refactor: implement parseBucketUrl() in baseDriver

Author: KSDaemon

packages/cubejs-base-driver/src/BaseDriver.ts

@@ -95,6 +95,21 @@ export type GoogleStorageClientConfig = {
   credentials: any,
 };
 
+export type ParsedBucketUrl = {
+  /**
+   * may be 's3', 'wasbs', 'gs', 'azure', etc
+   */
+  schema?: string;
+  bucketName: string;
+  /**
+   * prefix/path without leading and trailing / or empty string if not presented
+   */
+  path: string;
+  username?: string;
+  password?: string;
+  original: string;
+};
+
 const sortByKeys = (unordered: any) => {
   const ordered: any = {};
 
@@ -188,7 +203,7 @@ export abstract class BaseDriver implements DriverInterface {
   /**
    * Class constructor.
    */
-  public constructor(_options: {
+  protected constructor(_options: {
     /**
      * Time to wait for a response from a connection after validation
      * request before determining it as not valid. Default - 10000 ms.
@@ -695,6 +710,44 @@ export abstract class BaseDriver implements DriverInterface {
     query.query = `SELECT * FROM (${query.query}) AS t LIMIT ${query.limit}`;
   }
 
+  /**
+   * Returns parsed bucket structure.
+   * Supported variants:
+   *   s3://my-bucket-name/prefix/longer/
+   *   s3://my-bucket-name
+   *   my-bucket-name/some-path
+   *   my-bucket-name
+   *
+   */
+  protected parseBucketUrl(input: string | null | undefined): ParsedBucketUrl {
+    const original = input?.trim() || '';
+
+    if (!original) {
+      return {
+        bucketName: '',
+        path: '',
+        original,
+      };
+    }
+
+    const hasSchema = /^[a-zA-Z][a-zA-Z0-9+\-.]*:\/\//.test(original);
+    const normalized = hasSchema ? original : `schema://${original}`;
+
+    const url = new URL(normalized);
+
+    const path = url.pathname.replace(/^\/+|\/+$/g, '');
+    const schema = url.protocol.replace(/:$/, '');
+
+    return {
+      schema: schema || undefined,
+      bucketName: url.hostname,
+      path,
+      username: url.username || undefined,
+      password: url.password || undefined,
+      original,
+    };
+  }
+
   /**
    * Returns an array of signed AWS S3 URLs of the unloaded csv files.
    */

packages/cubejs-clickhouse-driver/src/ClickHouseDriver.ts

@@ -584,33 +584,19 @@ export class ClickHouseDriver extends BaseDriver implements DriverInterface {
     );
   }
 
-  /**
-   * Returns clean S3 bucket name and prefix path ending with / (if set)
-   */
-  private parseS3Path(input: string): { bucket: string; prefix: string | null } {
-    let trimmed = input.startsWith('s3://') ? input.slice(5) : input;
-    trimmed = trimmed.endsWith('/') ? trimmed.slice(0, -1) : trimmed;
-    const parts = trimmed.split('/');
-    const bucket = parts[0];
-    const prefixParts = parts.slice(1);
-    const prefix = prefixParts.length > 0 ? `${prefixParts.join('/')}/` : null;
-
-    return { bucket, prefix };
-  }
-
   public async unloadFromQuery(sql: string, params: unknown[], _options: UnloadOptions): Promise<DownloadTableCSVData> {
     if (!this.config.exportBucket) {
       throw new Error('Unload is not configured');
     }
 
     const types = await this.queryColumnTypes(`(${sql})`, params);
-    const { bucket, prefix } = this.parseS3Path(this.config.exportBucket.bucketName);
-    const exportPrefix = prefix ? `${prefix}${uuidv4()}` : uuidv4();
+    const { bucketName, path } = this.parseBucketUrl(this.config.exportBucket.bucketName);
+    const exportPrefix = path ? `${path}/${uuidv4()}` : uuidv4();
 
     const formattedQuery = sqlstring.format(`
       INSERT INTO FUNCTION
          s3(
-             'https://${bucket}.s3.${this.config.exportBucket.region}.amazonaws.com/${exportPrefix}/export.csv.gz',
+             'https://${bucketName}.s3.${this.config.exportBucket.region}.amazonaws.com/${exportPrefix}/export.csv.gz',
              '${this.config.exportBucket.keyId}',
              '${this.config.exportBucket.secretKey}',
              'CSV'
@@ -628,7 +614,7 @@ export class ClickHouseDriver extends BaseDriver implements DriverInterface {
         },
         region: this.config.exportBucket.region,
       },
-      bucket,
+      bucketName,
       exportPrefix,
     );