nv-ingest 2025.8.20.dev20250820__tar.gz → 2025.8.22.dev20250822__tar.gz

{nv_ingest-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/PKG-INFO

@@ -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 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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-2025.8.20.dev20250820 → nv_ingest-2025.8.22.dev20250822}/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 → nv_ingest-2025.8.22.dev20250822}/nv_ingest.egg-info/PKG-INFO

@@ -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