mirror of
https://github.com/simstudioai/sim.git
synced 2026-09-24 15:45:35 +08:00
feat(bigquery): add Google BigQuery integration (#3341)
* feat(bigquery): add Google BigQuery integration * fix(bigquery): add auth provider, fix docsLink and insertedRows count * fix(bigquery): set pageToken visibility to user-or-llm for pagination * fix(bigquery): use prefixed export names to avoid aliased imports * lint * improvement(bigquery): destructure tool outputs with structured array/object types * lint
This commit is contained in:
Waleed
@@ -3430,6 +3430,23 @@ export const ResendIcon = (props: SVGProps<SVGSVGElement>) => (
|
||||
</svg>
|
||||
)
|
||||
|
||||
export const GoogleBigQueryIcon = (props: SVGProps<SVGSVGElement>) => (
|
||||
<svg {...props} xmlns='http://www.w3.org/2000/svg' viewBox='0 0 64 64'>
|
||||
<path
|
||||
d='M14.48 58.196L.558 34.082c-.744-1.288-.744-2.876 0-4.164L14.48 5.805c.743-1.287 2.115-2.08 3.6-2.082h27.857c1.48.007 2.845.8 3.585 2.082l13.92 24.113c.744 1.288.744 2.876 0 4.164L49.52 58.196c-.743 1.287-2.115 2.08-3.6 2.082H18.07c-1.483-.005-2.85-.798-3.593-2.082z'
|
||||
fill='#4386fa'
|
||||
/>
|
||||
<path
|
||||
d='M40.697 24.235s3.87 9.283-1.406 14.545-14.883 1.894-14.883 1.894L43.95 60.27h1.984c1.486-.002 2.858-.796 3.6-2.082L58.75 42.23z'
|
||||
opacity='.1'
|
||||
/>
|
||||
<path
|
||||
d='M45.267 43.23L41 38.953a.67.67 0 0 0-.158-.12 11.63 11.63 0 1 0-2.032 2.037.67.67 0 0 0 .113.15l4.277 4.277a.67.67 0 0 0 .947 0l1.12-1.12a.67.67 0 0 0 0-.947zM31.64 40.464a8.75 8.75 0 1 1 8.749-8.749 8.75 8.75 0 0 1-8.749 8.749zm-5.593-9.216v3.616c.557.983 1.363 1.803 2.338 2.375v-6.013zm4.375-2.998v9.772a6.45 6.45 0 0 0 2.338 0V28.25zm6.764 6.606v-2.142H34.85v4.5a6.43 6.43 0 0 0 2.338-2.368z'
|
||||
fill='#fff'
|
||||
/>
|
||||
</svg>
|
||||
)
|
||||
|
||||
export const GoogleVaultIcon = (props: SVGProps<SVGSVGElement>) => (
|
||||
<svg {...props} xmlns='http://www.w3.org/2000/svg' viewBox='0 0 82 82'>
|
||||
<path
|
||||
|
||||
@@ -42,6 +42,7 @@ import {
|
||||
GitLabIcon,
|
||||
GmailIcon,
|
||||
GongIcon,
|
||||
GoogleBigQueryIcon,
|
||||
GoogleBooksIcon,
|
||||
GoogleCalendarIcon,
|
||||
GoogleDocsIcon,
|
||||
@@ -188,6 +189,7 @@ export const blockTypeToIconMap: Record<string, IconComponent> = {
|
||||
gitlab: GitLabIcon,
|
||||
gmail_v2: GmailIcon,
|
||||
gong: GongIcon,
|
||||
google_bigquery: GoogleBigQueryIcon,
|
||||
google_books: GoogleBooksIcon,
|
||||
google_calendar_v2: GoogleCalendarIcon,
|
||||
google_docs: GoogleDocsIcon,
|
||||
|
||||
@@ -0,0 +1,161 @@
|
||||
---
|
||||
title: Google BigQuery
|
||||
description: Query, list, and insert data in Google BigQuery
|
||||
---
|
||||
|
||||
import { BlockInfoCard } from "@/components/ui/block-info-card"
|
||||
|
||||
<BlockInfoCard
|
||||
type="google_bigquery"
|
||||
color="#E0E0E0"
|
||||
/>
|
||||
|
||||
## Usage Instructions
|
||||
|
||||
Connect to Google BigQuery to run SQL queries, list datasets and tables, get table metadata, and insert rows.
|
||||
|
||||
|
||||
|
||||
## Tools
|
||||
|
||||
### `google_bigquery_query`
|
||||
|
||||
Run a SQL query against Google BigQuery and return the results
|
||||
|
||||
#### Input
|
||||
|
||||
| Parameter | Type | Required | Description |
|
||||
| --------- | ---- | -------- | ----------- |
|
||||
| `projectId` | string | Yes | Google Cloud project ID |
|
||||
| `query` | string | Yes | SQL query to execute |
|
||||
| `useLegacySql` | boolean | No | Whether to use legacy SQL syntax \(default: false\) |
|
||||
| `maxResults` | number | No | Maximum number of rows to return |
|
||||
| `defaultDatasetId` | string | No | Default dataset for unqualified table names |
|
||||
| `location` | string | No | Processing location \(e.g., "US", "EU"\) |
|
||||
|
||||
#### Output
|
||||
|
||||
| Parameter | Type | Description |
|
||||
| --------- | ---- | ----------- |
|
||||
| `columns` | array | Array of column names from the query result |
|
||||
| `rows` | array | Array of row objects keyed by column name |
|
||||
| `totalRows` | string | Total number of rows in the complete result set |
|
||||
| `jobComplete` | boolean | Whether the query completed within the timeout |
|
||||
| `totalBytesProcessed` | string | Total bytes processed by the query |
|
||||
| `cacheHit` | boolean | Whether the query result was served from cache |
|
||||
| `jobReference` | object | Job reference \(useful when jobComplete is false\) |
|
||||
| ↳ `projectId` | string | Project ID containing the job |
|
||||
| ↳ `jobId` | string | Unique job identifier |
|
||||
| ↳ `location` | string | Geographic location of the job |
|
||||
| `pageToken` | string | Token for fetching additional result pages |
|
||||
|
||||
### `google_bigquery_list_datasets`
|
||||
|
||||
List all datasets in a Google BigQuery project
|
||||
|
||||
#### Input
|
||||
|
||||
| Parameter | Type | Required | Description |
|
||||
| --------- | ---- | -------- | ----------- |
|
||||
| `projectId` | string | Yes | Google Cloud project ID |
|
||||
| `maxResults` | number | No | Maximum number of datasets to return |
|
||||
| `pageToken` | string | No | Token for pagination |
|
||||
|
||||
#### Output
|
||||
|
||||
| Parameter | Type | Description |
|
||||
| --------- | ---- | ----------- |
|
||||
| `datasets` | array | Array of dataset objects |
|
||||
| ↳ `datasetId` | string | Unique dataset identifier |
|
||||
| ↳ `projectId` | string | Project ID containing this dataset |
|
||||
| ↳ `friendlyName` | string | Descriptive name for the dataset |
|
||||
| ↳ `location` | string | Geographic location where the data resides |
|
||||
| `nextPageToken` | string | Token for fetching next page of results |
|
||||
|
||||
### `google_bigquery_list_tables`
|
||||
|
||||
List all tables in a Google BigQuery dataset
|
||||
|
||||
#### Input
|
||||
|
||||
| Parameter | Type | Required | Description |
|
||||
| --------- | ---- | -------- | ----------- |
|
||||
| `projectId` | string | Yes | Google Cloud project ID |
|
||||
| `datasetId` | string | Yes | BigQuery dataset ID |
|
||||
| `maxResults` | number | No | Maximum number of tables to return |
|
||||
| `pageToken` | string | No | Token for pagination |
|
||||
|
||||
#### Output
|
||||
|
||||
| Parameter | Type | Description |
|
||||
| --------- | ---- | ----------- |
|
||||
| `tables` | array | Array of table objects |
|
||||
| ↳ `tableId` | string | Table identifier |
|
||||
| ↳ `datasetId` | string | Dataset ID containing this table |
|
||||
| ↳ `projectId` | string | Project ID containing this table |
|
||||
| ↳ `type` | string | Table type \(TABLE, VIEW, EXTERNAL, etc.\) |
|
||||
| ↳ `friendlyName` | string | User-friendly name for the table |
|
||||
| ↳ `creationTime` | string | Time when created, in milliseconds since epoch |
|
||||
| `totalItems` | number | Total number of tables in the dataset |
|
||||
| `nextPageToken` | string | Token for fetching next page of results |
|
||||
|
||||
### `google_bigquery_get_table`
|
||||
|
||||
Get metadata and schema for a Google BigQuery table
|
||||
|
||||
#### Input
|
||||
|
||||
| Parameter | Type | Required | Description |
|
||||
| --------- | ---- | -------- | ----------- |
|
||||
| `projectId` | string | Yes | Google Cloud project ID |
|
||||
| `datasetId` | string | Yes | BigQuery dataset ID |
|
||||
| `tableId` | string | Yes | BigQuery table ID |
|
||||
|
||||
#### Output
|
||||
|
||||
| Parameter | Type | Description |
|
||||
| --------- | ---- | ----------- |
|
||||
| `tableId` | string | Table ID |
|
||||
| `datasetId` | string | Dataset ID |
|
||||
| `projectId` | string | Project ID |
|
||||
| `type` | string | Table type \(TABLE, VIEW, SNAPSHOT, MATERIALIZED_VIEW, EXTERNAL\) |
|
||||
| `description` | string | Table description |
|
||||
| `numRows` | string | Total number of rows |
|
||||
| `numBytes` | string | Total size in bytes, excluding data in streaming buffer |
|
||||
| `schema` | array | Array of column definitions |
|
||||
| ↳ `name` | string | Column name |
|
||||
| ↳ `type` | string | Data type \(STRING, INTEGER, FLOAT, BOOLEAN, TIMESTAMP, RECORD, etc.\) |
|
||||
| ↳ `mode` | string | Column mode \(NULLABLE, REQUIRED, or REPEATED\) |
|
||||
| ↳ `description` | string | Column description |
|
||||
| `creationTime` | string | Table creation time \(milliseconds since epoch\) |
|
||||
| `lastModifiedTime` | string | Last modification time \(milliseconds since epoch\) |
|
||||
| `location` | string | Geographic location where the table resides |
|
||||
|
||||
### `google_bigquery_insert_rows`
|
||||
|
||||
Insert rows into a Google BigQuery table using streaming insert
|
||||
|
||||
#### Input
|
||||
|
||||
| Parameter | Type | Required | Description |
|
||||
| --------- | ---- | -------- | ----------- |
|
||||
| `projectId` | string | Yes | Google Cloud project ID |
|
||||
| `datasetId` | string | Yes | BigQuery dataset ID |
|
||||
| `tableId` | string | Yes | BigQuery table ID |
|
||||
| `rows` | string | Yes | JSON array of row objects to insert |
|
||||
| `skipInvalidRows` | boolean | No | Whether to insert valid rows even if some are invalid |
|
||||
| `ignoreUnknownValues` | boolean | No | Whether to ignore columns not in the table schema |
|
||||
|
||||
#### Output
|
||||
|
||||
| Parameter | Type | Description |
|
||||
| --------- | ---- | ----------- |
|
||||
| `insertedRows` | number | Number of rows successfully inserted |
|
||||
| `errors` | array | Array of per-row insertion errors \(empty if all succeeded\) |
|
||||
| ↳ `index` | number | Zero-based index of the row that failed |
|
||||
| ↳ `errors` | array | Error details for this row |
|
||||
| ↳ `reason` | string | Short error code summarizing the error |
|
||||
| ↳ `location` | string | Where the error occurred |
|
||||
| ↳ `message` | string | Human-readable error description |
|
||||
|
||||
|
||||
@@ -37,6 +37,7 @@
|
||||
"gitlab",
|
||||
"gmail",
|
||||
"gong",
|
||||
"google_bigquery",
|
||||
"google_books",
|
||||
"google_calendar",
|
||||
"google_docs",
|
||||
|
||||
Reference in New Issue
Block a user