feat(storage): paginate listSubDirectories and removeDir for S3 and GS

packages/oc-gs-storage-adapter/__mocks__/@google-cloud/storage.js

@@ -12,32 +12,61 @@ const googleStorage = jest.genMockFromModule('@google-cloud/storage');
 const _Storage = class {
   constructor() {
     this.bucket = jest.fn(bucket => ({
-      getFiles: () => {
-        const files =
-          bucket === 'my-empty-bucket'
-            ? []
-            : [
-                [
-                  {
-                    name: 'components/image/1.0.0/app.js'
-                  },
-                  {
-                    name: 'components/image/1.0.0/server.js'
-                  },
-                  {
-                    name: 'components/image/1.0.0/.env'
-                  },
-                  {
-                    name: 'components/image/1.0.1/new-server.js'
-                  },
-                  {
-                    name: 'components/image/1.0.1/new-.env'
-                  },
-                  {
-                    name: 'components/image/1.0.1/new-app.js'
-                  }
-                ]
-              ];
+      getFiles: (options = {}) => {
+        if (bucket === 'my-empty-bucket') {
+          return Promise.resolve([]);
+        }
+
+        if (bucket === 'paginated-bucket') {
+          const makeFile = name => ({
+            name,
+            delete: jest.fn(() => Promise.resolve())
+          });
+
+          if (!options.pageToken) {
+            return Promise.resolve([
+              [
+                makeFile('components/a/1.0.0/app.js'),
+                makeFile('components/a/1.0.0/server.js')
+              ],
+              { pageToken: 'page-2' }
+            ]);
+          }
+
+          return Promise.resolve([
+            [makeFile('components/a/2.0.0/app.js')],
+            null
+          ]);
+        }
+
+        const files = [
+          [
+            {
+              name: 'components/image/1.0.0/app.js',
+              delete: jest.fn(() => Promise.resolve())
+            },
+            {
+              name: 'components/image/1.0.0/server.js',
+              delete: jest.fn(() => Promise.resolve())
+            },
+            {
+              name: 'components/image/1.0.0/.env',
+              delete: jest.fn(() => Promise.resolve())
+            },
+            {
+              name: 'components/image/1.0.1/new-server.js',
+              delete: jest.fn(() => Promise.resolve())
+            },
+            {
+              name: 'components/image/1.0.1/new-.env',
+              delete: jest.fn(() => Promise.resolve())
+            },
+            {
+              name: 'components/image/1.0.1/new-app.js',
+              delete: jest.fn(() => Promise.resolve())
+            }
+          ]
+        ];
         return Promise.resolve(files);
       },
       upload: (filePath, { destination }) => {

packages/oc-gs-storage-adapter/__test__/gs.test.ts

@@ -185,6 +185,26 @@ test('test getJson force mode', async () => {
   });
 });
 
+test('listSubDirectories follows pageToken across pages', async () => {
+  const client = gs({ ...validOptions, bucket: 'paginated-bucket' });
+
+  const data = await client.listSubDirectories('components/a');
+
+  // Page 1 yields 1.0.0, page 2 yields 2.0.0. Both are only returned if the
+  // adapter follows the nextQuery pageToken across pages.
+  expect(data).toEqual(['1.0.0', '2.0.0']);
+});
+
+test('removeDir follows pageToken across pages and deletes all files', async () => {
+  const client = gs({ ...validOptions, bucket: 'paginated-bucket' });
+
+  const result = (await client.removeDir('components/a')) as unknown[];
+
+  // Page 1 lists two files, page 2 lists one file. Three deletes only happen
+  // if the adapter paginates using the nextQuery pageToken.
+  expect(result).toHaveLength(3);
+});
+
 test('test getUrl ', () => {
   const client = gs(validOptions);
   expect(client.getUrl('test', '1.0.0', 'test.js')).toBe('/test/1.0.0/test.js');

packages/oc-gs-storage-adapter/src/index.ts

@@ -118,19 +118,48 @@ export default function gsAdapter(conf: GsConfig): StorageAdapter {
         ? dir
         : dir + '/';
 
-    const options = {
-      prefix: normalisedPath
-    };
-
     try {
-      const results = await getClient().bucket(bucketName).getFiles(options);
+      const collected: { name: string }[] = [];
+      let pageToken: string | undefined;
+
+      do {
+        const requestPageToken = pageToken;
+        const options: {
+          prefix: string;
+          autoPaginate: false;
+          pageToken?: string;
+        } = {
+          prefix: normalisedPath,
+          autoPaginate: false
+        };
+        if (requestPageToken) {
+          options.pageToken = requestPageToken;
+        }
+
+        const results = await getClient()
+          .bucket(bucketName)
+          .getFiles(options);
+        const files = results[0] ?? [];
+        const nextQuery = results[1] as { pageToken?: string } | undefined;
+
+        for (const file of files) {
+          collected.push(file);
+        }
+
+        const nextPageToken = nextQuery?.pageToken;
+        if (nextPageToken && nextPageToken === requestPageToken) {
+          throw new Error(
+            'GCS getFiles returned an unchanged pageToken; aborting to avoid an infinite loop'
+          );
+        }
+        pageToken = nextPageToken;
+      } while (pageToken);
 
-      const files = results[0];
-      if (files.length === 0) {
+      if (collected.length === 0) {
         throw 'no files';
       }
 
-      const result = files
+      const result = collected
         //remove prefix
         .map(file => file.name.replace(normalisedPath, ''))
         // only get files that aren't in root directory
@@ -258,11 +287,41 @@ export default function gsAdapter(conf: GsConfig): StorageAdapter {
           ? dir
           : `${dir}/`;
 
-      const [files] = await getClient()
-        .bucket(bucketName)
-        .getFiles({ prefix: normalisedPath });
+      const collected: { delete: () => Promise<unknown> }[] = [];
+      let pageToken: string | undefined;
+
+      do {
+        const requestPageToken = pageToken;
+        const options: {
+          prefix: string;
+          autoPaginate: false;
+          pageToken?: string;
+        } = {
+          prefix: normalisedPath,
+          autoPaginate: false
+        };
+        if (requestPageToken) {
+          options.pageToken = requestPageToken;
+        }
+
+        const results = await getClient()
+          .bucket(bucketName)
+          .getFiles(options);
+        const files = results[0] ?? [];
+        const nextQuery = results[1] as { pageToken?: string } | undefined;
+
+        for (const file of files) {
+          collected.push(file as { delete: () => Promise<unknown> });
+        }
+
+        const nextPageToken = nextQuery?.pageToken;
+        if (nextPageToken && nextPageToken === requestPageToken) {
+          throw new Error('GCS getFiles returned an unchanged pageToken; aborting to avoid an infinite loop');
+        }
+        pageToken = nextPageToken;
+      } while (pageToken);
 
-      return Promise.all(files.map(file => file.delete()));
+      return Promise.all(collected.map(file => file.delete()));
     };
 
     return removeFromContainer();

packages/oc-s3-storage-adapter/__mocks__/@aws-sdk/client-s3.js

@@ -56,21 +56,67 @@ const _S3 = class {
     });
 
     this.listObjects = jest.fn(val => {
-      const CommonPrefixes =
-        val.Bucket === 'my-empty-bucket'
-          ? []
-          : [
-              {
-                Prefix: 'components/image/1.0.0/'
-              },
-              {
-                Prefix: 'components/image/1.0.1/'
-              }
-            ];
+      if (val.Bucket === 'my-empty-bucket') {
+        return Promise.resolve({ CommonPrefixes: [], Contents: [] });
+      }
+
+      if (val.Bucket === 'paginated-bucket') {
+        const isDelimited = val.Delimiter === '/';
+
+        if (isDelimited) {
+          // listSubDirectories path: S3 returns NextMarker when Delimiter is set.
+          if (!val.Marker) {
+            return Promise.resolve({
+              CommonPrefixes: [
+                { Prefix: 'components/a/1.0.0/' },
+                { Prefix: 'components/a/1.0.1/' }
+              ],
+              IsTruncated: true,
+              NextMarker: 'marker-2'
+            });
+          }
+
+          return Promise.resolve({
+            CommonPrefixes: [{ Prefix: 'components/a/2.0.0/' }],
+            IsTruncated: false
+          });
+        }
+
+        // removeDir path: no Delimiter, so S3 v1 does NOT return NextMarker;
+        // the client must use the last Key in Contents as the next Marker.
+        if (!val.Marker) {
+          return Promise.resolve({
+            Contents: [
+              { Key: 'components/a/1.0.0/file.js' },
+              { Key: 'components/a/1.0.0/package.json' }
+            ],
+            IsTruncated: true
+          });
+        }
+
+        return Promise.resolve({
+          Contents: [
+            { Key: 'components/a/2.0.0/file.js' },
+            { Key: 'components/a/2.0.0/package.json' }
+          ],
+          IsTruncated: false
+        });
+      }
+
+      const CommonPrefixes = [
+        {
+          Prefix: 'components/image/1.0.0/'
+        },
+        {
+          Prefix: 'components/image/1.0.1/'
+        }
+      ];
 
       return Promise.resolve({ CommonPrefixes });
     });
 
+    this.deleteObject = jest.fn(() => Promise.resolve({}));
+
     this.putObject = jest.fn(data => {
       if (data && data.Key && data.Key.indexOf('error') >= 0) {
         if (data.Key.indexOf('throw') >= 0) {

packages/oc-s3-storage-adapter/__test__/s3.test.ts

@@ -200,6 +200,26 @@ test('test getJson force mode', async () => {
   });
 });
 
+test('listSubDirectories follows continuation tokens across pages', async () => {
+  const client = s3({ ...validOptions, bucket: 'paginated-bucket' });
+
+  const data = await client.listSubDirectories('components/a');
+
+  // Page 1 returns 1.0.0 and 1.0.1, page 2 returns 2.0.0. All three are
+  // only returned if the adapter follows NextMarker across pages.
+  expect(data).toEqual(['1.0.0', '1.0.1', '2.0.0']);
+});
+
+test('removeDir follows continuation tokens across pages and deletes all keys', async () => {
+  const client = s3({ ...validOptions, bucket: 'paginated-bucket' });
+
+  const result = (await client.removeDir('components/a')) as unknown[];
+
+  // Page 1 lists two keys, page 2 lists two keys. Four deletes only happen
+  // if the adapter paginates using the last Key as the next Marker.
+  expect(result).toHaveLength(4);
+});
+
 test('test getUrl ', () => {
   const client = s3(validOptions);
   expect(client.getUrl('test', '1.0.0', 'test.js')).toBe('/test/1.0.0/test.js');