10165: Add FinalOutputPathMapper. Delete basename/size/listing from output object...

[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 40a64ffd20b5a769421f297c8a29cf35cec1d8d3..bd71a133550005c922ea1338012df5a9013a8b2c 100644 (file)
--- a/sdk/cwl/arvados_cwl/pathmapper.py
+++ b/sdk/cwl/arvados_cwl/pathmapper.py
@@ -6,7 +6,7 @@ import os
 import arvados.commands.run
 import arvados.collection
 
-from cwltool.pathmapper import PathMapper, MapperEnt, abspath
+from cwltool.pathmapper import PathMapper, MapperEnt, abspath, adjustFileObjs, adjustDirObjs
 from cwltool.workflow import WorkflowException
 
 logger = logging.getLogger('arvados.cwl-runner')
@@ -42,8 +42,11 @@ class ArvPathMapper(PathMapper):
                     uploadfiles.add((src, ab, st))
                 elif isinstance(st, arvados.commands.run.ArvFile):
                     self._pathmap[src] = MapperEnt(ab, st.fn, "File")
-                elif src.startswith("_:") and "contents" in srcobj:
-                    pass
+                elif src.startswith("_:"):
+                    if "contents" in srcobj:
+                        pass
+                    else:
+                        raise WorkflowException("File literal '%s' is missing contents" % src)
                 else:
                     raise WorkflowException("Input file path '%s' is invalid" % st)
             if "secondaryFiles" in srcobj:
@@ -52,7 +55,7 @@ class ArvPathMapper(PathMapper):
         elif srcobj["class"] == "Directory":
             if isinstance(src, basestring) and ArvPathMapper.pdh_dirpath.match(src):
                 self._pathmap[src] = MapperEnt(src, self.collection_pattern % src[5:], "Directory")
-            for l in srcobj["listing"]:
+            for l in srcobj.get("listing", []):
                 self.visit(l, uploadfiles)
 
     def addentry(self, obj, c, path, subdirs):
@@ -147,11 +150,11 @@ class InitialWorkDirPathMapper(PathMapper):
 
     def visit(self, obj, stagedir, basedir, copy=False):
         # type: (Dict[unicode, Any], unicode, unicode, bool) -> None
+        loc = obj["location"]
         if obj["class"] == "Directory":
-            self._pathmap[obj["location"]] = MapperEnt(obj["location"], stagedir, "Directory")
+            self._pathmap[loc] = MapperEnt(obj["location"], stagedir, "Directory")
             self.visitlisting(obj.get("listing", []), stagedir, basedir)
         elif obj["class"] == "File":
-            loc = obj["location"]
             if loc in self._pathmap:
                 return
             tgt = os.path.join(stagedir, obj["basename"])
@@ -169,10 +172,26 @@ class InitialWorkDirPathMapper(PathMapper):
 
         # Go through each file and set the target to its own directory along
         # with any secondary files.
-        stagedir = self.stagedir
-        for fob in referenced_files:
-            self.visit(fob, stagedir, basedir)
+        self.visitlisting(referenced_files, self.stagedir, basedir)
 
         for path, (ab, tgt, type) in self._pathmap.items():
             if type in ("File", "Directory") and ab.startswith("keep:"):
                 self._pathmap[path] = MapperEnt("$(task.keep)/%s" % ab[5:], tgt, type)
+
+
+class FinalOutputPathMapper(PathMapper):
+    def visit(self, obj, stagedir, basedir, copy=False):
+        # type: (Dict[unicode, Any], unicode, unicode, bool) -> None
+        loc = obj["location"]
+        if obj["class"] == "Directory":
+            self._pathmap[loc] = MapperEnt(loc, stagedir, "Directory")
+        elif obj["class"] == "File":
+            if loc in self._pathmap:
+                return
+            tgt = os.path.join(stagedir, obj["basename"])
+            self._pathmap[loc] = MapperEnt(loc, tgt, "File")
+            self.visitlisting(obj.get("secondaryFiles", []), stagedir, basedir)
+
+    def setup(self, referenced_files, basedir):
+        # type: (List[Any], unicode) -> None
+        self.visitlisting(referenced_files, self.stagedir, basedir)