From 2b9527602988e621096200b7c0be6485450efd8a Mon Sep 17 00:00:00 2001
From: Rob McMullen <feedback@playermissile.com>
Date: Mon, 30 Apr 2018 22:28:47 -0700
Subject: [PATCH] Added magic signature detection * appends detail to MIME type
 if match; e.g. application/vnd.atari8bit.atr.jumpman

---
 atrcopy/__init__.py |  4 +++
 atrcopy/magic.py    | 73 +++++++++++++++++++++++++++++++++++++++++++++
 atrcopy/parsers.py  |  3 ++
 3 files changed, 80 insertions(+)
 create mode 100644 atrcopy/magic.py

diff --git a/atrcopy/__init__.py b/atrcopy/__init__.py
index 28e416a..66c3a5c 100644
--- a/atrcopy/__init__.py
+++ b/atrcopy/__init__.py
@@ -24,6 +24,7 @@ from .segments import SegmentData, SegmentSaver, DefaultSegment, EmptySegment, O
 from .spartados import SpartaDosDiskImage
 from .cartridge import A8CartHeader, AtariCartImage
 from .parsers import SegmentParser, DefaultSegmentParser, guess_parser_for_mime, guess_parser_for_system, iter_parsers, iter_known_segment_parsers, mime_parse_order, parsers_for_filename
+from .magic import guess_detail_for_mime
 from .utils import to_numpy, text_to_int
 
 
@@ -70,6 +71,9 @@ def find_diskimage(filename):
                     continue
                 if options.verbose:
                     print("Found parser %s" % parser.menu_name)
+                mime2 = guess_detail_for_mime(mime, rawdata, parser)
+                if mime != mime2 and options.verbose:
+                    print("Signature match: %s" % mime2)
                 break
             if parser is None:
                 print("%s: Unknown disk image type" % filename)
diff --git a/atrcopy/magic.py b/atrcopy/magic.py
new file mode 100644
index 0000000..c72fc37
--- /dev/null
+++ b/atrcopy/magic.py
@@ -0,0 +1,73 @@
+import numpy as np
+
+import logging
+log = logging.getLogger(__name__)
+
+
+magic = [
+    {'mime': "application/vnd.atari8bit.atr.getaway_pd",
+     'name': "Getaway Public Domain ATR",
+     'signature': [
+         (slice(8, 10), [0x82, 0x39]),
+         (slice(12, 16), [0x67, 0x21, 0x70, 0x64]),
+         ],
+    },
+
+    {'mime': "application/vnd.atari8bit.xex.getaway",
+     'name': "Getaway XEX",
+     'signature': [
+         (slice(0, 6), [0xff, 0xff, 0x80, 0x2a, 0xff, 0x8a]),
+         ],
+    },
+
+    {'mime': "application/vnd.atari8bit.atr.getaway",
+     'name': "Getaway ATR",
+     'signature': [
+         (slice(0x10, 0x19), [0x00, 0xc1, 0x80, 0x0f, 0xcc, 0x22, 0x18, 0x60, 0x0e]),
+         ],
+    },
+
+    {'mime': "application/vnd.atari8bit.atr.jumpman_level_tester",
+     'name': "Jumpman Level Tester from Omnivore",
+     'signature': [
+         (slice(0, 5), [0x96, 0x02 , 0xd0 , 0x05 , 0x80]),
+         (0x0196 + 0x3f, 0x4c),
+         (0x0196 + 0x48, 0x20),
+         (0x0196 + 0x4b, 0x60),
+         (0x0196 + 0x4c, 0xff),
+         ],
+    },
+
+    {'mime': "application/vnd.atari8bit.atr.jumpman",
+     'name': "Jumpman",
+     'signature': [
+         (slice(0, 5), [0x96, 0x02 , 0x80 , 0x16 , 0x80]),
+         (0x0810 + 0x3f, 0x4c),
+         (0x0810 + 0x48, 0x20),
+         (0x0810 + 0x4b, 0x60),
+         (0x0810 + 0x4c, 0xff),
+         ],
+    },
+]
+
+
+def check_signature(raw, sig):
+    for index, expected in sig:
+        actual = raw.data[index].tolist()
+        if actual == expected:
+            log.debug(" match at %s: %s" % (str(index), str(expected)))
+        if actual != expected:
+            log.debug(" failed at %s: %s != %s" % (str(index), str(expected), str(raw.data[index])))
+            return False
+    return True
+
+
+def guess_detail_for_mime(mime, raw, parser):
+    for entry in magic:
+        if entry['mime'].startswith(mime):
+            log.debug("checking signature for %s" % entry['mime'])
+            if check_signature(raw, entry['signature']):
+                log.debug("found signature: %s" % entry['name'])
+                return entry['mime']
+    return mime
+
diff --git a/atrcopy/parsers.py b/atrcopy/parsers.py
index cc4823c..36a7cb4 100644
--- a/atrcopy/parsers.py
+++ b/atrcopy/parsers.py
@@ -12,6 +12,7 @@ from .mame import MameZipImage
 from .dos33 import Dos33DiskImage, ProdosDiskImage, Dos33BinFile
 from .standard_delivery import StandardDeliveryImage
 from .errors import *
+from .magic import guess_detail_for_mime
 
 import logging
 log = logging.getLogger(__name__)
@@ -178,6 +179,7 @@ def guess_parser_for_system(mime_base, r):
         if mime.startswith(mime_base):
             p = guess_parser_for_mime(mime, r)
             if p is not None:
+                mime = guess_detail_for_mime(mime, r, p)
                 return mime, p
     return None, None
 
@@ -186,6 +188,7 @@ def iter_parsers(r):
     for mime in mime_parse_order:
         p = guess_parser_for_mime(mime, r)
         if p is not None:
+            mime = guess_detail_for_mime(mime, r, p)
             return mime, p
     return None, None