Merge branch '9857-cwl-acceptlist-re' refs #9857

[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 12f7cd7021dc012e5a8945504552d404fd1e4a6b..0fd9a0ed332ececaf4474ca3939fe9b4785db82d 100644 (file)
--- a/sdk/cwl/arvados_cwl/pathmapper.py
+++ b/sdk/cwl/arvados_cwl/pathmapper.py
@@ -1,10 +1,12 @@
 import re
 import logging
+import uuid
+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')
@@ -40,8 +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:
@@ -50,13 +55,14 @@ 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")
-            else:
-                for l in srcobj["listing"]:
-                    self.visit(l, uploadfiles)
+            for l in srcobj.get("listing", []):
+                self.visit(l, uploadfiles)
 
     def addentry(self, obj, c, path, subdirs):
         if obj["location"] in self._pathmap:
             src, srcpath = self.arvrunner.fs_access.get_collection(self._pathmap[obj["location"]].resolved)
+            if srcpath == "":
+                srcpath = "."
             c.copy(srcpath, path + "/" + obj["basename"], source_collection=src, overwrite=True)
             for l in obj.get("secondaryFiles", []):
                 self.addentry(l, c, path, subdirs)
@@ -123,6 +129,9 @@ class ArvPathMapper(PathMapper):
 
                 ab = self.file_pattern % (c.portable_data_hash(), srcobj["basename"])
                 self._pathmap[srcobj["location"]] = MapperEnt(ab, ab, "File")
+                if srcobj.get("secondaryFiles"):
+                    ab = self.collection_pattern % c.portable_data_hash()
+                    self._pathmap["_:" + unicode(uuid.uuid4())] = MapperEnt(ab, ab, "Directory")
                 for loc, sub in subdirs:
                     ab = self.file_pattern % (c.portable_data_hash(), sub[2:])
                     self._pathmap[loc] = MapperEnt(ab, ab, "Directory")
@@ -136,3 +145,37 @@ class ArvPathMapper(PathMapper):
             return (target, "keep:" + target[len(self.keepdir)+1:])
         else:
             return super(ArvPathMapper, self).reversemap(target)
+
+class InitialWorkDirPathMapper(PathMapper):
+
+    def visit(self, obj, stagedir, basedir, copy=False):
+        # type: (Dict[unicode, Any], unicode, unicode, bool) -> None
+        if obj["class"] == "Directory":
+            self._pathmap[obj["location"]] = 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"])
+            if "contents" in obj and obj["location"].startswith("_:"):
+                self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile")
+            else:
+                if copy:
+                    self._pathmap[loc] = MapperEnt(obj["path"], tgt, "WritableFile")
+                else:
+                    self._pathmap[loc] = MapperEnt(obj["path"], tgt, "File")
+                self.visitlisting(obj.get("secondaryFiles", []), stagedir, basedir)
+
+    def setup(self, referenced_files, basedir):
+        # type: (List[Any], unicode) -> None
+
+        # 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)
+
+        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)