tests: git: test larger blob encode/decode

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

diff --git a/lib/bup/hashsplit.py b/lib/bup/hashsplit.py
index 33e91a64ba0830cb4d7e837312d54708b314efcb..dc3a538fe52d329537a3021c289dca16c9692de7 100644 (file)
--- a/lib/bup/hashsplit.py
+++ b/lib/bup/hashsplit.py
@@ -1,8 +1,12 @@
-import math, os
 
-from bup import _helpers, helpers
+from __future__ import absolute_import
+import io, math, os
+
+from bup import _helpers, compat, helpers
+from bup.compat import buffer, join_bytes
 from bup.helpers import sc_page_size
 
+
 _fmincore = getattr(helpers, 'fmincore', None)
 
 BLOB_MAX = 8192*4   # 8192 is the "typical" blob size for bupsplit
@@ -11,22 +15,21 @@ 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
+            self.data = join_bytes(buffer(self.data, self.start), s)
             self.start = 0
             
     def peek(self, count):
@@ -36,7 +39,10 @@ class Buf:
         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
 
@@ -95,13 +101,17 @@ def readfile_iter(files, progress=None):
         b = ''
         fd = rpr = rstart = rlen = None
         if _fmincore and hasattr(f, 'fileno'):
-            fd = f.fileno()
-            mcore = _fmincore(fd)
-            if mcore:
-                max_chunk = max(1, (8 * 1024 * 1024) / sc_page_size)
-                rpr = _nonresident_page_regions(mcore, helpers.MINCORE_INCORE,
-                                                max_chunk)
-                rstart, rlen = next(rpr, (None, None))
+            try:
+                fd = f.fileno()
+            except io.UnsupportedOperation:
+                pass
+            if fd:
+                mcore = _fmincore(fd)
+                if mcore:
+                    max_chunk = max(1, (8 * 1024 * 1024) / sc_page_size)
+                    rpr = _nonresident_page_regions(mcore, helpers.MINCORE_INCORE,
+                                                    max_chunk)
+                    rstart, rlen = next(rpr, (None, None))
         while 1:
             if progress:
                 progress(filenum, len(b))
@@ -183,10 +193,10 @@ 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)
@@ -234,7 +244,7 @@ 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))