Fixes KeyError when retrieving empty but existing object from S3 (#771)

* fix: Fixes KeyError when retrieving empty file from S3 * Add test
piskvorky · Sep 6, 2023 · 085b711 · 085b711
1 parent 3bbfc5d
commit 085b711
Show file tree

Hide file tree

Showing 2 changed files with 31 additions and 6 deletions.
diff --git a/smart_open/s3.py b/smart_open/s3.py
@@ -324,10 +324,13 @@ def open(
 
 def _get(client, bucket, key, version, range_string):
     try:
+        params = dict(Bucket=bucket, Key=key)
         if version:
-            return client.get_object(Bucket=bucket, Key=key, VersionId=version, Range=range_string)
-        else:
-            return client.get_object(Bucket=bucket, Key=key, Range=range_string)
+            params["VersionId"] = version
+        if range_string:
+            params["Range"] = range_string
+
+        return client.get_object(**params)
     except botocore.client.ClientError as error:
         wrapped_error = IOError(
             'unable to access bucket: %r key: %r version: %r error: %s' % (
@@ -447,8 +450,19 @@ def _open_body(self, start=None, stop=None):
             error_response = _unwrap_ioerror(ioe)
             if error_response is None or error_response.get('Code') != _OUT_OF_RANGE:
                 raise
-            self._position = self._content_length = int(error_response['ActualObjectSize'])
-            self._body = io.BytesIO()
+            try:
+                self._position = self._content_length = int(error_response['ActualObjectSize'])
+                self._body = io.BytesIO()
+            except KeyError:
+                response = _get(
+                    self._client,
+                    self._bucket,
+                    self._key,
+                    self._version_id,
+                    None,
+                )
+                self._position = self._content_length = response["ContentLength"]
+                self._body = response["Body"]
         else:
             #
             # Keep track of how many times boto3's built-in retry mechanism
@@ -461,7 +475,7 @@ def _open_body(self, start=None, stop=None):
                 self,
                 response['ResponseMetadata']['RetryAttempts'],
             )
-            units, start, stop, length = smart_open.utils.parse_content_range(response['ContentRange'])
+            _, start, stop, length = smart_open.utils.parse_content_range(response['ContentRange'])
             self._content_length = length
             self._position = start
             self._body = response['Body']

diff --git a/smart_open/tests/test_s3.py b/smart_open/tests/test_s3.py
@@ -73,6 +73,8 @@ def mock_get(*args, **kwargs):
                 error_response['ActualObjectSize'] = actual_size
                 error_response['Code'] = 'InvalidRange'
                 error_response['Message'] = 'The requested range is not satisfiable'
+            if actual_size is None:
+                error_response.pop('ActualObjectSize', None)
             raise
 
     with mock.patch('smart_open.s3._get', new=mock_get):
@@ -399,6 +401,15 @@ def test_read_empty_file(self):
 
         self.assertEqual(data, b'')
 
+    def test_read_empty_file_no_actual_size(self):
+        _resource('s3').Object(BUCKET_NAME, KEY_NAME).put(Body=b'')
+
+        with self.assertApiCalls(GetObject=2), patch_invalid_range_response(None):
+            with smart_open.s3.Reader(BUCKET_NAME, KEY_NAME) as fin:
+                data = fin.read()
+
+        self.assertEqual(data, b'')
+
 
 @moto.mock_s3
 class MultipartWriterTest(unittest.TestCase):