3699: code review

[arvados.git] / sdk / python / arvados / collection.py

diff --git a/sdk/python/arvados/collection.py b/sdk/python/arvados/collection.py
index 40b5889c317eef9a59c1d9acea51b93360847baa..686d694befec9d1aec84c43dcbe360b0a6008b09 100644 (file)
--- a/sdk/python/arvados/collection.py
+++ b/sdk/python/arvados/collection.py
@@ -98,7 +98,10 @@ class CollectionReader(object):
         if re.match(r'[a-f0-9]{32}(\+\d+)?(\+\S+)*$', manifest_locator_or_text):
             self._manifest_locator = manifest_locator_or_text
             self._manifest_text = None
-        elif re.match(r'(\S+)( [a-f0-9]{32}(\+\d+)(\+\S+)*)+( \d+:\d+:\S+)+\n', manifest_locator_or_text):
+        elif re.match(r'[a-z0-9]{5}-[a-z0-9]{5}-[a-z0-9]{15}$', manifest_locator_or_text):
+            self._manifest_locator = manifest_locator_or_text
+            self._manifest_text = None
+        elif re.match(r'((\S+)( [a-f0-9]{32}(\+\d+)(\+\S+)*)+( \d+:\d+:\S+)+$)+', manifest_locator_or_text, re.MULTILINE):
             self._manifest_text = manifest_locator_or_text
             self._manifest_locator = None
         else:
@@ -116,17 +119,29 @@ class CollectionReader(object):
         if self._streams is not None:
             return
         if not self._manifest_text:
-            if self._api_client is None:
-                self._api_client = arvados.api('v1')
-            if self._keep_client is None:
-                self._keep_client = KeepClient(api_client=self._api_client)
             try:
+                # As in KeepClient itself, we must wait until the last possible
+                # moment to instantiate an API client, in order to avoid
+                # tripping up clients that don't have access to an API server.
+                # If we do build one, make sure our Keep client uses it.
+                # If instantiation fails, we'll fall back to the except clause,
+                # just like any other Collection lookup failure.
+                if self._api_client is None:
+                    self._api_client = arvados.api('v1')
+                    self._keep_client = KeepClient(api_client=self._api_client)
+                if self._keep_client is None:
+                    self._keep_client = KeepClient(api_client=self._api_client)
                 c = self._api_client.collections().get(
                     uuid=self._manifest_locator).execute()
                 self._manifest_text = c['manifest_text']
             except Exception as e:
+                if not util.portable_data_hash_pattern.match(
+                      self._manifest_locator):
+                    raise
                 _logger.warning("API lookup failed for collection %s (%s: %s)",
                                 self._manifest_locator, type(e), str(e))
+                if self._keep_client is None:
+                    self._keep_client = KeepClient(api_client=self._api_client)
                 self._manifest_text = self._keep_client.get(self._manifest_locator)
         self._streams = []
         for stream_line in self._manifest_text.split("\n"):