Respect result limits in paginated tool, service and variable queries

CHANGELOG.md

@@ -32,6 +32,11 @@ adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
   having to write cumbersome code to get around the limitation.
   ([#106](https://github.com/nasa/python_cmr/issues/106))
 
+### Fixed
+
+- Respect the requested result limit when tool, service, and variable queries
+  retrieve a partial final page of JSON results with `get`.
+
 ## [0.13.0]
 
 ### Added

cmr/queries.py

@@ -1198,7 +1198,7 @@ def get(self, limit: int = 2000) -> Sequence[Any]:
             if len(latest) == 0:
                 break
 
-            results.extend(latest)
+            results.extend(latest[:limit - len(results)])
             page += 1
 
         return results

tests/test_get_limit.py

@@ -0,0 +1,32 @@
+from unittest.mock import Mock, patch
+
+import pytest
+
+from cmr import ServiceQuery, ToolQuery, VariableQuery
+
+
+@pytest.mark.parametrize("query_class", [ToolQuery, ServiceQuery, VariableQuery])
+@pytest.mark.parametrize("limit", [0, 2001, 4000, 4001, 5000])
+def test_get_respects_limit_across_pages(query_class, limit):
+    records = [{"id": index} for index in range(4500)]
+    pages = []
+
+    def get_page(url, params):
+        page_size = params["page_size"]
+        page = params["page_num"]
+        pages.append((page, page_size))
+        start = (page - 1) * page_size
+        response = Mock()
+        response.json.return_value = {"items": records[start:start + page_size]}
+        return response
+
+    with patch("cmr.queries.requests.get", side_effect=get_page):
+        results = query_class().get(limit)
+
+    assert results == records[:limit]
+    if limit:
+        # Changing page_size between page numbers would skip or repeat records.
+        assert all(page_size == 2000 for _, page_size in pages)
+        assert [page for page, _ in pages] == list(range(1, len(pages) + 1))
+    else:
+        assert pages == []