Fix: properly encode UTF-8 filenames in query results request (#4498)

* Fix: properly encode UTF-8 filenames in query results request

Ended up copying the implementation from Flask's send_file helper function, because send_file doesn't really fit our use case.

* Update tests/handlers/test_query_results.py

Co-Authored-By: Omer Lachish <omer@rauchy.net>

Co-authored-by: Omer Lachish <omer@rauchy.net>

redash/handlers/query_results.py

@@ -1,9 +1,11 @@
 import logging
 import time
 
+import unicodedata
 from flask import make_response, request
 from flask_login import current_user
 from flask_restful import abort
+from werkzeug.urls import url_quote
 from redash import models, settings
 from redash.handlers.base import BaseResource, get_object_or_404, record_event
 from redash.permissions import (
@@ -129,6 +131,25 @@ def get_download_filename(query_result, query, filetype):
     return "{}_{}.{}".format(filename, retrieved_at, filetype)
 
 
+def content_disposition_filenames(attachment_filename):
+    if not isinstance(attachment_filename, str):
+        attachment_filename = attachment_filename.decode("utf-8")
+
+    try:
+        attachment_filename = attachment_filename.encode("ascii")
+    except UnicodeEncodeError:
+        filenames = {
+            "filename": unicodedata.normalize("NFKD", attachment_filename).encode(
+                "ascii", "ignore"
+            ),
+            "filename*": "UTF-8''%s" % url_quote(attachment_filename, safe=b""),
+        }
+    else:
+        filenames = {"filename": attachment_filename}
+
+    return filenames
+
+
 class QueryResultListResource(BaseResource):
     @require_permission("execute_query")
     def post(self):
@@ -383,9 +404,8 @@ def get(self, query_id=None, query_result_id=None, filetype="json"):
 
             filename = get_download_filename(query_result, query, filetype)
 
-            response.headers.add_header(
-                "Content-Disposition", 'attachment; filename="{}"'.format(filename)
-            )
+            filenames = content_disposition_filenames(filename)
+            response.headers.add("Content-Disposition", "attachment", **filenames)
 
             return response
 

tests/handlers/test_query_results.py

@@ -29,6 +29,19 @@ def test_returns_404_if_no_cached_result_found(self):
         self.assertEqual(404, rv.status_code)
 
 
+class TestQueryResultsContentDispositionHeaders(BaseTestCase):
+    def test_supports_unicode(self):
+        query_result = self.factory.create_query_result()
+        query = self.factory.create_query(name="עברית", latest_query_data=query_result)
+
+        rv = self.make_request("get", "/api/queries/{}/results.json".format(query.id))
+        # This is what gunicorn will do with it
+        try:
+            rv.headers['Content-Disposition'].encode('ascii')
+        except Exception as e:
+            self.fail(repr(e))            
+
+
 class TestQueryResultListAPI(BaseTestCase):
     def test_get_existing_result(self):
         query_result = self.factory.create_query_result()