nv-ingest 25.6.4rc4__py3-none-any.whl → 25.6.4rc6__py3-none-any.whl

nv_ingest/framework/orchestration/ray/util/pipeline/stage_builders.py

@@ -230,11 +230,7 @@ def add_table_extractor_stage(pipeline, default_cpu_count, stage_name="table_ext
     )
 
     pipeline.add_stage(
-        name=stage_name,
-        stage_actor=TableExtractorStage,
-        config=table_extractor_config,
-        min_replicas=0,
-        max_replicas=int(max(1, (default_cpu_count // 7))),  # 14% of available CPU cores
+        name=stage_name, stage_actor=TableExtractorStage, config=table_extractor_config, min_replicas=1, max_replicas=1
     )
 
     return stage_name
@@ -262,8 +258,8 @@ def add_chart_extractor_stage(pipeline, default_cpu_count, stage_name="chart_ext
         name=stage_name,
         stage_actor=ChartExtractorStage,
         config=chart_extractor_config,
-        min_replicas=0,
-        max_replicas=int(max(1, (default_cpu_count // 7))),  # 14% of available CPU cores
+        min_replicas=1,
+        max_replicas=1,
     )
 
     return stage_name
@@ -286,8 +282,8 @@ def add_infographic_extractor_stage(pipeline, default_cpu_count, stage_name="inf
         name=stage_name,
         stage_actor=InfographicExtractorStage,
         config=infographic_content_extractor_config,
-        min_replicas=0,
-        max_replicas=int(max(1, (default_cpu_count // 14))),  # 7% of available CPU cores
+        min_replicas=1,
+        max_replicas=1,
     )
 
     return stage_name
@@ -308,8 +304,8 @@ def add_image_extractor_stage(pipeline, default_cpu_count, stage_name="image_ext
         name=stage_name,
         stage_actor=ImageExtractorStage,
         config=image_extractor_config,
-        min_replicas=0,
-        max_replicas=int(max(1, (default_cpu_count // 14))),  # 7% of available CPU cores
+        min_replicas=1,
+        max_replicas=1,
     )
 
     return stage_name
@@ -330,8 +326,8 @@ def add_docx_extractor_stage(pipeline, default_cpu_count, stage_name="docx_extra
         name=stage_name,
         stage_actor=DocxExtractorStage,
         config=DocxExtractorSchema(**docx_extractor_config),
-        min_replicas=0,
-        max_replicas=int(max(1, (default_cpu_count // 14))),  # 7% of available CPU cores
+        min_replicas=1,
+        max_replicas=1,
     )
 
     return stage_name
@@ -352,8 +348,8 @@ def add_pptx_extractor_stage(pipeline, default_cpu_count, stage_name="pptx_extra
         name=stage_name,
         stage_actor=PPTXExtractorStage,
         config=PPTXExtractorSchema(**pptx_extractor_config),
-        min_replicas=0,
-        max_replicas=int(max(1, (default_cpu_count // 14))),  # 7% of available CPU cores
+        min_replicas=1,
+        max_replicas=1,
     )
 
     return stage_name
@@ -378,7 +374,7 @@ def add_audio_extractor_stage(pipeline, default_cpu_count, stage_name="audio_ext
         name=stage_name,
         stage_actor=AudioExtractorStage,
         config=audio_extractor_config,
-        min_replicas=0,
+        min_replicas=1,
         max_replicas=1,  # Audio extraction is a heavy IO bound operation with minimal CPU usage
     )
 
@@ -391,8 +387,8 @@ def add_html_extractor_stage(pipeline, default_cpu_count, stage_name="html_extra
         name=stage_name,
         stage_actor=HtmlExtractorStage,
         config=HtmlExtractorSchema(),
-        min_replicas=0,
-        max_replicas=int(max(1, (default_cpu_count // 14))),  # 7% of available CPU cores
+        min_replicas=1,
+        max_replicas=1,
     )
 
     return stage_name
@@ -412,8 +408,8 @@ def add_otel_tracer_stage(pipeline, default_cpu_count, stage_name="otel_tracer")
         name=stage_name,
         stage_actor=OpenTelemetryTracerStage,
         config=otel_tracer_config,
-        min_replicas=0,
-        max_replicas=2,
+        min_replicas=1,
+        max_replicas=1,
     )
 
     return stage_name
@@ -426,7 +422,7 @@ def add_image_dedup_stage(pipeline, default_cpu_count, stage_name="image_dedup")
         name=stage_name,
         stage_actor=ImageDedupStage,
         config=config,
-        min_replicas=0,
+        min_replicas=1,
         max_replicas=1,
     )
 
@@ -440,7 +436,7 @@ def add_image_filter_stage(pipeline, default_cpu_count, stage_name="image_filter
         name=stage_name,
         stage_actor=ImageFilterStage,
         config=config,
-        min_replicas=0,
+        min_replicas=1,
         max_replicas=1,
     )
 
@@ -456,8 +452,8 @@ def add_text_splitter_stage(pipeline, default_cpu_count, stage_name="text_splitt
         name=stage_name,
         stage_actor=TextSplitterStage,
         config=config,
-        min_replicas=0,
-        max_replicas=int(max(1, (default_cpu_count // 14))),  # 7% of available CPU cores
+        min_replicas=1,
+        max_replicas=1,
     )
 
     return stage_name
@@ -488,7 +484,7 @@ def add_image_caption_stage(pipeline, default_cpu_count, stage_name="image_capti
         name=stage_name,
         stage_actor=ImageCaptionTransformStage,
         config=config,
-        min_replicas=0,
+        min_replicas=1,
         max_replicas=1,
     )
 
@@ -515,11 +511,7 @@ def add_text_embedding_stage(pipeline, default_cpu_count, stage_name="text_embed
     )
 
     pipeline.add_stage(
-        name=stage_name,
-        stage_actor=TextEmbeddingTransformStage,
-        config=config,
-        min_replicas=0,
-        max_replicas=int(max(1, (default_cpu_count // 14))),  # 7% of available CPU cores
+        name=stage_name, stage_actor=TextEmbeddingTransformStage, config=config, min_replicas=1, max_replicas=1
     )
 
     return stage_name
@@ -532,7 +524,7 @@ def add_embedding_storage_stage(pipeline, default_cpu_count, stage_name="embeddi
         name=stage_name,
         stage_actor=EmbeddingStorageStage,
         config=config,
-        min_replicas=0,
+        min_replicas=1,
         max_replicas=1,
     )
 
@@ -545,7 +537,7 @@ def add_image_storage_stage(pipeline, default_cpu_count, stage_name="image_stora
         name=stage_name,
         stage_actor=ImageStorageStage,
         config=config,
-        min_replicas=0,
+        min_replicas=1,
         max_replicas=1,
     )
 
@@ -583,8 +575,8 @@ def add_message_broker_response_stage(pipeline, default_cpu_count, stage_name="b
         name=stage_name,
         stage_actor=MessageBrokerTaskSinkStage,
         config=sink_config,
-        min_replicas=0,
-        max_replicas=2,
+        min_replicas=1,
+        max_replicas=1,
     )
 
     return stage_name

{nv_ingest-25.6.4rc4.dist-info → nv_ingest-25.6.4rc6.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest
-Version: 25.6.4rc4
+Version: 25.6.4rc6
 Summary: Python module for multimodal document ingestion
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License

{nv_ingest-25.6.4rc4.dist-info → nv_ingest-25.6.4rc6.dist-info}/RECORD

@@ -67,7 +67,7 @@ nv_ingest/framework/orchestration/ray/util/pipeline/__init__.py,sha256=wQSlVx3T1
 nv_ingest/framework/orchestration/ray/util/pipeline/pid_controller.py,sha256=AWyCFPP41vp1NOkO2urqm7vh-sTGKypJxwhdq8HxK6Q,50681
 nv_ingest/framework/orchestration/ray/util/pipeline/pipeline_builders.py,sha256=jMYnVe_0rb1OIO9mlB4LH3uXtgaXBbUG-rDPx6fe6J8,10456
 nv_ingest/framework/orchestration/ray/util/pipeline/pipeline_runners.py,sha256=IKQHlEwe0xsjr4MgQJVL0UtnKha1qaoPFc08DF5QzMM,14351
-nv_ingest/framework/orchestration/ray/util/pipeline/stage_builders.py,sha256=ZFJkeJNbDM_GsedUlfk2B8kI93L_MNK6gxPgeryZM6I,21463
+nv_ingest/framework/orchestration/ray/util/pipeline/stage_builders.py,sha256=rE1azhkVh5Gk_j4GdmUArSM76SByR44YRNFar4-BSkw,20803
 nv_ingest/framework/orchestration/ray/util/system_tools/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest/framework/orchestration/ray/util/system_tools/memory.py,sha256=ICqY0LLB3hFTZk03iX5yffMSKFH2q_aQomtDVzS_mKw,2228
 nv_ingest/framework/orchestration/ray/util/system_tools/visualizers.py,sha256=2oHZdO_3L1LGuzpyNmZBDh19n0E-APAaHk4MEwBwSHs,12895
@@ -95,8 +95,8 @@ nv_ingest/framework/util/service/meta/ingest/__init__.py,sha256=wQSlVx3T14ZgQAt-
 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-25.6.4rc4.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-nv_ingest-25.6.4rc4.dist-info/METADATA,sha256=FL8Q5cnYKQKGxGuURlNuGhOOUYjGlBkRz7BzAebV4e8,15167
-nv_ingest-25.6.4rc4.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nv_ingest-25.6.4rc4.dist-info/top_level.txt,sha256=sjb0ajIsgn3YgftSjZHlYO0HjYAIIhNuXG_AmywCvaU,10
-nv_ingest-25.6.4rc4.dist-info/RECORD,,
+nv_ingest-25.6.4rc6.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+nv_ingest-25.6.4rc6.dist-info/METADATA,sha256=ZbUiqF3zSJpLlk9P_uSQ3D_JcUUMHSVGSttvzyuZLlo,15167
+nv_ingest-25.6.4rc6.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nv_ingest-25.6.4rc6.dist-info/top_level.txt,sha256=sjb0ajIsgn3YgftSjZHlYO0HjYAIIhNuXG_AmywCvaU,10
+nv_ingest-25.6.4rc6.dist-info/RECORD,,