From 808e1cd3b68a77d9da56da7b06d8ec317f388200 Mon Sep 17 00:00:00 2001
From: CaroLS <5789321+kittylon@users.noreply.github.com>
Date: Mon, 2 Mar 2026 16:42:38 +0100
Subject: [PATCH 1/2] fix: dump the entire data into one JSON, to send the
 Content-Length header

---
 verdin/datasource.py | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/verdin/datasource.py b/verdin/datasource.py
index 03115cb..976a510 100644
--- a/verdin/datasource.py
+++ b/verdin/datasource.py
@@ -147,9 +147,7 @@ def append_ndjson(self, records: List[Dict]) -> requests.Response:
         :return: The HTTP response
         """
 
-        def _ndjson_iterator():
-            for record in records:
-                yield json.dumps(record) + "\n"
+        data = "".join(json.dumps(record) + "\n" for record in records)
 
         LOG.debug(
             "appending %d ndjson records to %s via %s",
@@ -160,7 +158,7 @@ def _ndjson_iterator():
         response = self._datasources_api.append(
             name=self.canonical_name,
             format="ndjson",
-            data=_ndjson_iterator(),
+            data=data,
         )
 
         return response._response

From 30e81892f8de10c96c8e4c4b9354d14eb80a4dde Mon Sep 17 00:00:00 2001
From: CaroLS <5789321+kittylon@users.noreply.github.com>
Date: Mon, 2 Mar 2026 19:27:22 +0100
Subject: [PATCH 2/2] feat: add test to the Datasource append_ndjson

---
 tests/test_datasource.py | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/tests/test_datasource.py b/tests/test_datasource.py
index 5510c5a..894f9f8 100644
--- a/tests/test_datasource.py
+++ b/tests/test_datasource.py
@@ -44,6 +44,28 @@ def handler(request):
         assert response.ok
 
 
+    def test_append_ndjson(self, httpserver: HTTPServer):
+        ds = Datasource("mydatasource", "123456", api=httpserver.url_for("/"))
+
+        records = [{"key": "foo", "value": "bar"}, {"key": "baz", "value": "ed"}]
+        expected_body = '{"key": "foo", "value": "bar"}\n{"key": "baz", "value": "ed"}\n'
+
+        httpserver.expect_request(
+            "/v0/datasources",
+            query_string={
+                "name": "mydatasource",
+                "mode": "append",
+                "format": "ndjson",
+            },
+            data=expected_body,
+            headers={"Content-Length": str(len(expected_body.encode()))},
+        ).respond_with_data("", 200)
+
+        response = ds.append_ndjson(records)
+        httpserver.check()
+        assert response.ok
+
+
 class TestFileDatasource:
     def test_append(self, tmp_path):
         file_path = tmp_path / "myfile.csv"