type):
+    """
+    Implements the AIFF audio type matcher.
+    """
+    MIME = 'audio/x-aiff'
+    EXTENSION = 'aiff'
+
+    def __init__(self):
+        super(Aiff, self).__init__(
+            mime=Aiff.MIME,
+            extension=Aiff.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 11 and
+                buf[0] == 0x46 and
+                buf[1] == 0x4F and
+                buf[2] == 0x52 and
+                buf[3] == 0x4D and
+                buf[8] == 0x41 and
+                buf[9] == 0x49 and
+                buf[10] == 0x46 and
+                buf[11] == 0x46)

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

@@ -0,0 +1,29 @@
+# -*- coding: utf-8 -*-
+
+
+class Type(object):
+    """
+    Represents the file type object inherited by
+    specific file type matchers.
+    Provides convenient accessor and helper methods.
+    """
+    def __init__(self, mime, extension):
+        self.__mime = mime
+        self.__extension = extension
+
+    @property
+    def mime(self):
+        return self.__mime
+
+    @property
+    def extension(self):
+        return self.__extension
+
+    def is_extension(self, extension):
+        return self.__extension is extension
+
+    def is_mime(self, mime):
+        return self.__mime is mime
+
+    def match(self, buf):
+        raise NotImplementedError

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

@@ -0,0 +1,256 @@
+# -*- coding: utf-8 -*-
+
+from __future__ import absolute_import
+
+from .base import Type
+
+
+class ZippedDocumentBase(Type):
+    def match(self, buf):
+        # start by checking for ZIP local file header signature
+        idx = self.search_signature(buf, 0, 6000)
+        if idx != 0:
+            return
+
+        return self.match_document(buf)
+
+    def match_document(self, buf):
+        raise NotImplementedError
+
+    def compare_bytes(self, buf, subslice, start_offset):
+        sl = len(subslice)
+
+        if start_offset + sl > len(buf):
+            return False
+
+        return buf[start_offset:start_offset + sl] == subslice
+
+    def search_signature(self, buf, start, rangeNum):
+        signature = b"PK\x03\x04"
+        length = len(buf)
+
+        end = start + rangeNum
+        end = length if end > length else end
+
+        if start >= end:
+            return -1
+
+        try:
+            return buf.index(signature, start, end)
+        except ValueError:
+            return -1
+
+
+class OpenDocument(ZippedDocumentBase):
+    def match_document(self, buf):
+        # Check if first file in archive is the identifying file
+        if not self.compare_bytes(buf, b"mimetype", 0x1E):
+            return
+
+        # Check content of mimetype file if it matches current mime
+        return self.compare_bytes(buf, bytes(self.mime, "ASCII"), 0x26)
+
+
+class OfficeOpenXml(ZippedDocumentBase):
+    def match_document(self, buf):
+        # Check if first file in archive is the identifying file
+        ft = self.match_filename(buf, 0x1E)
+        if ft:
+            return ft
+
+        # Otherwise check that the fist file is one of these
+        if (
+            not self.compare_bytes(buf, b"[Content_Types].xml", 0x1E)
+            and not self.compare_bytes(buf, b"_rels/.rels", 0x1E)
+            and not self.compare_bytes(buf, b"docProps", 0x1E)
+        ):
+            return
+
+        # Loop through next 3 files and check if they match
+        # NOTE: OpenOffice/Libreoffice orders ZIP entry differently, so check the 4th file
+        # https://github.com/h2non/filetype/blob/d730d98ad5c990883148485b6fd5adbdd378364a/matchers/document.go#L134
+        idx = 0
+        for i in range(4):
+            # Search for next file header
+            idx = self.search_signature(buf, idx + 4, 6000)
+            if idx == -1:
+                return
+
+            # Filename is at file header + 30
+            ft = self.match_filename(buf, idx + 30)
+            if ft:
+                return ft
+
+    def match_filename(self, buf, offset):
+        if self.compare_bytes(buf, b"word/", offset):
+            return (
+                self.mime
+                == "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+            )
+        if self.compare_bytes(buf, b"ppt/", offset):
+            return (
+                self.mime
+                == "application/vnd.openxmlformats-officedocument.presentationml.presentation"
+            )
+        if self.compare_bytes(buf, b"xl/", offset):
+            return (
+                self.mime
+                == "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
+            )
+
+
+class Doc(Type):
+    """
+    Implements the Microsoft Word (Office 97-2003) document type matcher.
+    """
+
+    MIME = "application/msword"
+    EXTENSION = "doc"
+
+    def __init__(self):
+        super(Doc, self).__init__(mime=Doc.MIME, extension=Doc.EXTENSION)
+
+    def match(self, buf):
+        if len(buf) > 515 and buf[0:8] == b"\xD0\xCF\x11\xE0\xA1\xB1\x1A\xE1":
+            if buf[512:516] == b"\xEC\xA5\xC1\x00":
+                return True
+            if (
+                len(buf) > 2142
+                and b"\x00\x0A\x00\x00\x00MSWordDoc\x00\x10\x00\x00\x00Word.Document.8\x00\xF49\xB2q"
+                in buf[2075:2142]
+            ):
+                return True
+
+        return False
+
+
+class Docx(OfficeOpenXml):
+    """
+    Implements the Microsoft Word OOXML (Office 2007+) document type matcher.
+    """
+
+    MIME = "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+    EXTENSION = "docx"
+
+    def __init__(self):
+        super(Docx, self).__init__(mime=Docx.MIME, extension=Docx.EXTENSION)
+
+
+class Odt(OpenDocument):
+    """
+    Implements the OpenDocument Text document type matcher.
+    """
+
+    MIME = "application/vnd.oasis.opendocument.text"
+    EXTENSION = "odt"
+
+    def __init__(self):
+        super(Odt, self).__init__(mime=Odt.MIME, extension=Odt.EXTENSION)
+
+
+class Xls(Type):
+    """
+    Implements the Microsoft Excel (Office 97-2003) document type matcher.
+    """
+
+    MIME = "application/vnd.ms-excel"
+    EXTENSION = "xls"
+
+    def __init__(self):
+        super(Xls, self).__init__(mime=Xls.MIME, extension=Xls.EXTENSION)
+
+    def match(self, buf):
+        if len(buf) > 520 and buf[0:8] == b"\xD0\xCF\x11\xE0\xA1\xB1\x1A\xE1":
+            if buf[512:516] == b"\xFD\xFF\xFF\xFF" and (
+                buf[518] == 0x00 or buf[518] == 0x02
+            ):
+                return True
+            if buf[512:520] == b"\x09\x08\x10\x00\x00\x06\x05\x00":
+                return True
+            if (
+                len(buf) > 2095
+                and b"\xE2\x00\x00\x00\x5C\x00\x70\x00\x04\x00\x00Calc"
+                in buf[1568:2095]
+            ):
+                return True
+
+        return False
+
+
+class Xlsx(OfficeOpenXml):
+    """
+    Implements the Microsoft Excel OOXML (Office 2007+) document type matcher.
+    """
+
+    MIME = "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
+    EXTENSION = "xlsx"
+
+    def __init__(self):
+        super(Xlsx, self).__init__(mime=Xlsx.MIME, extension=Xlsx.EXTENSION)
+
+
+class Ods(OpenDocument):
+    """
+    Implements the OpenDocument Spreadsheet document type matcher.
+    """
+
+    MIME = "application/vnd.oasis.opendocument.spreadsheet"
+    EXTENSION = "ods"
+
+    def __init__(self):
+        super(Ods, self).__init__(mime=Ods.MIME, extension=Ods.EXTENSION)
+
+
+class Ppt(Type):
+    """
+    Implements the Microsoft PowerPoint (Office 97-2003) document type matcher.
+    """
+
+    MIME = "application/vnd.ms-powerpoint"
+    EXTENSION = "ppt"
+
+    def __init__(self):
+        super(Ppt, self).__init__(mime=Ppt.MIME, extension=Ppt.EXTENSION)
+
+    def match(self, buf):
+        if len(buf) > 524 and buf[0:8] == b"\xD0\xCF\x11\xE0\xA1\xB1\x1A\xE1":
+            if buf[512:516] == b"\xA0\x46\x1D\xF0":
+                return True
+            if buf[512:516] == b"\x00\x6E\x1E\xF0":
+                return True
+            if buf[512:516] == b"\x0F\x00\xE8\x03":
+                return True
+            if buf[512:516] == b"\xFD\xFF\xFF\xFF" and buf[522:524] == b"\x00\x00":
+                return True
+            if (
+                len(buf) > 2096
+                and buf[2072:2096]
+                == b"\x00\xB9\x29\xE8\x11\x00\x00\x00MS PowerPoint 97"
+            ):
+                return True
+
+        return False
+
+
+class Pptx(OfficeOpenXml):
+    """
+    Implements the Microsoft PowerPoint OOXML (Office 2007+) document type matcher.
+    """
+
+    MIME = "application/vnd.openxmlformats-officedocument.presentationml.presentation"
+    EXTENSION = "pptx"
+
+    def __init__(self):
+        super(Pptx, self).__init__(mime=Pptx.MIME, extension=Pptx.EXTENSION)
+
+
+class Odp(OpenDocument):
+    """
+    Implements the OpenDocument Presentation document type matcher.
+    """
+
+    MIME = "application/vnd.oasis.opendocument.presentation"
+    EXTENSION = "odp"
+
+    def __init__(self):
+        super(Odp, self).__init__(mime=Odp.MIME, extension=Odp.EXTENSION)

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

@@ -0,0 +1,115 @@
+# -*- coding: utf-8 -*-
+
+from __future__ import absolute_import
+
+from .base import Type
+
+
+class Woff(Type):
+    """
+    Implements the WOFF font type matcher.
+    """
+    MIME = 'application/font-woff'
+    EXTENSION = 'woff'
+
+    def __init__(self):
+        super(Woff, self).__init__(
+            mime=Woff.MIME,
+            extension=Woff.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 7 and
+                buf[0] == 0x77 and
+                buf[1] == 0x4F and
+                buf[2] == 0x46 and
+                buf[3] == 0x46 and
+                ((buf[4] == 0x00 and
+                  buf[5] == 0x01 and
+                  buf[6] == 0x00 and
+                  buf[7] == 0x00) or
+                 (buf[4] == 0x4F and
+                  buf[5] == 0x54 and
+                  buf[6] == 0x54 and
+                  buf[7] == 0x4F) or
+                 (buf[4] == 0x74 and
+                  buf[5] == 0x72 and
+                  buf[6] == 0x75 and
+                  buf[7] == 0x65)))
+
+
+class Woff2(Type):
+    """
+    Implements the WOFF2 font type matcher.
+    """
+    MIME = 'application/font-woff'
+    EXTENSION = 'woff2'
+
+    def __init__(self):
+        super(Woff2, self).__init__(
+            mime=Woff2.MIME,
+            extension=Woff2.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 7 and
+                buf[0] == 0x77 and
+                buf[1] == 0x4F and
+                buf[2] == 0x46 and
+                buf[3] == 0x32 and
+                ((buf[4] == 0x00 and
+                  buf[5] == 0x01 and
+                  buf[6] == 0x00 and
+                  buf[7] == 0x00) or
+                 (buf[4] == 0x4F and
+                  buf[5] == 0x54 and
+                  buf[6] == 0x54 and
+                  buf[7] == 0x4F) or
+                 (buf[4] == 0x74 and
+                  buf[5] == 0x72 and
+                  buf[6] == 0x75 and
+                  buf[7] == 0x65)))
+
+
+class Ttf(Type):
+    """
+    Implements the TTF font type matcher.
+    """
+    MIME = 'application/font-sfnt'
+    EXTENSION = 'ttf'
+
+    def __init__(self):
+        super(Ttf, self).__init__(
+            mime=Ttf.MIME,
+            extension=Ttf.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 4 and
+                buf[0] == 0x00 and
+                buf[1] == 0x01 and
+                buf[2] == 0x00 and
+                buf[3] == 0x00 and
+                buf[4] == 0x00)
+
+
+class Otf(Type):
+    """
+    Implements the OTF font type matcher.
+    """
+    MIME = 'application/font-sfnt'
+    EXTENSION = 'otf'
+
+    def __init__(self):
+        super(Otf, self).__init__(
+            mime=Otf.MIME,
+            extension=Otf.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 4 and
+                buf[0] == 0x4F and
+                buf[1] == 0x54 and
+                buf[2] == 0x54 and
+                buf[3] == 0x4F and
+                buf[4] == 0x00)