4823: Add Collection.copy tests
[arvados.git] / sdk / python / tests / test_arvfile.py
index 9c561da6c00cb04af66be4c01a55e4260b39dcd5..06c70f004d3f8eaffe76fd0119475e022ab5b60e 100644 (file)
@@ -9,10 +9,11 @@ import unittest
 import hashlib
 
 import arvados
-from arvados import ArvadosFile, ArvadosFileReader, Range, import_manifest, export_manifest
+from arvados import ArvadosFile, ArvadosFileReader, Range, import_manifest, export_manifest, KeepLocator
+from arvados.arvfile import SYNC_READONLY, SYNC_EXPLICIT
 
 import arvados_testutil as tutil
-from test_stream import StreamFileReaderTestCase
+from test_stream import StreamFileReaderTestCase, StreamRetryTestMixin
 
 class ArvadosFileWriterTestCase(unittest.TestCase):
     class MockKeep(object):
@@ -56,7 +57,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
-                             api_client=api, keep=keep) as c:
+                             api_client=api, keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "r+")
             self.assertEqual(writer.size(), 10)
             writer.seek(5)
@@ -77,7 +78,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
-                             api_client=api, keep=keep) as c:
+                             api_client=api, keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "r+")
             writer.seek(5, os.SEEK_SET)
             self.assertEqual("56789", writer.read(8))
@@ -98,7 +99,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_write0(self):
         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
-                             keep=keep) as c:
+                             keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "r+")
             self.assertEqual("0123456789", writer.readfrom(0, 13))
             writer.seek(0, os.SEEK_SET)
@@ -110,7 +111,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_write1(self):
         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
-                             keep=keep) as c:
+                             keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "r+")
             self.assertEqual("0123456789", writer.readfrom(0, 13))
             writer.seek(3, os.SEEK_SET)
@@ -122,7 +123,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_write2(self):
         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
-                             keep=keep) as c:
+                             keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "r+")
             self.assertEqual("0123456789", writer.readfrom(0, 13))
             writer.seek(7, os.SEEK_SET)
@@ -134,7 +135,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_write3(self):
         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
-                             keep=keep) as c:
+                             keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "r+")
             self.assertEqual("012345678901234", writer.readfrom(0, 15))
             writer.seek(7, os.SEEK_SET)
@@ -146,7 +147,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_write4(self):
         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
-                             keep=keep) as c:
+                             keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "r+")
             self.assertEqual("012301230123", writer.readfrom(0, 15))
             writer.seek(2, os.SEEK_SET)
@@ -161,7 +162,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
         with import_manifest('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
-                             api_client=api, keep=keep) as c:
+                             api_client=api, keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "r+")
             text = ''.join(["0123456789" for a in xrange(0, 100)])
             for b in xrange(0, 100000):
@@ -177,7 +178,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_write_rewrite0(self):
         keep = ArvadosFileWriterTestCase.MockKeep({})
         with import_manifest('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
-                             keep=keep) as c:
+                             keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "r+")
             for b in xrange(0, 10):
                 writer.seek(0, os.SEEK_SET)
@@ -190,7 +191,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_write_rewrite1(self):
         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
-                             keep=keep) as c:
+                             keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "r+")
             for b in xrange(0, 10):
                 writer.seek(10, os.SEEK_SET)
@@ -203,7 +204,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_write_rewrite2(self):
         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
-                             keep=keep) as c:
+                             keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "r+")
             for b in xrange(0, 10):
                 writer.seek(5, os.SEEK_SET)
@@ -219,7 +220,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
         with import_manifest('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
-                             api_client=api, keep=keep) as c:
+                             api_client=api, keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "r+")
             text = ''.join(["0123456789" for a in xrange(0, 100)])
             for b in xrange(0, 100000):
@@ -239,7 +240,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
-        with arvados.Collection(api_client=api, keep_client=keep) as c:
+        with arvados.Collection(api_client=api, keep_client=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "w+")
             self.assertEqual(writer.size(), 0)
             writer.write("01234567")
@@ -259,7 +260,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
-        with arvados.Collection(api_client=api, keep_client=keep) as c:
+        with arvados.Collection(api_client=api, keep_client=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("foo/bar/count.txt", "w+")
             writer.write("01234567")
             c.save_as("test_create")
@@ -270,7 +271,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
-                             api_client=api, keep=keep) as c:
+                             api_client=api, keep=keep, sync=SYNC_EXPLICIT) as c:
             writer = c.open("count.txt", "w+")
             self.assertEqual(writer.size(), 0)
             writer.write("01234567")
@@ -297,7 +298,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
-        with arvados.Collection(api_client=api, keep_client=keep) as c:
+        with arvados.Collection(api_client=api, keep_client=keep, sync=SYNC_EXPLICIT) as c:
             w1 = c.open("count1.txt", "w")
             w2 = c.open("count2.txt", "w")
             w1.write("01234567")
@@ -313,31 +314,6 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
             self.assertEqual(False, c.modified())
             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 
-    def test_remove(self):
-        with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n') as c:
-            self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", export_manifest(c))
-            self.assertTrue("count1.txt" in c)
-            c.remove("count1.txt")
-            self.assertFalse("count1.txt" in c)
-            self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", export_manifest(c))
-
-    def test_remove_in_subdir(self):
-        with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n') as c:
-            c.remove("foo/count2.txt")
-            self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c))
-
-    def test_remove_subdir(self):
-        with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n') as c:
-            c.remove("foo")
-            self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c))
-
-    def test_prefetch(self):
-        keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
-        with import_manifest(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep=keep) as c:
-            r = c.open("count.txt", "r")
-            self.assertEqual("0123", r.read(4))
-        self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8", "2e9ec317e197819358fbc43afca7d837+8", "e8dc4081b13434b45189a720b77b6818+8"], keep.requests)
-
 
 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
     class MockParent(object):
@@ -361,6 +337,9 @@ class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
         def _my_block_manager(self):
             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
 
+        def sync_mode(self):
+            return SYNC_READONLY
+
     def make_count_reader(self, nocache=False):
         stream = []
         n = 0
@@ -388,20 +367,34 @@ class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
         sfile.read(5)
         self.assertEqual(3, sfile.tell())
 
+    def test_prefetch(self):
+        keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
+        with import_manifest(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep=keep) as c:
+            r = c.open("count.txt", "r")
+            self.assertEqual("0123", r.read(4))
+        self.assertTrue("2e9ec317e197819358fbc43afca7d837+8" in keep.requests)
+        self.assertTrue("e8dc4081b13434b45189a720b77b6818+8" in keep.requests)
+
 
 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
     def reader_for(self, coll_name, **kwargs):
+        stream = []
         segments = []
         n = 0
         for d in self.manifest_for(coll_name).split():
-            k = KeepLocator(loc)
-            segments.append(Range(d, n, k.size))
-            n += k.size
-        af = ArvadosFile(Collection(keep_client=self.keep_client()),
-                         stream=self.manifest_for(coll_name).split(),
-                         segments=segments,
-                         **kwargs)
-        return ArvadosFileReader(af, "test")
+            try:
+                k = KeepLocator(d)
+                segments.append(Range(n, n, k.size))
+                stream.append(Range(d, n, k.size))
+                n += k.size
+            except ValueError:
+                pass
+        col = arvados.Collection(keep_client=self.keep_client())
+        col._my_block_manager().prefetch_enabled = False
+        af = ArvadosFile(col,
+                         stream=stream,
+                         segments=segments)
+        return ArvadosFileReader(af, "test", **kwargs)
 
     def read_for_test(self, reader, byte_count, **kwargs):
         return reader.read(byte_count, **kwargs)