X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/05bea2c50474edeb9d0e3fb8daaf838b58ea9a54..af2c75821456e28875dc8006efb03f38ec1a74b8:/sdk/python/arvados/collection.py

diff --git a/sdk/python/arvados/collection.py b/sdk/python/arvados/collection.py
index 40b5889c31..496136ebe3 100644
--- a/sdk/python/arvados/collection.py
+++ b/sdk/python/arvados/collection.py
@@ -92,13 +92,16 @@ def normalize(collection):
 
 
 class CollectionReader(object):
-    def __init__(self, manifest_locator_or_text, api_client=None):
+    def __init__(self, manifest_locator_or_text, api_client=None, keep_client=None):
         self._api_client = api_client
-        self._keep_client = None
+        self._keep_client = keep_client
         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"):
@@ -138,7 +153,7 @@ class CollectionReader(object):
         # now regenerate the manifest text based on the normalized stream
 
         #print "normalizing", self._manifest_text
-        self._manifest_text = ''.join([StreamReader(stream).manifest_text() for stream in self._streams])
+        self._manifest_text = ''.join([StreamReader(stream, keep=self._keep_client).manifest_text() for stream in self._streams])
         #print "result", self._manifest_text
 
 
@@ -146,7 +161,7 @@ class CollectionReader(object):
         self._populate()
         resp = []
         for s in self._streams:
-            resp.append(StreamReader(s))
+            resp.append(StreamReader(s, keep=self._keep_client))
         return resp
 
     def all_files(self):
@@ -157,7 +172,7 @@ class CollectionReader(object):
     def manifest_text(self, strip=False):
         self._populate()
         if strip:
-            m = ''.join([StreamReader(stream).manifest_text(strip=True) for stream in self._streams])
+            m = ''.join([StreamReader(stream, keep=self._keep_client).manifest_text(strip=True) for stream in self._streams])
             return m
         else:
             return self._manifest_text