19385: Finally passing schemadef test.

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

diff --git a/sdk/cwl/arvados_cwl/arvcontainer.py b/sdk/cwl/arvados_cwl/arvcontainer.py
index fde9db384e3b8d67ed76761e721f988c37c1dea4..3703fc62173e42058b6032441c6e792b00518370 100644 (file)
--- a/sdk/cwl/arvados_cwl/arvcontainer.py
+++ b/sdk/cwl/arvados_cwl/arvcontainer.py
@@ -17,7 +17,7 @@ import uuid
 import math
 
 import arvados_cwl.util
-import ruamel.yaml as yaml
+import ruamel.yaml
 
 from cwltool.errors import WorkflowException
 from cwltool.process import UnsupportedRequirement, shortname
@@ -250,11 +250,7 @@ class ArvadosContainer(JobBase):
         container_request["container_image"] = arv_docker_get_image(self.arvrunner.api,
                                                                     docker_req,
                                                                     runtimeContext.pull_image,
-                                                                    runtimeContext.project_uuid,
-                                                                    runtimeContext.force_docker_pull,
-                                                                    runtimeContext.tmp_outdir_prefix,
-                                                                    runtimeContext.match_local_docker,
-                                                                    runtimeContext.copy_deps)
+                                                                    runtimeContext)
 
         network_req, _ = self.get_requirement("NetworkAccess")
         if network_req:
@@ -264,16 +260,19 @@ class ArvadosContainer(JobBase):
         if api_req:
             runtime_constraints["API"] = True
 
-        use_disk_cache = (self.arvrunner.api.config()["Containers"].get("DefaultKeepCacheDisk", 0) > 0)
+        use_disk_cache = (self.arvrunner.api.config()["Containers"].get("DefaultKeepCacheRAM", 0) == 0)
+
+        keep_cache_type_req, _ = self.get_requirement("http://arvados.org/cwl#KeepCacheTypeRequirement")
+        if keep_cache_type_req:
+            if "keepCacheType" in keep_cache_type_req:
+                if keep_cache_type_req["keepCacheType"] == "ram_cache":
+                    use_disk_cache = False
 
         runtime_req, _ = self.get_requirement("http://arvados.org/cwl#RuntimeConstraints")
         if runtime_req:
-            if "keepCacheType" in runtime_req:
-                if cache_type == "ram_cache":
-                    use_disk_cache = False
             if "keep_cache" in runtime_req:
                 if use_disk_cache:
-                    # If DefaultKeepCacheDisk is non-zero it means we should use disk cache.
+                    # If DefaultKeepCacheRAM is zero it means we should use disk cache.
                     runtime_constraints["keep_cache_disk"] = math.ceil(runtime_req["keep_cache"] * 2**20)
                 else:
                     runtime_constraints["keep_cache_ram"] = math.ceil(runtime_req["keep_cache"] * 2**20)
@@ -287,13 +286,6 @@ class ArvadosContainer(JobBase):
                         "writable": True
                     }
 
-        if use_disk_cache and "keep_cache_disk" not in runtime_constraints:
-            # Cache size wasn't explicitly set so calculate a default
-            # based on 2x RAM request or 1 GB per core, whichever is
-            # smaller.  This is to avoid requesting 100s of GB of disk
-            # cache when requesting a node with a huge amount of RAM.
-            runtime_constraints["keep_cache_disk"] = min(runtime_constraints["ram"] * 2, runtime_constraints["vcpus"] * (1024*1024*1024))
-
         partition_req, _ = self.get_requirement("http://arvados.org/cwl#PartitionRequirement")
         if partition_req:
             scheduling_parameters["partitions"] = aslist(partition_req["partition"])
@@ -540,20 +532,23 @@ class RunnerContainer(Runner):
                 "portable_data_hash": "%s" % workflowcollection
             }
         elif self.embedded_tool.tool.get("id", "").startswith("arvwf:"):
-            workflowpath = "/var/lib/cwl/workflow.json#main"
-            record = self.arvrunner.api.workflows().get(uuid=self.embedded_tool.tool["id"][6:33]).execute(num_retries=self.arvrunner.num_retries)
-            packed = yaml.safe_load(record["definition"])
+            uuid, frg = urllib.parse.urldefrag(self.embedded_tool.tool["id"])
+            workflowpath = "/var/lib/cwl/workflow.json#" + frg
+            packedtxt = self.loadingContext.loader.fetch_text(uuid)
+            yaml = ruamel.yaml.YAML(typ='safe', pure=True)
+            packed = yaml.load(packedtxt)
             container_req["mounts"]["/var/lib/cwl/workflow.json"] = {
                 "kind": "json",
                 "content": packed
             }
             container_req["properties"]["template_uuid"] = self.embedded_tool.tool["id"][6:33]
         else:
-            packed = packed_workflow(self.arvrunner, self.embedded_tool, self.merged_map, runtimeContext, git_info)
+            main = self.loadingContext.loader.idx["_:main"]
+            #print(json.dumps(main, indent=2))
             workflowpath = "/var/lib/cwl/workflow.json#main"
             container_req["mounts"]["/var/lib/cwl/workflow.json"] = {
                 "kind": "json",
-                "content": packed
+                "content": main
             }
 
         container_req["properties"].update({k.replace("http://arvados.org/cwl#", "arv:"): v for k, v in git_info.items()})
@@ -626,6 +621,9 @@ class RunnerContainer(Runner):
         if runtimeContext.prefer_cached_downloads:
             command.append("--prefer-cached-downloads")
 
+        if self.fast_parser:
+            command.append("--fast-parser")
+
         command.extend([workflowpath, "/var/lib/cwl/cwl.input.json"])
 
         container_req["command"] = command