Fix issue 68

bagit.py

@@ -766,9 +766,9 @@ def _make_manifest(manifest_file, data_dir, processes, algorithm='md5'):
 def _make_tagmanifest_file(alg, bag_dir):
     tagmanifest_file = join(bag_dir, "tagmanifest-%s.txt" % alg)
     LOGGER.info("writing %s", tagmanifest_file)
-    files = [f for f in listdir(bag_dir) if isfile(join(bag_dir, f))]
+
     checksums = []
-    for f in files:
+    for f in _find_tag_files(bag_dir):
         if re.match(r'^tagmanifest-.+\.txt$', f):
             continue
         with open(join(bag_dir, f), 'rb') as fh:
@@ -784,6 +784,15 @@ def _make_tagmanifest_file(alg, bag_dir):
         for digest, filename in checksums:
             tagmanifest.write('%s %s\n' % (digest, filename))
 
+def _find_tag_files(bag_dir):
+    for dir_name, _, filenames in os.walk(bag_dir):
+        if not re.match(r'.*data$', dir_name):
+            for filename in filenames:
+                if filename.startswith('tagmanifest-'):
+                    continue
+                #remove everything up to the bag_dir directory
+                p = join(dir_name, filename)
+                yield os.path.relpath(p, bag_dir)
 
 def _walk(data_dir):
     for dirpath, dirnames, filenames in os.walk(data_dir):

test.py

@@ -268,6 +268,37 @@ def test_validate_optional_tagfile(self):
         bag = bagit.Bag(self.tmpdir)
         self.assertRaises(bagit.BagValidationError, self.validate, bag)
 
+    def test_validate_optional_tagfile_in_directory(self):
+        bag = bagit.make_bag(self.tmpdir)
+        tagdir = tempfile.mkdtemp(dir=self.tmpdir)
+
+        if not os.path.exists(j(tagdir, "tagfolder")):
+            os.makedirs(j(tagdir, "tagfolder"))
+
+        with open(j(tagdir, "tagfolder", "tagfile"), "w") as tagfile:
+            tagfile.write("test")
+        relpath = j(tagdir, "tagfolder", "tagfile").replace(self.tmpdir + os.sep, "")
+        relpath.replace("\\", "/")
+        with open(j(self.tmpdir, "tagmanifest-md5.txt"), "w") as tagman:
+            # Incorrect checksum.
+            tagman.write("8e2af7a0143c7b8f4de0b3fc90f27354 " + relpath + "\n")
+        bag = bagit.Bag(self.tmpdir)
+        self.assertRaises(bagit.BagValidationError, self.validate, bag)
+
+        hasher = hashlib.new("md5")
+        with open(j(tagdir, "tagfolder", "tagfile"), "r") as tf:
+            contents = tf.read().encode('utf-8')
+        hasher.update(contents)
+        with open(j(self.tmpdir, "tagmanifest-md5.txt"), "w") as tagman:
+            tagman.write(hasher.hexdigest() + " " + relpath + "\n")
+        bag = bagit.Bag(self.tmpdir)
+        self.assertTrue(self.validate(bag))
+
+        # Missing tagfile.
+        os.remove(j(tagdir, "tagfolder", "tagfile"))
+        bag = bagit.Bag(self.tmpdir)
+        self.assertRaises(bagit.BagValidationError, self.validate, bag)
+
     def test_sha1_tagfile(self):
         info = {'Bagging-Date': '1970-01-01', 'Contact-Email': 'ehs@pobox.com'}
         bag = bagit.make_bag(self.tmpdir, checksum=['sha1'], bag_info=info)