X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/27dc00515f48ed69b4d5e26ff64805b8cda4ccd3..9e5d7b0dc43a85fc03740b39611545f470a63eb8:/sdk/python/arvados/collection.py

diff --git a/sdk/python/arvados/collection.py b/sdk/python/arvados/collection.py
index 70341d8d68..8450bd1ca0 100644
--- a/sdk/python/arvados/collection.py
+++ b/sdk/python/arvados/collection.py
@@ -474,6 +474,7 @@ class ResumableCollectionWriter(CollectionWriter):
 ADD = "add"
 DEL = "del"
 MOD = "mod"
+TOK = "tok"
 FILE = "file"
 COLLECTION = "collection"
 
@@ -986,6 +987,8 @@ class RichCollectionBase(CollectionBase):
                     changes.extend(self[k].diff(end_collection[k], os.path.join(prefix, k), holding_collection))
                 elif end_collection[k] != self[k]:
                     changes.append((MOD, os.path.join(prefix, k), self[k].clone(holding_collection, ""), end_collection[k].clone(holding_collection, "")))
+                else:
+                    changes.append((TOK, os.path.join(prefix, k), self[k].clone(holding_collection, ""), end_collection[k].clone(holding_collection, "")))
             else:
                 changes.append((ADD, os.path.join(prefix, k), end_collection[k].clone(holding_collection, "")))
         return changes
@@ -1016,7 +1019,7 @@ class RichCollectionBase(CollectionBase):
                     # There is already local file and it is different:
                     # save change to conflict file.
                     self.copy(initial, conflictpath)
-            elif event_type == MOD:
+            elif event_type == MOD or event_type == TOK:
                 final = change[3]
                 if local == initial:
                     # Local matches the "initial" item so it has not
@@ -1168,6 +1171,7 @@ class Collection(RichCollectionBase):
         self._manifest_locator = None
         self._manifest_text = None
         self._api_response = None
+        self._past_versions = set()
 
         self.lock = threading.RLock()
         self.events = None
@@ -1197,6 +1201,10 @@ class Collection(RichCollectionBase):
     def writable(self):
         return True
 
+    @synchronized
+    def known_past_version(self, modified_at_and_portable_data_hash):
+        return modified_at_and_portable_data_hash in self._past_versions
+
     @synchronized
     @retry_method
     def update(self, other=None, num_retries=None):
@@ -1206,6 +1214,15 @@ class Collection(RichCollectionBase):
             if self._manifest_locator is None:
                 raise errors.ArgumentError("`other` is None but collection does not have a manifest_locator uuid")
             response = self._my_api().collections().get(uuid=self._manifest_locator).execute(num_retries=num_retries)
+            if (self.known_past_version((response.get("modified_at"), response.get("portable_data_hash"))) and
+                response.get("portable_data_hash") != self.portable_data_hash()):
+                # The record on the server is different from our current one, but we've seen it before,
+                # so ignore it because it's already been merged.
+                # However, if it's the same as our current record, proceed with the update, because we want to update
+                # our tokens.
+                return
+            else:
+                self._past_versions.add((response.get("modified_at"), response.get("portable_data_hash")))
             other = CollectionReader(response["manifest_text"])
         baseline = CollectionReader(self._manifest_text)
         self.apply(baseline.diff(other))
@@ -1233,6 +1250,10 @@ class Collection(RichCollectionBase):
             self._block_manager = _BlockManager(self._my_keep())
         return self._block_manager
 
+    def _remember_api_response(self, response):
+        self._api_response = response
+        self._past_versions.add((response.get("modified_at"), response.get("portable_data_hash")))
+
     def _populate_from_api_server(self):
         # As in KeepClient itself, we must wait until the last
         # possible moment to instantiate an API client, in order to
@@ -1242,9 +1263,9 @@ class Collection(RichCollectionBase):
         # clause, just like any other Collection lookup
         # failure. Return an exception, or None if successful.
         try:
-            self._api_response = self._my_api().collections().get(
+            self._remember_api_response(self._my_api().collections().get(
                 uuid=self._manifest_locator).execute(
-                    num_retries=self.num_retries)
+                    num_retries=self.num_retries))
             self._manifest_text = self._api_response['manifest_text']
             return None
         except Exception as e:
@@ -1401,11 +1422,11 @@ class Collection(RichCollectionBase):
                 self.update()
 
             text = self.manifest_text(strip=False)
-            self._api_response = self._my_api().collections().update(
+            self._remember_api_response(self._my_api().collections().update(
                 uuid=self._manifest_locator,
                 body={'manifest_text': text}
                 ).execute(
-                    num_retries=num_retries)
+                    num_retries=num_retries))
             self._manifest_text = self._api_response["manifest_text"]
             self.set_committed()
 
@@ -1452,7 +1473,7 @@ class Collection(RichCollectionBase):
 
         if create_collection_record:
             if name is None:
-                name = "Collection created %s" % (time.strftime("%Y-%m-%d %H:%M:%S %Z", time.localtime()))
+                name = "New collection"
                 ensure_unique_name = True
 
             body = {"manifest_text": text,
@@ -1460,7 +1481,7 @@ class Collection(RichCollectionBase):
             if owner_uuid:
                 body["owner_uuid"] = owner_uuid
 
-            self._api_response = self._my_api().collections().create(ensure_unique_name=ensure_unique_name, body=body).execute(num_retries=num_retries)
+            self._remember_api_response(self._my_api().collections().create(ensure_unique_name=ensure_unique_name, body=body).execute(num_retries=num_retries))
             text = self._api_response["manifest_text"]
 
             self._manifest_locator = self._api_response["uuid"]
@@ -1543,7 +1564,7 @@ class Subcollection(RichCollectionBase):
     """This is a subdirectory within a collection that doesn't have its own API
     server record.
 
-    It falls under the umbrella of the root collection.
+    Subcollection locking falls under the umbrella lock of its root collection.
 
     """