feat: Add SQL queries support in /v1/sql endpoint#9301
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #9301 +/- ##
==========================================
- Coverage 83.75% 83.75% -0.01%
==========================================
Files 229 229
Lines 82604 82613 +9
==========================================
+ Hits 69186 69191 +5
- Misses 13418 13422 +4
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
mcheshkov
force-pushed
the
sql-endpoint-for-sql-api
branch
from
ba68574 to
29970e1
Compare
| return [queryType, normalizedQueries, queryNormalizationResult.map((it) => remapToQueryAdapterFormat(it.normalizedQuery))]; | ||
| } | ||
|
|
||
| public async sql4sql({ |
There was a problem hiding this comment.
TBH, I'm not sure.
I just copy-pasted public async sql({, but I see no reason to keep this public
There was a problem hiding this comment.
AFAIK, these methods are public because it can be called from SubscriptionServer, it's used in websocket protocol.
| Ok(promise.upcast::<JsValue>()) | ||
| } | ||
|
|
||
| fn sql4sql(mut cx: FunctionContext) -> JsResult<JsValue> { |
There was a problem hiding this comment.
Don't you mind moving all these functions to sub module (like it was done for orchestrator/planner/python)? - To keep this file clean and hold only kind of entry points, register interfaces, while having all the handlers in a separate sub modules.
| "sql": "SELECT | ||
| sum(\\"orders\\".amount) \\"total\\" | ||
| FROM | ||
| ( | ||
| select 1 as id, 100 as amount, 'new' status, '2024-01-01'::timestamptz created_at | ||
| UNION ALL | ||
| select 2 as id, 200 as amount, 'new' status, '2024-01-02'::timestamptz created_at | ||
| UNION ALL | ||
| select 3 as id, 300 as amount, 'processed' status, '2024-01-03'::timestamptz created_at | ||
| UNION ALL | ||
| select 4 as id, 500 as amount, 'processed' status, '2024-01-04'::timestamptz created_at | ||
| UNION ALL | ||
| select 5 as id, 600 as amount, 'shipped' status, '2024-01-05'::timestamptz created_at | ||
| ) AS \\"orders\\" WHERE (\\"orders\\".status = $1)", | ||
| "values": Array [ | ||
| "foo", | ||
| ], |
There was a problem hiding this comment.
@mcheshkov I think we're breaking the existing contract here. I thought that sql.sql will be a two element array, with first element being a string, second being an array of parameters.
Here's what /sql currently returns:
{
"sql": {
"sql": [
"SELECT\n `products`.product_category `orders__product_category`, date_trunc('year', from_utc_timestamp(`base_orders`.created_at::TIMESTAMP, 'UTC')) `orders__created_at_year`\n FROM\n hive_metastore.default.orders AS `base_orders`\nLEFT JOIN hive_metastore.default.line_items AS `line_items` ON `base_orders`.id = `line_items`.order_id\nLEFT JOIN hive_metastore.default.products AS `products` ON `line_items`.product_id = `products`.id WHERE (`base_orders`.created_at::TIMESTAMP >= from_utc_timestamp(replace(replace(?, 'T', ' '), 'Z', ''), 'UTC') AND `base_orders`.created_at::TIMESTAMP <= from_utc_timestamp(replace(replace(?, 'T', ' '), 'Z', ''), 'UTC')) GROUP BY 1, 2 ORDER BY 2 ASC LIMIT 10000",
[
"2025-03-06T00:00:00.000Z",
"2025-03-06T23:59:59.999Z"
]
],
// skippedThe current way is definitely ugly but I bet we should not break it for no good reason.
mcheshkov
force-pushed
the
sql-endpoint-for-sql-api
branch
from
dd2efc4 to
ebca109
Compare
mcheshkov
force-pushed
the
sql-endpoint-for-sql-api
branch
from
ebca109 to
6e68b86
Compare
* refactor(cubesql): Use &str instead of &String * refactor(backend-native): Extract create_session function * refactor(backend-native): Extract with_session function * refactor(cubesql): Extract QueryPlan::try_as_logical_plan * feat: Add SQL queries support in /v1/sql endpoint * Add docs * Remove mention of data_source and query_plan response fields from /v1/sql docs --------- Co-authored-by: Igor Lukanin <igor@cube.dev>
5 participants
Check List
Description of Changes Made (if issue reference is not provided)
/v1/sqlcan now generate SQL for SQL API queries. It errors out for meta-only queries (likeCREATE TEMPORARY TABLE) and post-processing queries (likeSELECT version();), and works only when logical plan root is eitherCubeScanorCubeScanWrappedSql