nv-ingest 2025.8.20.dev20250820__py3-none-any.whl → 2025.8.22.dev20250822__py3-none-any.whl

nv_ingest/framework/orchestration/ray/stages/extractors/audio_extractor.py

@@ -17,6 +17,7 @@ from nv_ingest_api.internal.schemas.extract.extract_audio_schema import AudioExt
 from nv_ingest_api.util.exception_handlers.decorators import (
     nv_ingest_node_failure_try_except,
 )
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 from nv_ingest.framework.util.flow_control.udf_intercept import udf_intercept_hook
 
@@ -69,7 +70,7 @@ class AudioExtractorStage(RayActorStage):
 
         # Remove the "audio_data_extract" task from the message to obtain task-specific configuration.
         task_config = remove_task_by_type(control_message, "extract")
-        self._logger.debug("Extracted task config: %s", task_config)
+        self._logger.debug("Extracted task config: %s", sanitize_for_logging(task_config))
 
         # Perform audio text extraction.
         new_df, extraction_info = extract_text_from_audio_internal(

nv_ingest/framework/orchestration/ray/stages/extractors/chart_extractor.py

@@ -13,8 +13,11 @@ from nv_ingest.framework.util.flow_control import filter_by_task
 from nv_ingest.framework.util.flow_control.udf_intercept import udf_intercept_hook
 from nv_ingest_api.internal.primitives.tracing.tagging import traceable
 from nv_ingest_api.internal.schemas.extract.extract_chart_schema import ChartExtractorSchema
-from nv_ingest_api.util.exception_handlers.decorators import nv_ingest_node_failure_try_except
 from nv_ingest.framework.orchestration.ray.stages.meta.ray_actor_stage_base import RayActorStage
+from nv_ingest_api.util.exception_handlers.decorators import (
+    nv_ingest_node_failure_try_except,
+)
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 logger = logging.getLogger(__name__)
 
@@ -66,7 +69,7 @@ class ChartExtractorStage(RayActorStage):
 
         # Remove the "chart_data_extract" task to obtain task-specific configuration.
         task_config = remove_task_by_type(control_message, "chart_data_extract")
-        logger.debug("ChartExtractorStage: Task config extracted: %s", task_config)
+        logger.debug("ChartExtractorStage: Task config extracted: %s", sanitize_for_logging(task_config))
 
         # Perform chart data extraction.
         execution_trace_log = {}

nv_ingest/framework/orchestration/ray/stages/extractors/docx_extractor.py

@@ -16,6 +16,7 @@ from nv_ingest_api.internal.schemas.extract.extract_docx_schema import DocxExtra
 from nv_ingest_api.util.exception_handlers.decorators import (
     nv_ingest_node_failure_try_except,
 )
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 from nv_ingest.framework.util.flow_control.udf_intercept import udf_intercept_hook
 
@@ -68,7 +69,7 @@ class DocxExtractorStage(RayActorStage):
 
         # Remove the "docx-extract" task from the message to obtain task-specific configuration.
         task_config = remove_task_by_type(control_message, "extract")
-        self._logger.debug("Extracted task config: %s", task_config)
+        self._logger.debug("Extracted task config: %s", sanitize_for_logging(task_config))
 
         # Perform DOCX content extraction.
         new_df, extraction_info = extract_primitives_from_docx_internal(

nv_ingest/framework/orchestration/ray/stages/extractors/html_extractor.py

@@ -17,6 +17,7 @@ from nv_ingest_api.internal.schemas.extract.extract_html_schema import HtmlExtra
 from nv_ingest_api.util.exception_handlers.decorators import (
     nv_ingest_node_failure_try_except,
 )
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 from nv_ingest.framework.util.flow_control.udf_intercept import udf_intercept_hook
 
@@ -69,7 +70,7 @@ class HtmlExtractorStage(RayActorStage):
 
         # Remove the "html_content_extract" task from the message to obtain task-specific configuration.
         task_config = remove_task_by_type(control_message, "extract")
-        self._logger.debug("Extracted task config: %s", task_config)
+        self._logger.debug("Extracted task config: %s", sanitize_for_logging(task_config))
 
         # Perform html content extraction.
         new_df, extraction_info = extract_markdown_from_html_internal(

nv_ingest/framework/orchestration/ray/stages/extractors/image_extractor.py

@@ -16,6 +16,7 @@ from nv_ingest_api.internal.schemas.extract.extract_image_schema import ImageExt
 from nv_ingest_api.util.exception_handlers.decorators import (
     nv_ingest_node_failure_try_except,
 )
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 from nv_ingest.framework.util.flow_control.udf_intercept import udf_intercept_hook
 
@@ -68,7 +69,7 @@ class ImageExtractorStage(RayActorStage):
 
             # Remove the "extract" task from the message to obtain task-specific configuration.
             task_config = remove_task_by_type(control_message, "extract")
-            logger.debug("Extracted task config: %s", task_config)
+            logger.debug("Extracted task config: %s", sanitize_for_logging(task_config))
 
             # Perform image primitives extraction.
             new_df, extraction_info = extract_primitives_from_image_internal(

nv_ingest/framework/orchestration/ray/stages/extractors/pdf_extractor.py

@@ -15,7 +15,10 @@ from nv_ingest_api.internal.primitives.tracing.tagging import set_trace_timestam
 from nv_ingest.framework.orchestration.ray.stages.meta.ray_actor_stage_base import RayActorStage
 from nv_ingest.framework.util.flow_control import filter_by_task
 from nv_ingest.framework.util.flow_control.udf_intercept import udf_intercept_hook
-from nv_ingest_api.util.exception_handlers.decorators import nv_ingest_node_failure_try_except
+from nv_ingest_api.util.exception_handlers.decorators import (
+    nv_ingest_node_failure_try_except,
+)
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 logger = logging.getLogger(__name__)
 
@@ -87,7 +90,7 @@ class PDFExtractorStage(RayActorStage):
 
         # Remove the "extract" task from the message to obtain task-specific configuration.
         task_config = remove_task_by_type(control_message, "extract")
-        logger.debug("Extracted task config: %s", task_config)
+        logger.debug("Extracted task config: %s", sanitize_for_logging(task_config))
 
         # Perform PDF extraction.
         execution_trace_log = {}

nv_ingest/framework/orchestration/ray/stages/extractors/table_extractor.py

@@ -16,6 +16,7 @@ from nv_ingest_api.internal.schemas.extract.extract_table_schema import TableExt
 from nv_ingest_api.util.exception_handlers.decorators import (
     nv_ingest_node_failure_try_except,
 )
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 logger = logging.getLogger(__name__)
 
@@ -65,7 +66,7 @@ class TableExtractorStage(RayActorStage):
 
         # Remove the "table_data_extract" task to obtain task-specific configuration.
         task_config = remove_task_by_type(control_message, "table_data_extract")
-        logger.debug("Extracted task configuration: %s", task_config)
+        logger.debug("Extracted task configuration: %s", sanitize_for_logging(task_config))
 
         # Perform table data extraction.
         execution_trace_log = {}

nv_ingest/framework/orchestration/ray/stages/injectors/metadata_injector.py

@@ -25,6 +25,7 @@ from nv_ingest_api.util.exception_handlers.decorators import (
     nv_ingest_node_failure_try_except,
 )
 from nv_ingest.framework.util.flow_control.udf_intercept import udf_intercept_hook
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 logger = logging.getLogger(__name__)
 
@@ -42,7 +43,7 @@ class MetadataInjectionStage(RayActorStage):
         # Call the base initializer to set attributes like self._running.
         super().__init__(config, stage_name=stage_name)
         # Additional initialization can be added here if necessary.
-        self._logger.debug("MetadataInjectionStage initialized with config: %s", config)
+        self._logger.debug("MetadataInjectionStage initialized with config: %s", sanitize_for_logging(config))
 
     @nv_ingest_node_failure_try_except()
     @traceable()

nv_ingest/framework/orchestration/ray/stages/mutate/image_dedup.py

@@ -18,6 +18,7 @@ from nv_ingest_api.internal.schemas.mutate.mutate_image_dedup_schema import Imag
 from nv_ingest_api.util.exception_handlers.decorators import (
     nv_ingest_node_failure_try_except,
 )
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 logger = logging.getLogger(__name__)
 
@@ -68,7 +69,7 @@ class ImageDedupStage(RayActorStage):
 
             # Remove the "dedup" task from the message to obtain task-specific configuration.
             task_config = remove_task_by_type(control_message, "dedup")
-            logger.debug("Extracted task config: %s", task_config)
+            logger.debug("Extracted task config: %s", sanitize_for_logging(task_config))
 
             # Perform image deduplication.
             new_df = deduplicate_images_internal(

nv_ingest/framework/orchestration/ray/stages/mutate/image_filter.py

@@ -17,6 +17,7 @@ from nv_ingest_api.internal.schemas.transform.transform_image_filter_schema impo
 from nv_ingest_api.util.exception_handlers.decorators import (
     nv_ingest_node_failure_try_except,
 )
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 logger = logging.getLogger(__name__)
 
@@ -67,7 +68,7 @@ class ImageFilterStage(RayActorStage):
 
         # Remove the "filter" task from the message to obtain task-specific configuration.
         task_config = remove_task_by_type(control_message, "filter")
-        logger.debug("Extracted task config: %s", task_config)
+        logger.debug("Extracted task config: %s", sanitize_for_logging(task_config))
 
         task_params: Dict[str, Any] = task_config.get("params", {})
 

nv_ingest/framework/orchestration/ray/stages/sources/message_broker_task_source.py

@@ -29,6 +29,7 @@ from nv_ingest_api.internal.schemas.meta.ingest_job_schema import validate_inges
 # Import clients
 from nv_ingest_api.util.message_brokers.simple_message_broker.simple_client import SimpleClient
 from nv_ingest_api.util.service_clients.redis.redis_client import RedisClient
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 logger = logging.getLogger(__name__)
 
@@ -104,8 +105,11 @@ class MessageBrokerTaskSourceStage(RayActorSourceStage):
     def __init__(self, config: MessageBrokerTaskSourceConfig, stage_name: Optional[str] = None) -> None:
         super().__init__(config, log_to_stdout=False, stage_name=stage_name)
         self.config: MessageBrokerTaskSourceConfig  # Add a type hint for self.config
+
+        # Sanitize config before logging to avoid leaking secrets
+        _sanitized = sanitize_for_logging(config)
         self._logger.debug(
-            "Initializing MessageBrokerTaskSourceStage with config: %s", config.model_dump()
+            "Initializing MessageBrokerTaskSourceStage with config: %s", _sanitized
         )  # Log validated config
 
         # Access validated configuration directly via self.config

nv_ingest/framework/orchestration/ray/stages/storage/store_embeddings.py

@@ -16,6 +16,7 @@ from nv_ingest_api.internal.store.embed_text_upload import store_text_embeddings
 from nv_ingest_api.util.exception_handlers.decorators import (
     nv_ingest_node_failure_try_except,
 )
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 from nv_ingest.framework.util.flow_control.udf_intercept import udf_intercept_hook
 
@@ -68,7 +69,7 @@ class EmbeddingStorageStage(RayActorStage):
 
         # Remove the "store_embedding" task from the message to obtain task-specific configuration.
         task_config = remove_task_by_type(control_message, "store_embedding")
-        logger.debug("Extracted task config: %s", task_config)
+        logger.debug("Extracted task config: %s", sanitize_for_logging(task_config))
 
         # Perform embedding storage.
         new_df = store_text_embeddings_internal(

nv_ingest/framework/orchestration/ray/stages/transforms/image_caption.py

@@ -18,6 +18,7 @@ from nv_ingest_api.internal.transform.caption_image import transform_image_creat
 from nv_ingest_api.util.exception_handlers.decorators import (
     nv_ingest_node_failure_try_except,
 )
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 logger = logging.getLogger(__name__)
 
@@ -67,7 +68,10 @@ class ImageCaptionTransformStage(RayActorStage):
 
         # Remove the "caption" task to obtain task-specific configuration.
         task_config = remove_task_by_type(control_message, "caption")
-        logger.debug("ImageCaptionTransformStage: Task configuration extracted: %s", pprint.pformat(task_config))
+        logger.debug(
+            "ImageCaptionTransformStage: Task configuration extracted: %s",
+            pprint.pformat(sanitize_for_logging(task_config)),
+        )
 
         # Call the caption extraction function.
         new_df = transform_image_create_vlm_caption_internal(

nv_ingest/framework/orchestration/ray/stages/transforms/text_embed.py

@@ -15,6 +15,7 @@ from nv_ingest_api.internal.transform.embed_text import transform_create_text_em
 from nv_ingest_api.util.exception_handlers.decorators import (
     nv_ingest_node_failure_try_except,
 )
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 from nv_ingest.framework.util.flow_control.udf_intercept import udf_intercept_hook
 
@@ -62,7 +63,10 @@ class TextEmbeddingTransformStage(RayActorStage):
 
         # Remove the "embed" task to obtain task-specific configuration.
         task_config = remove_task_by_type(control_message, "embed")
-        self._logger.debug("TextEmbeddingTransformStage: Task configuration extracted: %s", pprint.pformat(task_config))
+        self._logger.debug(
+            "TextEmbeddingTransformStage: Task configuration extracted: %s",
+            pprint.pformat(sanitize_for_logging(task_config)),
+        )
 
         # Call the text embedding extraction function.
         new_df, execution_trace_log = transform_create_text_embeddings_internal(

nv_ingest/framework/orchestration/ray/stages/transforms/text_splitter.py

@@ -16,6 +16,7 @@ from nv_ingest_api.internal.transform.split_text import transform_text_split_and
 from nv_ingest_api.util.exception_handlers.decorators import (
     nv_ingest_node_failure_try_except,
 )
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 from nv_ingest.framework.util.flow_control.udf_intercept import udf_intercept_hook
 
@@ -36,7 +37,7 @@ class TextSplitterStage(RayActorStage):
         super().__init__(config, stage_name=stage_name)
         # Store the validated configuration (assumed to be an instance of TextSplitterSchema)
         self.validated_config: TextSplitterSchema = config
-        logger.debug("TextSplitterStage initialized with config: %s", config)
+        logger.info("TextSplitterStage initialized with config: %s", sanitize_for_logging(config))
 
     @nv_ingest_node_failure_try_except()
     @traceable()
@@ -63,7 +64,7 @@ class TextSplitterStage(RayActorStage):
 
         # Remove the "split" task to obtain task-specific configuration.
         task_config = remove_task_by_type(message, "split")
-        logger.debug("Extracted task config: %s", task_config)
+        logger.debug("Extracted task config: %s", sanitize_for_logging(task_config))
 
         # Transform the DataFrame (split text and tokenize).
         df_updated = transform_text_split_and_tokenize_internal(
@@ -107,7 +108,7 @@ def text_splitter_fn(control_message: IngestControlMessage, stage_config: TextSp
 
     # Remove the "split" task to obtain task-specific configuration.
     task_config = remove_task_by_type(control_message, "split")
-    logger.debug("Extracted task config: %s", task_config)
+    logger.debug("Extracted task config: %s", sanitize_for_logging(task_config))
 
     # Transform the DataFrame (split text and tokenize).
     df_updated = transform_text_split_and_tokenize_internal(

{nv_ingest-2025.8.20.dev20250820.dist-info → nv_ingest-2025.8.22.dev20250822.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest
-Version: 2025.8.20.dev20250820
+Version: 2025.8.22.dev20250822
 Summary: Python module for multimodal document ingestion
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License

{nv_ingest-2025.8.20.dev20250820.dist-info → nv_ingest-2025.8.22.dev20250822.dist-info}/RECORD

@@ -34,41 +34,41 @@ nv_ingest/framework/orchestration/ray/primitives/ray_pipeline.py,sha256=t9lf6zTj
 nv_ingest/framework/orchestration/ray/primitives/ray_stat_collector.py,sha256=GGY6_i6_g5xTFzdo9Qmsu9i4knMTq6pJfgm-aaPEt_o,17226
 nv_ingest/framework/orchestration/ray/stages/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/ray/stages/extractors/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest/framework/orchestration/ray/stages/extractors/audio_extractor.py,sha256=MdJA79_XECTmvksmxRZiIj5cZ46HzvIYOFNi5-S9o1k,3540
-nv_ingest/framework/orchestration/ray/stages/extractors/chart_extractor.py,sha256=D_clXtJe63CkeOcqor6C-X3Gbny7rNlp8PRMp9sRuPw,4281
-nv_ingest/framework/orchestration/ray/stages/extractors/docx_extractor.py,sha256=8DQOivEfvptUY7x7I6NsdJMTkiLdq2e_xPpPl3_bLJY,3536
-nv_ingest/framework/orchestration/ray/stages/extractors/html_extractor.py,sha256=fp0nzRQ21PmIInxtXjTYhOiYytmgBypDlY0e-WHCmGA,3551
-nv_ingest/framework/orchestration/ray/stages/extractors/image_extractor.py,sha256=CMIfHLCE4DUIedTxedZWWJKWlrLbWZzsuEE3j_ix_nY,3889
+nv_ingest/framework/orchestration/ray/stages/extractors/audio_extractor.py,sha256=4SdgvzI9oJ_OK5oWGir9wXVIPV4Pont2EKv9mwcWMC0,3631
+nv_ingest/framework/orchestration/ray/stages/extractors/chart_extractor.py,sha256=rfaDx6PqRCguhSYkJI6iVmMMtAlJNxzKfUrLmw_fKqs,4381
+nv_ingest/framework/orchestration/ray/stages/extractors/docx_extractor.py,sha256=R4vshPcAUN2U6BIv8BCZQ862wLx8RJhCGXfpQ3K09Bs,3627
+nv_ingest/framework/orchestration/ray/stages/extractors/html_extractor.py,sha256=7JrZSVIrK4_wr2s7TOTss7pgTY2F9GPQ7Ze3F_WFlKU,3642
+nv_ingest/framework/orchestration/ray/stages/extractors/image_extractor.py,sha256=iY9fEfucfgCmO2ixX6qwn418J97nJz_FQGh7B6yziVo,3980
 nv_ingest/framework/orchestration/ray/stages/extractors/infographic_extractor.py,sha256=v5J7dnJBEaDfjoTz_N_yC3RAt6lwMLgLT28V-ahquLE,3261
-nv_ingest/framework/orchestration/ray/stages/extractors/pdf_extractor.py,sha256=5gpbf2gJiUmL0-XR6nxWih3E0_FqkMRDaLdgP_MQs-c,4892
+nv_ingest/framework/orchestration/ray/stages/extractors/pdf_extractor.py,sha256=QagIA99AsHLihjRbXm-2BphdoQGHwzOHlqLyz7oDOSk,4992
 nv_ingest/framework/orchestration/ray/stages/extractors/pptx_extractor.py,sha256=RMbbl7Cuj4BT-TcgUx_0k8R-DLdw-o3fHxcIBIgrWt4,3776
-nv_ingest/framework/orchestration/ray/stages/extractors/table_extractor.py,sha256=YvCsBD4hxzgY-vGQmDwXVhleClMj2PHzsX71I5_VM_o,4079
+nv_ingest/framework/orchestration/ray/stages/extractors/table_extractor.py,sha256=p71ktv6v5T-9npYpCbgbwW6-fS-65UWS7rCm8OWr2Bc,4170
 nv_ingest/framework/orchestration/ray/stages/injectors/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest/framework/orchestration/ray/stages/injectors/metadata_injector.py,sha256=b4yve-J7ZMURSSuOPzSvfdIcyZ9fX8d-q7cYnXuxgsM,7025
+nv_ingest/framework/orchestration/ray/stages/injectors/metadata_injector.py,sha256=gTPRFOoUGcwNrmPeqv4n5VmU-LBkha9QTYHO0ntiaIk,7116
 nv_ingest/framework/orchestration/ray/stages/meta/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/ray/stages/meta/ray_actor_edge_base.py,sha256=LnVqBJmpfCmcI-eJLbkwK-7SS-hpEp98P4iCRv_Zhb0,1726
 nv_ingest/framework/orchestration/ray/stages/meta/ray_actor_sink_stage_base.py,sha256=HQJXIuU7VjiQ6fQjHjbNNmIJX5f30cXFB0CJGixgwVo,3633
 nv_ingest/framework/orchestration/ray/stages/meta/ray_actor_source_stage_base.py,sha256=hP25MLTP2bOEEncrYdxPPqeRyRVbij8aEurR1F1ZmhE,1811
 nv_ingest/framework/orchestration/ray/stages/meta/ray_actor_stage_base.py,sha256=qiB_ZU5_3bXgvE9C2rvnXIS0Alm6M5PWLCeQm8ZxOy4,29812
 nv_ingest/framework/orchestration/ray/stages/mutate/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest/framework/orchestration/ray/stages/mutate/image_dedup.py,sha256=ipH-3dctKaXRbjfkcXfpKRPXJBHXEosqPpm_Gb-5PZ0,3618
-nv_ingest/framework/orchestration/ray/stages/mutate/image_filter.py,sha256=-UkLArpgP1cr433dQDzO0g5Wn8cpvjqK6RGamufO5gQ,3432
+nv_ingest/framework/orchestration/ray/stages/mutate/image_dedup.py,sha256=cPLG3ZEqhZkRiSsbL7cbF1zsvOAimd8K5O-qadUR9Mg,3709
+nv_ingest/framework/orchestration/ray/stages/mutate/image_filter.py,sha256=f1CS8x9uifY1FJ_1lUF0fNNMExvM4zBIF012gxnSpqU,3523
 nv_ingest/framework/orchestration/ray/stages/sinks/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/ray/stages/sinks/default_drain.py,sha256=_USW1Vq8G2Wn-QFdPfFQCrtKG46hHeJvkEGbBxdpbVM,1488
 nv_ingest/framework/orchestration/ray/stages/sinks/message_broker_task_sink.py,sha256=QcvMQXIJ7EWIxty76Mo5Xv38Oj6X2KuS8qXQlf7E1uA,11676
 nv_ingest/framework/orchestration/ray/stages/sources/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest/framework/orchestration/ray/stages/sources/message_broker_task_source.py,sha256=A_D17mOqTh1a-4_FbcEUzPOhCxK7gqrXdEHXHpnHTcU,22163
+nv_ingest/framework/orchestration/ray/stages/sources/message_broker_task_source.py,sha256=Qm9XtTNX2CcUAlZRw33BS3Ql0djcsMGp52FPA2zHu3Q,22340
 nv_ingest/framework/orchestration/ray/stages/storage/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/ray/stages/storage/image_storage.py,sha256=WZN_-3Li-izDaPtk8IMrtn2os1ckT3U8Rb2PsfOWrcI,4009
-nv_ingest/framework/orchestration/ray/stages/storage/store_embeddings.py,sha256=4-bV4lIAdeylBeBHMO-6bVa2xilhPdZqkwfM8OwI2rM,3456
+nv_ingest/framework/orchestration/ray/stages/storage/store_embeddings.py,sha256=EUtwhSDf-qGLVEhWEInr1VaLsvpcHUSyzCmHQVai-Ps,3547
 nv_ingest/framework/orchestration/ray/stages/telemetry/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/ray/stages/telemetry/job_counter.py,sha256=jEtEUibqs6IS6QakrzWY9zmxSUzuBpg_hzXy2R-I10Y,2870
 nv_ingest/framework/orchestration/ray/stages/telemetry/otel_meter.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/ray/stages/telemetry/otel_tracer.py,sha256=GqFwRpTlIu2qgb08B3OqVIKUG8QQAwSOHhjvnyL_9UE,8553
 nv_ingest/framework/orchestration/ray/stages/transforms/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest/framework/orchestration/ray/stages/transforms/image_caption.py,sha256=Ghg09z5fSH0uyGplJPybE1e6sCD9IFpkGjzU3x8zd9o,3616
-nv_ingest/framework/orchestration/ray/stages/transforms/text_embed.py,sha256=RT4SkyjHPSJS7TuQhIcNW3B5a2tahEEtlxwB7M2UBYk,3309
-nv_ingest/framework/orchestration/ray/stages/transforms/text_splitter.py,sha256=D8HGPwOCqgCA6PMoeBAHaxKfLUgPhTlSP84uiujskYM,4829
+nv_ingest/framework/orchestration/ray/stages/transforms/image_caption.py,sha256=GYF41y_teXMIzB24JQXuesVmvksmgNvTGYU3uU4TzbM,3742
+nv_ingest/framework/orchestration/ray/stages/transforms/text_embed.py,sha256=o6QPd3GCPxbAvQFeb1oByOP5LII-FVkKbmAkBjCioB0,3435
+nv_ingest/framework/orchestration/ray/stages/transforms/text_splitter.py,sha256=FAScWaZijrJHc5F5jgulHP_cdX2telS4pq3spwliFXw,4963
 nv_ingest/framework/orchestration/ray/stages/utility/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/ray/stages/utility/throughput_monitor.py,sha256=J7Qs745rt7cQlR3L2K9U6Mb_BDKUNpl_xoqCZMEPlks,2674
 nv_ingest/framework/orchestration/ray/util/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
@@ -113,8 +113,8 @@ nv_ingest/pipeline/pipeline_schema.py,sha256=rLZZz2It2o2hVNWrZUJU8CarrqRei1fho3Z
 nv_ingest/pipeline/config/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/pipeline/config/loaders.py,sha256=75Yr9WYO7j7ghvKTnYLfZXQZEH3J3VEZo5J4TunC_Us,7590
 nv_ingest/pipeline/config/replica_resolver.py,sha256=3zjh8gmepEYORFZRM4inq7GoBW0YL3gzUDiixUugjzQ,8899
-nv_ingest-2025.8.20.dev20250820.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-nv_ingest-2025.8.20.dev20250820.dist-info/METADATA,sha256=1ebyXL2wfHlv6B2_21Jvawg4mMmicYmf_GCKLMk5aNo,15061
-nv_ingest-2025.8.20.dev20250820.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nv_ingest-2025.8.20.dev20250820.dist-info/top_level.txt,sha256=sjb0ajIsgn3YgftSjZHlYO0HjYAIIhNuXG_AmywCvaU,10
-nv_ingest-2025.8.20.dev20250820.dist-info/RECORD,,
+nv_ingest-2025.8.22.dev20250822.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+nv_ingest-2025.8.22.dev20250822.dist-info/METADATA,sha256=8YzCR5EKProTRxaNYN9EiuR0Y5F9hJfFs1_Z6euw40s,15061
+nv_ingest-2025.8.22.dev20250822.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nv_ingest-2025.8.22.dev20250822.dist-info/top_level.txt,sha256=sjb0ajIsgn3YgftSjZHlYO0HjYAIIhNuXG_AmywCvaU,10
+nv_ingest-2025.8.22.dev20250822.dist-info/RECORD,,