diff --git a/tools/preprocess_data.py b/tools/preprocess_data.py
index a81fe8ca7e..9a7fecfd76 100644
--- a/tools/preprocess_data.py
+++ b/tools/preprocess_data.py
@@ -146,7 +146,11 @@ def split_sentences(self, file_name):
     def process_json_file(self, file_name):
         input_file_name, output_prefix = file_name
         print("Opening", input_file_name)
-        fin = open(input_file_name, 'r', encoding='utf-8')
+        # support for gzip files
+        if input_file_name.endswith(".gz"):
+            fin = gzip.open(input_file_name, 'rt')
+        else:
+            fin = open(input_file_name, 'r', encoding='utf-8')
 
         startup_start = time.time()
         encoder = Encoder(self.args)