Working decompression for gzip files

Author: rhpvorderman

src/isal/igzip.py

@@ -24,6 +24,7 @@
 import argparse
 import io
 import os
+import _compression
 
 # We import a copy of the gzip module so we can use that to define the classes
 # here and replace  the zlib module with the isal_zlib module. This way the
@@ -142,15 +143,12 @@ def write(self,data):
         return length
 
 
-class _IGzipReader(gzip._GzipReader):
+class _IGzipReader(_compression.DecompressReader):
     def __init__(self, fp):
-        super().__init__(fp)
-        self._decompfactory = isal_zlib.decompressobj
-        self._decompressor = self._decompfactory(**self._decomp_args)
+        super().__init__(fp, isal_zlib.decompressobj,
+                         wbits=16 + isal_zlib.MAX_WBITS)
 
-    def _add_read_data(self, data):
-        self._crc = isal_zlib.crc32(data, self._crc)
-        self._stream_size += len(data)
+    
 
 # Plagiarized from gzip.py from python's stdlib.
 def compress(data, compresslevel=_COMPRESS_LEVEL_BEST, *, mtime=None):

src/isal/isal_zlib.pyx

@@ -353,6 +353,7 @@ cdef class Decompress:
     cdef public bytes unused_data
     cdef public unconsumed_tail
     cdef public bint eof
+    cdef public bint needs_input
     cdef bint is_initialised
     cdef inflate_state stream
     cdef unsigned char * obuf
@@ -379,6 +380,7 @@ cdef class Decompress:
         self.unconsumed_tail = b""
         self.eof = 0
         self.is_initialised = 1
+        self.needs_input = 1
 
     def __dealloc__(self):
         if self.obuf is not NULL:
@@ -396,6 +398,8 @@ cdef class Decompress:
         elif max_length < 0:
             raise ValueError("max_length can not be smaller than 0")
 
+        if data == b"":
+            data = self.unconsumed_tail
         cdef Py_ssize_t total_length = len(data)
         if total_length > UINT32_MAX:
             # Zlib allows a maximum of 64 KB (16-bit length) and python has
@@ -454,6 +458,7 @@ cdef class Decompress:
             # 2. All input data was consumed. Clear unconsumed_tail.
             unused_bytes = self.stream.avail_in
             self.unconsumed_tail = data[-unused_bytes:]
+            self.needs_input = 0 if unused_bytes > 0 else 1
         return b"".join(out)
 
     def flush(self, Py_ssize_t length = DEF_BUF_SIZE):