datatoolpack 0.6.0__tar.gz → 0.7.0__tar.gz

{datatoolpack-0.6.0 → datatoolpack-0.7.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datatoolpack
-Version: 0.6.0
+Version: 0.7.0
 Summary: Official Python SDK for the AutoData ML data preparation pipeline API
 Home-page: https://autodata.datatoolpack.com
 Author: AutoData Team

{datatoolpack-0.6.0 → datatoolpack-0.7.0}/datatoolpack/__init__.py

@@ -1,4 +1,4 @@
 from .client import AutoDataClient, AutoDataError
 
-__version__ = "0.6.0"
+__version__ = "0.7.0"
 __all__ = ["AutoDataClient", "AutoDataError", "__version__"]

{datatoolpack-0.6.0 → datatoolpack-0.7.0}/datatoolpack/client.py

@@ -214,6 +214,7 @@ class AutoDataClient:
         auto_download: bool = True,
         output_preferences: Optional[List[str]] = None,
         compressed: bool = True,
+        return_dataframes: bool = False,
     ) -> Dict:
         """Upload a data file and start the AutoData pipeline.
 
@@ -307,6 +308,12 @@ class AutoDataClient:
                 output_preferences=output_preferences,
                 compressed=compressed,
             )
+        # New in 0.7.0: return the output files as DataFrames so callers can
+        # skip the "download to disk, then pd.read_csv()" two-step. Gated on
+        # an opt-in flag to keep the default light (no pandas import).
+        if return_dataframes:
+            final = dict(final)  # don't mutate the cached result object
+            final["dataframes"] = self.get_dataframes(session_id, names=output_preferences)
         return final
 
     def get_status(self, session_id: str) -> Dict:
@@ -564,6 +571,125 @@ class AutoDataClient:
         if last_err:
             raise last_err
 
+    # ------------------------------------------------------------------
+    # DataFrame-native helpers
+    # ------------------------------------------------------------------
+    # These let callers skip the "download to disk, then pd.read_csv()" two-step.
+    # pandas is imported lazily so the SDK stays light when a caller doesn't
+    # need it (the package install_requires only `requests`).
+
+    @staticmethod
+    def _require_pandas():
+        try:
+            import pandas as pd  # noqa: F401
+            return pd
+        except ImportError as e:
+            raise AutoDataError(
+                "pandas is required for DataFrame helpers — install it with `pip install pandas` "
+                "or use download_file()/download_results() to get raw files instead."
+            ) from e
+
+    def get_dataframe(
+        self,
+        session_id: str,
+        output_name: str,
+        *,
+        timeout: Optional[float] = None,
+        **read_csv_kwargs,
+    ):
+        """Fetch one output file and return it as a pandas DataFrame.
+
+        Args:
+            session_id:      Completed processing session.
+            output_name:     File base name (e.g. ``"dsg_output"``, ``"mdh_output"``)
+                             or the full filename (``"dsg_output.csv"``). ``.csv``
+                             is appended if missing.
+            timeout:         Per-request timeout; defaults to the SDK's normal
+                             download timeout (15 min).
+            read_csv_kwargs: Forwarded to ``pandas.read_csv`` (e.g. ``dtype``,
+                             ``parse_dates``, ``usecols``).
+
+        Returns:
+            A pandas.DataFrame.
+
+        Raises:
+            AutoDataError: if pandas isn't installed, the file is missing or
+                           empty (server failed to generate it), or the
+                           download was truncated.
+        """
+        pd = self._require_pandas()
+        fname = output_name if output_name.lower().endswith(
+            (".csv", ".json", ".parquet", ".feather", ".orc")
+        ) else f"{output_name}.csv"
+        url = self._url(f"/download/{session_id}/{fname}")
+        if not url.startswith("http"):
+            url = f"{self.base_url}{url}"
+        r = self._request(
+            "GET", url, stream=True,
+            timeout=timeout if timeout is not None else max(self.timeout, 900),
+        )
+        self._raise_for_error(r)
+        cl = r.headers.get("Content-Length")
+        if cl == "0":
+            raise AutoDataError(
+                f"Server returned an empty file for {fname} — the output likely "
+                "failed to generate. Retry the run or check /result/<session_id>."
+            )
+        # Buffer into memory — pandas.read_csv needs a seekable stream and
+        # output files are bounded by the server-side row ceiling anyway.
+        body = r.content
+        if not body:
+            raise AutoDataError(f"Server returned 0 bytes for {fname}")
+        import io as _io
+        buf = _io.BytesIO(body)
+        low = fname.lower()
+        if low.endswith(".parquet"):
+            return pd.read_parquet(buf)
+        if low.endswith(".feather"):
+            return pd.read_feather(buf)
+        if low.endswith(".orc"):
+            return pd.read_orc(buf)
+        if low.endswith(".json"):
+            return pd.read_json(buf, **read_csv_kwargs)
+        return pd.read_csv(buf, **read_csv_kwargs)
+
+    def get_dataframes(
+        self,
+        session_id: str,
+        names: Optional[List[str]] = None,
+        *,
+        timeout: Optional[float] = None,
+    ) -> Dict:
+        """Fetch multiple output files and return a ``{name: DataFrame}`` dict.
+
+        Args:
+            session_id: Completed processing session.
+            names:      Output base names to fetch. ``None`` means every file
+                        the server lists for the session (skipping PDF
+                        reports and .json metadata files, which aren't
+                        naturally a DataFrame).
+            timeout:    Per-file timeout; see ``get_dataframe``.
+
+        Returns:
+            Dict of ``{file_stem: pandas.DataFrame}``.
+        """
+        self._require_pandas()
+        if names is None:
+            result = self.get_result(session_id)
+            names = []
+            for f in result.get("files", []):
+                fn = f.get("name", "")
+                low = fn.lower()
+                if not low or low.endswith((".pdf", ".json", ".md")):
+                    continue
+                names.append(fn)
+        out: Dict = {}
+        for n in names:
+            # Strip the extension so callers can access via the logical name.
+            key = n.rsplit(".", 1)[0] if "." in n else n
+            out[key] = self.get_dataframe(session_id, n, timeout=timeout)
+        return out
+
     # ------------------------------------------------------------------
     # Account / API key management
     # ------------------------------------------------------------------
@@ -699,6 +825,7 @@ class AutoDataClient:
         compressed: bool = True,
         incremental: bool = False,
         type_casts: Optional[Dict[str, str]] = None,
+        return_dataframes: bool = False,
     ) -> Dict:
         """Run the AutoData pipeline on data from a connector source.
 
@@ -799,6 +926,9 @@ class AutoDataClient:
                 output_preferences=output_preferences,
                 compressed=compressed,
             )
+        if return_dataframes:
+            final = dict(final)
+            final["dataframes"] = self.get_dataframes(session_id, names=output_preferences)
         return final
 
     def write_output(

{datatoolpack-0.6.0 → datatoolpack-0.7.0}/datatoolpack.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datatoolpack
-Version: 0.6.0
+Version: 0.7.0
 Summary: Official Python SDK for the AutoData ML data preparation pipeline API
 Home-page: https://autodata.datatoolpack.com
 Author: AutoData Team

{datatoolpack-0.6.0 → datatoolpack-0.7.0}/setup.py

@@ -7,7 +7,7 @@ with open(os.path.join(here, "README.md"), encoding="utf-8") as f:
 
 setup(
     name="datatoolpack",
-    version="0.6.0",
+    version="0.7.0",
     description="Official Python SDK for the AutoData ML data preparation pipeline API",
     long_description=long_description,
     long_description_content_type="text/markdown",