X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/9828e9218084856240fdeafa2d388d8bf322e655..09cbdc3074b3f1e69c9c537875146f6da0a6ed8f:/sdk/cwl/tests/test_http.py

diff --git a/sdk/cwl/tests/test_http.py b/sdk/cwl/tests/test_http.py
index 650b5f0598..5598b1f138 100644
--- a/sdk/cwl/tests/test_http.py
+++ b/sdk/cwl/tests/test_http.py
@@ -58,7 +58,7 @@ class TestHttpToKeep(unittest.TestCase):
         r = arvados_cwl.http.http_to_keep(api, None, "http://example.com/file1.txt", utcnow=utcnow)
         self.assertEqual(r, "keep:99999999999999999999999999999998+99/file1.txt")
 
-        getmock.assert_called_with("http://example.com/file1.txt", stream=True, allow_redirects=True)
+        getmock.assert_called_with("http://example.com/file1.txt", stream=True, allow_redirects=True, headers={})
 
         cm.open.assert_called_with("file1.txt", "wb")
         cm.save_new.assert_called_with(name="Downloaded from http%3A%2F%2Fexample.com%2Ffile1.txt",
@@ -186,7 +186,7 @@ class TestHttpToKeep(unittest.TestCase):
         r = arvados_cwl.http.http_to_keep(api, None, "http://example.com/file1.txt", utcnow=utcnow)
         self.assertEqual(r, "keep:99999999999999999999999999999997+99/file1.txt")
 
-        getmock.assert_called_with("http://example.com/file1.txt", stream=True, allow_redirects=True)
+        getmock.assert_called_with("http://example.com/file1.txt", stream=True, allow_redirects=True, headers={})
 
         cm.open.assert_called_with("file1.txt", "wb")
         cm.save_new.assert_called_with(name="Downloaded from http%3A%2F%2Fexample.com%2Ffile1.txt",
@@ -212,7 +212,7 @@ class TestHttpToKeep(unittest.TestCase):
                     'http://example.com/file1.txt': {
                         'Date': 'Tue, 15 May 2018 00:00:00 GMT',
                         'Expires': 'Tue, 16 May 2018 00:00:00 GMT',
-                        'ETag': '123456'
+                        'ETag': '"123456"'
                     }
                 }
             }]
@@ -229,7 +229,7 @@ class TestHttpToKeep(unittest.TestCase):
         req.headers = {
             'Date': 'Tue, 17 May 2018 00:00:00 GMT',
             'Expires': 'Tue, 19 May 2018 00:00:00 GMT',
-            'ETag': '123456'
+            'ETag': '"123456"'
         }
         headmock.return_value = req
 
@@ -247,7 +247,7 @@ class TestHttpToKeep(unittest.TestCase):
                       body={"collection":{"properties": {'http://example.com/file1.txt': {
                           'Date': 'Tue, 17 May 2018 00:00:00 GMT',
                           'Expires': 'Tue, 19 May 2018 00:00:00 GMT',
-                          'ETag': '123456'
+                          'ETag': '"123456"'
                       }}}})
                       ])
 
@@ -277,7 +277,7 @@ class TestHttpToKeep(unittest.TestCase):
         r = arvados_cwl.http.http_to_keep(api, None, "http://example.com/download?fn=/file1.txt", utcnow=utcnow)
         self.assertEqual(r, "keep:99999999999999999999999999999998+99/file1.txt")
 
-        getmock.assert_called_with("http://example.com/download?fn=/file1.txt", stream=True, allow_redirects=True)
+        getmock.assert_called_with("http://example.com/download?fn=/file1.txt", stream=True, allow_redirects=True, headers={})
 
         cm.open.assert_called_with("file1.txt", "wb")
         cm.save_new.assert_called_with(name="Downloaded from http%3A%2F%2Fexample.com%2Fdownload%3Ffn%3D%2Ffile1.txt",
@@ -287,3 +287,156 @@ class TestHttpToKeep(unittest.TestCase):
             mock.call(uuid=cm.manifest_locator(),
                       body={"collection":{"properties": {"http://example.com/download?fn=/file1.txt": {'Date': 'Tue, 15 May 2018 00:00:00 GMT'}}}})
         ])
+
+    @mock.patch("requests.get")
+    @mock.patch("requests.head")
+    @mock.patch("arvados.collection.CollectionReader")
+    def test_http_etag_if_none_match(self, collectionmock, headmock, getmock):
+        api = mock.MagicMock()
+
+        api.collections().list().execute.return_value = {
+            "items": [{
+                "uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz3",
+                "portable_data_hash": "99999999999999999999999999999998+99",
+                "properties": {
+                    'http://example.com/file1.txt': {
+                        'Date': 'Tue, 15 May 2018 00:00:00 GMT',
+                        'Expires': 'Tue, 16 May 2018 00:00:00 GMT',
+                        'ETag': '"123456"'
+                    }
+                }
+            }]
+        }
+
+        cm = mock.MagicMock()
+        cm.manifest_locator.return_value = "zzzzz-4zz18-zzzzzzzzzzzzzz3"
+        cm.portable_data_hash.return_value = "99999999999999999999999999999998+99"
+        cm.keys.return_value = ["file1.txt"]
+        collectionmock.return_value = cm
+
+        # Head request fails, will try a conditional GET instead
+        req = mock.MagicMock()
+        req.status_code = 403
+        req.headers = {
+        }
+        headmock.return_value = req
+
+        utcnow = mock.MagicMock()
+        utcnow.return_value = datetime.datetime(2018, 5, 17)
+
+        req = mock.MagicMock()
+        req.status_code = 304
+        req.headers = {
+            'Date': 'Tue, 17 May 2018 00:00:00 GMT',
+            'Expires': 'Tue, 19 May 2018 00:00:00 GMT',
+            'ETag': '"123456"'
+        }
+        getmock.return_value = req
+
+        r = arvados_cwl.http.http_to_keep(api, None, "http://example.com/file1.txt", utcnow=utcnow)
+        self.assertEqual(r, "keep:99999999999999999999999999999998+99/file1.txt")
+
+        getmock.assert_called_with("http://example.com/file1.txt", stream=True, allow_redirects=True, headers={"If-None-Match": '"123456"'})
+        cm.open.assert_not_called()
+
+        api.collections().update.assert_has_calls([
+            mock.call(uuid=cm.manifest_locator(),
+                      body={"collection":{"properties": {'http://example.com/file1.txt': {
+                          'Date': 'Tue, 17 May 2018 00:00:00 GMT',
+                          'Expires': 'Tue, 19 May 2018 00:00:00 GMT',
+                          'ETag': '"123456"'
+                      }}}})
+                      ])
+
+
+    @mock.patch("requests.get")
+    @mock.patch("requests.head")
+    @mock.patch("arvados.collection.CollectionReader")
+    def test_http_prefer_cached_downloads(self, collectionmock, headmock, getmock):
+        api = mock.MagicMock()
+
+        api.collections().list().execute.return_value = {
+            "items": [{
+                "uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz3",
+                "portable_data_hash": "99999999999999999999999999999998+99",
+                "properties": {
+                    'http://example.com/file1.txt': {
+                        'Date': 'Tue, 15 May 2018 00:00:00 GMT',
+                        'Expires': 'Tue, 16 May 2018 00:00:00 GMT',
+                        'ETag': '"123456"'
+                    }
+                }
+            }]
+        }
+
+        cm = mock.MagicMock()
+        cm.manifest_locator.return_value = "zzzzz-4zz18-zzzzzzzzzzzzzz3"
+        cm.portable_data_hash.return_value = "99999999999999999999999999999998+99"
+        cm.keys.return_value = ["file1.txt"]
+        collectionmock.return_value = cm
+
+        utcnow = mock.MagicMock()
+        utcnow.return_value = datetime.datetime(2018, 5, 17)
+
+        r = arvados_cwl.http.http_to_keep(api, None, "http://example.com/file1.txt", utcnow=utcnow, prefer_cached_downloads=True)
+        self.assertEqual(r, "keep:99999999999999999999999999999998+99/file1.txt")
+
+        headmock.assert_not_called()
+        getmock.assert_not_called()
+        cm.open.assert_not_called()
+        api.collections().update.assert_not_called()
+
+    @mock.patch("requests.get")
+    @mock.patch("requests.head")
+    @mock.patch("arvados.collection.CollectionReader")
+    def test_http_varying_url_params(self, collectionmock, headmock, getmock):
+        for prurl in ("http://example.com/file1.txt", "http://example.com/file1.txt?KeyId=123&Signature=456&Expires=789"):
+            api = mock.MagicMock()
+
+            api.collections().list().execute.return_value = {
+                "items": [{
+                    "uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz3",
+                    "portable_data_hash": "99999999999999999999999999999998+99",
+                    "properties": {
+                        prurl: {
+                            'Date': 'Tue, 15 May 2018 00:00:00 GMT',
+                            'Expires': 'Tue, 16 May 2018 00:00:00 GMT',
+                            'ETag': '"123456"'
+                        }
+                    }
+                }]
+            }
+
+            cm = mock.MagicMock()
+            cm.manifest_locator.return_value = "zzzzz-4zz18-zzzzzzzzzzzzzz3"
+            cm.portable_data_hash.return_value = "99999999999999999999999999999998+99"
+            cm.keys.return_value = ["file1.txt"]
+            collectionmock.return_value = cm
+
+            req = mock.MagicMock()
+            req.status_code = 200
+            req.headers = {
+                'Date': 'Tue, 17 May 2018 00:00:00 GMT',
+                'Expires': 'Tue, 19 May 2018 00:00:00 GMT',
+                'ETag': '"123456"'
+            }
+            headmock.return_value = req
+
+            utcnow = mock.MagicMock()
+            utcnow.return_value = datetime.datetime(2018, 5, 17)
+
+            r = arvados_cwl.http.http_to_keep(api, None, "http://example.com/file1.txt?KeyId=123&Signature=456&Expires=789",
+                                              utcnow=utcnow, varying_url_params="KeyId,Signature,Expires")
+            self.assertEqual(r, "keep:99999999999999999999999999999998+99/file1.txt")
+
+            getmock.assert_not_called()
+            cm.open.assert_not_called()
+
+            api.collections().update.assert_has_calls([
+                mock.call(uuid=cm.manifest_locator(),
+                          body={"collection":{"properties": {'http://example.com/file1.txt': {
+                              'Date': 'Tue, 17 May 2018 00:00:00 GMT',
+                              'Expires': 'Tue, 19 May 2018 00:00:00 GMT',
+                              'ETag': '"123456"'
+                          }}}})
+                          ])