X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/aaa964ac44f923985d6a6eb40c179f62c13ca8ce..a6be53f6913862b3884c57e9e93a891f3cfec471:/sdk/cwl/arvados_cwl/pathmapper.py

diff --git a/sdk/cwl/arvados_cwl/pathmapper.py b/sdk/cwl/arvados_cwl/pathmapper.py
index 74d9481ff5..a8619a8598 100644
--- a/sdk/cwl/arvados_cwl/pathmapper.py
+++ b/sdk/cwl/arvados_cwl/pathmapper.py
@@ -2,10 +2,13 @@ import re
 import logging
 import uuid
 import os
+import urllib
 
 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
 
@@ -15,7 +18,7 @@ class ArvPathMapper(PathMapper):
     """Convert container-local paths to and from Keep collection ids."""
 
     pdh_path = re.compile(r'^keep:[0-9a-f]{32}\+\d+/.+$')
-    pdh_dirpath = re.compile(r'^keep:[0-9a-f]{32}\+\d+(/.+)?$')
+    pdh_dirpath = re.compile(r'^keep:[0-9a-f]{32}\+\d+(/.*)?$')
 
     def __init__(self, arvrunner, referenced_files, input_basedir,
                  collection_pattern, file_pattern, name=None, **kwargs):
@@ -32,31 +35,32 @@ class ArvPathMapper(PathMapper):
             if "#" in src:
                 src = src[:src.index("#")]
             if isinstance(src, basestring) and ArvPathMapper.pdh_path.match(src):
-                self._pathmap[src] = MapperEnt(src, self.collection_pattern % src[5:], "File")
+                self._pathmap[src] = MapperEnt(src, self.collection_pattern % urllib.unquote(src[5:]), "File")
             if src not in self._pathmap:
                 # Local FS ref, may need to be uploaded or may be on keep
                 # 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 % urllib.unquote(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)
-                elif src.startswith("arvwf:"):
-                    self._pathmap[src] = MapperEnt(src, src, "File")
-                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)
         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")
+                self._pathmap[src] = MapperEnt(src, self.collection_pattern % urllib.unquote(src[5:]), "Directory")
             for l in srcobj.get("listing", []):
                 self.visit(l, uploadfiles)
 
@@ -76,14 +80,18 @@ 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 % urllib.unquote(v.resolved[5:]), "File")
 
         for srcobj in referenced_files:
             self.visit(srcobj, uploadfiles)
@@ -98,7 +106,7 @@ class ArvPathMapper(PathMapper):
                                              project=self.arvrunner.project_uuid)
 
         for src, ab, st in uploadfiles:
-            self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.fn[5:], "File")
+            self._pathmap[src] = MapperEnt(urllib.quote(st.fn, "/:+@"), self.collection_pattern % st.fn[5:], "File")
             self.arvrunner.add_uploaded(src, self._pathmap[src])
 
         for srcobj in referenced_files:
@@ -176,7 +184,7 @@ class StagingPathMapper(PathMapper):
                 self.visitlisting(obj.get("secondaryFiles", []), stagedir, basedir)
 
 
-class InitialWorkDirPathMapper(StagingPathMapper):
+class VwdPathMapper(StagingPathMapper):
     def setup(self, referenced_files, basedir):
         # type: (List[Any], unicode) -> None
 
@@ -189,7 +197,7 @@ class InitialWorkDirPathMapper(StagingPathMapper):
                 self._pathmap[path] = MapperEnt("$(task.keep)/%s" % ab[5:], tgt, type)
 
 
-class FinalOutputPathMapper(StagingPathMapper):
+class NoFollowPathMapper(StagingPathMapper):
     _follow_dirs = False
     def setup(self, referenced_files, basedir):
         # type: (List[Any], unicode) -> None