feat: add dataset.statistics (#621)

Part of apify/apify-core#18807 - Implementation of new API endpoint `v2/datasets/{datasetId}/statistics`
apify · Jan 15, 2025 · 6aeb2b7 · 6aeb2b7
1 parent 6327f3b
commit 6aeb2b7
Show file tree

Hide file tree

Showing 3 changed files with 45 additions and 2 deletions.
diff --git a/src/resource_clients/dataset.ts b/src/resource_clients/dataset.ts
@@ -1,9 +1,10 @@
 import ow from 'ow';
 
+import { ApifyApiError } from '../apify_api_error';
 import { ApiClientSubResourceOptions } from '../base/api_client';
 import { ResourceClient } from '../base/resource_client';
-import { ApifyResponse } from '../http_client';
-import { cast, PaginatedList } from '../utils';
+import { ApifyRequestConfig, ApifyResponse } from '../http_client';
+import { cast, catchNotFoundOrThrow, PaginatedList, pluckData } from '../utils';
 
 export class DatasetClient<
     Data extends Record<string | number, any> = Record<string | number, unknown>,
@@ -130,6 +131,24 @@ export class DatasetClient<
         });
     }
 
+    /**
+     * https://docs.apify.com/api/v2#tag/DatasetsStatistics/operation/dataset_statistics_get
+     */
+    async getStatistics(): Promise<DatasetStatistics | undefined> {
+        const requestOpts: ApifyRequestConfig = {
+            url: this._url('statistics'),
+            method: 'GET',
+            params: this._params(),
+        };
+        try {
+            const response = await this.httpClient.call(requestOpts);
+            return cast(pluckData(response.data));
+        } catch (err) {
+            catchNotFoundOrThrow(err as ApifyApiError);
+        }
+        return undefined;
+    }
+
     private _createPaginationList(response: ApifyResponse, userProvidedDesc: boolean): PaginatedList<Data> {
         return {
             items: response.data,
@@ -210,3 +229,14 @@ export interface DatasetClientDownloadItemsOptions extends DatasetClientListItem
     xmlRoot?: string;
     xmlRow?: string;
 }
+
+export interface DatasetStatistics {
+    fieldStatistics: Record<string, FieldStatistics>;
+}
+
+export interface FieldStatistics {
+    min?: number;
+    max?: number;
+    nullCount?: number;
+    emptyCount?: number;
+}
diff --git a/test/datasets.test.js b/test/datasets.test.js
@@ -309,5 +309,17 @@ describe('Dataset methods', () => {
             expect(res).toBeUndefined();
             validateRequest({}, { datasetId }, data, expectedHeaders);
         });
+
+        test('statistics() works', async () => {
+            const datasetId = 'some-id';
+
+            const res = await client.dataset(datasetId).getStatistics();
+            expect(res.id).toEqual('get-statistics');
+            validateRequest({}, { datasetId });
+
+            const browserRes = await page.evaluate((id) => client.dataset(id).getStatistics(), datasetId);
+            expect(browserRes).toEqual(res);
+            validateRequest({}, { datasetId });
+        });
     });
 });
diff --git a/test/mock_server/routes/datasets.js b/test/mock_server/routes/datasets.js
@@ -12,6 +12,7 @@ const ROUTES = [
     { id: 'update-dataset', method: 'PUT', path: '/:datasetId' },
     { id: 'list-items', method: 'GET', path: '/:datasetId/items', type: 'responseJsonMock' },
     { id: 'push-items', method: 'POST', path: '/:datasetId/items' },
+    { id: 'get-statistics', method: 'GET', path: '/:datasetId/statistics' },
 ];
 
 addRoutes(datasets, ROUTES);