nv-ingest-api 2025.10.15.dev20251015__py3-none-any.whl → 2025.10.16.dev20251016__py3-none-any.whl

nv_ingest_api/internal/primitives/nim/model_interface/ocr.py

@@ -758,8 +758,8 @@ def get_ocr_model_name(ocr_grpc_endpoint=None, default_model_name=DEFAULT_OCR_MO
     if ocr_model_name is not None:
         return ocr_model_name
 
-    # 2. If no gRPC endpoint is provided, fall back to the default immediately.
-    if not ocr_grpc_endpoint:
+    # 2. If no gRPC endpoint is provided or the endpoint is a NVCF endpoint, fall back to the default immediately.
+    if (not ocr_grpc_endpoint) or ("grpc.nvcf.nvidia.com" in ocr_grpc_endpoint):
         logger.debug(f"No OCR gRPC endpoint provided. Falling back to default model name '{default_model_name}'.")
         return default_model_name
 

{nv_ingest_api-2025.10.15.dev20251015.dist-info → nv_ingest_api-2025.10.16.dev20251016.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest-api
-Version: 2025.10.15.dev20251015
+Version: 2025.10.16.dev20251016
 Summary: Python module with core document ingestion functions.
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License

{nv_ingest_api-2025.10.15.dev20251015.dist-info → nv_ingest_api-2025.10.16.dev20251016.dist-info}/RECORD

@@ -58,7 +58,7 @@ nv_ingest_api/internal/primitives/nim/model_interface/decorators.py,sha256=qwubk
 nv_ingest_api/internal/primitives/nim/model_interface/deplot.py,sha256=TvKdk6PTuI1WNhRmNNrvygaI_DIutkJkDL-XdtLZQac,10787
 nv_ingest_api/internal/primitives/nim/model_interface/helpers.py,sha256=jqbEbavvr9giODpzsGQSRDu5yZ4YfNfKAQfqUm9yUDI,11698
 nv_ingest_api/internal/primitives/nim/model_interface/nemoretriever_parse.py,sha256=WysjDZeegclO3mZgVcGOwzWbr8wSI4pWRiYD4iC2EXo,7098
-nv_ingest_api/internal/primitives/nim/model_interface/ocr.py,sha256=jFPB0h2SAq4RrZqiHdh28bNHvrkm0RWKZEx-ei-ahlU,29521
+nv_ingest_api/internal/primitives/nim/model_interface/ocr.py,sha256=WuX-veTC510TbvMWYGYP6WCzjYCbUBAUc5ovJUWCrFU,29607
 nv_ingest_api/internal/primitives/nim/model_interface/parakeet.py,sha256=3XXJkeJaVFe_iIfNn_bDYn79JN20besjZHiNZ5dEnZQ,12778
 nv_ingest_api/internal/primitives/nim/model_interface/text_embedding.py,sha256=lFhppNqrq5X_fzbCWKphvZQMzaJd3gHrkWsyJORzFrU,5010
 nv_ingest_api/internal/primitives/nim/model_interface/vlm.py,sha256=sUDKKlVqKjiHEGr2D04I7S4pDfnLR8b-NplV1pf5GVQ,6240
@@ -164,10 +164,10 @@ nv_ingest_api/util/string_processing/configuration.py,sha256=2HS08msccuPCT0fn_jf
 nv_ingest_api/util/string_processing/yaml.py,sha256=4Zdmc4474lUZn6kznqaNTlQJwsmRnnJQZ-DvAWLu-zo,2678
 nv_ingest_api/util/system/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nv_ingest_api/util/system/hardware_info.py,sha256=1UFM8XE6M3pgQcpbVsCsqDQ7Dj-zzptL-XRE-DEu9UA,27213
-nv_ingest_api-2025.10.15.dev20251015.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+nv_ingest_api-2025.10.16.dev20251016.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
 udfs/__init__.py,sha256=pXFqPgXIUqHDfj7SAR1Q19tt8KwGv_iMvhHyziz4AYM,205
-udfs/llm_summarizer_udf.py,sha256=t_ZFoz0e03uECYcRw4IabRj0GBlwAoJkJn13NL2wbsI,7217
-nv_ingest_api-2025.10.15.dev20251015.dist-info/METADATA,sha256=C3OVAyEQD5iSPzluBsDanslOq8aNRojyt3QDA8yXwGQ,14086
-nv_ingest_api-2025.10.15.dev20251015.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nv_ingest_api-2025.10.15.dev20251015.dist-info/top_level.txt,sha256=I1lseG9FF0CH93SPx4kFblsxFuv190cfzaas_CLNIiw,19
-nv_ingest_api-2025.10.15.dev20251015.dist-info/RECORD,,
+udfs/llm_summarizer_udf.py,sha256=9kM3W08FVnNHIRrDDZEuQRnPJmOdU2N-YTcnaBy8NdU,7267
+nv_ingest_api-2025.10.16.dev20251016.dist-info/METADATA,sha256=fsVHTpz4y-WxO1gDJOu87u8fSfAxOFrabEkcYf-Wvg8,14086
+nv_ingest_api-2025.10.16.dev20251016.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nv_ingest_api-2025.10.16.dev20251016.dist-info/top_level.txt,sha256=I1lseG9FF0CH93SPx4kFblsxFuv190cfzaas_CLNIiw,19
+nv_ingest_api-2025.10.16.dev20251016.dist-info/RECORD,,

udfs/llm_summarizer_udf.py

@@ -2,8 +2,9 @@
 """
 LLM Content Summarizer UDF for NV-Ingest Pipeline
 
-This UDF uses an LLM to generate concise summaries of text content chunks. These summaries are added to the metadata
-for enhanced downstream processing and search capabilities.
+Generates document summaries using NVIDIA-hosted LLMs. This production UDF demonstrates how to extract the pipeline
+payload, run custom code (summarization), and inject results into the metadata for downstream usecases (such as
+retrieval).
 
 These variables can be set in the environment before running the pipeline. These can be treated as kwargs.
 - NVIDIA_API_KEY: API key for NVIDIA NIM endpoints (required)
@@ -14,16 +15,14 @@ These variables can be set in the environment before running the pipeline. These
 - MAX_CONTENT_LENGTH: Maximum content length to send to API (default: 12000)
 TODO: Implement this
 - NUM_CHUNKS: (Optional) Number of first and last pages to summarize. default=1
+
+More info can be found in `examples/udfs/README.md`
 """
 
 import logging
 import os
 import time
 
-# REMOVE BEFORE MERGING
-# import yaml
-# from pathlib import Path
-
 
 logger = logging.getLogger(__name__)