datatoolpack 0.7.4__tar.gz → 0.10.0__tar.gz

{datatoolpack-0.7.4 → datatoolpack-0.10.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datatoolpack
-Version: 0.7.4
+Version: 0.10.0
 Summary: Official Python SDK for the AutoData ML data preparation pipeline API
 Home-page: https://autodata.datatoolpack.com
 Author: AutoData Team

{datatoolpack-0.7.4 → datatoolpack-0.10.0}/datatoolpack/__init__.py

@@ -1,4 +1,4 @@
 from .client import AutoDataClient, AutoDataError
 
-__version__ = "0.7.4"
+__version__ = "0.10.0"
 __all__ = ["AutoDataClient", "AutoDataError", "__version__"]

{datatoolpack-0.7.4 → datatoolpack-0.10.0}/datatoolpack/client.py

@@ -74,7 +74,7 @@ class AutoDataClient:
     Args:
         api_key:      API key string starting with ``dtpk_``.
         base_url:     Base URL of the AutoData server (no trailing slash).
-        timeout:      HTTP request timeout in seconds (default 120).
+        timeout:      HTTP request timeout in seconds (default 300).
         max_retries:  Max automatic retries for transient errors
                       (429 / 502 / 503 / 504).  Default 3.
 
@@ -89,7 +89,7 @@ class AutoDataClient:
         api_key: Optional[str] = None,
         passcode: Optional[str] = None,
         base_url: Optional[str] = None,
-        timeout: int = 120,
+        timeout: int = 300,
         max_retries: int = 3,
     ):
         # Resolve from env vars if not provided explicitly
@@ -195,7 +195,10 @@ class AutoDataClient:
                     return response  # let caller handle the error
                 retry_after = int(response.headers.get("Retry-After", 0))
                 wait = max(retry_after, 2 ** attempt)
-            except requests.ConnectionError as exc:
+            except (requests.ConnectionError, requests.Timeout) as exc:
+                # Retry transient read/connect timeouts too: a long-running
+                # server-side stage can briefly delay a poll past the per-request
+                # timeout; that should not fail the whole job.
                 last_exc = exc
                 if attempt == self.max_retries:
                     raise
@@ -364,7 +367,7 @@ class AutoDataClient:
         session_id: str,
         poll_interval: int = 2,
         timeout: Optional[float] = None,
-        stall_timeout: float = 600.0,
+        stall_timeout: float = 3600.0,
     ) -> Dict:
         """Block until a session reaches *completed*, *error*, or *cancelled*.
 
@@ -405,6 +408,8 @@ class AutoDataClient:
                 raise AutoDataError(f"Processing failed: {msg}")
             if status == "cancelled":
                 raise AutoDataError("Processing was cancelled")
+            if status == "failed":
+                raise AutoDataError(f"Processing failed: {msg}")
 
             # Detect stalled workers: status stays 'unknown'/'processing' with
             # no progress change. Prevents infinite loops when cpu_worker died.
@@ -771,20 +776,29 @@ class AutoDataClient:
         Args:
             connector_type: One of the supported types:
 
-                Databases / warehouses:
+                Databases:
                   ``sql`` (Postgres/MySQL/MSSQL via SQLAlchemy URL),
-                  ``snowflake``, ``bigquery``, ``redshift``, ``oracle``,
-                  ``databricks``, ``fabric`` (Microsoft Fabric).
+                  ``mongodb``, ``oracle``, ``cassandra``, ``dynamodb``,
+                  ``elasticsearch``.
 
-                NoSQL:
-                  ``mongodb``, ``cassandra``, ``dynamodb``, ``elasticsearch``.
+                Data warehouses:
+                  ``snowflake``, ``bigquery``, ``redshift``, ``clickhouse``,
+                  ``databricks``, ``fabric`` (Microsoft Fabric),
+                  ``synapse`` (Azure Synapse), ``sap_hana`` (SAP HANA).
 
-                Object storage / data lakes:
+                Object storage / data lakes / files:
                   ``s3``, ``gcs``, ``azure_blob``, ``adls_gen2`` (Azure
-                  Data Lake Storage Gen2), ``delta`` (Delta Lake).
+                  Data Lake Storage Gen2), ``delta`` (Delta Lake),
+                  ``google_sheets``.
+
+                Streaming & IoT / time-series:
+                  ``kafka``, ``kinesis``, ``mqtt``, ``opcua`` (OPC-UA),
+                  ``influxdb``, ``timescaledb``, ``pi_web_api``
+                  (OSIsoft/AVEVA PI).
 
-                Streaming:
-                  ``kafka``, ``kinesis``.
+                SaaS & apps:
+                  ``salesforce``, ``hubspot``, ``stripe``, ``ga4``
+                  (Google Analytics 4), ``netsuite``, ``sharepoint``.
 
                 Generic:
                   ``http`` (REST/HTTP API).
@@ -929,7 +943,11 @@ class AutoDataClient:
             custom_query:   Optional SQL query (overrides *table*).
             output_rows:    Target row count in output dataset.
             tools:          Toggle pipeline steps (same as :meth:`process`).
+                            Set ``tools["feature_selection"]=True`` to run F4
+                            feature selection between DTC and MDH.
             advanced_params: Fine-grained parameters (same as :meth:`process`).
+                            ``feature_selection_config`` = ``{method, top_k,
+                            threshold}`` configures the F4 stage when enabled.
             wait:           Block until processing completes.
             poll_interval:  Seconds between status polls.
             download_path:  Directory to save results.
@@ -977,6 +995,11 @@ class AutoDataClient:
             body["enable_cds"] = bool(tls.get("cds", True))
             body["enable_dsm"] = bool(tls.get("dsm", True))
             body["enable_dsg"] = bool(tls.get("dsg", True))
+            body["enable_feature_selection"] = bool(tls.get("feature_selection", False))  # F4 (opt-in)
+        # F4 — feature selection config: {method, top_k, threshold}
+        fs_cfg = adv.get("feature_selection_config")
+        if fs_cfg:
+            body["feature_selection_config"] = fs_cfg
         if credential_id:
             body["credential_id"] = credential_id
         if secrets:
@@ -1019,6 +1042,110 @@ class AutoDataClient:
             final["dataframes"] = self.get_dataframes(session_id, names=output_preferences)
         return final
 
+    def recommend_features(
+        self,
+        session_id: str,
+        target_columns: List[str],
+        methods: Optional[List[str]] = None,
+        top_k: int = 20,
+    ) -> Dict:
+        """Rank a completed session's columns (F4) without re-running the pipeline.
+
+        Args:
+            session_id:      A completed session owned by the caller.
+            target_columns:  Target column name(s) to rank features against.
+            methods:         Ranking methods — any of ``variance_threshold``,
+                             ``correlation_pruning``, ``mutual_information``,
+                             ``rfe``, ``shap_importance``, ``pca``. Defaults to
+                             ``["mutual_information"]``.
+            top_k:           Number of top features to return (default 20).
+
+        Returns:
+            Dict with ``ranking`` (combined), ``per_method``, and ``warnings``.
+        """
+        body = {
+            "session_id": session_id,
+            "target_columns": target_columns,
+            "methods": methods or ["mutual_information"],
+            "top_k": top_k,
+        }
+        r = self._request("POST", self._url("/feature-selection/recommend"), json=body)
+        self._raise_for_error(r)
+        return r.json()
+
+    # ── F7 — Triggers ─────────────────────────────────────────────────────
+    def list_triggers(self) -> List[Dict]:
+        """List the caller's triggers (webhook secrets are never returned)."""
+        r = self._request("GET", self._url("/triggers"))
+        self._raise_for_error(r)
+        return r.json().get("triggers", [])
+
+    def create_trigger(
+        self,
+        name: str,
+        trigger_type: str,
+        config: Optional[Dict] = None,
+        target_pipeline_config: Optional[Dict] = None,
+        enabled: bool = True,
+    ) -> Dict:
+        """Create a trigger that auto-starts a pipeline on an external event.
+
+        Args:
+            name:           Human-readable trigger name.
+            trigger_type:   One of ``inbound_webhook``, ``watermark_advance``,
+                            ``pipeline_completion``, ``quality_alert``.
+            config:         Type-specific config. ``watermark_advance`` needs
+                            ``{credential_id, table, watermark_column,
+                            min_advance}``; ``inbound_webhook`` accepts an
+                            optional ``{hmac_key}``.
+            target_pipeline_config: The pipeline to run when fired — a
+                            connector data source + pipeline params
+                            (``{source_type, credential_id, table,
+                            target_columns, output_rows, ...}``).
+            enabled:        Whether the trigger is active (default True).
+
+        Returns:
+            The created trigger. For ``inbound_webhook`` the response
+            includes ``webhook_secret`` and ``inbound_url`` — this is the
+            ONLY time the secret is returned, so save it now.
+        """
+        body = {
+            "name": name,
+            "trigger_type": trigger_type,
+            "config": config or {},
+            "target_pipeline_config": target_pipeline_config or {},
+            "enabled": enabled,
+        }
+        r = self._request("POST", self._url("/triggers"), json=body)
+        self._raise_for_error(r)
+        return r.json().get("trigger", r.json())
+
+    def get_trigger(self, trigger_id: str) -> Dict:
+        """Fetch one trigger by id (secret not included)."""
+        r = self._request("GET", self._url(f"/triggers/{trigger_id}"))
+        self._raise_for_error(r)
+        return r.json().get("trigger", r.json())
+
+    def update_trigger(self, trigger_id: str, **fields) -> Dict:
+        """Update a trigger. Accepts ``name``, ``enabled``, ``config``,
+        ``target_pipeline_config``. The webhook secret is immutable."""
+        r = self._request("PUT", self._url(f"/triggers/{trigger_id}"), json=fields)
+        self._raise_for_error(r)
+        return r.json().get("trigger", r.json())
+
+    def delete_trigger(self, trigger_id: str) -> bool:
+        """Delete a trigger. Returns True on success."""
+        r = self._request("DELETE", self._url(f"/triggers/{trigger_id}"))
+        self._raise_for_error(r)
+        return True
+
+    def test_trigger(self, trigger_id: str) -> Dict:
+        """Dry-run a trigger: validate its config + target without firing.
+        Returns a report with ``would_fire`` and per-check details."""
+        r = self._request("POST", self._url(f"/triggers/{trigger_id}/test"))
+        self._raise_for_error(r)
+        return r.json()
+
     def write_output(
         self,
         session_id: str,

{datatoolpack-0.7.4 → datatoolpack-0.10.0}/datatoolpack.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datatoolpack
-Version: 0.7.4
+Version: 0.10.0
 Summary: Official Python SDK for the AutoData ML data preparation pipeline API
 Home-page: https://autodata.datatoolpack.com
 Author: AutoData Team

{datatoolpack-0.7.4 → datatoolpack-0.10.0}/setup.py

@@ -7,7 +7,7 @@ with open(os.path.join(here, "README.md"), encoding="utf-8") as f:
 
 setup(
     name="datatoolpack",
-    version="0.7.4",
+    version="0.10.0",
     description="Official Python SDK for the AutoData ML data preparation pipeline API",
     long_description=long_description,
     long_description_content_type="text/markdown",