nv-ingest-client 2025.10.31.dev20251031__py3-none-any.whl → 2025.11.5.dev20251105__py3-none-any.whl

nv_ingest_client/client/client.py

@@ -1705,9 +1705,6 @@ class NvIngestClient:
                     )
                     logger.error(error_msg)
                     failures.append((self._job_index_to_job_spec[job_id].source_id, str(e)))
-                finally:
-                    # Clean up the job spec mapping
-                    del self._job_index_to_job_spec[job_id]
 
         if return_failures:
             return results, failures

nv_ingest_client/client/interface.py

@@ -1359,3 +1359,85 @@ class Ingestor:
         terminal_jobs = self.completed_jobs() + self.failed_jobs() + self.cancelled_jobs()
 
         return len(self._job_states) - terminal_jobs
+
+    def get_status(self) -> Dict[str, str]:
+        """
+        Returns a dictionary mapping document identifiers to their current status in the pipeline.
+
+        This method is designed for use with async ingestion to poll the status of submitted jobs.
+        For each document submitted to the ingestor, the method returns its current processing state.
+
+        Returns
+        -------
+        Dict[str, str]
+            A dictionary where:
+            - Keys are document identifiers (source names or source IDs)
+            - Values are status strings representing the current state:
+              * "pending": Job created but not yet submitted
+              * "submitted": Job submitted and waiting for processing
+              * "processing": Job is currently being processed
+              * "completed": Job finished successfully
+              * "failed": Job encountered an error
+              * "cancelled": Job was cancelled
+              * "unknown": Job state could not be determined (initial state)
+
+        Examples
+        --------
+        >>> ingestor = Ingestor(documents=["doc1.pdf", "doc2.pdf"], client=client)
+        >>> ingestor.extract().embed()
+        >>> future = ingestor.ingest_async()
+        >>>
+        >>> # Poll status while processing
+        >>> status = ingestor.get_status()
+        >>> print(status)
+        {'doc1.pdf': 'processing', 'doc2.pdf': 'submitted'}
+        >>>
+        >>> # Check again after some time
+        >>> status = ingestor.get_status()
+        >>> print(status)
+        {'doc1.pdf': 'completed', 'doc2.pdf': 'processing'}
+
+        Notes
+        -----
+        - This method is most useful when called after `ingest_async()` to track progress
+        - If called before any jobs are submitted, returns an empty dictionary or
+          documents with "unknown" status
+        - The method accesses internal job state from the client, so it reflects
+          the most current known state
+        """
+        status_dict = {}
+
+        if not self._job_states:
+            # If job states haven't been initialized yet (before ingest_async is called)
+            # Return unknown status for all documents
+            for doc in self._documents:
+                doc_name = os.path.basename(doc) if isinstance(doc, str) else str(doc)
+                status_dict[doc_name] = "unknown"
+            return status_dict
+
+        # Map job IDs to their states and source identifiers
+        for job_id, job_state in self._job_states.items():
+            # Get the job spec to find the source identifier
+            job_spec = self._client._job_index_to_job_spec.get(job_id)
+
+            if job_spec:
+                # Use source_name as the key (the document name)
+                source_identifier = job_spec.source_name
+            else:
+                # Fallback to job_id if we can't find the spec
+                source_identifier = f"job_{job_id}"
+
+            # Map the JobStateEnum to a user-friendly string
+            state_mapping = {
+                JobStateEnum.PENDING: "pending",
+                JobStateEnum.SUBMITTED_ASYNC: "submitted",
+                JobStateEnum.SUBMITTED: "submitted",
+                JobStateEnum.PROCESSING: "processing",
+                JobStateEnum.COMPLETED: "completed",
+                JobStateEnum.FAILED: "failed",
+                JobStateEnum.CANCELLED: "cancelled",
+            }
+
+            status_dict[source_identifier] = state_mapping.get(job_state.state, "unknown")
+
+        return status_dict

{nv_ingest_client-2025.10.31.dev20251031.dist-info → nv_ingest_client-2025.11.5.dev20251105.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest-client
-Version: 2025.10.31.dev20251031
+Version: 2025.11.5.dev20251105
 Summary: Python client for the nv-ingest service
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License

{nv_ingest_client-2025.10.31.dev20251031.dist-info → nv_ingest_client-2025.11.5.dev20251105.dist-info}/RECORD

@@ -6,9 +6,9 @@ nv_ingest_client/cli/util/click.py,sha256=YjQU1uF148FU5D3ozC2m1kkfOOJxO1U8U552-T
 nv_ingest_client/cli/util/processing.py,sha256=ULGCYQF1RTDQV_b35YM1WQRqIjR2wQRMJWu41DogagE,6259
 nv_ingest_client/cli/util/system.py,sha256=AQLq0DD2Ns8jRanrKu1tmVBKPA9rl-F3-ZsGI6FXLqE,1105
 nv_ingest_client/client/__init__.py,sha256=eEX9l1qmkLH2lAAZU3eP17SCV06ZjjrshHAB_xbboHA,375
-nv_ingest_client/client/client.py,sha256=WH2KRuaqoRm0qe3XAomAJQUetCDXp84xqcsMdumICbk,77505
+nv_ingest_client/client/client.py,sha256=3uA54D4Y6lSS-Nvz8R8uzkHkoV8vJu8GPQQRPoc-Uxk,77368
 nv_ingest_client/client/ingest_job_handler.py,sha256=4exvMwXbzwC-tb0dWleXE-AwhJkvxvhkf_u_1bJt30U,18387
-nv_ingest_client/client/interface.py,sha256=vmRdooNkaMVBv6RSxcgMYHfmMs0E3ZBnyrp5mmmhCOI,51247
+nv_ingest_client/client/interface.py,sha256=OCbH_5Q-cv1V4HpLBxLdaPCeaNKNkdEYi1JS4Tu6DGY,54745
 nv_ingest_client/client/util/processing.py,sha256=Ky7x7QbLn3BlgYwmrmoIc-o1VwmlmrcP9tn7GVTi0t0,2502
 nv_ingest_client/primitives/__init__.py,sha256=3rbpLCI7Bl0pntGatAxXD_V01y6dcLhHFheI3wqet-I,269
 nv_ingest_client/primitives/jobs/__init__.py,sha256=-yohgHv3LcCtSleHSaxjv1oO7nNcMCjN3ZYoOkIypIk,469
@@ -47,9 +47,9 @@ nv_ingest_client/util/vdb/__init__.py,sha256=ZmoEzeM9LzwwrVvu_DVUnjRNx-x8ahkNeIr
 nv_ingest_client/util/vdb/adt_vdb.py,sha256=UubzAMSfyrqqpD-OQErpBs25hC2Mw8zGZ4waenGXPOk,515
 nv_ingest_client/util/vdb/milvus.py,sha256=6XWRh2SDJlgVZOKZVXG3cZTB4L-ZHIiiTenuIzkxp2Y,78704
 nv_ingest_client/util/vdb/opensearch.py,sha256=I4FzF95VWCOkyzhfm-szdfK1Zd9ugUc8AxxpAdEMWGE,7538
-nv_ingest_client-2025.10.31.dev20251031.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-nv_ingest_client-2025.10.31.dev20251031.dist-info/METADATA,sha256=CR6KYTU_zty8JnF4vDkN8K_7yjOhR5FGy0MxxMOXvN4,30627
-nv_ingest_client-2025.10.31.dev20251031.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nv_ingest_client-2025.10.31.dev20251031.dist-info/entry_points.txt,sha256=3uQVZkTZIjO08_bjTV-g0CDF5H1nrP1zWXU9gJOweuI,137
-nv_ingest_client-2025.10.31.dev20251031.dist-info/top_level.txt,sha256=1eMhBFD3SiWmpXnod2LM66C1HrSLSk96ninZi5XX-cE,17
-nv_ingest_client-2025.10.31.dev20251031.dist-info/RECORD,,
+nv_ingest_client-2025.11.5.dev20251105.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+nv_ingest_client-2025.11.5.dev20251105.dist-info/METADATA,sha256=rB92S3YltqT5qi70cDN7VK1wtRDiOKFMe0vU7Av8tQ4,30626
+nv_ingest_client-2025.11.5.dev20251105.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nv_ingest_client-2025.11.5.dev20251105.dist-info/entry_points.txt,sha256=3uQVZkTZIjO08_bjTV-g0CDF5H1nrP1zWXU9gJOweuI,137
+nv_ingest_client-2025.11.5.dev20251105.dist-info/top_level.txt,sha256=1eMhBFD3SiWmpXnod2LM66C1HrSLSk96ninZi5XX-cE,17
+nv_ingest_client-2025.11.5.dev20251105.dist-info/RECORD,,