11510: Fix longstanding bug in replace_range() when appending data from a buffer...
[arvados.git] / sdk / python / tests / test_arvfile.py
index ea8661437c04f3c0d2dec8d4a01c0c8176a2e87c..1a2c034802312eb2c3e4ea0f89b377e2ec927bc4 100644 (file)
@@ -1,6 +1,7 @@
 #!/usr/bin/env python
 
 import bz2
+import datetime
 import gzip
 import io
 import mock
@@ -28,7 +29,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
         def get_from_cache(self, locator):
             self.requests.append(locator)
             return self.blocks.get(locator)
-        def put(self, data, num_retries=None):
+        def put(self, data, num_retries=None, copies=None):
             pdh = tutil.str_keep_locator(data)
             self.blocks[pdh] = str(data)
             return pdh
@@ -37,6 +38,11 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
         def __init__(self, b, r):
             self.body = b
             self.response = r
+            self._schema = ArvadosFileWriterTestCase.MockApi.MockSchema()
+            self._rootDesc = {}
+        class MockSchema(object):
+            def __init__(self):
+                self.schemas = {'Collection': {'properties': {'replication_desired': {'type':'integer'}}}}
         class MockCollections(object):
             def __init__(self, b, r):
                 self.body = b
@@ -59,9 +65,11 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_truncate(self):
         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
         api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate",
-                                                 "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"},
+                                                 "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
+                                                 "replication_desired":None},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
-                                                 "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"})
+                                                 "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
+                                                 "portable_data_hash":"7fcd0eaac3aad4c31a6a0e756475da92+52"})
         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
                              api_client=api, keep_client=keep) as c:
             writer = c.open("count.txt", "r+")
@@ -83,12 +91,42 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
             self.assertFalse(c.modified())
 
+
+    def test_truncate2(self):
+        keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
+        api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate2",
+                                                 "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 7f614da9329cd3aebf59b91aadc30bf0+67108864 0:12:count.txt\n",
+                                                 "replication_desired":None},
+                                                {"uuid":"zzzzz-4zz18-mockcollection0",
+                                                 "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 7f614da9329cd3aebf59b91aadc30bf0+67108864 0:12:count.txt\n",
+                                                 "portable_data_hash":"272da898abdf86ddc71994835e3155f8+95"})
+        with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
+                             api_client=api, keep_client=keep) as c:
+            writer = c.open("count.txt", "r+")
+            self.assertEqual(writer.size(), 10)
+            self.assertEqual("0123456789", writer.read(12))
+
+            # extend file size
+            writer.truncate(12)
+
+            self.assertEqual(writer.size(), 12)
+            writer.seek(0, os.SEEK_SET)
+            self.assertEqual(b"0123456789\x00\x00", writer.read(12))
+
+            self.assertIsNone(c.manifest_locator())
+            self.assertTrue(c.modified())
+            c.save_new("test_truncate2")
+            self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
+            self.assertFalse(c.modified())
+
     def test_write_to_end(self):
         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
         api = ArvadosFileWriterTestCase.MockApi({"name":"test_append",
-                                                 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"},
+                                                 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
+                                                 "replication_desired":None},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
-                                                 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"})
+                                                 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
+                                                 "portable_data_hash":"c5c3af76565c8efb6a806546bcf073f3+88"})
         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
                              api_client=api, keep_client=keep) as c:
             writer = c.open("count.txt", "r+")
@@ -222,9 +260,11 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_write_large(self):
         keep = ArvadosFileWriterTestCase.MockKeep({})
         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
-                                                 "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"},
+                                                 "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
+                                                 "replication_desired":None},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
-                                                 "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"})
+                                                 "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
+                                                 "portable_data_hash":"9132ca8e3f671c76103a38f5bc24328c+108"})
         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
                              api_client=api, keep_client=keep) as c:
             writer = c.open("count.txt", "r+")
@@ -256,6 +296,57 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
 
             self.assertEqual(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n")
 
+    def test_sparse_write(self):
+        keep = ArvadosFileWriterTestCase.MockKeep({})
+        api = ArvadosFileWriterTestCase.MockApi({}, {})
+        with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
+                             api_client=api, keep_client=keep) as c:
+            writer = c.open("count.txt", "r+")
+            self.assertEqual(writer.size(), 0)
+
+            text = "0123456789"
+            writer.seek(2)
+            writer.write(text)
+            self.assertEqual(writer.size(), 12)
+            writer.seek(0, os.SEEK_SET)
+            self.assertEqual(writer.read(), b"\x00\x00"+text)
+
+            self.assertEqual(c.manifest_text(), ". 7f614da9329cd3aebf59b91aadc30bf0+67108864 781e5e245d69b566979b86e28d23f2c7+10 0:2:count.txt 67108864:10:count.txt\n")
+
+
+    def test_sparse_write2(self):
+        keep = ArvadosFileWriterTestCase.MockKeep({})
+        api = ArvadosFileWriterTestCase.MockApi({}, {})
+        with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
+                             api_client=api, keep_client=keep) as c:
+            writer = c.open("count.txt", "r+")
+            self.assertEqual(writer.size(), 0)
+
+            text = "0123456789"
+            writer.seek((arvados.config.KEEP_BLOCK_SIZE*2) + 2)
+            writer.write(text)
+            self.assertEqual(writer.size(), (arvados.config.KEEP_BLOCK_SIZE*2) + 12)
+            writer.seek(0, os.SEEK_SET)
+
+            self.assertEqual(c.manifest_text(), ". 7f614da9329cd3aebf59b91aadc30bf0+67108864 781e5e245d69b566979b86e28d23f2c7+10 0:67108864:count.txt 0:67108864:count.txt 0:2:count.txt 67108864:10:count.txt\n")
+
+
+    def test_sparse_write3(self):
+        keep = ArvadosFileWriterTestCase.MockKeep({})
+        api = ArvadosFileWriterTestCase.MockApi({}, {})
+        for r in [[0, 1, 2, 3, 4], [4, 3, 2, 1, 0], [3, 2, 0, 4, 1]]:
+            with Collection() as c:
+                writer = c.open("count.txt", "r+")
+                self.assertEqual(writer.size(), 0)
+
+                for i in r:
+                    w = ("%s" % i) * 10
+                    writer.seek(i*10)
+                    writer.write(w)
+                writer.seek(0)
+                self.assertEqual(writer.read(), "00000000001111111111222222222233333333334444444444")
+
+
     def test_rewrite_on_empty_file(self):
         keep = ArvadosFileWriterTestCase.MockKeep({})
         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
@@ -313,9 +404,11 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_write_large_rewrite(self):
         keep = ArvadosFileWriterTestCase.MockKeep({})
         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
-                                                 "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"},
+                                                 "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n",
+                                                 "replication_desired":None},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
-                                                 "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"})
+                                                 "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n",
+                                                 "portable_data_hash":"217665c6b713e1b78dfba7ebd42344db+156"})
         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
                              api_client=api, keep_client=keep) as c:
             writer = c.open("count.txt", "r+")
@@ -335,9 +428,11 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_create(self):
         keep = ArvadosFileWriterTestCase.MockKeep({})
         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
-                                                 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
+                                                 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
+                                                 "replication_desired":None},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
-                                                 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
+                                                 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
+                                                 "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51"})
         with Collection(api_client=api, keep_client=keep) as c:
             writer = c.open("count.txt", "w+")
             self.assertEqual(writer.size(), 0)
@@ -356,9 +451,11 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_create_subdir(self):
         keep = ArvadosFileWriterTestCase.MockKeep({})
         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
-                                                 "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
+                                                 "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
+                                                 "replication_desired":None},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
-                                                 "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
+                                                 "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
+                                                 "portable_data_hash":"1b02aaa62528d28a5be41651cbb9d7c7+59"})
         with Collection(api_client=api, keep_client=keep) as c:
             self.assertIsNone(c.api_response())
             writer = c.open("foo/bar/count.txt", "w+")
@@ -371,9 +468,11 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_overwrite(self):
         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
-                                                 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
+                                                 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
+                                                 "replication_desired":None},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
-                                                 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
+                                                 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
+                                                 "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51"})
         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
                              api_client=api, keep_client=keep) as c:
             writer = c.open("count.txt", "w+")
@@ -400,9 +499,11 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_create_multiple(self):
         keep = ArvadosFileWriterTestCase.MockKeep({})
         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
-                                                 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"},
+                                                 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
+                                                 "replication_desired":None},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
-                                                 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"})
+                                                 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
+                                                 "portable_data_hash":"71e7bb6c00d31fc2b4364199fd97be08+102"})
         with Collection(api_client=api, keep_client=keep) as c:
             w1 = c.open("count1.txt", "w")
             w2 = c.open("count2.txt", "w")
@@ -557,6 +658,26 @@ class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
     def read_for_test(self, reader, byte_count, **kwargs):
         return ''.join(reader.readlines(**kwargs))
 
+
+class ArvadosFileTestCase(unittest.TestCase):
+    def datetime_to_hex(self, dt):
+        return hex(int(time.mktime(dt.timetuple())))[2:]
+
+    def test_permission_expired(self):
+        base_manifest = ". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@{} 0:10:count.txt\n"
+        now = datetime.datetime.now()
+        a_week_ago = now - datetime.timedelta(days=7)
+        a_month_ago = now - datetime.timedelta(days=30)
+        a_week_from_now = now + datetime.timedelta(days=7)
+        with Collection(base_manifest.format(self.datetime_to_hex(a_week_from_now))) as c:
+            self.assertFalse(c.find('count.txt').permission_expired())
+        with Collection(base_manifest.format(self.datetime_to_hex(a_week_ago))) as c:
+            f = c.find('count.txt')
+            self.assertTrue(f.permission_expired())
+            self.assertTrue(f.permission_expired(a_week_from_now))
+            self.assertFalse(f.permission_expired(a_month_ago))
+
+
 class BlockManagerTest(unittest.TestCase):
     def test_bufferblock_append(self):
         keep = ArvadosFileWriterTestCase.MockKeep({})