X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/454ee2b8f0385c542b6f1165a3baf2820425e1a3..c340eecc7a03dd066792e5f046f087b8b3dfced6:/sdk/cwl/arvados_cwl/pathmapper.py

diff --git a/sdk/cwl/arvados_cwl/pathmapper.py b/sdk/cwl/arvados_cwl/pathmapper.py
index 56c15a4a43..4a91a7a836 100644
--- a/sdk/cwl/arvados_cwl/pathmapper.py
+++ b/sdk/cwl/arvados_cwl/pathmapper.py
@@ -21,7 +21,9 @@ import arvados.collection
 from schema_salad.sourceline import SourceLine
 
 from arvados.errors import ApiError
-from cwltool.pathmapper import PathMapper, MapperEnt, abspath, adjustFileObjs, adjustDirObjs
+from cwltool.pathmapper import PathMapper, MapperEnt
+from cwltool.utils import adjustFileObjs, adjustDirObjs
+from cwltool.stdfsaccess import abspath
 from cwltool.workflow import WorkflowException
 
 from .http import http_to_keep
@@ -42,13 +44,13 @@ def trim_listing(obj):
     if obj.get("location", "").startswith("keep:") and "listing" in obj:
         del obj["listing"]
 
+collection_pdh_path = re.compile(r'^keep:[0-9a-f]{32}\+\d+/.+$')
+collection_pdh_pattern = re.compile(r'^keep:([0-9a-f]{32}\+\d+)(/.*)?')
+collection_uuid_pattern = re.compile(r'^keep:([a-z0-9]{5}-4zz18-[a-z0-9]{15})(/.*)?$')
 
 class ArvPathMapper(PathMapper):
     """Convert container-local paths to and from Keep collection ids."""
 
-    pdh_path = re.compile(r'^keep:[0-9a-f]{32}\+\d+/.+$')
-    pdh_dirpath = re.compile(r'^keep:[0-9a-f]{32}\+\d+(/.*)?$')
-
     def __init__(self, arvrunner, referenced_files, input_basedir,
                  collection_pattern, file_pattern, name=None, single_collection=False):
         self.arvrunner = arvrunner
@@ -66,13 +68,17 @@ class ArvPathMapper(PathMapper):
         if "#" in src:
             src = src[:src.index("#")]
 
-        if isinstance(src, basestring) and ArvPathMapper.pdh_dirpath.match(src):
-            self._pathmap[src] = MapperEnt(src, self.collection_pattern % urllib.parse.unquote(src[5:]), srcobj["class"], True)
-            if arvados_cwl.util.collectionUUID in srcobj:
-                self.pdh_to_uuid[src.split("/", 1)[0][5:]] = srcobj[arvados_cwl.util.collectionUUID]
-
         debug = logger.isEnabledFor(logging.DEBUG)
 
+        if isinstance(src, basestring) and src.startswith("keep:"):
+            if collection_pdh_pattern.match(src):
+                self._pathmap[src] = MapperEnt(src, self.collection_pattern % urllib.parse.unquote(src[5:]), srcobj["class"], True)
+                if arvados_cwl.util.collectionUUID in srcobj:
+                    self.pdh_to_uuid[src.split("/", 1)[0][5:]] = srcobj[arvados_cwl.util.collectionUUID]
+            elif not collection_uuid_pattern.match(src):
+                with SourceLine(srcobj, "location", WorkflowException, debug):
+                    raise WorkflowException("Invalid keep reference '%s'" % src)
+
         if src not in self._pathmap:
             if src.startswith("file:"):
                 # Local FS ref, may need to be uploaded or may be on keep
@@ -95,9 +101,12 @@ class ArvPathMapper(PathMapper):
                 if srcobj["class"] == "Directory" and "listing" not in srcobj:
                     raise WorkflowException("Directory literal '%s' is missing `listing`" % src)
             elif src.startswith("http:") or src.startswith("https:"):
-                keepref = http_to_keep(self.arvrunner.api, self.arvrunner.project_uuid, src)
-                logger.info("%s is %s", src, keepref)
-                self._pathmap[src] = MapperEnt(keepref, keepref, srcobj["class"], True)
+                try:
+                    keepref = http_to_keep(self.arvrunner.api, self.arvrunner.project_uuid, src)
+                    logger.info("%s is %s", src, keepref)
+                    self._pathmap[src] = MapperEnt(keepref, keepref, srcobj["class"], True)
+                except Exception as e:
+                    logger.warning(str(e))
             else:
                 self._pathmap[src] = MapperEnt(src, src, srcobj["class"], True)
 
@@ -281,6 +290,7 @@ class StagingPathMapper(PathMapper):
     def visit(self, obj, stagedir, basedir, copy=False, staged=False):
         # type: (Dict[unicode, Any], unicode, unicode, bool) -> None
         loc = obj["location"]
+        stagedir = obj.get("dirname") or stagedir
         tgt = os.path.join(stagedir, obj["basename"])
         basetgt, baseext = os.path.splitext(tgt)