[keshavdandeva]

b/570669133

Summary

Fixes BigQueryBaseStruct.toString() (and ResultSet.getString() on STRUCT / ARRAY<STRUCT> columns) so that nested ARRAY fields inside a STRUCT have their elements formatted as valid JSON values instead of unquoted Arrays.deepToString() strings.

Problem

BigQueryBaseStruct.toString() constructs a JSON object string ({...}) and quotes scalar STRING, DATE, TIME, DATETIME, TIMESTAMP, GEOGRAPHY, and Base64-encoded BYTES attributes. However, when a STRUCT contained a nested ARRAY attribute (BigQueryBaseArray), BigQueryBaseStruct.toString() fell through to value.toString(), which called BigQueryBaseArray.toString() (Arrays.deepToString). This produced unquoted array elements inside the JSON object (e.g., {"str_arr":[hello, world],"date_arr":[2026-01-01, 2026-01-02]}), resulting in syntactically invalid JSON.

Solution

• Extracted the existing scalar JSON value formatting logic in BigQueryBaseStruct.toString() into a shared helper method appendJsonValue(...).
• When a STRUCT attribute is a BigQueryBaseArray, iterate over its elements and format each element using appendJsonValue(...) inside [...].
• Top-level BigQueryBaseArray.toString() behavior remains unchanged for backwards compatibility.

Example Output

For SELECT STRUCT(1 AS id, ['hello', 'world'] AS str_arr, [DATE '2026-01-01'] AS date_arr, [TIME '12:34:56.789012'] AS time_arr, [TIMESTAMP '2026-01-01 00:00:00 UTC'] AS ts_arr, [b'abc'] AS bytes_arr):

• Before: {"id":1,"str_arr":[hello, world],"date_arr":[2026-01-01],"time_arr":[12:34:56.789012],"ts_arr":[2026-01-01 00:00:00.0],"bytes_arr":[YWJj]} (Invalid JSON)
• After: {"id":1,"str_arr":["hello","world"],"date_arr":["2026-01-01"],"time_arr":["12:34:56.789012"],"ts_arr":["2026-01-01 00:00:00.0"],"bytes_arr":["YWJj"]} (Valid JSON)

Testing

• Added struct.toString() assertions covering all BigQuery array types to BigQueryJsonStructTest#structOfArrays and BigQueryArrowStructTest#structOfArrays.
• Added integration test ITBigQueryJDBCTest#testStructWithNestedArraysValidJson verifying both REST and Storage Read API (EnableHighThroughputAPI=1) paths.

[gemini-code-assist]

Code Review

This pull request refactors the toString method in BigQueryBaseStruct to properly serialize nested arrays (BigQueryBaseArray) into valid JSON format. It extracts the JSON formatting logic into a new helper method appendJsonValue and adds unit and integration tests to verify the JSON serialization of structs with nested arrays. There are no review comments, and I have no additional feedback to provide.