10401: Add integration tests for directory listing behavior.

[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 9cb73d3582dc29ffe0b585bc97fca873ebcbdfa0..3a3d16073833a6367876b5833b54cbb8f35584e7 100644 (file)
--- a/sdk/cwl/arvados_cwl/fsaccess.py
+++ b/sdk/cwl/arvados_cwl/fsaccess.py
@@ -2,16 +2,24 @@ import fnmatch
 import os
 import errno
 import urlparse
+import re
+import logging
+
+import ruamel.yaml as yaml
 
 import cwltool.stdfsaccess
 from cwltool.pathmapper import abspath
+import cwltool.resolver
 
 import arvados.util
 import arvados.collection
 import arvados.arvfile
+import arvados.errors
 
 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."""
 
@@ -22,13 +30,14 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
         self.collections = {}
 
     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:]
+        sp = path.split("/", 1)
+        p = sp[0]
+        if p.startswith("keep:") and arvados.util.keep_locator_pattern.match(p[5:]):
+            pdh = p[5:]
             if pdh not in self.collections:
                 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], sp[1] if len(sp) == 2 else None)
         else:
             return (None, path)
 
@@ -71,7 +80,10 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
     def exists(self, fn):
         collection, rest = self.get_collection(fn)
         if collection:
-            return collection.exists(rest)
+            if rest:
+                return collection.exists(rest)
+            else:
+                return True
         else:
             return super(CollectionFsAccess, self).exists(fn)
 
@@ -125,19 +137,36 @@ class CollectionFsAccess(cwltool.stdfsaccess.StdFsAccess):
             return os.path.realpath(path)
 
 class CollectionFetcher(DefaultFetcher):
-    def __init__(self, cache, session, api_client=None, keep_client=None):
+    def __init__(self, cache, session, api_client=None, keep_client=None, num_retries=4):
         super(CollectionFetcher, self).__init__(cache, session)
+        self.api_client = api_client
         self.fsaccess = CollectionFsAccess("", api_client=api_client, keep_client=keep_client)
+        self.num_retries = num_retries
 
     def fetch_text(self, url):
         if url.startswith("keep:"):
-            with self.fsaccess.open(url) as f:
+            with self.fsaccess.open(url, "r") as f:
                 return f.read()
+        if url.startswith("arvwf:"):
+            record = self.api_client.workflows().get(uuid=url[6:]).execute(num_retries=self.num_retries)
+            definition = record["definition"] + ('\nlabel: "%s"\n' % record["name"].replace('"', '\\"'))
+            return definition
         return super(CollectionFetcher, self).fetch_text(url)
 
     def check_exists(self, url):
-        if url.startswith("keep:"):
-            return self.fsaccess.exists(url)
+        try:
+            if url.startswith("http://arvados.org/cwl"):
+                return True
+            if url.startswith("keep:"):
+                return self.fsaccess.exists(url)
+            if url.startswith("arvwf:"):
+                if self.fetch_text(url):
+                    return True
+        except arvados.errors.NotFoundError:
+            return False
+        except:
+            logger.exception("Got unexpected exception checking if file exists:")
+            return False
         return super(CollectionFetcher, self).check_exists(url)
 
     def urljoin(self, base_url, url):
@@ -149,7 +178,7 @@ class CollectionFetcher(DefaultFetcher):
             return url
 
         basesp = urlparse.urlsplit(base_url)
-        if basesp.scheme == "keep":
+        if basesp.scheme in ("keep", "arvwf"):
             if not basesp.path:
                 raise IOError(errno.EINVAL, "Invalid Keep locator", base_url)
 
@@ -158,7 +187,7 @@ class CollectionFetcher(DefaultFetcher):
 
             pdh = baseparts.pop(0)
 
-            if not arvados.util.keep_locator_pattern.match(pdh):
+            if basesp.scheme == "keep" and not arvados.util.keep_locator_pattern.match(pdh):
                 raise IOError(errno.EINVAL, "Invalid Keep locator", base_url)
 
             if urlsp.path.startswith("/"):
@@ -169,6 +198,28 @@ class CollectionFetcher(DefaultFetcher):
                 baseparts.pop()
 
             path = "/".join([pdh] + baseparts + urlparts)
-            return urlparse.urlunsplit(("keep", "", path, "", urlsp.fragment))
+            return urlparse.urlunsplit((basesp.scheme, "", path, "", urlsp.fragment))
 
         return super(CollectionFetcher, self).urljoin(base_url, url)
+
+workflow_uuid_pattern = re.compile(r'[a-z0-9]{5}-7fd4e-[a-z0-9]{15}')
+pipeline_template_uuid_pattern = re.compile(r'[a-z0-9]{5}-p5p6p-[a-z0-9]{15}')
+
+def collectionResolver(api_client, document_loader, uri, num_retries=4):
+    if workflow_uuid_pattern.match(uri):
+        return "arvwf:%s#main" % (uri)
+
+    if pipeline_template_uuid_pattern.match(uri):
+        pt = api_client.pipeline_templates().get(uuid=uri).execute(num_retries=num_retries)
+        return "keep:" + pt["components"].values()[0]["script_parameters"]["cwl:tool"]
+
+    p = uri.split("/")
+    if arvados.util.keep_locator_pattern.match(p[0]):
+        return "keep:%s" % (uri)
+
+    if arvados.util.collection_uuid_pattern.match(p[0]):
+        return "keep:%s%s" % (api_client.collections().
+                              get(uuid=p[0]).execute()["portable_data_hash"],
+                              uri[len(p[0]):])
+
+    return cwltool.resolver.tool_resolver(document_loader, uri)