Merge branch '18947-githttpd'

[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 0cbf9eb56bfa3836e3d024b79ecbee50e4bf56e8..64fdfa0d04032e97235dc581144d9cb74494c597 100644 (file)
--- a/sdk/cwl/arvados_cwl/pathmapper.py
+++ b/sdk/cwl/arvados_cwl/pathmapper.py
@@ -74,7 +74,6 @@ class ArvPathMapper(PathMapper):
 
         if isinstance(src, basestring) and src.startswith("keep:"):
             if collection_pdh_pattern.match(src):
-                #self._pathmap[src] = MapperEnt(src, self.collection_pattern % 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:
@@ -94,12 +93,9 @@ class ArvPathMapper(PathMapper):
                                                    raiseOSError=True)
                 with SourceLine(srcobj, "location", WorkflowException, debug):
                     if isinstance(st, arvados.commands.run.UploadFile):
-                        print("VV", (src, ab, st))
                         uploadfiles.add((src, ab, st))
                     elif isinstance(st, arvados.commands.run.ArvFile):
                         self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % urllib.parse.unquote(st.fn[5:]), "File", True)
-
-                        #self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % st.fn[5:], "File", True)
                     else:
                         raise WorkflowException("Input file path '%s' is invalid" % st)
             elif src.startswith("_:"):
@@ -125,7 +121,6 @@ class ArvPathMapper(PathMapper):
                 self.visit(l, uploadfiles)
 
     def addentry(self, obj, c, path, remap):
-        print(obj["location"], self._pathmap)
         if obj["location"] in self._pathmap:
             src, srcpath = self.arvrunner.fs_access.get_collection(self._pathmap[obj["location"]].resolved)
             if srcpath == "":
@@ -161,18 +156,23 @@ class ArvPathMapper(PathMapper):
         if loc.startswith("_:"):
             return True
 
-        if not prefix:
-            i = loc.rfind("/")
-            if i > -1:
-                prefix = loc[:i+1]
-                suffix = urllib.parse.quote(urllib.parse.unquote(loc[i+1:]), "/+@")
-            else:
-                prefix = loc+"/"
-                suffix = ""
+        i = loc.rfind("/")
+        if i > -1:
+            loc_prefix = loc[:i+1]
+            if not prefix:
+                prefix = loc_prefix
+            # quote/unquote to ensure consistent quoting
+            suffix = urllib.parse.quote(urllib.parse.unquote(loc[i+1:]), "/+@")
+        else:
+            # no '/' found
+            loc_prefix = loc+"/"
+            prefix = loc+"/"
+            suffix = ""
 
-        print("LLL", prefix+suffix, prefix+urllib.parse.quote(srcobj["basename"], "/+@"))
-        if prefix+suffix != prefix+urllib.parse.quote(srcobj["basename"], "/+@"):
-            print("LLL -> needs new collection")
+        if prefix != loc_prefix:
+            return True
+
+        if "basename" in srcobj and suffix != urllib.parse.quote(srcobj["basename"], "/+@"):
             return True
 
         if srcobj["class"] == "File" and loc not in self._pathmap:
@@ -181,7 +181,7 @@ class ArvPathMapper(PathMapper):
             if self.needs_new_collection(s, prefix):
                 return True
         if srcobj.get("listing"):
-            prefix = "%s%s/" % (prefix, srcobj["basename"])
+            prefix = "%s%s/" % (prefix, urllib.parse.quote(srcobj.get("basename", suffix), "/+@"))
             for l in srcobj["listing"]:
                 if self.needs_new_collection(l, prefix):
                     return True
@@ -211,12 +211,9 @@ class ArvPathMapper(PathMapper):
                                          packed=False)
 
         for src, ab, st in uploadfiles:
-            print("BBBBB", src, ab, st.fn, urllib.parse.quote(st.fn, "/:+@"))
             self._pathmap[src] = MapperEnt(urllib.parse.quote(st.fn, "/:+@"), urllib.parse.quote(self.collection_pattern % st.fn[5:], "/:+@"),
                                            "Directory" if os.path.isdir(ab) else "File", True)
 
-        print("CCCCC", self._pathmap)
-
         for srcobj in referenced_files:
             remap = []
             if srcobj["class"] == "Directory" and srcobj["location"] not in self._pathmap: