Merge remote-tracking branch 'origin/master' into 14645-fuse-operations-reporting

[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 c844311d55d0c9c4db853c8d3036bf4d10a0f9b9..38135899dca7c66a10a135c85d2c0f8db43b0f2a 100644 (file)
--- a/sdk/cwl/arvados_cwl/pathmapper.py
+++ b/sdk/cwl/arvados_cwl/pathmapper.py
@@ -2,11 +2,17 @@
 #
 # SPDX-License-Identifier: Apache-2.0
 
+from future import standard_library
+standard_library.install_aliases()
+from builtins import str
+from past.builtins import basestring
+from future.utils import viewitems
+
 import re
 import logging
 import uuid
 import os
-import urllib
+import urllib.request, urllib.parse, urllib.error
 
 import arvados_cwl.util
 import arvados.commands.run
@@ -52,6 +58,7 @@ class ArvPathMapper(PathMapper):
         self.name = name
         self.referenced_files = [r["location"] for r in referenced_files]
         self.single_collection = single_collection
+        self.pdh_to_uuid = {}
         super(ArvPathMapper, self).__init__(referenced_files, input_basedir, None)
 
     def visit(self, srcobj, uploadfiles):
@@ -60,7 +67,9 @@ class ArvPathMapper(PathMapper):
             src = src[:src.index("#")]
 
         if isinstance(src, basestring) and ArvPathMapper.pdh_dirpath.match(src):
-            self._pathmap[src] = MapperEnt(src, self.collection_pattern % urllib.unquote(src[5:]), srcobj["class"], True)
+            self._pathmap[src] = MapperEnt(src, self.collection_pattern % urllib.parse.unquote(src[5:]), srcobj["class"], True)
+            if arvados_cwl.util.collectionUUID in srcobj:
+                self.pdh_to_uuid[src.split("/", 1)[0][5:]] = srcobj[arvados_cwl.util.collectionUUID]
 
         debug = logger.isEnabledFor(logging.DEBUG)
 
@@ -77,7 +86,7 @@ class ArvPathMapper(PathMapper):
                     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", True)
+                        self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % urllib.parse.unquote(st.fn[5:]), "File", True)
                     else:
                         raise WorkflowException("Input file path '%s' is invalid" % st)
             elif src.startswith("_:"):
@@ -114,12 +123,20 @@ class ArvPathMapper(PathMapper):
             remap.append((obj["location"], path + "/" + obj["basename"]))
         elif obj["location"].startswith("_:") and "contents" in obj:
             with c.open(path + "/" + obj["basename"], "w") as f:
-                f.write(obj["contents"].encode("utf-8"))
+                f.write(obj["contents"])
             remap.append((obj["location"], path + "/" + obj["basename"]))
         else:
             raise SourceLine(obj, "location", WorkflowException).makeError("Don't know what to do with '%s'" % obj["location"])
 
     def needs_new_collection(self, srcobj, prefix=""):
+        """Check if files need to be staged into a new collection.
+
+        If all the files are in the same collection and in the same
+        paths they would be staged to, return False.  Otherwise, a new
+        collection is needed with files copied/created in the
+        appropriate places.
+        """
+
         loc = srcobj["location"]
         if loc.startswith("_:"):
             return True
@@ -134,10 +151,9 @@ class ArvPathMapper(PathMapper):
                 prefix = loc+"/"
         if srcobj["class"] == "File" and loc not in self._pathmap:
             return True
-        if srcobj.get("secondaryFiles"):
-            for s in srcobj["secondaryFiles"]:
-                if self.needs_new_collection(s, prefix):
-                    return True
+        for s in srcobj.get("secondaryFiles", []):
+            if self.needs_new_collection(s, prefix):
+                return True
         if srcobj.get("listing"):
             prefix = "%s%s/" % (prefix, srcobj["basename"])
             for l in srcobj["listing"]:
@@ -169,7 +185,7 @@ class ArvPathMapper(PathMapper):
                                          packed=False)
 
         for src, ab, st in uploadfiles:
-            self._pathmap[src] = MapperEnt(urllib.quote(st.fn, "/:+@"), self.collection_pattern % st.fn[5:],
+            self._pathmap[src] = MapperEnt(urllib.parse.quote(st.fn, "/:+@"), self.collection_pattern % st.fn[5:],
                                            "Directory" if os.path.isdir(ab) else "File", True)
 
         for srcobj in referenced_files:
@@ -195,6 +211,10 @@ class ArvPathMapper(PathMapper):
             elif srcobj["class"] == "File" and (srcobj.get("secondaryFiles") or
                 (srcobj["location"].startswith("_:") and "contents" in srcobj)):
 
+                # If all secondary files/directories are located in
+                # the same collection as the primary file and the
+                # paths and names that are consistent with staging,
+                # don't create a new collection.
                 if not self.needs_new_collection(srcobj):
                     continue
 
@@ -217,7 +237,7 @@ class ArvPathMapper(PathMapper):
                                                               ab, "File", True)
                 if srcobj.get("secondaryFiles"):
                     ab = self.collection_pattern % c.portable_data_hash()
-                    self._pathmap["_:" + unicode(uuid.uuid4())] = MapperEnt("keep:"+c.portable_data_hash(), ab, "Directory", True)
+                    self._pathmap["_:" + str(uuid.uuid4())] = MapperEnt("keep:"+c.portable_data_hash(), ab, "Directory", True)
 
             if remap:
                 for loc, sub in remap:
@@ -290,7 +310,7 @@ class VwdPathMapper(StagingPathMapper):
         # with any secondary files.
         self.visitlisting(referenced_files, self.stagedir, basedir)
 
-        for path, (ab, tgt, type, staged) in self._pathmap.items():
+        for path, (ab, tgt, type, staged) in viewitems(self._pathmap):
             if type in ("File", "Directory") and ab.startswith("keep:"):
                 self._pathmap[path] = MapperEnt("$(task.keep)/%s" % ab[5:], tgt, type, staged)