From c2a89b89689c68b4e12877186780ae367dcc570e Mon Sep 17 00:00:00 2001
From: flgc <abotes@gmail.com>
Date: Sun, 18 Aug 2019 11:52:44 +0900
Subject: [PATCH 1/6] enh: Add low hanging fruit from GO version

---
 filetype/types/application.py |  34 ++++++++++
 filetype/types/archive.py     |  88 +++++++++++++++++++++++++
 filetype/types/audio.py       |  21 ++++++
 filetype/types/document.py    | 118 ++++++++++++++++++++++++++++++++++
 filetype/types/image.py       |  27 ++++----
 filetype/types/video.py       |  94 +++++++++++++++++----------
 6 files changed, 335 insertions(+), 47 deletions(-)
 create mode 100644 filetype/types/application.py
 create mode 100644 filetype/types/document.py

diff --git a/filetype/types/application.py b/filetype/types/application.py
new file mode 100644
index 0000000..b9ac5f9
--- /dev/null
+++ b/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)
diff --git a/filetype/types/archive.py b/filetype/types/archive.py
index 5cde4b3..8a5eff7 100644
--- a/filetype/types/archive.py
+++ b/filetype/types/archive.py
@@ -513,3 +513,91 @@ 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)
diff --git a/filetype/types/audio.py b/filetype/types/audio.py
index 5dafba5..c8c5644 100644
--- a/filetype/types/audio.py
+++ b/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)))
diff --git a/filetype/types/document.py b/filetype/types/document.py
new file mode 100644
index 0000000..afad633
--- /dev/null
+++ b/filetype/types/document.py
@@ -0,0 +1,118 @@
+# -*- coding: utf-8 -*-
+
+from __future__ import absolute_import
+
+from .base import Type
+
+
+class Doc(Type):
+    """
+    Implements Doc file type matcher.
+    """
+
+    MIME = 'application/msword'
+    EXTENSION = 'doc'
+
+    def __init__(self):
+        super(Doc, self).__init__(
+            mime=Doc.MIME,
+            extension=Doc.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 7 and
+                buf[0] == 0xD0 and
+                buf[1] == 0xCF and
+                buf[2] == 0x11 and
+                buf[3] == 0xE0 and
+                buf[4] == 0xA1 and
+                buf[5] == 0xB1 and
+                buf[6] == 0x1A and
+                buf[7] == 0xE1)
+
+
+# todo: Docx
+
+
+class Xls(Type):
+    """
+    Implements Xls file type matcher.
+    Which looks just like the Doc file 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):
+        return (len(buf) > 7 and
+                buf[0] == 0xD0 and
+                buf[1] == 0xCF and
+                buf[2] == 0x11 and
+                buf[3] == 0xE0 and
+                buf[4] == 0xA1 and
+                buf[5] == 0xB1 and
+                buf[6] == 0x1A and
+                buf[7] == 0xE1)
+
+
+class Xls(Type):
+    """
+    Implements Xls file type matcher.
+    Which looks like the Doc file 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):
+        return (len(buf) > 7 and
+                buf[0] == 0xD0 and
+                buf[1] == 0xCF and
+                buf[2] == 0x11 and
+                buf[3] == 0xE0 and
+                buf[4] == 0xA1 and
+                buf[5] == 0xB1 and
+                buf[6] == 0x1A and
+                buf[7] == 0xE1)
+
+
+# todo: Xlsx
+
+
+class Ppt(Type):
+    """
+    Implements Ppt file type matcher.
+    Which looks like the Doc and Xls file type matchers..
+    """
+
+    MIME = 'application/vnd.ms-powerpoint'
+    EXTENSION = 'ppt'
+
+    def __init__(self):
+        super(Ppt, self).__init__(
+            mime=Ppt.MIME,
+            extension=Ppt.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 7 and
+                buf[0] == 0xD0 and
+                buf[1] == 0xCF and
+                buf[2] == 0x11 and
+                buf[3] == 0xE0 and
+                buf[4] == 0xA1 and
+                buf[5] == 0xB1 and
+                buf[6] == 0x1A and
+                buf[7] == 0xE1)
diff --git a/filetype/types/image.py b/filetype/types/image.py
index 1fd6e17..7fe107d 100644
--- a/filetype/types/image.py
+++ b/filetype/types/image.py
@@ -259,21 +259,22 @@ def match(self, buf):
         return False
 
 
-class Dcm(Type):
-
-    MIME = 'application/dicom'
-    EXTENSION = 'dcm'
-    OFFSET = 128
+class Dwg(Type):
+    """
+    Implements the DWG image type matcher.
+    """
+    MIME = 'image/vnd.dwg'
+    EXTENSION = 'dwg'
 
     def __init__(self):
-        super(Dcm, self).__init__(
-            mime=Dcm.MIME,
-            extension=Dcm.EXTENSION
+        super(Dwg, self).__init__(
+            mime=Dwg.MIME,
+            extension=Dwg.EXTENSION,
         )
 
     def match(self, buf):
-        return (len(buf) > Dcm.OFFSET + 4 and
-                buf[Dcm.OFFSET + 0] == 0x44 and
-                buf[Dcm.OFFSET + 1] == 0x49 and
-                buf[Dcm.OFFSET + 2] == 0x43 and
-                buf[Dcm.OFFSET + 3] == 0x4D)
+        return (len(buf) > 3 and
+                buf[0] == 0x41 and
+                buf[1] == 0x43 and
+                buf[2] == 0x31 and
+                buf[3] == 0x30)
diff --git a/filetype/types/video.py b/filetype/types/video.py
index 9955397..2e370d1 100644
--- a/filetype/types/video.py
+++ b/filetype/types/video.py
@@ -6,27 +6,6 @@
 from .isobmff import IsoBmff
 
 
-class Mp4(IsoBmff):
-    """
-    Implements the MP4 video type matcher.
-    """
-    MIME = 'video/mp4'
-    EXTENSION = 'mp4'
-
-    def __init__(self):
-        super(Mp4, self).__init__(
-            mime=Mp4.MIME,
-            extension=Mp4.EXTENSION
-        )
-
-    def match(self, buf):
-        if not self._is_isobmff(buf):
-            return False
-
-        major_brand, minor_version, compatible_brands = self._get_ftyp(buf)
-        return major_brand in ['mp41', 'mp42']
-
-
 class M4v(Type):
     """
     Implements the M4V video type matcher.
@@ -64,6 +43,8 @@ def __init__(self):
         )
 
     def match(self, buf):
+        # todo: GO source checks first 4 bytes, then searches next 4096 for 'matroska',
+        # and finally checks if bytes 5 & 6 below are located prior to 'matroska'.
         return ((len(buf) > 15 and
                 buf[0] == 0x1A and buf[1] == 0x45 and
                 buf[2] == 0xDF and buf[3] == 0xA3 and
@@ -173,6 +154,28 @@ def match(self, buf):
                 buf[9] == 0xD9)
 
 
+class Mpeg(Type):
+    """
+    Implements the MPEG video type matcher.
+    """
+    MIME = 'video/mpeg'
+    EXTENSION = 'mpg'
+
+    def __init__(self):
+        super(Mpeg, self).__init__(
+            mime=Mpeg.MIME,
+            extension=Mpeg.EXTENSION
+        )
+
+    def match(self, buf):
+        return (len(buf) > 3 and
+                buf[0] == 0x0 and
+                buf[1] == 0x0 and
+                buf[2] == 0x1 and
+                buf[3] >= 0xb0 and
+                buf[3] <= 0xbf)
+
+
 class Flv(Type):
     """
     Implements the FLV video type matcher.
@@ -194,23 +197,46 @@ def match(self, buf):
                 buf[3] == 0x01)
 
 
-class Mpeg(Type):
+class Mp4(IsoBmff):
     """
-    Implements the MPEG video type matcher.
+    Implements the MP4 video type matcher.
     """
-    MIME = 'video/mpeg'
-    EXTENSION = 'mpg'
+    MIME = 'video/mp4'
+    EXTENSION = 'mp4'
 
     def __init__(self):
-        super(Mpeg, self).__init__(
-            mime=Mpeg.MIME,
-            extension=Mpeg.EXTENSION
+        super(Mp4, self).__init__(
+            mime=Mp4.MIME,
+            extension=Mp4.EXTENSION
         )
 
     def match(self, buf):
-        return (len(buf) > 3 and
-                buf[0] == 0x0 and
-                buf[1] == 0x0 and
-                buf[2] == 0x1 and
-                buf[3] >= 0xb0 and
-                buf[3] <= 0xbf)
+        if not self._is_isobmff(buf):
+            return False
+
+        major_brand, minor_version, compatible_brands = self._get_ftyp(buf)
+        return major_brand in ['mp41', 'mp42']
+
+
+class Match3gp(Type):
+    """
+    Implements the 3GB video type matcher.
+    """
+    MIME = 'video/3gpp'
+    EXTENSION = '3gp'
+
+    def __init__(self):
+        super(Match3gp, self).__init__(
+            mime=Match3gp.MIME,
+            extension=Match3gp.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] == 0x33 and
+                buf[9] == 0x67 and
+                buf[10] == 0x70)

From fa1d9af62e5ed776ea4c44fd3f760145d302374c Mon Sep 17 00:00:00 2001
From: flgc <abotes@gmail.com>
Date: Sun, 18 Aug 2019 12:03:15 +0900
Subject: [PATCH 2/6] bug: Replace Dcm with Dwg under IMAGES

---
 filetype/types/__init__.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/filetype/types/__init__.py b/filetype/types/__init__.py
index 87d95b8..8457328 100644
--- a/filetype/types/__init__.py
+++ b/filetype/types/__init__.py
@@ -23,7 +23,7 @@
     image.Psd(),
     image.Ico(),
     image.Heic(),
-    image.Dcm(),
+    image.Dwg(),
 )
 
 # Supported video types

From 08bee2021bd1f938b0b19989e8627e58b5bcc281 Mon Sep 17 00:00:00 2001
From: flgc <abotes@gmail.com>
Date: Sun, 18 Aug 2019 12:19:09 +0900
Subject: [PATCH 3/6] bug: Update __init__ for new types

---
 filetype/types/__init__.py | 27 ++++++++++++++++++++++++---
 filetype/types/archive.py  |  3 ---
 filetype/types/document.py | 27 ---------------------------
 3 files changed, 24 insertions(+), 33 deletions(-)

diff --git a/filetype/types/__init__.py b/filetype/types/__init__.py
index 8457328..4836c20 100644
--- a/filetype/types/__init__.py
+++ b/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
@@ -28,15 +30,16 @@
 
 # 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)
diff --git a/filetype/types/archive.py b/filetype/types/archive.py
index 8a5eff7..2e10ec9 100644
--- a/filetype/types/archive.py
+++ b/filetype/types/archive.py
@@ -536,7 +536,6 @@ def match(self, buf):
                 buf[3] == 0xDB)
 
 
-
 class Elf(Type):
     """
     Implements the Elf archive type matcher.
@@ -558,7 +557,6 @@ def match(self, buf):
                 buf[3] == 0x46)
 
 
-
 class Dcm(Type):
     """
     Implements the Dcm archive type matcher.
@@ -580,7 +578,6 @@ def match(self, buf):
                 buf[131] == 0x4D)
 
 
-
 class Iso(Type):
     """
     Implements the ISO archive type matcher.
diff --git a/filetype/types/document.py b/filetype/types/document.py
index afad633..aea313b 100644
--- a/filetype/types/document.py
+++ b/filetype/types/document.py
@@ -34,33 +34,6 @@ def match(self, buf):
 # todo: Docx
 
 
-class Xls(Type):
-    """
-    Implements Xls file type matcher.
-    Which looks just like the Doc file 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):
-        return (len(buf) > 7 and
-                buf[0] == 0xD0 and
-                buf[1] == 0xCF and
-                buf[2] == 0x11 and
-                buf[3] == 0xE0 and
-                buf[4] == 0xA1 and
-                buf[5] == 0xB1 and
-                buf[6] == 0x1A and
-                buf[7] == 0xE1)
-
-
 class Xls(Type):
     """
     Implements Xls file type matcher.

From e65c1117fdd0bbc3c6e7b4e8c545263568dda0ee Mon Sep 17 00:00:00 2001
From: flgc <abotes@gmail.com>
Date: Sun, 18 Aug 2019 12:28:03 +0900
Subject: [PATCH 4/6] bug: Import types in match.py

---
 filetype/match.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/filetype/match.py b/filetype/match.py
index f1d9891..8a60802 100644
--- a/filetype/match.py
+++ b/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

From 5091580fe03b38f63084c6ca0276d2f291c812d9 Mon Sep 17 00:00:00 2001
From: flgc <abotes@gmail.com>
Date: Sun, 18 Aug 2019 12:37:09 +0900
Subject: [PATCH 5/6] enh: Add new helpers and matchers for new types

---
 filetype/helpers.py | 32 ++++++++++++++++++++++++++++++++
 filetype/match.py   | 34 ++++++++++++++++++++++++++++++++++
 2 files changed, 66 insertions(+)

diff --git a/filetype/helpers.py b/filetype/helpers.py
index 9aafa29..a3e170f 100644
--- a/filetype/helpers.py
+++ b/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
diff --git a/filetype/match.py b/filetype/match.py
index 8a60802..de2d1fe 100644
--- a/filetype/match.py
+++ b/filetype/match.py
@@ -119,3 +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)

From 1937d2b22b6852655e79f40550306a5e7282e9c5 Mon Sep 17 00:00:00 2001
From: flgc <abotes@gmail.com>
Date: Sun, 18 Aug 2019 12:46:23 +0900
Subject: [PATCH 6/6] chores: Add new detected file types to docs

---
 README.rst | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)

diff --git a/README.rst b/README.rst
index 529ac4a..b57a54e 100644
--- a/README.rst
+++ b/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