Merge pull request #11797 from internetarchive/copilot/fix-fulltext-search-pagination

Fix fulltext search pagination: all pages returned identical results

Author: RayBB

openlibrary/core/fulltext.py

@@ -46,7 +46,9 @@ async def fulltext_search_api(params):
         return {'error': 'Error converting search engine data to JSON'}
 
 
-async def fulltext_search_async(q, page=1, offset=0, limit=100, js=False, facets=False):
+async def fulltext_search_async(
+    q, page=1, offset=None, limit=100, js=False, facets=False
+):
     if offset is None:
         offset = (page - 1) * limit
     params = {

openlibrary/tests/core/test_fulltext.py

@@ -39,3 +39,31 @@ async def test_bad_json(self, httpx_mock, monkeypatch):
 
         response = await fulltext.fulltext_search_api({"q": "hello"})
         assert response == {"error": "Error converting search engine data to JSON"}
+
+    @pytest.mark.asyncio
+    @pytest.mark.parametrize(
+        ("page", "limit", "offset_kwarg", "expected_from"),
+        [
+            (1, 20, "NOT_PASSED", 0),  # offset not passed at all
+            (1, 20, None, 0),  # offset=None explicitly
+            (10, 20, "NOT_PASSED", 180),  # offset not passed, page 10
+            (5, 20, 100, 100),  # explicit offset provided
+        ],
+    )
+    async def test_pagination_offset_calculation(
+        self, httpx_mock, monkeypatch, page, limit, offset_kwarg, expected_from
+    ):
+        url = "http://mock"
+        monkeypatch.setattr(
+            config, "plugin_inside", {"search_endpoint": url}, raising=False
+        )
+        httpx_mock.add_response(json={"hits": {"hits": []}})
+
+        # Conditionally build kwargs to test "not passed" scenario
+        kwargs = {"page": page, "limit": limit}
+        if offset_kwarg != "NOT_PASSED":
+            kwargs["offset"] = offset_kwarg
+
+        await fulltext.fulltext_search_async("test", **kwargs)
+        request = httpx_mock.get_request()
+        assert f"from={expected_from}" in request.url.query.decode()