test hashing files larger than blocksize

Co-authored-by: Ace Medlock <ace.medlock@gmail.com>
Co-authored-by: Kendrick Shaw <kms15@case.edu>

Author: 3 people

meshfssync/__init__.py

@@ -2,18 +2,17 @@
 
 import hashlib
 
-def sha256_for_path(fp, blocksize = (4 * 1024 * 1024)):
+def sha256_for_file(fp, blocksize = (4 * 1024 * 1024)):
     hasher = hashlib.sha256()
     buf = fp.read(blocksize)
-    # while len(buf) > 0:
-    assert len(buf) < blocksize
-    hasher.update(buf)
-    #    buf = fp.read(blocksize)
+    while len(buf) > 0:
+        hasher.update(buf)
+        buf = fp.read(blocksize)
     return hasher.hexdigest()
 
 def get_dir_state(dirpath):
     ds = {}
     for path in dirpath.iterdir():
         with open(path, 'rb') as fp:
-            ds[path.name] = { 'sha256': sha256_for_path(fp) }
+            ds[path.name] = { 'sha256': sha256_for_file(fp) }
     return ds

test/test_init.py

@@ -2,8 +2,18 @@
 import tempfile
 from pathlib import Path
 
-
-
+def test_sha_for_file():
+    # $ sha256sum COPYING
+    # 8ceb4b9ee5adedde47b31e975c1d90c73ad27b6b165a1dcd80c7c545eb65b903 COPYING
+    sha = '8ceb4b9ee5adedde47b31e975c1d90c73ad27b6b165a1dcd80c7c545eb65b903'
+    blocksize_4k = (4 * 1024)
+    with open('./COPYING', 'rb') as fp:
+        class WrappedFp:
+            def read(self, blocksize):
+                assert blocksize <= blocksize_4k
+                return fp.read(blocksize_4k);
+        h = meshfssync.sha256_for_file(WrappedFp(), blocksize_4k)
+        assert h == sha
 
 def test_get_dir_state():
     # $ echo 'foo' > foo.txt