From: Brett Smith <brett@curoverse.com>
Date: Fri, 30 May 2014 13:08:50 +0000 (-0400)
Subject: 2752: Avoid ResumeCache name collisions from concatenating paths.
X-Git-Tag: 1.1.0~2587^2~1^2~4
X-Git-Url: https://git.arvados.org/arvados.git/commitdiff_plain/83b6a3fdd9a0fe432758dd44acf252bbd129a930?hp=06cf8e35c69540ae44890f8e96961e31cb7fbe66

2752: Avoid ResumeCache name collisions from concatenating paths.
---

diff --git a/sdk/python/arvados/commands/put.py b/sdk/python/arvados/commands/put.py
index 3229125887..12f464e605 100644
--- a/sdk/python/arvados/commands/put.py
+++ b/sdk/python/arvados/commands/put.py
@@ -172,7 +172,7 @@ class ResumeCache(object):
         md5 = hashlib.md5()
         md5.update(arvados.config.get('ARVADOS_API_HOST', '!nohost'))
         realpaths = sorted(os.path.realpath(path) for path in args.paths)
-        md5.update(''.join(realpaths))
+        md5.update('\0'.join(realpaths))
         if any(os.path.isdir(path) for path in realpaths):
             md5.update(str(max(args.max_manifest_depth, -1)))
         elif args.filename:
diff --git a/sdk/python/tests/test_arv-put.py b/sdk/python/tests/test_arv-put.py
index 41fb33041a..b19e1c5b38 100644
--- a/sdk/python/tests/test_arv-put.py
+++ b/sdk/python/tests/test_arv-put.py
@@ -65,6 +65,12 @@ class ArvadosPutResumeCacheTest(ArvadosBaseTestCase):
             self.cache_path_from_arglist(['-', '--filename', 'stdin']),
             self.cache_path_from_arglist(['--filename', 'stdin', '-']))
 
+    def test_cache_names_differ_for_similar_paths(self):
+        # This test needs names at / that don't exist on the real filesystem.
+        self.assertNotEqual(
+            self.cache_path_from_arglist(['/_arvputtest1', '/_arvputtest2']),
+            self.cache_path_from_arglist(['/_arvputtest1/_arvputtest2']))
+
     def test_cache_names_ignore_irrelevant_arguments(self):
         # Workaround: parse_arguments bails on --filename with a directory.
         path1 = self.cache_path_from_arglist(['/tmp'])