13931: un-uriquote keep: URIs when converting to (collection, path) tuple

[arvados.git] / sdk / cwl / arvados_cwl / fsaccess.py

diff --git a/sdk/cwl/arvados_cwl/fsaccess.py b/sdk/cwl/arvados_cwl/fsaccess.py
index 47749eea5fa5bd4f46ef8f6f3e7cbdf858f80b06..316a652529b384205661827e2c46d056025d5506 100644 (file)
--- a/sdk/cwl/arvados_cwl/fsaccess.py
+++ b/sdk/cwl/arvados_cwl/fsaccess.py
@@ -22,6 +22,8 @@ import arvados.collection
 import arvados.arvfile
 import arvados.errors
 
+from googleapiclient.errors import HttpError
+
 from schema_salad.ref_resolver import DefaultFetcher
 
 logger = logging.getLogger('arvados.cwl-runner')
@@ -32,6 +34,7 @@ class CollectionCache(object):
                  min_entries=2):
         self.api_client = api_client
         self.keep_client = keep_client
+        self.num_retries = num_retries
         self.collections = OrderedDict()
         self.lock = threading.Lock()
         self.total = 0
@@ -54,7 +57,8 @@ class CollectionCache(object):
             if pdh not in self.collections:
                 logger.debug("Creating collection reader for %s", pdh)
                 cr = arvados.collection.CollectionReader(pdh, api_client=self.api_client,
-                                                         keep_client=self.keep_client)
+                                                         keep_client=self.keep_client,
+                                                         num_retries=self.num_retries)
                 sz = len(cr.manifest_text()) * 128
                 self.collections[pdh] = (cr, sz)
                 self.total += sz
@@ -79,7 +83,7 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
         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)
+            return (self.collection_cache.get(pdh), urlparse.unquote(sp[1]) if len(sp) == 2 else None)
         else:
             return (None, path)
 
@@ -120,7 +124,13 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
             return super(CollectionFsAccess, self).open(self._abs(fn), mode)
 
     def exists(self, fn):
-        collection, rest = self.get_collection(fn)
+        try:
+            collection, rest = self.get_collection(fn)
+        except HttpError as err:
+            if err.resp.status == 404:
+                return False
+            else:
+                raise
         if collection is not None:
             if rest:
                 return collection.exists(rest)
@@ -129,6 +139,17 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
         else:
             return super(CollectionFsAccess, self).exists(fn)
 
+    def size(self, fn):  # type: (unicode) -> bool
+        collection, rest = self.get_collection(fn)
+        if collection is not None:
+            if rest:
+                arvfile = collection.find(rest)
+                if isinstance(arvfile, arvados.arvfile.ArvadosFile):
+                    return arvfile.size()
+            raise IOError(errno.EINVAL, "Not a path to a file %s" % (fn))
+        else:
+            return super(CollectionFsAccess, self).size(fn)
+
     def isfile(self, fn):  # type: (unicode) -> bool
         collection, rest = self.get_collection(fn)
         if collection is not None:
@@ -179,16 +200,13 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
             return os.path.realpath(path)
 
 class CollectionFetcher(DefaultFetcher):
-    def __init__(self, cache, session, api_client=None, fs_access=None, num_retries=4, overrides=None):
+    def __init__(self, cache, session, api_client=None, fs_access=None, num_retries=4):
         super(CollectionFetcher, self).__init__(cache, session)
         self.api_client = api_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()
@@ -199,8 +217,6 @@ 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