Updated file read to not fail early with a boto empty file exception (#…

…614)

papermill/s3.py

@@ -304,6 +304,12 @@ def cat(
                 elif size != obj.content_length:
                     raise AwsError('key size unexpectedly changed while reading')
 
+                # For an empty file, 0 (first-bytes-pos) is equal to the length of the object
+                # hence the range is "unsatisfiable", and botocore correctly handles it by
+                # raising an exception. We'd rather just return with empty file contents here.
+                if size == 0:
+                    break
+
                 r = obj.get(Range="bytes={}-".format(bytes_read))
 
                 try:

papermill/tests/test_s3.py

@@ -151,6 +151,7 @@ def test_s3_defaults():
 test_bucket_name = 'test-pm-bucket'
 test_string = 'Hello'
 test_file_path = 'notebooks/s3/s3_in/s3-simple_notebook.ipynb'
+test_empty_file_path = 'notebooks/s3/s3_in/s3-empty.ipynb'
 
 with open(os.path.join(local_dir, test_file_path)) as f:
     test_nb_content = f.read()
@@ -169,10 +170,12 @@ def s3_client():
     client = boto3.client('s3')
     client.create_bucket(Bucket=test_bucket_name)
     client.put_object(Bucket=test_bucket_name, Key=test_file_path, Body=test_nb_content)
+    client.put_object(Bucket=test_bucket_name, Key=test_empty_file_path, Body='')
     yield S3()
     try:
         client.delete_object(Bucket=test_bucket_name, Key=test_file_path)
         client.delete_object(Bucket=test_bucket_name, Key=test_file_path + '.txt')
+        client.delete_object(Bucket=test_bucket_name, Key=test_empty_file_path)
     except Exception:
         pass
     mock_s3.stop()
@@ -184,6 +187,12 @@ def test_s3_read(s3_client):
     assert data == test_clean_nb_content
 
 
+def test_s3_read_empty(s3_client):
+    s3_path = "s3://{}/{}".format(test_bucket_name, test_empty_file_path)
+    data = read_from_gen(s3_client.read(s3_path))
+    assert data == ''
+
+
 def test_s3_write(s3_client):
     s3_path = "s3://{}/{}.txt".format(test_bucket_name, test_file_path)
     s3_client.cp_string(test_string, s3_path)
@@ -205,5 +214,5 @@ def test_s3_listdir(s3_client):
     s3_dir = "s3://{}/{}".format(test_bucket_name, dir_name)
     s3_path = "s3://{}/{}".format(test_bucket_name, test_file_path)
     dir_listings = s3_client.listdir(s3_dir)
-    assert len(dir_listings) == 1
+    assert len(dir_listings) == 2
     assert s3_path in dir_listings