Byte/string handling fixes for Python 3.

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

diff --git a/python/cumulus/__init__.py b/python/cumulus/__init__.py
index 8dc4c98..6e03bd2 100644 (file)
--- a/python/cumulus/__init__.py
+++ b/python/cumulus/__init__.py
@@ -26,19 +26,30 @@ various parts of a Cumulus archive:
   - reading and maintaining the local object database
 """
 
+from __future__ import division, print_function, unicode_literals
 
+import codecs
 import hashlib
 import itertools
 import os
 import re
 import sqlite3
+import sys
 import tarfile
 import tempfile
-import _thread
+try:
+    import _thread
+except ImportError:
+    import thread as _thread
 
 import cumulus.store
 import cumulus.store.file
 
+if sys.version < "3":
+    StringTypes = (str, unicode)
+else:
+    StringTypes = (str,)
+
 # The largest supported snapshot format that can be understood.
 FORMAT_VERSION = (0, 11)        # Cumulus Snapshot v0.11
 
@@ -58,6 +69,12 @@ 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))
@@ -264,11 +281,8 @@ class BackendWrapper(object):
 
         store may either be a Store object or URL.
         """
-        if type(backend) in (str, str):
-            if backend.find(":") >= 0:
-                self._backend = cumulus.store.open(backend)
-            else:
-                self._backend = cumulus.store.file.FileStore(backend)
+        if type(backend) in StringTypes:
+            self._backend = cumulus.store.open(backend)
         else:
             self._backend = backend
 
@@ -295,7 +309,7 @@ class BackendWrapper(object):
     def prefetch_generic(self):
         """Calls scan on directories to prefetch file metadata."""
         directories = set()
-        for typeinfo in list(SEARCH_PATHS.values()):
+        for typeinfo in SEARCH_PATHS.values():
             directories.update(typeinfo.directories())
         for d in directories:
             print("Prefetch", d)
@@ -359,7 +373,7 @@ 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):
@@ -506,7 +520,7 @@ 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)
 
@@ -566,7 +580,7 @@ class MetadataItem:
     @staticmethod
     def decode_device(s):
         """Decode a device major/minor number."""
-        (major, minor) = list(map(MetadataItem.decode_int, s.split("/")))
+        (major, minor) = map(MetadataItem.decode_int, s.split("/"))
         return (major, minor)
 
     class Items: pass
@@ -578,7 +592,7 @@ class MetadataItem:
         self.object_store = object_store
         self.keys = []
         self.items = self.Items()
-        for (k, v) in list(fields.items()):
+        for (k, v) in fields.items():
             if k in self.field_types:
                 decoder = self.field_types[k]
                 setattr(self.items, k, decoder(v))