14539: Persists empty directories by adding a '\056' empty file.

Also:
* Fixes test
* Errors out when trying to open a file/dir with '\056' on its path.

Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <ldipentima@veritasgenetics.com>

diff --git a/sdk/python/arvados/collection.py b/sdk/python/arvados/collection.py
index 627f0346db2c6760710db3edaf356f4cb724bf91..c2517c618923c6f6e2cd36961f2d358a5daf792b 100644 (file)
--- a/sdk/python/arvados/collection.py
+++ b/sdk/python/arvados/collection.py
@@ -600,6 +600,9 @@ class RichCollectionBase(CollectionBase):
 
         pathcomponents = path.split("/", 1)
         if pathcomponents[0]:
+            # Don't allow naming files/dirs \\056
+            if pathcomponents[0] == "\\056":
+                raise IOError(errno.EINVAL, "Invalid name", pathcomponents[0])
             item = self._items.get(pathcomponents[0])
             if len(pathcomponents) == 1:
                 if item is None:
@@ -1058,7 +1061,9 @@ class RichCollectionBase(CollectionBase):
             if stream:
                 buf.append(" ".join(normalize_stream(stream_name, stream)) + "\n")
             for dirname in [s for s in sorted_keys if isinstance(self[s], RichCollectionBase)]:
-                buf.append(self[dirname].manifest_text(stream_name=os.path.join(stream_name, dirname), strip=strip, normalize=True, only_committed=only_committed))
+                buf.append(self[dirname].manifest_text(
+                    stream_name=os.path.join(stream_name, dirname),
+                    strip=strip, normalize=True, only_committed=only_committed))
             return "".join(buf)
         else:
             if strip:
@@ -1833,6 +1838,15 @@ class Subcollection(RichCollectionBase):
         self.name = newname
         self.lock = self.parent.root_collection().lock
 
+    @synchronized
+    def _get_manifest_text(self, stream_name, strip, normalize, only_committed=False):
+        """Encode empty directories by using an \056-named (".") empty file"""
+        if len(self._items) == 0:
+            return "%s %s 0:0:\\056\n" % (stream_name, config.EMPTY_BLOCK_LOCATOR)
+        return super(Subcollection, self)._get_manifest_text(stream_name,
+                                                             strip, normalize,
+                                                             only_committed)
+
 
 class CollectionReader(Collection):
     """A read-only collection object.

diff --git a/sdk/python/tests/test_collections.py b/sdk/python/tests/test_collections.py
index de01006741e91b12047f70d6b82dfff04f80bfdc..b5c0c1c524968322779c9663bbd0e1c1c3ff2162 100644 (file)
--- a/sdk/python/tests/test_collections.py
+++ b/sdk/python/tests/test_collections.py
@@ -955,7 +955,7 @@ class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
     def test_remove_in_subdir(self):
         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
         c.remove("foo/count2.txt")
-        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
+        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n", c.portable_manifest_text())
 
     def test_remove_empty_subdir(self):
         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')