tinybird 4.6.11.dev0__tar.gz → 4.6.12.dev0__tar.gz

{tinybird-4.6.11.dev0 → tinybird-4.6.12.dev0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: tinybird
-Version: 4.6.11.dev0
+Version: 4.6.12.dev0
 Summary: Tinybird Command Line Tool
 Home-page: https://www.tinybird.co/docs/forward/commands
 Author: Tinybird
@@ -52,6 +52,11 @@ The Tinybird command-line tool allows you to use all the Tinybird functionality
 Changelog
 ----------
 
+4.6.11
+********
+
+- `Added` `tb datasource sample` support for DynamoDB connectors.
+
 4.6.10
 ********
 

{tinybird-4.6.11.dev0 → tinybird-4.6.12.dev0}/tinybird/tb/__cli__.py

@@ -4,5 +4,5 @@ __description__ = 'Tinybird Command Line Tool'
 __url__ = 'https://www.tinybird.co/docs/forward/commands'
 __author__ = 'Tinybird'
 __author_email__ = 'support@tinybird.co'
-__version__ = '4.6.11.dev0'
-__revision__ = 'd98671a'
+__version__ = '4.6.12.dev0'
+__revision__ = '7bdea3d'

{tinybird-4.6.11.dev0 → tinybird-4.6.12.dev0}/tinybird/tb/client.py

@@ -546,20 +546,36 @@ class TinyB:
     def datasource_sync(self, datasource_id: str):
         return self._req(f"/v0/datasources/{datasource_id}/scheduling/runs", method="POST", data="")
 
-    def datasource_sample(self, datasource_name: str, max_files: int = 1) -> Dict[str, Any]:
-        """Start a sample import job for an S3/GCS connected datasource.
+    def datasource_sample(
+        self,
+        datasource_name: str,
+        max_files: int = 1,
+        rows: Optional[int] = None,
+        max_bytes: Optional[str] = None,
+        full_export: bool = False,
+    ) -> Dict[str, Any]:
+        """Start a sample import job for an S3/GCS/DynamoDB connected datasource.
 
         Args:
             datasource_name: Name of the datasource to import sample data into
-            max_files: Maximum number of files to import (default 1, max 10)
+            max_files: Maximum number of files to import for blob storage connectors (default 1, max 10)
+            rows: For DynamoDB, the maximum number of rows to scan and import (mutually exclusive with max_bytes)
+            max_bytes: For DynamoDB, the maximum approximate JSONEachRow bytes to import, capped by the server's
+                workspace limit (mutually exclusive with rows)
+            full_export: For DynamoDB, trigger a full PITR export instead of a bounded scan
 
         Returns:
             dict with job info including id, job_id, job_url, job, status
         """
+        payload: Dict[str, Any] = {"max_files": max_files, "full_export": full_export}
+        if rows is not None:
+            payload["rows"] = rows
+        if max_bytes is not None:
+            payload["max_bytes"] = max_bytes
         return self._req(
             f"/v0/datasources/{datasource_name}/sample",
             method="POST",
-            data=json.dumps({"max_files": max_files}),
+            data=json.dumps(payload),
         )
 
     def datasource_scheduling_state(self, datasource_id: str):

{tinybird-4.6.11.dev0 → tinybird-4.6.12.dev0}/tinybird/tb/modules/common.py

@@ -1614,6 +1614,7 @@ def run_aws_iamrole_connection_flow(
     connection_name: str,
     policy: str,
     local_unavailable: bool = False,
+    env: str = "local",
 ) -> Tuple[str, str, Optional[TinyB], Optional[TinyB]]:
     """
     Run the interactive AWS IAM Role connection flow for S3 or DynamoDB.
@@ -1628,6 +1629,9 @@ def run_aws_iamrole_connection_flow(
         connection_name: The name for the connection being created.
         policy: The access policy type ('read' or 'write').
         local_unavailable: If True, local environment is unavailable (e.g., missing AWS credentials).
+        env: The active target environment ('local' or 'cloud'). When not local, the
+            passed-in `client` is reused as the cloud client so secrets land in the
+            workspace the command targets (including branches).
 
     Returns:
         A tuple containing:
@@ -1736,12 +1740,15 @@ def run_aws_iamrole_connection_flow(
 
     if use_cloud:
         try:
-            cloud_client = TinyB(
-                token=config.get("token", ""),
-                host=config.get("host", ""),
-                staging=False,
-                request_from=getattr(client, "request_from", None),
-            )
+            if env != "local":
+                cloud_client = client
+            else:
+                cloud_client = TinyB(
+                    token=config.get("token", ""),
+                    host=config.get("host", ""),
+                    staging=False,
+                    request_from=getattr(client, "request_from", None),
+                )
         except Exception as e:
             click.echo(FeedbackManager.warning(message=f"Failed to initialize cloud client: {e}"))
             click.echo(FeedbackManager.warning(message="Continuing without cloud environment."))

{tinybird-4.6.11.dev0 → tinybird-4.6.12.dev0}/tinybird/tb/modules/connection_dynamodb.py

@@ -158,6 +158,7 @@ def connection_create_dynamodb(
         connection_name=connection_name,
         policy="read",
         local_unavailable=local_aws_unavailable,
+        env=env,
     )
 
     unique_suffix = uuid.uuid4().hex[:8]

{tinybird-4.6.11.dev0 → tinybird-4.6.12.dev0}/tinybird/tb/modules/connection_s3.py

@@ -363,6 +363,7 @@ def connection_create_s3(
         connection_name=connection_name,
         policy=access_type.lower(),
         local_unavailable=local_aws_unavailable,
+        env=env,
     )
     unique_suffix = uuid.uuid4().hex[:8]  # Use first 8 chars of a UUID for brevity
     secret_name = f"s3_role_arn_{connection_name}_{unique_suffix}"

{tinybird-4.6.11.dev0 → tinybird-4.6.12.dev0}/tinybird/tb/modules/datasource.py

@@ -862,20 +862,50 @@ def datasource_sync(ctx: Context, datasource_name: str, yes: bool):
     default=False,
     help="Wait for the import job to finish",
 )
+@click.option(
+    "--rows",
+    default=None,
+    type=int,
+    help="For DynamoDB, the maximum number of rows to scan and import (default 1500; mutually exclusive with --max-bytes)",
+)
+@click.option(
+    "--max-bytes",
+    default=None,
+    type=str,
+    help="For DynamoDB, the maximum approximate JSONEachRow bytes to import, e.g. 500MB (capped at 10GB by default; mutually exclusive with --rows)",
+)
+@click.option(
+    "--full-export",
+    is_flag=True,
+    default=False,
+    help="For DynamoDB, trigger a full PITR export instead of a bounded sample (mutually exclusive with --rows and --max-bytes)",
+)
 @click.pass_context
-def datasource_sample(ctx: Context, datasource_name: str, max_files: int, wait: bool) -> None:
-    """Import sample data from a datasource connected to an S3 or GCS bucket.
-
-    This command only works with datasources that have IMPORT_CONNECTION_NAME and
-    IMPORT_BUCKET_URI configured (i.e., datasources ingesting from S3 or GCS).
-    It imports a limited number of files from the bucket URI pattern, useful for
-    testing data pipelines in branches without importing the entire dataset.
+def datasource_sample(
+    ctx: Context,
+    datasource_name: str,
+    max_files: int,
+    wait: bool,
+    rows: Optional[int],
+    max_bytes: Optional[str],
+    full_export: bool,
+) -> None:
+    """Import sample data from a datasource connected to S3, GCS, or DynamoDB.
+
+    For S3 and GCS, this imports a limited number of files from the bucket URI
+    pattern. For DynamoDB, this scans and imports a bounded sample limited by
+    either --rows or --max-bytes (defaulting to 1500 rows), or --full-export to
+    trigger a PITR export of the whole table.
 
     By default, returns immediately with job info. Use --wait to block until complete.
 
     Examples:
         tb --branch=my_branch datasource sample my_s3_ds
         tb --branch=my_branch datasource sample my_s3_ds --max-files 3 --wait
+        tb --branch=my_branch datasource sample my_dynamodb_ds --wait
+        tb --branch=my_branch datasource sample my_dynamodb_ds --rows 100000 --wait
+        tb --branch=my_branch datasource sample my_dynamodb_ds --max-bytes 1GB --wait
+        tb --branch=my_branch datasource sample my_dynamodb_ds --full-export --wait
     """
     from tinybird.tb.modules.common import wait_job
     from tinybird.tb.modules.job_common import echo_job_url
@@ -884,10 +914,32 @@ def datasource_sample(ctx: Context, datasource_name: str, max_files: int, wait:
         client: TinyB = ctx.obj["client"]
         config = ctx.obj.get("config", {})
 
+        if rows is not None and max_bytes is not None:
+            raise CLIDatasourceException(
+                FeedbackManager.error(message="--rows and --max-bytes are mutually exclusive; pass only one.")
+            )
+
+        if full_export and (rows is not None or max_bytes is not None):
+            raise CLIDatasourceException(
+                FeedbackManager.error(message="--full-export cannot be combined with --rows or --max-bytes.")
+            )
+
+        if full_export:
+            click.echo(
+                FeedbackManager.warning(
+                    message=(
+                        "DynamoDB full export samples will import the whole table. "
+                        "Use --rows or --max-bytes without --full-export for a bounded sample."
+                    )
+                )
+            )
+
         click.echo(FeedbackManager.info(message=f"Starting sample import for {datasource_name}..."))
 
         # Start the job
-        result = client.datasource_sample(datasource_name, max_files=max_files)
+        result = client.datasource_sample(
+            datasource_name, max_files=max_files, rows=rows, max_bytes=max_bytes, full_export=full_export
+        )
 
         job_id = result.get("job_id") or result.get("id")
         if not job_id:

{tinybird-4.6.11.dev0 → tinybird-4.6.12.dev0}/tinybird.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: tinybird
-Version: 4.6.11.dev0
+Version: 4.6.12.dev0
 Summary: Tinybird Command Line Tool
 Home-page: https://www.tinybird.co/docs/forward/commands
 Author: Tinybird
@@ -52,6 +52,11 @@ The Tinybird command-line tool allows you to use all the Tinybird functionality
 Changelog
 ----------
 
+4.6.11
+********
+
+- `Added` `tb datasource sample` support for DynamoDB connectors.
+
 4.6.10
 ********