nv-ingest-api 2025.6.3.dev20250603__py3-none-any.whl → 2025.6.4.dev20250604__py3-none-any.whl

nv_ingest_api/internal/extract/audio/audio_extraction.py

@@ -5,6 +5,8 @@
 import logging
 
 import pandas as pd
+import functools
+import uuid
 from typing import Any
 from typing import Dict
 from typing import Optional
@@ -21,7 +23,7 @@ logger = logging.getLogger(__name__)
 
 
 @unified_exception_handler
-def _update_audio_metadata(row: pd.Series, audio_client: Any, trace_info: Dict) -> Dict:
+def _extract_from_audio(row: pd.Series, audio_client: Any, trace_info: Dict, segment_audio: bool = False) -> Dict:
     """
     Modifies the metadata of a row if the conditions for table extraction are met.
 
@@ -56,24 +58,42 @@ def _update_audio_metadata(row: pd.Series, audio_client: Any, trace_info: Dict)
     base64_audio = metadata.pop("content")
     content_metadata = metadata.get("content_metadata", {})
 
-    # Only modify if content type is audio
+    # Only extract transcript if content type is audio
     if (content_metadata.get("type") != ContentTypeEnum.AUDIO) or (base64_audio in (None, "")):
-        return metadata
+        return [row.to_list()]
 
-    # Modify audio metadata with the result from the inference model
-    audio_result = audio_client.infer(
+    # Get the result from the inference model
+    segments, transcript = audio_client.infer(
         base64_audio,
         model_name="parakeet",
         trace_info=trace_info,  # traceable_func arg
         stage_name="audio_extraction",
     )
 
-    row["document_type"] = ContentTypeEnum.AUDIO
-    audio_metadata = {"audio_transcript": audio_result}
-    metadata["audio_metadata"] = validate_schema(audio_metadata, AudioMetadataSchema).model_dump()
-    row["metadata"] = validate_schema(metadata, MetadataSchema).model_dump()
+    extracted_data = []
+    if segment_audio:
+        for segment in segments:
+            segment_metadata = metadata.copy()
+            audio_metadata = {"audio_transcript": segment["text"]}
+            segment_metadata["audio_metadata"] = validate_schema(audio_metadata, AudioMetadataSchema).model_dump()
+            segment_metadata["content_metadata"]["start_time"] = segment["start"]
+            segment_metadata["content_metadata"]["end_time"] = segment["end"]
+
+            extracted_data.append(
+                [
+                    ContentTypeEnum.AUDIO,
+                    validate_schema(segment_metadata, MetadataSchema).model_dump(),
+                    str(uuid.uuid4()),
+                ]
+            )
+    else:
+        audio_metadata = {"audio_transcript": transcript}
+        metadata["audio_metadata"] = validate_schema(audio_metadata, AudioMetadataSchema).model_dump()
+        extracted_data.append(
+            [ContentTypeEnum.AUDIO, validate_schema(metadata, MetadataSchema).model_dump(), str(uuid.uuid4())]
+        )
 
-    return metadata
+    return extracted_data
 
 
 def extract_text_from_audio_internal(
@@ -121,6 +141,7 @@ def extract_text_from_audio_internal(
     function_id = extract_params.get("function_id") or audio_extraction_config.function_id
     use_ssl = extract_params.get("use_ssl") or audio_extraction_config.use_ssl
     ssl_cert = extract_params.get("ssl_cert") or audio_extraction_config.ssl_cert
+    segment_audio = extract_params.get("segment_audio") or audio_extraction_config.segment_audio
 
     parakeet_client = create_audio_inference_client(
         (grpc_endpoint, http_endpoint),
@@ -136,12 +157,27 @@ def extract_text_from_audio_internal(
         logger.debug("No trace_info provided. Initialized empty trace_info dictionary.")
 
     try:
-        # Apply the _update_metadata function to each row in the DataFrame
-        df_extraction_ledger["metadata"] = df_extraction_ledger.apply(
-            _update_audio_metadata, axis=1, args=(parakeet_client, execution_trace_log)
+        # Create a partial function to extract using the provided configurations.
+        _extract_from_audio_partial = functools.partial(
+            _extract_from_audio,
+            audio_client=parakeet_client,
+            trace_info=execution_trace_log,
+            segment_audio=segment_audio,
         )
 
-        return df_extraction_ledger, execution_trace_log
+        # Apply the _extract_from_audio_partial function to each row in the DataFrame
+        extraction_series = df_extraction_ledger.apply(_extract_from_audio_partial, axis=1)
+
+        # Explode the results if the extraction returns lists.
+        extraction_series = extraction_series.explode().dropna()
+
+        # Convert the extracted results into a DataFrame.
+        if not extraction_series.empty:
+            extracted_df = pd.DataFrame(extraction_series.to_list(), columns=["document_type", "metadata", "uuid"])
+        else:
+            extracted_df = pd.DataFrame({"document_type": [], "metadata": [], "uuid": []})
+
+        return extracted_df, execution_trace_log
 
     except Exception as e:
         logger.exception(f"Error occurred while extracting audio data: {e}", exc_info=True)

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

@@ -101,7 +101,7 @@ class ParakeetClient:
         segments, transcript = process_transcription_response(response)
         logger.debug("Processing Parakeet inference results (pass-through).")
 
-        return transcript
+        return segments, transcript
 
     def transcribe(
         self,

nv_ingest_api/internal/schemas/extract/extract_audio_schema.py

@@ -48,6 +48,7 @@ class AudioConfigSchema(BaseModel):
     function_id: Optional[str] = None
     use_ssl: Optional[bool] = None
     ssl_cert: Optional[str] = None
+    segment_audio: Optional[bool] = None
 
     @root_validator(pre=True)
     def validate_endpoints(cls, values):

nv_ingest_api/internal/schemas/meta/ingest_job_schema.py

@@ -124,6 +124,7 @@ class IngestTaskAudioExtraction(BaseModelNoExt):
     function_id: Optional[str] = None
     use_ssl: Optional[bool] = None
     ssl_cert: Optional[str] = None
+    segment_audio: Optional[bool] = None
 
 
 class IngestTaskTableExtraction(BaseModelNoExt):

nv_ingest_api/internal/schemas/meta/metadata_schema.py

@@ -97,6 +97,8 @@ class ContentMetadataSchema(BaseModelNoExt):
     page_number: int = -1
     hierarchy: ContentHierarchySchema = ContentHierarchySchema()
     subtype: Union[ContentTypeEnum, str] = ""
+    start_time: int = -1
+    end_time: int = -1
 
 
 class TextMetadataSchema(BaseModelNoExt):

{nv_ingest_api-2025.6.3.dev20250603.dist-info → nv_ingest_api-2025.6.4.dev20250604.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest-api
-Version: 2025.6.3.dev20250603
+Version: 2025.6.4.dev20250604
 Summary: Python module with core document ingestion functions.
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License

{nv_ingest_api-2025.6.3.dev20250603.dist-info → nv_ingest_api-2025.6.4.dev20250604.dist-info}/RECORD

@@ -10,7 +10,7 @@ nv_ingest_api/internal/enums/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8y
 nv_ingest_api/internal/enums/common.py,sha256=HSj7qqNr6KXu_FIyK_Wvel24R-r8lV7dLA173z5XFBc,12321
 nv_ingest_api/internal/extract/__init__.py,sha256=uLsBITo_XfgbwpzqXUm1IYX6XlZrTfx6T1cIhdILwG8,140
 nv_ingest_api/internal/extract/audio/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest_api/internal/extract/audio/audio_extraction.py,sha256=L8cK7xB6QTaSx8gsrdyaYHYh0HpW6lycGfduCk7XSMg,5364
+nv_ingest_api/internal/extract/audio/audio_extraction.py,sha256=_jf_UC_FTqZr-xEpwG8edwBzdDjM01gGhqm9ulOsDcY,6973
 nv_ingest_api/internal/extract/docx/__init__.py,sha256=HIHfzSig66GT0Uk8qsGBm_f13fKYcPtItBicRUWOOVA,183
 nv_ingest_api/internal/extract/docx/docx_extractor.py,sha256=jjbL12F5dtpbqHRbhL0uomSiQ90bcQq3N7M43XYsq34,8356
 nv_ingest_api/internal/extract/docx/engines/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -57,7 +57,7 @@ nv_ingest_api/internal/primitives/nim/model_interface/deplot.py,sha256=TvKdk6PTu
 nv_ingest_api/internal/primitives/nim/model_interface/helpers.py,sha256=x35a9AyTYxpESQflLo_YnhVOKblQKVen6vGGFaXmNiE,9927
 nv_ingest_api/internal/primitives/nim/model_interface/nemoretriever_parse.py,sha256=MFWPqMTXs_MZG3ripRR21o7f_mVeoE46Q10yvJ8KNr0,7023
 nv_ingest_api/internal/primitives/nim/model_interface/paddle.py,sha256=rSUPwl5XOrqneoS6aKhatVjrNBg_LhP3nwUWS_aTwz0,17950
-nv_ingest_api/internal/primitives/nim/model_interface/parakeet.py,sha256=OYg4AGki_wm--Np9VlSm0eZC-r54GbDOISbe9v0B9fw,12967
+nv_ingest_api/internal/primitives/nim/model_interface/parakeet.py,sha256=5PqD2JuHY2rwd-6SSB4axr2Dd79vm95sAEkcmI3U7ME,12977
 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=qJ382PU1ZrIM-SR3cqIhtY_W2rmHec2HIa2aUB2SvaU,6031
 nv_ingest_api/internal/primitives/nim/model_interface/yolox.py,sha256=uYXqdvqgkyS4Yfr9ZoikRDX4e94OV3ch3Xhv3JVg-3s,49581
@@ -67,7 +67,7 @@ nv_ingest_api/internal/primitives/tracing/logging.py,sha256=SSzIgS7afLH-e1C7VagY
 nv_ingest_api/internal/primitives/tracing/tagging.py,sha256=O5dD7Z7j43nrjqn0AxhxOPm5zIyMFo0akxaWU_FguAM,7866
 nv_ingest_api/internal/schemas/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/schemas/extract/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest_api/internal/schemas/extract/extract_audio_schema.py,sha256=VVppZgV1lnyJCTfADexzoj3V0lOSq3t6Dw_6VhIxZ7k,3771
+nv_ingest_api/internal/schemas/extract/extract_audio_schema.py,sha256=W-nEBriqiNkjpaQ5AT_8LhtVXlW8AhlcftmoeQQtKAs,3812
 nv_ingest_api/internal/schemas/extract/extract_chart_schema.py,sha256=iu8lHQC0zbBB9VRK7PZisAVzpeSpFqjcXRAnwZ9OzoM,4301
 nv_ingest_api/internal/schemas/extract/extract_docx_schema.py,sha256=M2N7WjMNvSemHcJHWeNUD_kFG0wC5VE2W3K6SVrJqvA,3761
 nv_ingest_api/internal/schemas/extract/extract_html_schema.py,sha256=lazpONTGZ6Fl420BGBAr6rogFGtlzBiZTc1uA694OIs,841
@@ -82,8 +82,8 @@ nv_ingest_api/internal/schemas/message_brokers/request_schema.py,sha256=LZX_wXDx
 nv_ingest_api/internal/schemas/message_brokers/response_schema.py,sha256=4b275HlzBSzpmuE2wdoeaGKPCdKki3wuWldtRIfrj8w,727
 nv_ingest_api/internal/schemas/meta/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/schemas/meta/base_model_noext.py,sha256=8hXU1uuiqZ6t8EsoZ8vlC5EFf2zSZrKEX133FcfZMwI,316
-nv_ingest_api/internal/schemas/meta/ingest_job_schema.py,sha256=pkMOKIyfAmVcimqZ-zDXngW_lzYxdXYHv8C8cdduUvA,8083
-nv_ingest_api/internal/schemas/meta/metadata_schema.py,sha256=_FAE-yeb01hxq05SXrV3NLM4DPUPSfnIbH6ZMliWsEg,6625
+nv_ingest_api/internal/schemas/meta/ingest_job_schema.py,sha256=JqlK66fg_eRQE9kwzAX4b15FqOSbpgrzLWkcWrWbKdE,8124
+nv_ingest_api/internal/schemas/meta/metadata_schema.py,sha256=KqCUMeBNNgPtpBWzzNhsMtlfr_XvUxnALBbYkI-kfT4,6673
 nv_ingest_api/internal/schemas/mutate/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/schemas/mutate/mutate_image_dedup_schema.py,sha256=k1JOdlPPpsipc0XhHf-9YxJ_-W0HvpVE1ZhYmr7fzj0,395
 nv_ingest_api/internal/schemas/store/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
@@ -150,8 +150,8 @@ nv_ingest_api/util/service_clients/rest/rest_client.py,sha256=dZ-jrk7IK7oNtHoXFS
 nv_ingest_api/util/string_processing/__init__.py,sha256=mkwHthyS-IILcLcL1tJYeF6mpqX3pxEw5aUzDGjTSeU,1411
 nv_ingest_api/util/system/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nv_ingest_api/util/system/hardware_info.py,sha256=ORZeKpH9kSGU_vuPhyBwkIiMyCViKUX2CP__MCjrfbU,19463
-nv_ingest_api-2025.6.3.dev20250603.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-nv_ingest_api-2025.6.3.dev20250603.dist-info/METADATA,sha256=A_3Cb69Ptv-NHBK1-GtK-W1txbWwJcQ0c1-11_gknDc,13918
-nv_ingest_api-2025.6.3.dev20250603.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nv_ingest_api-2025.6.3.dev20250603.dist-info/top_level.txt,sha256=abjYMlTJGoG5tOdfIB-IWvLyKclw6HLaRSc8MxX4X6I,14
-nv_ingest_api-2025.6.3.dev20250603.dist-info/RECORD,,
+nv_ingest_api-2025.6.4.dev20250604.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+nv_ingest_api-2025.6.4.dev20250604.dist-info/METADATA,sha256=kDYJOI53puRP8V4DtVLQ1knOiOeqXsnzArjuqywQhNs,13918
+nv_ingest_api-2025.6.4.dev20250604.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nv_ingest_api-2025.6.4.dev20250604.dist-info/top_level.txt,sha256=abjYMlTJGoG5tOdfIB-IWvLyKclw6HLaRSc8MxX4X6I,14
+nv_ingest_api-2025.6.4.dev20250604.dist-info/RECORD,,