X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/df5c912a9eb5af7222e5446bc437ee97262542c8..7bbb0940b0398d17bb85fe410c3f23e4cd6ec8d6:/sdk/python/tests/test_collections.py

diff --git a/sdk/python/tests/test_collections.py b/sdk/python/tests/test_collections.py
index ac18c44c68..66f062c167 100644
--- a/sdk/python/tests/test_collections.py
+++ b/sdk/python/tests/test_collections.py
@@ -836,17 +836,58 @@ class CollectionOpenModes(run_test_server.TestCaseWithServers):
         with c.open('foo', 'wb') as f:
             f.write('foo')
         for mode in ['r', 'rt', 'r+', 'rt+', 'w', 'wt', 'a', 'at']:
-            if sys.version_info >= (3, 0):
-                with self.assertRaises(NotImplementedError):
-                    c.open('foo', mode)
-            else:
-                with c.open('foo', mode) as f:
-                    if mode[0] == 'r' and '+' not in mode:
-                        self.assertEqual('foo', f.read(3))
-                    else:
-                        f.write('bar')
-                        f.seek(-3, os.SEEK_CUR)
-                        self.assertEqual('bar', f.read(3))
+            with c.open('foo', mode) as f:
+                if mode[0] == 'r' and '+' not in mode:
+                    self.assertEqual('foo', f.read(3))
+                else:
+                    f.write('bar')
+                    f.seek(0, os.SEEK_SET)
+                    self.assertEqual('bar', f.read(3))
+
+
+class TextModes(run_test_server.TestCaseWithServers):
+
+    def setUp(self):
+        arvados.config.KEEP_BLOCK_SIZE = 4
+        if sys.version_info < (3, 0):
+            import unicodedata
+            self.sailboat = unicodedata.lookup('SAILBOAT')
+            self.snowman = unicodedata.lookup('SNOWMAN')
+        else:
+            self.sailboat = '\N{SAILBOAT}'
+            self.snowman = '\N{SNOWMAN}'
+
+    def tearDown(self):
+        arvados.config.KEEP_BLOCK_SIZE = 2 ** 26
+
+    def test_read_sailboat_across_block_boundary(self):
+        c = Collection()
+        f = c.open('sailboats', 'wb')
+        data = self.sailboat.encode('utf-8')
+        f.write(data)
+        f.write(data[:1])
+        f.write(data[1:])
+        f.write(b'\n')
+        f.close()
+        self.assertRegex(c.portable_manifest_text(), r'\+4 .*\+3 ')
+
+        f = c.open('sailboats', 'r')
+        string = f.readline()
+        self.assertEqual(string, self.sailboat+self.sailboat+'\n')
+        f.close()
+
+    def test_write_snowman_across_block_boundary(self):
+        c = Collection()
+        f = c.open('snowmany', 'w')
+        data = self.snowman
+        f.write(data+data+'\n'+data+'\n')
+        f.close()
+        self.assertRegex(c.portable_manifest_text(), r'\+4 .*\+4 .*\+3 ')
+
+        f = c.open('snowmany', 'r')
+        self.assertEqual(f.readline(), self.snowman+self.snowman+'\n')
+        self.assertEqual(f.readline(), self.snowman+'\n')
+        f.close()
 
 
 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
@@ -911,10 +952,49 @@ class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
         self.assertIs(c.find("./nonexistant.txt"), None)
         self.assertIs(c.find("./nonexistantsubdir/nonexistant.txt"), None)
 
+    def test_escaped_paths_dont_get_unescaped_on_manifest(self):
+        # Dir & file names are literally '\056' (escaped form: \134056)
+        manifest = './\\134056\\040Test d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\134056\n'
+        c = Collection(manifest)
+        self.assertEqual(c.portable_manifest_text(), manifest)
+
+    def test_other_special_chars_on_file_token(self):
+        cases = [
+            ('\\000', '\0'),
+            ('\\011', '\t'),
+            ('\\012', '\n'),
+            ('\\072', ':'),
+            ('\\134400', '\\400'),
+        ]
+        for encoded, decoded in cases:
+            manifest = '. d41d8cd98f00b204e9800998ecf8427e+0 0:0:some%sfile.txt\n' % encoded
+            c = Collection(manifest)
+            self.assertEqual(c.portable_manifest_text(), manifest)
+            self.assertIn('some%sfile.txt' % decoded, c.keys())
+
+    def test_escaped_paths_do_get_unescaped_on_listing(self):
+        # Dir & file names are literally '\056' (escaped form: \134056)
+        manifest = './\\134056\\040Test d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\134056\n'
+        c = Collection(manifest)
+        self.assertIn('\\056 Test', c.keys())
+        self.assertIn('\\056', c['\\056 Test'].keys())
+
+    def test_make_empty_dir_with_escaped_chars(self):
+        c = Collection()
+        c.mkdirs('./Empty\\056Dir')
+        self.assertEqual(c.portable_manifest_text(),
+                         './Empty\\134056Dir d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n')
+
+    def test_make_empty_dir_with_spaces(self):
+        c = Collection()
+        c.mkdirs('./foo bar/baz waz')
+        self.assertEqual(c.portable_manifest_text(),
+                         './foo\\040bar/baz\\040waz d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n')
+
     def test_remove_in_subdir(self):
         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
         c.remove("foo/count2.txt")
-        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
+        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n", c.portable_manifest_text())
 
     def test_remove_empty_subdir(self):
         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')