Use the subprocess module, not os.popen2, for Python 3 compatibility.

[cumulus.git] / python / cumulus / __init__.py

diff --git a/python/cumulus/__init__.py b/python/cumulus/__init__.py
index 4b7bd6c..c5961c7 100644 (file)
--- a/python/cumulus/__init__.py
+++ b/python/cumulus/__init__.py
@@ -28,11 +28,13 @@ various parts of a Cumulus archive:
 
 from __future__ import division, print_function, unicode_literals
 
+import codecs
 import hashlib
 import itertools
 import os
 import re
 import sqlite3
+import subprocess
 import sys
 import tarfile
 import tempfile
@@ -44,7 +46,7 @@ except ImportError:
 import cumulus.store
 import cumulus.store.file
 
-if sys.version < '3':
+if sys.version < "3":
     StringTypes = (str, unicode)
 else:
     StringTypes = (str,)
@@ -68,6 +70,12 @@ SEGMENT_FILTERS = [
     ("", None),
 ]
 
+def to_lines(data):
+    """Decode binary data from a file into a sequence of lines.
+
+    Newline markers are retained."""
+    return list(codecs.iterdecode(data.splitlines(True), "utf-8"))
+
 def uri_decode(s):
     """Decode a URI-encoded (%xx escapes) string."""
     def hex_decode(m): return chr(int(m.group(1), 16))
@@ -366,13 +374,15 @@ class CumulusStore:
 
     def load_snapshot(self, snapshot):
         snapshot_file = self.backend.open_snapshot(snapshot)[0]
-        return snapshot_file.read().splitlines(True)
+        return to_lines(snapshot_file.read())
 
     @staticmethod
     def filter_data(filehandle, filter_cmd):
         if filter_cmd is None:
             return filehandle
-        (input, output) = os.popen2(filter_cmd)
+        p = subprocess.Popen(filter_cmd, shell=True, stdin=subprocess.PIPE,
+                             stdout=subprocess.PIPE, close_fds=True)
+        input, output = p.stdin, p.stdout
         def copy_thread(src, dst):
             BLOCK_SIZE = 4096
             while True:
@@ -381,6 +391,7 @@ class CumulusStore:
                 dst.write(block)
             src.close()
             dst.close()
+            p.wait()
         _thread.start_new_thread(copy_thread, (filehandle, input))
         return output
 
@@ -513,7 +524,7 @@ def read_metadata(object_store, root):
 
     def follow_ref(refstr):
         if len(stack) >= MAX_RECURSION_DEPTH: raise OverflowError
-        lines = object_store.get(refstr).splitlines(True)
+        lines = to_lines(object_store.get(refstr))
         lines.reverse()
         stack.append(lines)