actually use the offset in CompressedFPSource

Author: mmohrhard

bloscpack/append.py

@@ -112,11 +112,9 @@ def append_fp(original_fp, new_content_fp, new_size, blosc_args=None):
                   (METADATA_HEADER_LENGTH + metadata_header['max_meta_size'] +
                       CHECKSUMS_LOOKUP[metadata_header['meta_checksum']].size
                    if metadata is not None else 0))
-    # seek to the final offset
-    original_fp.seek(offsets[-1], 0)
     # decompress the last chunk
     compressed, blosc_header, digest = _read_compressed_chunk_fp(original_fp,
-                                                         checksum_impl)
+                                                         checksum_impl, offsets[-1])
     # TODO check digest
     decompressed = blosc.decompress(compressed)
     # figure out how many bytes we need to read to rebuild the last chunk

bloscpack/cli.py

@@ -511,7 +511,7 @@ def main():
                 checksum_impl = bloscpack_header.checksum_impl
                 # get the header of the first chunk
                 _, blosc_header, _ = _read_compressed_chunk_fp(
-                    fp, checksum_impl)
+                    fp, checksum_impl, offsets[0])
         except ValueError as ve:
             log.error(str(ve) + "\n" +
                       "This might not be a bloscpack compressed file.")

bloscpack/file_io.py

@@ -298,7 +298,7 @@ def _write_offsets(output_fp, offsets):
     output_fp.write(encoded_offsets)
 
 
-def _read_compressed_chunk_fp(input_fp, checksum_impl):
+def _read_compressed_chunk_fp(input_fp, checksum_impl, offset):
     """ Read a compressed chunk from a file pointer.
 
     Parameters
@@ -318,6 +318,7 @@ def _read_compressed_chunk_fp(input_fp, checksum_impl):
         the checksum of the chunk
     """
     # read blosc header
+    input_fp.seek(offset)
     blosc_header_raw = input_fp.read(BLOSC_HEADER_LENGTH)
     blosc_header = decode_blosc_header(blosc_header_raw)
     if log.LEVEL == log.DEBUG:
@@ -365,7 +366,7 @@ def __init__(self, input_fp):
 
     def __iter__(self):
         for i in xrange(self.nchunks):
-            compressed, header, digest = _read_compressed_chunk_fp(self.input_fp, self.checksum_impl)
+            compressed, header, digest = _read_compressed_chunk_fp(self.input_fp, self.checksum_impl, self.offsets[i])
             yield compressed, digest
 
 

test/test_append.py

@@ -366,14 +366,12 @@ def test_append_mix_shuffle():
     # now get the first and the last chunk and check that the shuffle doesn't
     # match
     bloscpack_header, offsets = reset_read_beginning(orig)[0:4:3]
-    orig.seek(offsets[0])
     checksum_impl = CHECKSUMS_LOOKUP[bloscpack_header['checksum']]
     compressed_zero,  blosc_header_zero, digest = \
-        _read_compressed_chunk_fp(orig, checksum_impl)
+        _read_compressed_chunk_fp(orig, checksum_impl, offsets[0])
     decompressed_zero = blosc.decompress(compressed_zero)
-    orig.seek(offsets[-1])
     compressed_last,  blosc_header_last, digest = \
-        _read_compressed_chunk_fp(orig, checksum_impl)
+        _read_compressed_chunk_fp(orig, checksum_impl, offsets[-1])
     decompressed_last = blosc.decompress(compressed_last)
     # first chunk has shuffle active
     nt.assert_equal(blosc_header_zero['flags'], 1)