nv-ingest 2025.8.16.dev20250816__py3-none-any.whl → 2025.11.21.dev20251121__py3-none-any.whl

nv_ingest/pipeline/default_pipeline_impl.py

@@ -7,17 +7,17 @@
 # pylint: disable=line-too-long
 
 """
-Default pipeline implementation for libmode.
+Default pipeline implementation (runtime default).
 
-This module contains the default libmode pipeline configuration as a string,
-allowing the pipeline to be loaded without requiring external YAML files.
+This module embeds the exact contents of config/default_pipeline.yaml so code can
+load the default pipeline without reading the YAML file at runtime.
 """
 
-DEFAULT_LIBMODE_PIPELINE_YAML = """# Default Ingestion Pipeline Configuration for Library Mode
+DEFAULT_PIPELINE_YAML = """# Default Ingestion Pipeline Configuration
 # This file replicates the static pipeline defined in pipeline_builders.py
 
-name: "NVIngest default libmode pipeline"
-description: "This is the default ingestion pipeline for NVIngest in library mode"
+name: "NVIngest default pipeline"
+description: "This is the default ingestion pipeline for NVIngest"
 stages:
   # Source
   - name: "source_stage"
@@ -26,20 +26,19 @@ stages:
     actor: "nv_ingest.framework.orchestration.ray.stages.sources.message_broker_task_source:MessageBrokerTaskSourceStage"
     config:
       broker_client:
-        client_type: "simple"
-        host: $MESSAGE_CLIENT_HOST|"0.0.0.0"
-        port: $MESSAGE_CLIENT_PORT|7671
+        client_type: $MESSAGE_CLIENT_TYPE|"redis"
+        host: $MESSAGE_CLIENT_HOST|"redis"
+        port: $MESSAGE_CLIENT_PORT|6379
       task_queue: "ingest_task_queue"
       poll_interval: 0.1
     replicas:
-      min_replicas: 0
+      min_replicas: 1
       max_replicas:
         strategy: "static"
         value: 1
       static_replicas:
         strategy: "static"
         value: 1
-    runs_after: []
 
   # Pre-processing
   - name: "metadata_injector"
@@ -65,25 +64,25 @@ stages:
     actor: "nv_ingest.framework.orchestration.ray.stages.extractors.pdf_extractor:PDFExtractorStage"
     config:
       pdfium_config:
-        auth_token: $NGC_API_KEY|""
+        auth_token: $NGC_API_KEY|$NVIDIA_API_KEY
         yolox_endpoints: [
-          $YOLOX_GRPC_ENDPOINT|"",
-          $YOLOX_HTTP_ENDPOINT|"https://ai.api.nvidia.com/v1/cv/nvidia/nemoretriever-page-elements-v2"
+          $YOLOX_GRPC_ENDPOINT|"page-elements:8001",
+          $YOLOX_HTTP_ENDPOINT|"http://page-elements:8000/v1/infer",
         ]
-        yolox_infer_protocol: $YOLOX_INFER_PROTOCOL|http
+        yolox_infer_protocol: $YOLOX_INFER_PROTOCOL|grpc
       nemoretriever_parse_config:
-        auth_token: $NGC_API_KEY|""
+        auth_token: $NGC_API_KEY|$NVIDIA_API_KEY
         nemoretriever_parse_endpoints: [
           $NEMORETRIEVER_PARSE_GRPC_ENDPOINT|"",
-          $NEMORETRIEVER_PARSE_HTTP_ENDPOINT|"https://integrate.api.nvidia.com/v1/chat/completions"
+          $NEMORETRIEVER_PARSE_HTTP_ENDPOINT|"http://nemoretriever-parse:8000/v1/chat/completions",
         ]
         nemoretriever_parse_infer_protocol: $NEMORETRIEVER_PARSE_INFER_PROTOCOL|http
         nemoretriever_parse_model_name: $NEMORETRIEVER_PARSE_MODEL_NAME|"nvidia/nemoretriever-parse"
         yolox_endpoints: [
-          $YOLOX_GRPC_ENDPOINT|"",
-          $YOLOX_HTTP_ENDPOINT|"https://ai.api.nvidia.com/v1/cv/nvidia/nemoretriever-page-elements-v2"
+          $YOLOX_GRPC_ENDPOINT|"page-elements:8001",
+          $YOLOX_HTTP_ENDPOINT|"http://page-elements:8000/v1/infer",
         ]
-        yolox_infer_protocol: $YOLOX_INFER_PROTOCOL|http
+        yolox_infer_protocol: $YOLOX_INFER_PROTOCOL|grpc
     replicas:
       min_replicas: 0
       max_replicas:
@@ -101,12 +100,12 @@ stages:
     config:
       audio_extraction_config:
         audio_endpoints: [
-          $AUDIO_GRPC_ENDPOINT|"grpc.nvcf.nvidia.com:443",
-          $AUDIO_HTTP_ENDPOINT|""
+          $AUDIO_GRPC_ENDPOINT|"audio:50051",
+          $AUDIO_HTTP_ENDPOINT|"",
         ]
-        function_id: $AUDIO_FUNCTION_ID|"1598d209-5e27-4d3c-8079-4751568b1081"
+        function_id: $AUDIO_FUNCTION_ID|""
         audio_infer_protocol: $AUDIO_INFER_PROTOCOL|grpc
-        auth_token: $NGC_API_KEY|""
+        auth_token: $NGC_API_KEY|$NVIDIA_API_KEY
     replicas:
       min_replicas: 0
       max_replicas:
@@ -123,11 +122,11 @@ stages:
     config:
       docx_extraction_config:
         yolox_endpoints: [
-          $YOLOX_GRPC_ENDPOINT|"",
-          $YOLOX_HTTP_ENDPOINT|"https://ai.api.nvidia.com/v1/cv/nvidia/nemoretriever-page-elements-v2"
+          $YOLOX_GRPC_ENDPOINT|"page-elements:8001",
+          $YOLOX_HTTP_ENDPOINT|"",
         ]
-        yolox_infer_protocol: $YOLOX_INFER_PROTOCOL|http
-        auth_token: $NGC_API_KEY|""
+        yolox_infer_protocol: $YOLOX_INFER_PROTOCOL|grpc
+        auth_token: $NGC_API_KEY|$NVIDIA_API_KEY
     replicas:
       min_replicas: 0
       max_replicas:
@@ -144,11 +143,11 @@ stages:
     config:
       pptx_extraction_config:
         yolox_endpoints: [
-          $YOLOX_GRPC_ENDPOINT|"",
-          $YOLOX_HTTP_ENDPOINT|"https://ai.api.nvidia.com/v1/cv/nvidia/nemoretriever-page-elements-v2"
+          $YOLOX_GRPC_ENDPOINT|"page-elements:8001",
+          $YOLOX_HTTP_ENDPOINT|"http://page-elements:8000/v1/infer",
         ]
-        yolox_infer_protocol: $YOLOX_INFER_PROTOCOL|http
-        auth_token: $NGC_API_KEY|""
+        yolox_infer_protocol: $YOLOX_INFER_PROTOCOL|grpc
+        auth_token: $NGC_API_KEY|$NVIDIA_API_KEY
     replicas:
       min_replicas: 0
       max_replicas:
@@ -165,11 +164,11 @@ stages:
     config:
       image_extraction_config:
         yolox_endpoints: [
-          $YOLOX_GRPC_ENDPOINT|"",
-          $YOLOX_HTTP_ENDPOINT|"https://ai.api.nvidia.com/v1/cv/nvidia/nemoretriever-page-elements-v2"
+          $YOLOX_GRPC_ENDPOINT|"page-elements:8001",
+          $YOLOX_HTTP_ENDPOINT|"http://page-elements:8000/v1/infer",
         ]
-        yolox_infer_protocol: $YOLOX_INFER_PROTOCOL|http
-        auth_token: $NGC_API_KEY|""
+        yolox_infer_protocol: $YOLOX_INFER_PROTOCOL|grpc
+        auth_token: $NGC_API_KEY|$NVIDIA_API_KEY
     replicas:
       min_replicas: 0
       max_replicas:
@@ -193,6 +192,27 @@ stages:
         strategy: "static"
         value: 1
 
+  - name: "ocr_extractor"
+    type: "stage"
+    phase: 1  # EXTRACTION
+    actor: "nv_ingest.framework.orchestration.ray.stages.extractors.ocr_extractor:OCRExtractorStage"
+    config:
+      endpoint_config:
+        ocr_endpoints: [
+          $OCR_GRPC_ENDPOINT|"ocr:8001",
+          $OCR_HTTP_ENDPOINT|"http://ocr:8000/v1/infer",
+        ]
+        ocr_infer_protocol: $OCR_INFER_PROTOCOL|grpc
+        auth_token: $NGC_API_KEY|$NVIDIA_API_KEY
+    replicas:
+      min_replicas: 0
+      max_replicas:
+        strategy: "static"
+        value: 4
+      static_replicas:
+        strategy: "static"
+        value: 3
+
   - name: "infographic_extractor"
     type: "stage"
     phase: 1  # EXTRACTION
@@ -200,11 +220,11 @@ stages:
     config:
       endpoint_config:
         ocr_endpoints: [
-          $OCR_GRPC_ENDPOINT|"grpc.nvcf.nvidia.com:443",
-          $OCR_HTTP_ENDPOINT|""
+          $OCR_GRPC_ENDPOINT|"ocr:8001",
+          $OCR_HTTP_ENDPOINT|"http://ocr:8000/v1/infer",
         ]
         ocr_infer_protocol: $OCR_INFER_PROTOCOL|grpc
-        auth_token: $NGC_API_KEY|""
+        auth_token: $NGC_API_KEY|$NVIDIA_API_KEY
     replicas:
       min_replicas: 0
       max_replicas:
@@ -221,16 +241,16 @@ stages:
     config:
       endpoint_config:
         yolox_endpoints: [
-          $YOLOX_TABLE_STRUCTURE_GRPC_ENDPOINT|"",
-          $YOLOX_TABLE_STRUCTURE_HTTP_ENDPOINT|"https://ai.api.nvidia.com/v1/cv/nvidia/nemoretriever-table-structure-v1"
+          $YOLOX_TABLE_STRUCTURE_GRPC_ENDPOINT|"table-structure:8001",
+          $YOLOX_TABLE_STRUCTURE_HTTP_ENDPOINT|"http://table-structure:8000/v1/infer",
         ]
-        yolox_infer_protocol: $YOLOX_TABLE_STRUCTURE_INFER_PROTOCOL|"http"
+        yolox_infer_protocol: $YOLOX_TABLE_STRUCTURE_INFER_PROTOCOL|grpc
         ocr_endpoints: [
-          $OCR_GRPC_ENDPOINT|"",
-          $OCR_HTTP_ENDPOINT|"https://ai.api.nvidia.com/v1/cv/baidu/paddleocr"
+          $OCR_GRPC_ENDPOINT|"ocr:8001",
+          $OCR_HTTP_ENDPOINT|"http://ocr:8000/v1/infer",
         ]
-        ocr_infer_protocol: $PADDLE_INFER_PROTOCOL|"http"
-        auth_token: $NGC_API_KEY|""
+        ocr_infer_protocol: $OCR_INFER_PROTOCOL|grpc
+        auth_token: $NGC_API_KEY|$NVIDIA_API_KEY
     replicas:
       min_replicas: 0
       max_replicas:
@@ -248,16 +268,16 @@ stages:
     config:
       endpoint_config:
         yolox_endpoints: [
-          $YOLOX_GRAPHIC_ELEMENTS_GRPC_ENDPOINT|"",
-          $YOLOX_GRAPHIC_ELEMENTS_HTTP_ENDPOINT|"https://ai.api.nvidia.com/v1/cv/nvidia/nemoretriever-graphic-elements-v1"
+          $YOLOX_GRAPHIC_ELEMENTS_GRPC_ENDPOINT|"graphic-elements:8001",
+          $YOLOX_GRAPHIC_ELEMENTS_HTTP_ENDPOINT|""
         ]
-        yolox_infer_protocol: $YOLOX_GRAPHIC_ELEMENTS_INFER_PROTOCOL|"http"
+        yolox_infer_protocol: $YOLOX_GRAPHIC_ELEMENTS_INFER_PROTOCOL|grpc
         ocr_endpoints: [
-          $OCR_GRPC_ENDPOINT|"",
-          $OCR_HTTP_ENDPOINT|"https://ai.api.nvidia.com/v1/cv/baidu/paddleocr"
+          $OCR_GRPC_ENDPOINT|"ocr:8001",
+          $OCR_HTTP_ENDPOINT|""
         ]
-        ocr_infer_protocol: $OCR_INFER_PROTOCOL|"http"
-        auth_token: $NGC_API_KEY|""
+        ocr_infer_protocol: $OCR_INFER_PROTOCOL|grpc
+        auth_token: $NGC_API_KEY|$NVIDIA_API_KEY
     replicas:
       min_replicas: 0
       max_replicas:
@@ -317,9 +337,9 @@ stages:
     phase: 4  # TRANSFORM
     actor: "nv_ingest.framework.orchestration.ray.stages.transforms.image_caption:ImageCaptionTransformStage"
     config:
-      api_key: $NGC_API_KEY|""
-      endpoint_url: $VLM_CAPTION_ENDPOINT|"https://ai.api.nvidia.com/v1/gr/nvidia/llama-3.1-nemotron-nano-vl-8b-v1/chat/completions"
-      model_name: $VLM_CAPTION_MODEL_NAME|"nvidia/llama-3.1-nemotron-nano-vl-8b-v1"
+      api_key: $NGC_API_KEY|$NVIDIA_API_KEY
+      model_name: $VLM_CAPTION_MODEL_NAME|"nvidia/nemotron-nano-12b-v2-vl"
+      endpoint_url: $VLM_CAPTION_ENDPOINT|"http://vlm:8000/v1/chat/completions"
       prompt: "Caption the content of this image:"
     replicas:
       min_replicas: 0
@@ -335,17 +355,17 @@ stages:
     phase: 4  # TRANSFORM
     actor: "nv_ingest.framework.orchestration.ray.stages.transforms.text_embed:TextEmbeddingTransformStage"
     config:
-      api_key: $NGC_API_KEY|""
+      api_key: $NGC_API_KEY|$NVIDIA_API_KEY
       embedding_model: $EMBEDDING_NIM_MODEL_NAME|"nvidia/llama-3.2-nv-embedqa-1b-v2"
-      embedding_nim_endpoint: $EMBEDDING_NIM_ENDPOINT|"https://integrate.api.nvidia.com/v1"
+      embedding_nim_endpoint: $EMBEDDING_NIM_ENDPOINT|"http://embedding:8000/v1"
     replicas:
       min_replicas: 0
       max_replicas:
         strategy: "static"
-        value: 2
+        value: 4
       static_replicas:
         strategy: "static"
-        value: 1
+        value: 3
 
   # Storage and Output
   - name: "image_storage"
@@ -380,9 +400,10 @@ stages:
     actor: "nv_ingest.framework.orchestration.ray.stages.sinks.message_broker_task_sink:MessageBrokerTaskSinkStage"
     config:
       broker_client:
-        client_type: "simple"
-        host: "localhost"
-        port: 7671
+        client_type: $MESSAGE_CLIENT_TYPE|"redis"
+        host: $MESSAGE_CLIENT_HOST|localhost
+        port: $MESSAGE_CLIENT_PORT|6379
+      poll_interval: 0.1
     replicas:
       min_replicas: 1
       max_replicas:
@@ -416,7 +437,7 @@ stages:
     actor: "nv_ingest.framework.orchestration.ray.stages.sinks.default_drain:DefaultDrainSink"
     config: {}
     replicas:
-      min_replicas: 0
+      min_replicas: 1
       max_replicas:
         strategy: "static"
         value: 1
@@ -428,83 +449,85 @@ edges:
   # Intake
   - from: "source_stage"
     to: "metadata_injector"
-    queue_size: 32
+    queue_size: 4
 
   # Document Extractors
   - from: "metadata_injector"
     to: "pdf_extractor"
-    queue_size: 32
+    queue_size: 8
   - from: "pdf_extractor"
     to: "audio_extractor"
-    queue_size: 32
+    queue_size: 4
   - from: "audio_extractor"
     to: "docx_extractor"
-    queue_size: 32
+    queue_size: 4
   - from: "docx_extractor"
     to: "pptx_extractor"
-    queue_size: 32
+    queue_size: 4
   - from: "pptx_extractor"
     to: "image_extractor"
-    queue_size: 32
+    queue_size: 4
   - from: "image_extractor"
     to: "html_extractor"
-    queue_size: 32
+    queue_size: 4
   - from: "html_extractor"
     to: "infographic_extractor"
-    queue_size: 32
+    queue_size: 4
 
   # Primitive Extractors
   - from: "infographic_extractor"
     to: "table_extractor"
-    queue_size: 32
+    queue_size: 4
   - from: "table_extractor"
     to: "chart_extractor"
-    queue_size: 32
+    queue_size: 4
   - from: "chart_extractor"
+    to: "ocr_extractor"
+    queue_size: 8
+  - from: "ocr_extractor"
     to: "image_filter"
-    queue_size: 32
+    queue_size: 4
 
   # Primitive Mutators
   - from: "image_filter"
     to: "image_dedup"
-    queue_size: 32
+    queue_size: 4
   - from: "image_dedup"
     to: "text_splitter"
-    queue_size: 32
+    queue_size: 4
 
   # Primitive Transforms
   - from: "text_splitter"
     to: "image_caption"
-    queue_size: 32
+    queue_size: 4
   - from: "image_caption"
     to: "text_embedder"
-    queue_size: 32
+    queue_size: 4
   - from: "text_embedder"
     to: "image_storage"
-    queue_size: 32
+    queue_size: 4
 
   # Primitive Storage
   - from: "image_storage"
     to: "embedding_storage"
-    queue_size: 32
+    queue_size: 4
   - from: "embedding_storage"
     to: "broker_response"
-    queue_size: 32
+    queue_size: 4
 
   # Response and Telemetry
   - from: "broker_response"
     to: "otel_tracer"
-    queue_size: 32
+    queue_size: 4
   - from: "otel_tracer"
     to: "default_drain"
-    queue_size: 32
+    queue_size: 4
 
 # Pipeline Runtime Configuration
-# These parameters control dynamic scaling and PID controller behavior
-# All values can be overridden by environment variables from env_config.py
 pipeline:
-  disable_dynamic_scaling: $INGEST_DISABLE_DYNAMIC_SCALING|true
+  disable_dynamic_scaling: $INGEST_DISABLE_DYNAMIC_SCALING|false
   dynamic_memory_threshold: $INGEST_DYNAMIC_MEMORY_THRESHOLD|0.75
+  static_memory_threshold: $INGEST_STATIC_MEMORY_THRESHOLD|0.75
   pid_controller:
     kp: $INGEST_DYNAMIC_MEMORY_KP|0.2
     ki: $INGEST_DYNAMIC_MEMORY_KI|0.01
@@ -513,5 +536,5 @@ pipeline:
     penalty_factor: $INGEST_DYNAMIC_MEMORY_PENALTY_FACTOR|0.1
     error_boost_factor: $INGEST_DYNAMIC_MEMORY_ERROR_BOOST_FACTOR|1.5
     rcm_memory_safety_buffer_fraction: $INGEST_DYNAMIC_MEMORY_RCM_MEMORY_SAFETY_BUFFER_FRACTION|0.15
-  launch_simple_broker: true
+  launch_simple_broker: $INGEST_LAUNCH_SIMPLE_BROKER|false
 """

{nv_ingest-2025.8.16.dev20250816.dist-info → nv_ingest-2025.11.21.dev20251121.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest
-Version: 2025.8.16.dev20250816
+Version: 2025.11.21.dev20251121
 Summary: Python module for multimodal document ingestion
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License
@@ -225,10 +225,11 @@ Requires-Dist: httpx>=0.28.1
 Requires-Dist: isodate>=0.7.2
 Requires-Dist: langdetect>=1.0.9
 Requires-Dist: minio>=7.2.12
-Requires-Dist: openai>=1.82.0
+Requires-Dist: librosa>=0.10.2
 Requires-Dist: opentelemetry-api>=1.27.0
 Requires-Dist: opentelemetry-exporter-otlp>=1.27.0
 Requires-Dist: opentelemetry-sdk>=1.27.0
+Requires-Dist: psutil>=7.1.0
 Requires-Dist: pydantic>2.0.0
 Requires-Dist: pydantic-settings>2.0.0
 Requires-Dist: pypdfium2==4.30.0
@@ -240,7 +241,7 @@ Requires-Dist: python-docx>=1.1.2
 Requires-Dist: python-dotenv>=1.0.1
 Requires-Dist: python-pptx>=1.0.2
 Requires-Dist: prometheus-client
-Requires-Dist: ray[all]>=2.37.0
+Requires-Dist: ray[all]>=2.49.0
 Requires-Dist: redis>=5.2.1
 Requires-Dist: requests>=2.28.2
 Requires-Dist: scikit-learn>=1.6.0

{nv_ingest-2025.8.16.dev20250816.dist-info → nv_ingest-2025.11.21.dev20251121.dist-info}/RECORD

@@ -1,73 +1,79 @@
 nv_ingest/__init__.py,sha256=vJLPeuxiIHqbxXPJSu9qe3MS-GPavbOUExyRq83DxxM,895
 nv_ingest/version.py,sha256=MG7DxlzpnoJI56vqxwzs9WeMAEI3uPhfDiNLs6GN6wI,986
-nv_ingest/api/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest/api/main.py,sha256=XE-p4lJp1E7CCDOB8ENtYFrf63Dtq2bzQiGxpRfL2LA,1603
+nv_ingest/api/__init__.py,sha256=ED07QUqwVyJalH0ahhnnjvc2W_in6TpZZ5nJ6NWU9-Y,271
+nv_ingest/api/main.py,sha256=uCCkUNLS1xE9TDYKDOdxEfo_9jQWumpQAPWrxj5m9Go,1706
+nv_ingest/api/tracing.py,sha256=NkqMuUiB6ixGU5MYp3TrODsZDQepJ1kbH8JFHsYjuE0,2940
 nv_ingest/api/v1/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/api/v1/health.py,sha256=pV-RoVq5y0iBPp0qZoLzd1xKpd0JiHAi0UMyMj99LqU,4740
 nv_ingest/api/v1/ingest.py,sha256=LWk3LN4lBd3uO8h30EN42g3LHCVcO00avVd5ohVK7NI,19392
 nv_ingest/api/v1/metrics.py,sha256=ZGVRApYLnzc2f2C7wRgGd7deqiXan-jxfA-33a16clY,981
+nv_ingest/api/v2/README.md,sha256=VhpdjEmCyr3qIOhwqISFx9C5WezJFcxYc-NB9S98HMg,7562
+nv_ingest/api/v2/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
+nv_ingest/api/v2/ingest.py,sha256=ikbZE2eAjSnFmt5CcpTduY1t9DsUQBhnBQlsd3HaBww,53103
 nv_ingest/framework/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/execution/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/execution/helpers.py,sha256=-F8SZh7ISWtzJz6X1O2LQ133t-17Jxi8lL-NHz4rwj0,2818
 nv_ingest/framework/orchestration/execution/options.py,sha256=Ms1t4591EIv4ZrMRdhsCYPgLnMVXJosG3MURCbPXUoA,3983
 nv_ingest/framework/orchestration/process/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest/framework/orchestration/process/dependent_services.py,sha256=GpYYQ-YPKJbAFTcMHk4wCQUvtkJCfMOqCss5j845x0g,1979
-nv_ingest/framework/orchestration/process/execution.py,sha256=YkeNKk2H7DpiF2-9ygjKH2DeIM7dpvS4a74gDVBC0Eo,19540
-nv_ingest/framework/orchestration/process/lifecycle.py,sha256=xM6zjMJfoI1KyFlr76fcE9N8GINV3oG_5n1aw4P7AlY,4160
-nv_ingest/framework/orchestration/process/strategies.py,sha256=LzJDx9g_QdHtfBzg-edNMSqdYRZfF71i9S3Fv1ghK_M,6377
+nv_ingest/framework/orchestration/process/dependent_services.py,sha256=s0j_rsFtCKHFIuvOkBe9NEAkPNPhSYse_ApeHka8gyg,3032
+nv_ingest/framework/orchestration/process/execution.py,sha256=P1kzpYV23e4QYrKw9Td1TCZK3CK1ENVqqnI_axRCqBk,19814
+nv_ingest/framework/orchestration/process/lifecycle.py,sha256=L5NDwnzSMQPGjqJDC8jC75L1YqWey-dtK8N_HgBzb0E,8001
+nv_ingest/framework/orchestration/process/strategies.py,sha256=Q1Q04PPseF775omeS0FoXfK187NiS_bbqTaaJRwzKn8,7972
+nv_ingest/framework/orchestration/process/termination.py,sha256=PAogFeW0FATFS6Mcp_UkZgq_SbWV18RtdZN-0NbComw,5042
 nv_ingest/framework/orchestration/ray/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/ray/edges/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/ray/edges/async_queue_edge.py,sha256=PQliU_kyGbO9o42njpb8FrDMLrbLqwZzmBNXifxyG5Y,2312
 nv_ingest/framework/orchestration/ray/edges/ray_queue_edge.py,sha256=VFii2yxJuikimOxie3edKq5JN06g78AF8bdHSHVX8p8,2677
 nv_ingest/framework/orchestration/ray/edges/threaded_queue_edge.py,sha256=N6NH4KgZJ60e_JkGRcSmfQtX37qtX4TMcavOR-n3heE,2549
 nv_ingest/framework/orchestration/ray/examples/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest/framework/orchestration/ray/examples/pipeline_test_harness.py,sha256=hnRLybIpVTj3mXkLW0ErWVn4vRsInjNZmA80JqDiQuw,16473
+nv_ingest/framework/orchestration/ray/examples/pipeline_test_harness.py,sha256=Bn4rjkO14BwvvUNG_HBCSVXetYk7DKqRRsYHJADWqjc,16455
 nv_ingest/framework/orchestration/ray/examples/task_source_harness.py,sha256=Yt7uxThg7s8WuMiaHLKC8r1XAG7QixegfkT-juE5oNw,1953
 nv_ingest/framework/orchestration/ray/examples/task_source_sink_harness.py,sha256=XkvsoIzH5ftXvAZ4ox7mxbx7ESVx6D8Xupcwbqgd52w,3277
 nv_ingest/framework/orchestration/ray/primitives/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/ray/primitives/dataclasses.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nv_ingest/framework/orchestration/ray/primitives/pipeline_monitor.py,sha256=L8ENPiF-lxqhIXVEQwQD5CCqQMb710ynj5D_Y4ixGhs,11077
 nv_ingest/framework/orchestration/ray/primitives/pipeline_topology.py,sha256=yd2sb_q_FbBLDWiYgdKRhpPVAIl26Wg8w3yelZ7c5zQ,15741
-nv_ingest/framework/orchestration/ray/primitives/ray_pipeline.py,sha256=ncVOYTCrvqcipUdLBmBgH5vKpGTRrX6MrgGyBNPVknM,66656
+nv_ingest/framework/orchestration/ray/primitives/ray_pipeline.py,sha256=t9lf6zTjl_P5pe8mW-5F5pbZoC5mlcvEn2RCaLn_3Tk,66521
 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=UVp_kDmkaBlfO0Mbl_IxKq6imzLvs4-DKHgUHJIh3mo,3629
+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/ocr_extractor.py,sha256=pwVoA5-CF9GVWusoFZOMGBvSyW5udD9bdxVJXA_SghE,3188
+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=LrqaWpWyuiAHlpXWKYSyHZJBFegGXfNlpCXrucbK5NM,24067
 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
@@ -98,21 +104,22 @@ nv_ingest/framework/util/flow_control/udf_intercept.py,sha256=zQ9uuCcHLEd0P52Eiw
 nv_ingest/framework/util/service/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/util/service/impl/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/util/service/impl/ingest/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest/framework/util/service/impl/ingest/redis_ingest_service.py,sha256=KbzQFo7qVbCITiKYVPcGN0x4NI8piJy70Dz-8jf59Xs,15415
+nv_ingest/framework/util/service/impl/ingest/redis_ingest_service.py,sha256=59P-BMWnFY37GJm5w23-TMxgLhiZGZpJogC0gjDBaTA,23835
 nv_ingest/framework/util/service/meta/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/util/service/meta/ingest/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/util/service/meta/ingest/ingest_service_meta.py,sha256=QS3uNxWBl5dIcmIpJKNe8_TLcTUuN2vcKyHeAwa-eSo,1589
 nv_ingest/framework/util/telemetry/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/util/telemetry/global_stats.py,sha256=nq65pEEdiwjAfGiqsxG1CeQMC96O3CfQxsZuGFCY-ds,4554
 nv_ingest/pipeline/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest/pipeline/default_pipeline_impl.py,sha256=TV3WvmQL7DulE54rbF5-apKboVagqI-xqPFx24Sq_8Q,15582
+nv_ingest/pipeline/default_libmode_pipeline_impl.py,sha256=yNJtjfHQyxtasGa1hQrvgX7UrPa7BAd0oog8EIN8Y_w,15592
+nv_ingest/pipeline/default_pipeline_impl.py,sha256=DhClC17lWUvtBIi2mCC4WkLWT0lxY-CFY0n6nriAxas,16017
 nv_ingest/pipeline/ingest_pipeline.py,sha256=wHAJhqAM2s8nbY-8itVogmSU-yVN4PZONGWcKnhzgfg,17794
 nv_ingest/pipeline/pipeline_schema.py,sha256=rLZZz2It2o2hVNWrZUJU8CarrqRei1fho3ZEMkkoBcg,17940
 nv_ingest/pipeline/config/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest/pipeline/config/loaders.py,sha256=Zt9QOEd6g59pyN0zJZjh4VfXpjCHg2AR_4Wsbu4GkbM,6599
+nv_ingest/pipeline/config/loaders.py,sha256=75Yr9WYO7j7ghvKTnYLfZXQZEH3J3VEZo5J4TunC_Us,7590
 nv_ingest/pipeline/config/replica_resolver.py,sha256=3zjh8gmepEYORFZRM4inq7GoBW0YL3gzUDiixUugjzQ,8899
-nv_ingest-2025.8.16.dev20250816.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-nv_ingest-2025.8.16.dev20250816.dist-info/METADATA,sha256=Z49se75SyHtLkkUP00ovk0ePfb6FduPK_rCMJt4o3fk,15061
-nv_ingest-2025.8.16.dev20250816.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nv_ingest-2025.8.16.dev20250816.dist-info/top_level.txt,sha256=sjb0ajIsgn3YgftSjZHlYO0HjYAIIhNuXG_AmywCvaU,10
-nv_ingest-2025.8.16.dev20250816.dist-info/RECORD,,
+nv_ingest-2025.11.21.dev20251121.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+nv_ingest-2025.11.21.dev20251121.dist-info/METADATA,sha256=tY61Ze6cR62q-06kGMCBpCaRpqeY5IdmxfAKPie_Qjc,15092
+nv_ingest-2025.11.21.dev20251121.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nv_ingest-2025.11.21.dev20251121.dist-info/top_level.txt,sha256=sjb0ajIsgn3YgftSjZHlYO0HjYAIIhNuXG_AmywCvaU,10
+nv_ingest-2025.11.21.dev20251121.dist-info/RECORD,,