add possibility to uncompress bz2 files - fix #25

Snakefile

@@ -2,6 +2,7 @@ import os
 import shutil
 import gzip
 import json
+import bz2
 from copy import deepcopy
 
 

rules/Preprocessing/master.rules

@@ -8,6 +8,9 @@ include:
     "MT.rules"
 
 
+CHUNK_SIZE = 100 * 1024
+
+
 def preprocessing_output_files():
     """
     Dynamically generate output files names based on parameters
@@ -74,10 +77,15 @@ def prepare_input_files(input, rtype):
         print(inp, '=>', outfilename)
         # ungunzip
         if os.path.splitext(fname)[-1] in ['.gz', '.gzip']:
-            with open(outfilename, 'wb') as whandle:
-                with gzip.open(inp, 'rb') as rhandle:
-                    whandle.write(rhandle.read())
-                    # copy
+            with open(outfilename, 'wb') as whandle, gzip.open(inp, 'rb') as rhandle:
+                whandle.write(rhandle.read())
+        # unbzip2
+        elif os.path.splitext(fname)[-1] in ['.bz2', '.bzip2']:
+            dec = bz2.BZ2Decompressor()
+            with open(outfilename, 'wb') as whandle, open(inp, 'rb') as rhandle:
+                for data in iter(lambda: rhandle.read(CHUNK_SIZE), b''):
+                    whandle.write(dec.decompress(data))
+        # copy
         else:
             shutil.copy(inp, outfilename)