put client.search() back in

Author: jason-raitz

CHANGELOG.md

@@ -10,15 +10,16 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Added
 - client method to write search results to an XML file, with validation against expected number of records to be written
 - client method to return an iterator of XML records from a search, to support streaming results for large result sets
+- xml fixture files for future tests
 
 ### Changed
-- N/A
+- README examples
 
 ### Deprecated
 - N/A
 
 ### Removed
-- client method to return raw XML or pyMARC records from a search, in favor of separate methods for each result format and a streaming iterator method for XML results
+- N/A
 
 ### Fixed
 - N/A

README.md

@@ -74,6 +74,11 @@ ids = client.fetch_ids_search("collection:'Disabled Students Program Photos'")
 # return PyMARC records matching a query
 records = client.fetch_search_metadata("collection:'Disabled Students Program Photos'")
 
+# return raw XML or PyMARC records from a paginated search
+# NOTE: for large result sets, use the write_search_results_to_file() method and then parse that file
+xml_results = client.search("collection:'Disabled Students Program Photos'", result_format="xml")
+pymarc_results = client.search("collection:'Disabled Students Program Photos'", result_format="pymarc")
+
 # search Tind with a query and write results to an XML file in the default storage directory
 records_written = client.write_search_results_to_file("Old Emperor Norton", "full_norton_results.xml")
 ```

tests/test_fetch.py

@@ -156,27 +156,27 @@ def test_fetch_ids_search_error(
 # ---------------------------------------------------------------------------
 
 
-# def test_search_invalid_format(client: TINDClient) -> None:
-#     """search raises ValueError for unsupported result_format values."""
-#     with pytest.raises(ValueError, match="Unexpected result format"):
-#         client.search("title:test", result_format="csv")
-
-
-# def test_search_returns_xml(
-#     sample_marc_xml: str,
-#     requests_mock: req_mock.Mocker,
-#     client: TINDClient,
-# ) -> None:
-#     """search returns a list of XML strings when result_format='xml'."""
-#     # Wrap sample XML in a search_id element so pagination terminates correctly.
-#     wrapped = sample_marc_xml.replace(
-#         "<collection",
-#         "<root><search_id></search_id><collection",
-#     ).replace("</collection>", "</collection></root>")
-
-#     requests_mock.get(f"{BASE_URL}/search", text=wrapped, status_code=200)
-
-#     results = client.search("title:sample", result_format="xml")
-#     assert isinstance(results, list)
-#     assert len(results) >= 1
-#     assert requests_mock.call_count == 1
+def test_search_invalid_format(client: TINDClient) -> None:
+    """search raises ValueError for unsupported result_format values."""
+    with pytest.raises(ValueError, match="Unexpected result format"):
+        client.search("title:test", result_format="csv")
+
+
+def test_search_returns_xml(
+    sample_marc_xml: str,
+    requests_mock: req_mock.Mocker,
+    client: TINDClient,
+) -> None:
+    """search returns a list of XML strings when result_format='xml'."""
+    # Wrap sample XML in a search_id element so pagination terminates correctly.
+    wrapped = sample_marc_xml.replace(
+        "<collection",
+        "<root><search_id></search_id><collection",
+    ).replace("</collection>", "</collection></root>")
+
+    requests_mock.get(f"{BASE_URL}/search", text=wrapped, status_code=200)
+
+    results = client.search("title:sample", result_format="xml")
+    assert isinstance(results, list)
+    assert len(results) >= 1
+    assert requests_mock.call_count == 1

tind_client/client.py

@@ -146,6 +146,39 @@ def fetch_search_metadata(self, query: str) -> list[Record]:
         ids = self.fetch_ids_search(query)
         return self.fetch_marc_by_ids(ids)
 
+    def search(self, query: str, result_format: str = "xml") -> list[Any]:
+        """Search TIND and return results as XML strings or PyMARC records.
+
+        :param str query: A TIND search string.
+        :param str result_format: ``'xml'`` for XML strings, ``'pymarc'`` for PyMARC records.
+        :raises ValueError: When ``result_format`` is neither ``'xml'`` nor ``'pymarc'``.
+        :returns list: Records as XML strings or PyMARC Record objects.
+        """
+        if result_format not in ("xml", "pymarc"):
+            raise ValueError(
+                f"Unexpected result format: {result_format} is neither 'xml' nor 'pymarc'"
+            )
+
+        recs: list[Any] = []
+        search_id = None
+
+        while True:
+            response = self._search_request(query, search_id=search_id)
+            xml, search_id = self._retrieve_xml_search_id(response)
+            collection = xml.find("{http://www.loc.gov/MARC21/slim}collection")
+            records = list(collection) if collection is not None else []
+
+            if result_format == "pymarc":
+                recs = recs + parse_xml_to_array(StringIO(response))
+            else:
+                for record in records:
+                    recs.append(E.tostring(record, encoding="unicode"))
+
+            if not search_id or not records:
+                break
+
+        return recs
+
     def write_search_results_to_file(self, query: str, output_file_name: str = "tind.xml") -> int:
         """Search TIND and stream results to an XML file.