From 26ab767faa9942c3964743baac4a8163c77e3a1d Mon Sep 17 00:00:00 2001
From: Jack Cushman <jcushman@law.harvard.edu>
Date: Thu, 15 Oct 2020 13:49:42 -0400
Subject: [PATCH] Fix seeking past end of file for s3 backend

---
 smart_open/s3.py            | 15 +++------------
 smart_open/tests/test_s3.py | 38 +++++++++++++++++++++++++++++++++++++
 2 files changed, 41 insertions(+), 12 deletions(-)

diff --git a/smart_open/s3.py b/smart_open/s3.py
index c8c28a15..e29366e2 100644
--- a/smart_open/s3.py
+++ b/smart_open/s3.py
@@ -48,7 +48,7 @@
 _SLEEP_SECONDS = 10
 
 # Returned by AWS when we try to seek beyond EOF.
-_OUT_OF_RANGE = 'Requested Range Not Satisfiable'
+_OUT_OF_RANGE = 'InvalidRange'
 
 
 def parse_uri(uri_as_string):
@@ -385,18 +385,9 @@ def _open_body(self, start=None, stop=None):
         except IOError as ioe:
             # Handle requested content range exceeding content size.
             error_response = _unwrap_ioerror(ioe)
-            if error_response is None or error_response.get('Message') != _OUT_OF_RANGE:
+            if error_response is None or error_response.get('Code') != _OUT_OF_RANGE:
                 raise
-            try:
-                self._position = self._content_length = int(error_response['ActualObjectSize'])
-            except KeyError:
-                # This shouldn't happen with real S3, but moto lacks ActualObjectSize.
-                # Reported at https://github.com/spulec/moto/issues/2981
-                self._position = self._content_length = _get(
-                    self._object,
-                    version=self._version_id,
-                    **self._object_kwargs,
-                )['ContentLength']
+            self._position = self._content_length = int(error_response['ActualObjectSize'])
             self._body = io.BytesIO()
         else:
             units, start, stop, length = smart_open.utils.parse_content_range(response['ContentRange'])
diff --git a/smart_open/tests/test_s3.py b/smart_open/tests/test_s3.py
index 857a83b0..164942ba 100644
--- a/smart_open/tests/test_s3.py
+++ b/smart_open/tests/test_s3.py
@@ -108,6 +108,26 @@ def mock_make_request(self, operation_model, *args, **kwargs):
         finally:
             patcher.stop()
 
+    @contextmanager
+    def patch_invalid_range_response(self, actual_size):
+        """ Work around a bug in moto (https://github.com/spulec/moto/issues/2981) where the
+         API response doesn't match when requesting an invalid range of bytes from an S3 GetObject. """
+        _real_get = smart_open.s3._get
+
+        def mock_get(*args, **kwargs):
+            try:
+                return _real_get(*args, **kwargs)
+            except IOError as ioe:
+                error_response = smart_open.s3._unwrap_ioerror(ioe)
+                if error_response and error_response.get('Message') == 'Requested Range Not Satisfiable':
+                    error_response['ActualObjectSize'] = actual_size
+                    error_response['Code'] = 'InvalidRange'
+                    error_response['Message'] = 'The requested range is not satisfiable'
+                raise
+
+        with patch('smart_open.s3._get', new=mock_get):
+            yield
+
 
 @unittest.skipUnless(
     ENABLE_MOTO_SERVER,
@@ -236,6 +256,15 @@ def test_seek_end(self):
             self.assertEqual(seek, len(content) - 4)
             self.assertEqual(fin.read(), b'you?')
 
+    def test_seek_past_end(self):
+        content = u"hello wořld\nhow are you?".encode('utf8')
+        put_to_bucket(contents=content)
+
+        with self.assertApiCalls(GetObject=1), self.patch_invalid_range_response(str(len(content))):
+            fin = smart_open.s3.SeekableBufferedInputBase(BUCKET_NAME, KEY_NAME, defer_seek=True)
+            seek = fin.seek(60)
+            self.assertEqual(seek, len(content))
+
     def test_detect_eof(self):
         content = u"hello wořld\nhow are you?".encode('utf8')
         put_to_bucket(contents=content)
@@ -352,6 +381,15 @@ def test_defer_seek(self):
             fin.seek(10)
             self.assertEqual(fin.read(), content[10:])
 
+    def test_read_empty_file(self):
+        put_to_bucket(contents=b'')
+
+        with self.assertApiCalls(GetObject=1), self.patch_invalid_range_response('0'):
+            with smart_open.s3.SeekableBufferedInputBase(BUCKET_NAME, KEY_NAME) as fin:
+                data = fin.read()
+
+        self.assertEqual(data, b'')
+
 
 @moto.mock_s3
 class MultipartWriterTest(unittest.TestCase):