nv-ingest-client 2025.10.22.dev20251022__py3-none-any.whl → 2025.10.24.dev20251024__py3-none-any.whl

nv_ingest_client/client/ingest_job_handler.py

@@ -329,12 +329,33 @@ class IngestJobHandler:
                         job_id: str = futures_dict[future]
                         trace_ids[job_id_map[job_id]] = trace_id
 
-                        first_page_metadata = future_response["data"][0]["metadata"]
-                        file_page_counts: Dict[str, int] = {
-                            first_page_metadata["source_metadata"]["source_name"]: first_page_metadata[
-                                "content_metadata"
-                            ]["hierarchy"]["page_count"]
-                        }
+                        # Extract page count: prefer V2 metadata location, fall back to V1
+                        page_count = None
+                        source_name = None
+
+                        # Try V2 metadata location first (top-level metadata.total_pages)
+                        if "metadata" in future_response and future_response["metadata"]:
+                            response_metadata = future_response["metadata"]
+                            page_count = response_metadata.get("total_pages")
+                            source_name = response_metadata.get("original_source_name")
+
+                        # Fall back to V1 location (first data element's hierarchy.page_count)
+                        if page_count is None and future_response.get("data"):
+                            try:
+                                first_page_metadata = future_response["data"][0]["metadata"]
+                                page_count = first_page_metadata["content_metadata"]["hierarchy"]["page_count"]
+                                source_name = first_page_metadata["source_metadata"]["source_name"]
+                            except (KeyError, IndexError, TypeError):
+                                # If we can't extract from V1 location, use defaults
+                                pass
+
+                        # Use extracted values or defaults
+                        if page_count is None:
+                            page_count = 0  # Default if not found
+                        if source_name is None:
+                            source_name = "unknown_source"
+
+                        file_page_counts: Dict[str, int] = {source_name: page_count}
 
                         if self.output_directory:
                             self._save_response_data(

nv_ingest_client/primitives/tasks/filter.py

@@ -31,7 +31,7 @@ class FilterTask(Task):
         min_size: int = 128,
         max_aspect_ratio: Union[int, float] = 5.0,
         min_aspect_ratio: Union[int, float] = 0.2,
-        filter: bool = False,
+        filter: bool = True,
     ) -> None:
         """
         Setup Filter Task Config

{nv_ingest_client-2025.10.22.dev20251022.dist-info → nv_ingest_client-2025.10.24.dev20251024.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest-client
-Version: 2025.10.22.dev20251022
+Version: 2025.10.24.dev20251024
 Summary: Python client for the nv-ingest service
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License

{nv_ingest_client-2025.10.22.dev20251022.dist-info → nv_ingest_client-2025.10.24.dev20251024.dist-info}/RECORD

@@ -7,7 +7,7 @@ nv_ingest_client/cli/util/processing.py,sha256=ULGCYQF1RTDQV_b35YM1WQRqIjR2wQRMJ
 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=Ic7FPXGN4o-qk0atcbVcofE0ytgW16-B-KqJtXjY8ws,74461
-nv_ingest_client/client/ingest_job_handler.py,sha256=lMk-yQ0b0aK5ucxfNPVhxofzORIIK0jDzCYTmfdMZFw,17059
+nv_ingest_client/client/ingest_job_handler.py,sha256=fwyF48ObgusbSaeMxIgwOG0cuAUC8LYJc7FfxxCLOi0,18346
 nv_ingest_client/client/interface.py,sha256=7G2M59FayYyxvTwP6YCSeB42l6bMfpBNmd4kadKK6iU,50890
 nv_ingest_client/client/util/processing.py,sha256=Ky7x7QbLn3BlgYwmrmoIc-o1VwmlmrcP9tn7GVTi0t0,2502
 nv_ingest_client/primitives/__init__.py,sha256=3rbpLCI7Bl0pntGatAxXD_V01y6dcLhHFheI3wqet-I,269
@@ -21,7 +21,7 @@ nv_ingest_client/primitives/tasks/chart_extraction.py,sha256=s5hsljgSXxQMZHGekpA
 nv_ingest_client/primitives/tasks/dedup.py,sha256=qort6p3t6ZJuK_74sfOOLp3vMT3hkB5DAu3467WenyY,1719
 nv_ingest_client/primitives/tasks/embed.py,sha256=I6Irmvm1Qj9oqzDGSgfykCtfz8pz9LNxiXO-t29nXv8,5916
 nv_ingest_client/primitives/tasks/extract.py,sha256=bRriVkQyXN-UwzprHIt4Lp0iwmAojLEXqBb-IUrf3vY,9328
-nv_ingest_client/primitives/tasks/filter.py,sha256=wjcfSBGhdEyPh2tf42NMcyKZziigm24CO9B4obpQytU,2618
+nv_ingest_client/primitives/tasks/filter.py,sha256=dr6fWnh94i50MsGbrz9m_oN6DJKWIWsp7sMwm6Mjz8A,2617
 nv_ingest_client/primitives/tasks/infographic_extraction.py,sha256=SyTjZQbdVA3QwM5yVm4fUzE4Gu4zm4tAfNLDZMvySV8,1537
 nv_ingest_client/primitives/tasks/split.py,sha256=8UkB3EialsOTEbsOZLxzmnDIfTJzC6uvjNv21IbgAVA,2332
 nv_ingest_client/primitives/tasks/store.py,sha256=nIOnCH8vw4FLCLVBJYnsS5Unc0QmuO_jEtUp7-E9FU4,4199
@@ -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=uJUnH9gv8JYKvmI3BbljEsyRhV3l9-jP4F4sKOcfsWE,78702
 nv_ingest_client/util/vdb/opensearch.py,sha256=I4FzF95VWCOkyzhfm-szdfK1Zd9ugUc8AxxpAdEMWGE,7538
-nv_ingest_client-2025.10.22.dev20251022.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-nv_ingest_client-2025.10.22.dev20251022.dist-info/METADATA,sha256=F8GG56dInsRlhxwtaFangHWMf6QvIPnkfIy6y4jX_qI,30627
-nv_ingest_client-2025.10.22.dev20251022.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nv_ingest_client-2025.10.22.dev20251022.dist-info/entry_points.txt,sha256=3uQVZkTZIjO08_bjTV-g0CDF5H1nrP1zWXU9gJOweuI,137
-nv_ingest_client-2025.10.22.dev20251022.dist-info/top_level.txt,sha256=1eMhBFD3SiWmpXnod2LM66C1HrSLSk96ninZi5XX-cE,17
-nv_ingest_client-2025.10.22.dev20251022.dist-info/RECORD,,
+nv_ingest_client-2025.10.24.dev20251024.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+nv_ingest_client-2025.10.24.dev20251024.dist-info/METADATA,sha256=Kytxu5r2qPRsaE8Z2UvAocrXbNEcQgxbEShBQ-ZW_KQ,30627
+nv_ingest_client-2025.10.24.dev20251024.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nv_ingest_client-2025.10.24.dev20251024.dist-info/entry_points.txt,sha256=3uQVZkTZIjO08_bjTV-g0CDF5H1nrP1zWXU9gJOweuI,137
+nv_ingest_client-2025.10.24.dev20251024.dist-info/top_level.txt,sha256=1eMhBFD3SiWmpXnod2LM66C1HrSLSk96ninZi5XX-cE,17
+nv_ingest_client-2025.10.24.dev20251024.dist-info/RECORD,,