Merge branch '11017-docker-migration'

[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 c15b289037210d8b3c98be08664d4c5713315b64..1f6aa577c180c65a9f8a31b0155c863f17e59b67 100644 (file)
--- a/sdk/cwl/arvados_cwl/pathmapper.py
+++ b/sdk/cwl/arvados_cwl/pathmapper.py
@@ -6,6 +6,8 @@ import os
 import arvados.commands.run
 import arvados.collection
 
+from schema_salad.sourceline import SourceLine
+
 from cwltool.pathmapper import PathMapper, MapperEnt, abspath, adjustFileObjs, adjustDirObjs
 from cwltool.workflow import WorkflowException
 
@@ -38,17 +40,20 @@ class ArvPathMapper(PathMapper):
                 # mount.
                 ab = abspath(src, self.input_basedir)
                 st = arvados.commands.run.statfile("", ab, fnPattern="keep:%s/%s")
-                if isinstance(st, arvados.commands.run.UploadFile):
-                    uploadfiles.add((src, ab, st))
-                elif isinstance(st, arvados.commands.run.ArvFile):
-                    self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.fn[5:], "File")
-                elif src.startswith("_:"):
-                    if "contents" in srcobj:
-                        pass
+                with SourceLine(srcobj, "location", WorkflowException):
+                    if isinstance(st, arvados.commands.run.UploadFile):
+                        uploadfiles.add((src, ab, st))
+                    elif isinstance(st, arvados.commands.run.ArvFile):
+                        self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.fn[5:], "File")
+                    elif src.startswith("_:"):
+                        if "contents" in srcobj:
+                            pass
+                        else:
+                            raise WorkflowException("File literal '%s' is missing contents" % src)
+                    elif src.startswith("arvwf:"):
+                        self._pathmap[src] = MapperEnt(src, src, "File")
                     else:
-                        raise WorkflowException("File literal '%s' is missing contents" % src)
-                else:
-                    raise WorkflowException("Input file path '%s' is invalid" % st)
+                        raise WorkflowException("Input file path '%s' is invalid" % st)
             if "secondaryFiles" in srcobj:
                 for l in srcobj["secondaryFiles"]:
                     self.visit(l, uploadfiles)
@@ -74,14 +79,18 @@ class ArvPathMapper(PathMapper):
             with c.open(path + "/" + obj["basename"], "w") as f:
                 f.write(obj["contents"].encode("utf-8"))
         else:
-            raise WorkflowException("Don't know what to do with '%s'" % obj["location"])
+            raise SourceLine(obj, "location", WorkflowException).makeError("Don't know what to do with '%s'" % obj["location"])
 
     def setup(self, referenced_files, basedir):
         # type: (List[Any], unicode) -> None
         uploadfiles = set()
 
-        for k,v in self.arvrunner.get_uploaded().iteritems():
-            self._pathmap[k] = MapperEnt(v.resolved, self.collection_pattern % v.resolved[5:], "File")
+        already_uploaded = self.arvrunner.get_uploaded()
+        for k in referenced_files:
+            loc = k["location"]
+            if loc in already_uploaded:
+                v = already_uploaded[loc]
+                self._pathmap[loc] = MapperEnt(v.resolved, self.collection_pattern % v.resolved[5:], "File")
 
         for srcobj in referenced_files:
             self.visit(srcobj, uploadfiles)
@@ -150,27 +159,31 @@ class ArvPathMapper(PathMapper):
         else:
             return super(ArvPathMapper, self).reversemap(target)
 
-class InitialWorkDirPathMapper(PathMapper):
+class StagingPathMapper(PathMapper):
+    _follow_dirs = True
 
     def visit(self, obj, stagedir, basedir, copy=False):
         # type: (Dict[unicode, Any], unicode, unicode, bool) -> None
         loc = obj["location"]
+        tgt = os.path.join(stagedir, obj["basename"])
         if obj["class"] == "Directory":
-            self._pathmap[loc] = MapperEnt(obj["location"], stagedir, "Directory")
-            self.visitlisting(obj.get("listing", []), stagedir, basedir)
+            self._pathmap[loc] = MapperEnt(loc, tgt, "Directory")
+            if loc.startswith("_:") or self._follow_dirs:
+                self.visitlisting(obj.get("listing", []), tgt, basedir)
         elif obj["class"] == "File":
             if loc in self._pathmap:
                 return
-            tgt = os.path.join(stagedir, obj["basename"])
-            if "contents" in obj and obj["location"].startswith("_:"):
+            if "contents" in obj and loc.startswith("_:"):
                 self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile")
             else:
                 if copy:
-                    self._pathmap[loc] = MapperEnt(obj["path"], tgt, "WritableFile")
+                    self._pathmap[loc] = MapperEnt(loc, tgt, "WritableFile")
                 else:
-                    self._pathmap[loc] = MapperEnt(obj["path"], tgt, "File")
+                    self._pathmap[loc] = MapperEnt(loc, tgt, "File")
                 self.visitlisting(obj.get("secondaryFiles", []), stagedir, basedir)
 
+
+class VwdPathMapper(StagingPathMapper):
     def setup(self, referenced_files, basedir):
         # type: (List[Any], unicode) -> None
 
@@ -183,24 +196,8 @@ class InitialWorkDirPathMapper(PathMapper):
                 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"]
-        tgt = os.path.join(stagedir, obj["basename"])
-        if obj["class"] == "Directory":
-            self._pathmap[loc] = MapperEnt(tgt, tgt, "Directory")
-            if loc.startswith("_:"):
-                self.visitlisting(obj.get("listing", []), tgt, basedir)
-        elif obj["class"] == "File":
-            if loc in self._pathmap:
-                return
-            if "contents" in obj and loc.startswith("_:"):
-                self._pathmap[loc] = MapperEnt(obj["contents"], tgt, "CreateFile")
-            else:
-                self._pathmap[loc] = MapperEnt(loc, tgt, "File")
-                self.visitlisting(obj.get("secondaryFiles", []), stagedir, basedir)
-
+class NoFollowPathMapper(StagingPathMapper):
+    _follow_dirs = False
     def setup(self, referenced_files, basedir):
         # type: (List[Any], unicode) -> None
         self.visitlisting(referenced_files, self.stagedir, basedir)