20343: Define Python regexps with raw strings

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

diff --git a/sdk/cwl/arvados_cwl/pathmapper.py b/sdk/cwl/arvados_cwl/pathmapper.py
index 64fdfa0d04032e97235dc581144d9cb74494c597..539188fddd995b9cda5c58c89f1f8ef1dd96293a 100644 (file)
--- a/sdk/cwl/arvados_cwl/pathmapper.py
+++ b/sdk/cwl/arvados_cwl/pathmapper.py
@@ -26,7 +26,7 @@ from cwltool.utils import adjustFileObjs, adjustDirObjs
 from cwltool.stdfsaccess import abspath
 from cwltool.workflow import WorkflowException
 
-from .http import http_to_keep
+from arvados.http_to_keep import http_to_keep
 
 logger = logging.getLogger('arvados.cwl-runner')
 
@@ -105,11 +105,17 @@ class ArvPathMapper(PathMapper):
                     raise WorkflowException("Directory literal '%s' is missing `listing`" % src)
             elif src.startswith("http:") or src.startswith("https:"):
                 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)
+                    if self.arvrunner.defer_downloads:
+                        # passthrough, we'll download it later.
+                        self._pathmap[src] = MapperEnt(src, src, srcobj["class"], True)
+                    else:
+                        keepref = "keep:%s/%s" % http_to_keep(self.arvrunner.api, self.arvrunner.project_uuid, src,
+                                                              varying_url_params=self.arvrunner.toplevel_runtimeContext.varying_url_params,
+                                                              prefer_cached_downloads=self.arvrunner.toplevel_runtimeContext.prefer_cached_downloads)
+                        logger.info("%s is %s", src, keepref)
+                        self._pathmap[src] = MapperEnt(keepref, keepref, srcobj["class"], True)
                 except Exception as e:
-                    logger.warning(str(e))
+                    logger.warning("Download error: %s", e)
             else:
                 self._pathmap[src] = MapperEnt(src, src, srcobj["class"], True)
 
@@ -141,7 +147,7 @@ class ArvPathMapper(PathMapper):
             for opt in self.optional_deps:
                 if obj["location"] == opt["location"]:
                     return
-            raise SourceLine(obj, "location", WorkflowException).makeError("Don't know what to do with '%s'" % obj["location"])
+            raise SourceLine(obj, "location", WorkflowException).makeError("Can't handle '%s'" % obj["location"])
 
     def needs_new_collection(self, srcobj, prefix=""):
         """Check if files need to be staged into a new collection.
@@ -156,6 +162,9 @@ class ArvPathMapper(PathMapper):
         if loc.startswith("_:"):
             return True
 
+        if self.arvrunner.defer_downloads and (loc.startswith("http:") or loc.startswith("https:")):
+            return False
+
         i = loc.rfind("/")
         if i > -1:
             loc_prefix = loc[:i+1]