pylint: check for trailing whitespace

[bup.git] / lib / bup / hashsplit.py

diff --git a/lib/bup/hashsplit.py b/lib/bup/hashsplit.py
index ddf8d0c9fa8002418381bbb8802d83d4026681c3..ff00e54b3cd46b4fda1b30827e03762c7ebce025 100644 (file)
--- a/lib/bup/hashsplit.py
+++ b/lib/bup/hashsplit.py
@@ -1,42 +1,56 @@
+
+from __future__ import absolute_import
 import io, math, os
 
 from bup import _helpers, helpers
+from bup._helpers import cat_bytes
+from bup.compat import buffer
 from bup.helpers import sc_page_size
 
+
 _fmincore = getattr(helpers, 'fmincore', None)
 
 BLOB_MAX = 8192*4   # 8192 is the "typical" blob size for bupsplit
-BLOB_READ_SIZE = 1024*1024
+BLOB_READ_SIZE = 8 * 1024 * 1024
 MAX_PER_TREE = 256
 progress_callback = None
 fanout = 16
 
-GIT_MODE_FILE = 0100644
-GIT_MODE_TREE = 040000
-GIT_MODE_SYMLINK = 0120000
-assert(GIT_MODE_TREE != 40000)  # 0xxx should be treated as octal
+GIT_MODE_FILE = 0o100644
+GIT_MODE_TREE = 0o40000
+GIT_MODE_SYMLINK = 0o120000
 
 # The purpose of this type of buffer is to avoid copying on peek(), get(),
 # and eat().  We do copy the buffer contents on put(), but that should
 # be ok if we always only put() large amounts of data at a time.
 class Buf:
     def __init__(self):
-        self.data = ''
+        self.data = b''
         self.start = 0
 
     def put(self, s):
-        if s:
-            self.data = buffer(self.data, self.start) + s
+        if not self.data:
+            self.data = s
+            self.start = 0
+        elif s:
+            remaining = len(self.data) - self.start
+            self.data = cat_bytes(self.data, self.start, remaining,
+                                  s, 0, len(s))
             self.start = 0
-            
+
     def peek(self, count):
+        if count <= 256:
+            return self.data[self.start : self.start + count]
         return buffer(self.data, self.start, count)
-    
+
     def eat(self, count):
         self.start += count
 
     def get(self, count):
-        v = buffer(self.data, self.start, count)
+        if count <= 256:
+            v = self.data[self.start : self.start + count]
+        else:
+            v = buffer(self.data, self.start, count)
         self.start += count
         return v
 
@@ -187,10 +201,10 @@ def _make_shalist(l):
     ofs = 0
     l = list(l)
     total = sum(size for mode,sha,size, in l)
-    vlen = len('%x' % total)
+    vlen = len(b'%x' % total)
     shalist = []
     for (mode, sha, size) in l:
-        shalist.append((mode, '%0*x' % (vlen,ofs), sha))
+        shalist.append((mode, b'%0*x' % (vlen,ofs), sha))
         ofs += size
     assert(ofs == total)
     return (shalist, total)
@@ -238,7 +252,7 @@ def split_to_blob_or_tree(makeblob, maketree, files,
     if len(shalist) == 1:
         return (shalist[0][0], shalist[0][2])
     elif len(shalist) == 0:
-        return (GIT_MODE_FILE, makeblob(''))
+        return (GIT_MODE_FILE, makeblob(b''))
     else:
         return (GIT_MODE_TREE, maketree(shalist))