X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/e502060ffe4f68d33e2cca8f8d7544ce40d53eb7..85bd5f3272cfc8ba8e54f4f47ea9865d7de62779:/sdk/cwl/arvados_cwl/fsaccess.py

diff --git a/sdk/cwl/arvados_cwl/fsaccess.py b/sdk/cwl/arvados_cwl/fsaccess.py
index a99b2a7414..93e2819084 100644
--- a/sdk/cwl/arvados_cwl/fsaccess.py
+++ b/sdk/cwl/arvados_cwl/fsaccess.py
@@ -1,9 +1,14 @@
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
+
 import fnmatch
 import os
 import errno
 import urlparse
 import re
 import logging
+import threading
 
 import ruamel.yaml as yaml
 
@@ -20,23 +25,35 @@ from schema_salad.ref_resolver import DefaultFetcher
 
 logger = logging.getLogger('arvados.cwl-runner')
 
-class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
-    """Implement the cwltool FsAccess interface for Arvados Collections."""
-
-    def __init__(self, basedir, api_client=None, keep_client=None):
-        super(CollectionFsAccess, self).__init__(basedir)
+class CollectionCache(object):
+    def __init__(self, api_client, keep_client, num_retries):
         self.api_client = api_client
         self.keep_client = keep_client
         self.collections = {}
+        self.lock = threading.Lock()
 
-    def get_collection(self, path):
-        p = path.split("/")
-        if p[0].startswith("keep:") and arvados.util.keep_locator_pattern.match(p[0][5:]):
-            pdh = p[0][5:]
+    def get(self, pdh):
+        with self.lock:
             if pdh not in self.collections:
+                logger.debug("Creating collection reader for %s", pdh)
                 self.collections[pdh] = arvados.collection.CollectionReader(pdh, api_client=self.api_client,
                                                                             keep_client=self.keep_client)
-            return (self.collections[pdh], "/".join(p[1:]))
+            return self.collections[pdh]
+
+
+class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
+    """Implement the cwltool FsAccess interface for Arvados Collections."""
+
+    def __init__(self, basedir, collection_cache=None):
+        super(CollectionFsAccess, self).__init__(basedir)
+        self.collection_cache = collection_cache
+
+    def get_collection(self, path):
+        sp = path.split("/", 1)
+        p = sp[0]
+        if p.startswith("keep:") and arvados.util.keep_locator_pattern.match(p[5:]):
+            pdh = p[5:]
+            return (self.collection_cache.get(pdh), sp[1] if len(sp) == 2 else None)
         else:
             return (None, path)
 
@@ -78,8 +95,11 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
 
     def exists(self, fn):
         collection, rest = self.get_collection(fn)
-        if collection:
-            return collection.exists(rest)
+        if collection is not None:
+            if rest:
+                return collection.exists(rest)
+            else:
+                return True
         else:
             return super(CollectionFsAccess, self).exists(fn)
 
@@ -133,13 +153,16 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
             return os.path.realpath(path)
 
 class CollectionFetcher(DefaultFetcher):
-    def __init__(self, cache, session, api_client=None, keep_client=None, num_retries=4):
+    def __init__(self, cache, session, api_client=None, fs_access=None, num_retries=4, overrides=None):
         super(CollectionFetcher, self).__init__(cache, session)
         self.api_client = api_client
-        self.fsaccess = CollectionFsAccess("", api_client=api_client, keep_client=keep_client)
+        self.fsaccess = fs_access
         self.num_retries = num_retries
+        self.overrides = overrides if overrides else {}
 
     def fetch_text(self, url):
+        if url in self.overrides:
+            return self.overrides[url]
         if url.startswith("keep:"):
             with self.fsaccess.open(url, "r") as f:
                 return f.read()
@@ -150,7 +173,11 @@ class CollectionFetcher(DefaultFetcher):
         return super(CollectionFetcher, self).fetch_text(url)
 
     def check_exists(self, url):
+        if url in self.overrides:
+            return True
         try:
+            if url.startswith("http://arvados.org/cwl"):
+                return True
             if url.startswith("keep:"):
                 return self.fsaccess.exists(url)
             if url.startswith("arvwf:"):