datatoolpack 0.7.3__tar.gz → 0.8.0__tar.gz

{datatoolpack-0.7.3 → datatoolpack-0.8.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datatoolpack
-Version: 0.7.3
+Version: 0.8.0
 Summary: Official Python SDK for the AutoData ML data preparation pipeline API
 Home-page: https://autodata.datatoolpack.com
 Author: AutoData Team

{datatoolpack-0.7.3 → datatoolpack-0.8.0}/datatoolpack/__init__.py

@@ -1,4 +1,4 @@
 from .client import AutoDataClient, AutoDataError
 
-__version__ = "0.7.3"
+__version__ = "0.8.0"
 __all__ = ["AutoDataClient", "AutoDataError", "__version__"]

{datatoolpack-0.7.3 → datatoolpack-0.8.0}/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.
@@ -445,8 +450,14 @@ class AutoDataClient:
             session_id:         Session to download.
             download_path:      Directory to save files (created if needed).
                                 Defaults to ``./auto_data_outputs/<session_id>/``.
-            output_preferences: Subset of filenames to include.  ``None``
-                                downloads everything.
+            output_preferences: Subset of outputs to include.  ``None``
+                                downloads everything.  Each entry can be
+                                either the full filename
+                                (``'dsg_output.csv'``) or the short key
+                                (``'dsg_output'``); the SDK appends ``.csv``
+                                to extensionless entries automatically.
+                                Use ``'pipeline_report'`` (or
+                                ``'pipeline_report.pdf'``) for the PDF report.
             compressed:         If ``True`` (default), downloads a single ZIP
                                 archive and extracts it.  If ``False``,
                                 downloads each file individually.
@@ -463,7 +474,29 @@ class AutoDataClient:
         if compressed:
             body: Dict = {}
             if output_preferences:
-                body["files"] = output_preferences
+                # Server <= 0.7.3 used exact filename match — a user passing
+                # the key-form name ('dsg_output' instead of 'dsg_output.csv')
+                # got back an empty archive. Server >= 0.7.4 accepts both
+                # forms, but to keep older deployments working we also
+                # canonicalize on the client: if an entry has no extension,
+                # append the conventional .csv (which is correct for every
+                # AutoData output except pipeline_report → .pdf and the
+                # *_artifacts.pkl files, which users rarely cherry-pick).
+                EXTENSIONLESS_DEFAULTS = {
+                    "pipeline_report": "pipeline_report.pdf",
+                }
+                normalized = []
+                for item in output_preferences:
+                    item = str(item).strip()
+                    if not item:
+                        continue
+                    if "." in item:
+                        normalized.append(item)
+                    elif item in EXTENSIONLESS_DEFAULTS:
+                        normalized.append(EXTENSIONLESS_DEFAULTS[item])
+                    else:
+                        normalized.append(f"{item}.csv")
+                body["files"] = normalized
             r = self._request(
                 "POST",
                 self._url(f"/download-archive/{session_id}"),

{datatoolpack-0.7.3 → datatoolpack-0.8.0}/datatoolpack.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datatoolpack
-Version: 0.7.3
+Version: 0.8.0
 Summary: Official Python SDK for the AutoData ML data preparation pipeline API
 Home-page: https://autodata.datatoolpack.com
 Author: AutoData Team

{datatoolpack-0.7.3 → datatoolpack-0.8.0}/setup.py

@@ -7,7 +7,7 @@ with open(os.path.join(here, "README.md"), encoding="utf-8") as f:
 
 setup(
     name="datatoolpack",
-    version="0.7.3",
+    version="0.8.0",
     description="Official Python SDK for the AutoData ML data preparation pipeline API",
     long_description=long_description,
     long_description_content_type="text/markdown",