Refs #1: added gzip container as reference

atrcopy/__init__.py

@@ -58,13 +58,16 @@ def find_diskimage(filename):
         with open(filename, "rb") as fh:
             if options.verbose:
                 print("Loading file %s" % filename)
-            rawdata = SegmentData(fh.read())
+            data = to_numpy(fh.read())
             parser = None
             try:
-                unpacked = guess_container(rawdata.data)
+                container = guess_container(data, options.verbose)
+                if container is not None:
+                    data = container.unpacked
             except errors.UnsupportedContainer as e:
                 print(f"{filename}: {e}")
                 return None
+            rawdata = SegmentData(data)
             for mime in mime_parse_order:
                 if options.verbose:
                     print("Trying MIME type %s" % mime)

atrcopy/container.py

@@ -1,12 +1,32 @@
+import gzip
+import io
+
 import numpy as np
 
 from . import errors
 from .segments import SegmentData
+from .utils import to_numpy
 
 
 class DiskImageContainer:
     def __init__(self, data):
-        self.unpacked = self.unpack_raw_data(data)
+        self.unpacked = self.__unpack_raw_data(data)
 
-    def unpack_raw_data(self, data):
+    def __unpack_raw_data(self, data):
+        raw = data.tobytes()
+        unpacked = self.unpack_bytes(raw)
+        return to_numpy(unpacked)
+
+    def unpack_bytes(self, byte_data):
         pass
+
+
+class GZipContainer(DiskImageContainer):
+    def unpack_bytes(self, byte_data):
+        try:
+            buf = io.BytesIO(byte_data)
+            with gzip.GzipFile(mode='rb', fileobj=buf) as f:
+                unpacked = f.read()
+        except OSError as e:
+            raise errors.InvalidContainer(e)
+        return unpacked

atrcopy/dcm.py

@@ -21,7 +21,7 @@ class DCMContainer(DiskImageContainer):
             self.index += 1
         return data
 
-    def unpack_raw_data(self, data):
+    def unpack_bytes(self, data):
         self.index = 0
         self.count = len(data)
         self.raw = data

atrcopy/parsers.py

@@ -10,6 +10,7 @@ from .dos33 import Dos33DiskImage, ProdosDiskImage, Dos33BinFile
 from .standard_delivery import StandardDeliveryImage
 from . import errors
 from .magic import guess_detail_for_mime
+from .container import GZipContainer
 from .dcm import DCMContainer
 
 import logging
@@ -159,17 +160,22 @@ class ProdosSegmentParser(SegmentParser):
 
 
 known_containers = [
+    GZipContainer,
     DCMContainer,
 ]
 
 
-def guess_container(r):
+def guess_container(r, verbose=False):
     for c in known_containers:
+        if verbose:
+            log.info(f"trying container {c}")
         try:
             found = c(r)
-        except errors.InvalidContainer:
+        except errors.InvalidContainer as e:
             continue
         else:
+            if verbose:
+                log.info(f"found container {c}")
             return found
     return None