Add some file types from the Go version

README.rst

@@ -74,6 +74,7 @@ Image
 -  **psd** - ``image/vnd.adobe.photoshop``
 -  **ico** - ``image/x-icon``
 -  **heic** - ``image/heic``
+-  **dwg** - ``image/vnd.dwg``
 
 Video
 ^^^^^
@@ -87,6 +88,7 @@ Video
 -  **wmv** - ``video/x-ms-wmv``
 -  **mpg** - ``video/mpeg``
 -  **flv** - ``video/x-flv``
+-  **3gp** - ``video/3gpp``
 
 Audio
 ^^^^^
@@ -98,6 +100,13 @@ Audio
 -  **flac** - ``audio/x-flac``
 -  **wav** - ``audio/x-wav``
 -  **amr** - ``audio/amr``
+-  **aac** - ``audio/aac``
+
+Application
+^^^^^^^^^^^
+
+-  **wasm** - ``application/wasm``
+
 
 Archive
 ^^^^^^^
@@ -126,6 +135,11 @@ Archive
 -  **Z** - ``application/x-compress``
 -  **lz** - ``application/x-lzip``
 
+-  **rpm** - ``application/x-rpm``
+-  **elf** - ``application/x-executable``
+-  **dcm** - ``application/dicom``
+-  **iso** - ``application/x-iso9660-image``
+
 Font
 ^^^^
 
@@ -134,6 +148,13 @@ Font
 -  **ttf** - ``application/font-sfnt``
 -  **otf** - ``application/font-sfnt``
 
+Document
+^^^^^^^^
+-  **doc** - ``application/msword``
+-  **xls** - ``application/vnd.ms-excel``
+-  **ppt** - ``application/vnd.ms-powerpoint``
+
+
 .. _Python: http://python.org
 .. _magic numbers: https://en.wikipedia.org/wiki/Magic_number_(programming)#Magic_numbers_in_files
 .. _filetype: https://github.com/h2non/filetype

filetype/helpers.py

@@ -120,3 +120,35 @@ def is_font(obj):
         TypeError: if obj is not a supported type.
     """
     return match.font(obj) is not None
+
+
+def is_application(obj):
+    """
+    Checks if a given input is a supported type application.
+
+    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 match.application(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 font. Otherwise False.
+
+    Raises:
+        TypeError: if obj is not a supported type.
+    """
+    return match.document(obj) is not None

filetype/match.py

@@ -2,8 +2,10 @@
 
 from __future__ import absolute_import
 
+from .types import APPLICATION as application_matchers
 from .types import ARCHIVE as archive_matchers
 from .types import AUDIO as audio_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
@@ -117,3 +119,37 @@ def archive(obj):
         TypeError: if obj is not a supported type.
     """
     return match(obj, archive_matchers)
+
+
+def application(obj):
+    """
+    Matches the given input againts 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(obj):
+    """
+    Matches the given input againts 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)

filetype/types/__init__.py

@@ -2,8 +2,10 @@
 
 from __future__ import absolute_import
 
+from . import application
 from . import archive
 from . import audio
+from . import document
 from . import font
 from . import image
 from . import video
@@ -23,20 +25,21 @@
     image.Psd(),
     image.Ico(),
     image.Heic(),
-    image.Dcm(),
+    image.Dwg(),
 )
 
 # Supported video types
 VIDEO = (
-    video.Mp4(),
     video.M4v(),
     video.Mkv(),
+    video.Webm(),
     video.Mov(),
     video.Avi(),
     video.Wmv(),
     video.Mpeg(),
-    video.Webm(),
     video.Flv(),
+    video.Mp4(),
+    video.Match3gp(),
 )
 
 # Supported audio types
@@ -48,6 +51,7 @@
     audio.Flac(),
     audio.Wav(),
     audio.Amr(),
+    audio.Aac(),
 )
 
 # Supported font types
@@ -77,7 +81,24 @@
     archive.Ar(),
     archive.Z(),
     archive.Lz(),
+    archive.Rpm(),
+    archive.Elf(),
+    archive.Dcm(),
+    archive.Iso(),
 )
 
+# Supported application types
+APPLICATION = (
+    application.Wasm(),
+)
+
+# Supported application types
+DOCUMENT = (
+    document.Doc(),
+    document.Xls(),
+    document.Ppt(),
+)
+
+
 # Expose supported type matchers
-TYPES = list(VIDEO + IMAGE + AUDIO + FONT + ARCHIVE)
+TYPES = list(VIDEO + IMAGE + AUDIO + FONT + ARCHIVE + APPLICATION + DOCUMENT)

filetype/types/application.py

@@ -0,0 +1,34 @@
+# -*- coding: utf-8 -*-
+
+from __future__ import absolute_import
+
+from .base import Type
+
+
+class Wasm(Type):
+    """
+    Implements the WASM Web Assembly 1.0 filetype.
+
+    WASM has starts with `\0asm`, followed by the version.
+    http://webassembly.github.io/spec/core/binary/modules.html#binary-magic
+    """
+
+    MIME = 'application/wasm'
+    EXTENSION = 'wasm'
+
+    def __init__(self):
+        super(Wasm, self).__init__(
+            mime=Wasm.MIME,
+            extension=Wasm.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 8 and
+                buf[0] == 0x00 and
+                buf[1] == 0x61 and
+                buf[2] == 0x73 and
+                buf[3] == 0x6D and
+                buf[4] == 0x01 and
+                buf[5] == 0x00 and
+                buf[6] == 0x00 and
+                buf[7] == 0x00)

filetype/types/archive.py

@@ -513,3 +513,88 @@ def match(self, buf):
                 buf[1] == 0x5A and
                 buf[2] == 0x49 and
                 buf[3] == 0x50)
+
+
+class Rpm(Type):
+    """
+    Implements the RPM archive type matcher.
+    """
+    MIME = 'application/x-rpm'
+    EXTENSION = 'rpm'
+
+    def __init__(self):
+        super(Rpm, self).__init__(
+            mime=Rpm.MIME,
+            extension=Rpm.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 96 and
+                buf[0] == 0xED and
+                buf[1] == 0xAB and
+                buf[2] == 0xEE and
+                buf[3] == 0xDB)
+
+
+class Elf(Type):
+    """
+    Implements the Elf archive type matcher.
+    """
+    MIME = 'application/x-executable'
+    EXTENSION = 'elf'
+
+    def __init__(self):
+        super(Elf, self).__init__(
+            mime=Elf.MIME,
+            extension=Elf.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 52 and
+                buf[0] == 0x7F and
+                buf[1] == 0x45 and
+                buf[2] == 0x4C and
+                buf[3] == 0x46)
+
+
+class Dcm(Type):
+    """
+    Implements the Dcm archive type matcher.
+    """
+    MIME = 'application/dicom'
+    EXTENSION = 'dcm'
+
+    def __init__(self):
+        super(Dcm, self).__init__(
+            mime=Dcm.MIME,
+            extension=Dcm.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 131 and
+                buf[128] == 0x44 and
+                buf[129] == 0x49 and
+                buf[130] == 0x43 and
+                buf[131] == 0x4D)
+
+
+class Iso(Type):
+    """
+    Implements the ISO archive type matcher.
+    """
+    MIME = 'application/x-iso9660-image'
+    EXTENSION = 'iso'
+
+    def __init__(self):
+        super(Iso, self).__init__(
+            mime=Iso.MIME,
+            extension=Iso.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 32773 and
+                buf[32769] == 0x43 and
+                buf[32770] == 0x44 and
+                buf[32771] == 0x30 and
+                buf[32772] == 0x30 and
+                buf[32773] == 0x31)

filetype/types/audio.py

@@ -164,3 +164,24 @@ def match(self, buf):
                 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 (len(buf) > 1 and
+                ((buf[0] == 0xFF and
+                    buf[1] == 0xF1) or
+                (buf[0] == 0xFF and
+                    buf[1] == 0xF9)))