From f49bc2790a18de14e22e69fb39f93805488fdbb3 Mon Sep 17 00:00:00 2001
From: Peter Amstutz <peter.amstutz@curoverse.com>
Date: Mon, 10 Apr 2017 10:07:23 -0400
Subject: [PATCH] 11332: Use urllib quote/unquote to handle special chars (e.g.
 '#')

---
 sdk/cwl/arvados_cwl/arvcontainer.py |  3 ++-
 sdk/cwl/arvados_cwl/pathmapper.py   | 11 ++++++-----
 sdk/cwl/tests/arvados-tests.sh      |  2 +-
 3 files changed, 9 insertions(+), 7 deletions(-)

diff --git a/sdk/cwl/arvados_cwl/arvcontainer.py b/sdk/cwl/arvados_cwl/arvcontainer.py
index 3090936121..28107b491c 100644
--- a/sdk/cwl/arvados_cwl/arvcontainer.py
+++ b/sdk/cwl/arvados_cwl/arvcontainer.py
@@ -1,6 +1,7 @@
 import logging
 import json
 import os
+import urllib
 
 import ruamel.yaml as yaml
 
@@ -77,7 +78,7 @@ class ArvadosContainer(object):
                     "portable_data_hash": pdh
                 }
                 if len(sp) == 2:
-                    mounts[p]["path"] = sp[1]
+                    mounts[p]["path"] = urllib.unquote(sp[1])
 
         with Perf(metrics, "generatefiles %s" % self.name):
             if self.generatefiles["listing"]:
diff --git a/sdk/cwl/arvados_cwl/pathmapper.py b/sdk/cwl/arvados_cwl/pathmapper.py
index 713162f66a..74401ca85a 100644
--- a/sdk/cwl/arvados_cwl/pathmapper.py
+++ b/sdk/cwl/arvados_cwl/pathmapper.py
@@ -2,6 +2,7 @@ import re
 import logging
 import uuid
 import os
+import urllib
 
 import arvados.commands.run
 import arvados.collection
@@ -34,7 +35,7 @@ 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.
@@ -44,7 +45,7 @@ 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 % st.fn[5:], "File")
+                        self._pathmap[src] = MapperEnt(st.fn, self.collection_pattern % urllib.unquote(st.fn[5:]), "File")
                     elif src.startswith("_:"):
                         if "contents" in srcobj:
                             pass
@@ -59,7 +60,7 @@ class ArvPathMapper(PathMapper):
                     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)
 
@@ -90,7 +91,7 @@ class ArvPathMapper(PathMapper):
             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")
+                self._pathmap[loc] = MapperEnt(v.resolved, self.collection_pattern % urllib.unquote(v.resolved[5:]), "File")
 
         for srcobj in referenced_files:
             self.visit(srcobj, uploadfiles)
@@ -105,7 +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:
diff --git a/sdk/cwl/tests/arvados-tests.sh b/sdk/cwl/tests/arvados-tests.sh
index 86467040a5..2c03812ed3 100755
--- a/sdk/cwl/tests/arvados-tests.sh
+++ b/sdk/cwl/tests/arvados-tests.sh
@@ -2,4 +2,4 @@
 if ! arv-get d7514270f356df848477718d58308cc4+94 > /dev/null ; then
     arv-put --portable-data-hash testdir
 fi
-exec cwltest --test arvados-tests.yml --tool $PWD/runner.sh
+exec cwltest --test arvados-tests.yml --tool $PWD/runner.sh $@
-- 
2.30.2