nv-ingest-api 2025.5.13.dev20250513__py3-none-any.whl → 2025.5.14.dev20250514__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.

This version of nv-ingest-api might be problematic. Click here for more details.

Files changed (13) hide show
  1. nv_ingest_api/internal/extract/image/chart_extractor.py +3 -3
  2. nv_ingest_api/internal/extract/image/image_helpers/common.py +1 -1
  3. nv_ingest_api/internal/extract/image/infographic_extractor.py +1 -1
  4. nv_ingest_api/internal/extract/image/table_extractor.py +2 -2
  5. nv_ingest_api/internal/extract/pdf/engines/nemoretriever.py +1 -1
  6. nv_ingest_api/internal/extract/pdf/engines/pdfium.py +1 -1
  7. nv_ingest_api/util/image_processing/processing.py +1 -1
  8. nv_ingest_api/util/pdf/pdfium.py +1 -1
  9. {nv_ingest_api-2025.5.13.dev20250513.dist-info → nv_ingest_api-2025.5.14.dev20250514.dist-info}/METADATA +1 -1
  10. {nv_ingest_api-2025.5.13.dev20250513.dist-info → nv_ingest_api-2025.5.14.dev20250514.dist-info}/RECORD +13 -13
  11. {nv_ingest_api-2025.5.13.dev20250513.dist-info → nv_ingest_api-2025.5.14.dev20250514.dist-info}/WHEEL +1 -1
  12. {nv_ingest_api-2025.5.13.dev20250513.dist-info → nv_ingest_api-2025.5.14.dev20250514.dist-info}/licenses/LICENSE +0 -0
  13. {nv_ingest_api-2025.5.13.dev20250513.dist-info → nv_ingest_api-2025.5.14.dev20250514.dist-info}/top_level.txt +0 -0
nv_ingest_api/internal/extract/image/chart_extractor.py CHANGED
@@ -27,7 +27,7 @@ from nv_ingest_api.util.nim import create_inference_client
27
27
  PADDLE_MIN_WIDTH = 32
28
28
  PADDLE_MIN_HEIGHT = 32
29
29
 
30
- logger = logging.getLogger(f"morpheus.{__name__}")
30
+ logger = logging.getLogger(f"ray.{__name__}")
31
31
 
32
32
 
33
33
  def _filter_valid_chart_images(
@@ -80,7 +80,7 @@ def _run_chart_inference(
80
80
  yolox_client.infer,
81
81
  data=data_yolox,
82
82
  model_name="yolox",
83
- stage_name="chart_data_extraction",
83
+ stage_name="chart_extraction",
84
84
  max_batch_size=8,
85
85
  trace_info=trace_info,
86
86
  )
@@ -88,7 +88,7 @@ def _run_chart_inference(
88
88
  paddle_client.infer,
89
89
  data=data_paddle,
90
90
  model_name="paddle",
91
- stage_name="chart_data_extraction",
91
+ stage_name="chart_extraction",
92
92
  max_batch_size=1 if paddle_client.protocol == "grpc" else 2,
93
93
  trace_info=trace_info,
94
94
  )
nv_ingest_api/internal/extract/image/image_helpers/common.py CHANGED
@@ -223,7 +223,7 @@ def extract_page_elements_from_images(
223
223
  model_name="yolox",
224
224
  max_batch_size=YOLOX_MAX_BATCH_SIZE,
225
225
  trace_info=trace_info,
226
- stage_name="pdf_content_extractor",
226
+ stage_name="pdf_extraction",
227
227
  )
228
228
 
229
229
  # Process each result along with its corresponding image.
nv_ingest_api/internal/extract/image/infographic_extractor.py CHANGED
@@ -100,7 +100,7 @@ def _update_infographic_metadata(
100
100
  paddle_results = paddle_client.infer(
101
101
  data=data_paddle,
102
102
  model_name="paddle",
103
- stage_name="infographic_data_extraction",
103
+ stage_name="infographic_extraction",
104
104
  max_batch_size=1 if paddle_client.protocol == "grpc" else 2,
105
105
  trace_info=trace_info,
106
106
  )
nv_ingest_api/internal/extract/image/table_extractor.py CHANGED
@@ -81,7 +81,7 @@ def _run_inference(
81
81
  yolox_client.infer,
82
82
  data=data_yolox,
83
83
  model_name="yolox",
84
- stage_name="table_data_extraction",
84
+ stage_name="table_extraction",
85
85
  max_batch_size=8,
86
86
  trace_info=trace_info,
87
87
  )
@@ -89,7 +89,7 @@ def _run_inference(
89
89
  paddle_client.infer,
90
90
  data=data_paddle,
91
91
  model_name="paddle",
92
- stage_name="table_data_extraction",
92
+ stage_name="table_extraction",
93
93
  max_batch_size=1 if paddle_client.protocol == "grpc" else 2,
94
94
  trace_info=trace_info,
95
95
  )
nv_ingest_api/internal/extract/pdf/engines/nemoretriever.py CHANGED
@@ -466,7 +466,7 @@ def _extract_text_and_bounding_boxes(
466
466
  inference_results = nemoretriever_parse_client.infer(
467
467
  data=data,
468
468
  model_name="nemoretriever_parse",
469
- stage_name="pdf_content_extractor",
469
+ stage_name="pdf_extraction",
470
470
  max_batch_size=NEMORETRIEVER_PARSE_MAX_BATCH_SIZE,
471
471
  execution_trace_log=execution_trace_log,
472
472
  )
nv_ingest_api/internal/extract/pdf/engines/pdfium.py CHANGED
@@ -105,7 +105,7 @@ def _extract_page_elements_using_image_ensemble(
105
105
  model_name="yolox",
106
106
  max_batch_size=YOLOX_MAX_BATCH_SIZE,
107
107
  trace_info=execution_trace_log,
108
- stage_name="pdf_content_extractor",
108
+ stage_name="pdf_extraction",
109
109
  )
110
110
 
111
111
  # Process results: iterate over each image's inference output.
nv_ingest_api/util/image_processing/processing.py CHANGED
@@ -150,7 +150,7 @@ def extract_tables_and_charts_yolox(
150
150
  min_score=YOLOX_MIN_SCORE,
151
151
  final_thresh=YOLOX_FINAL_SCORE,
152
152
  trace_info=trace_info,
153
- stage_name="pdf_content_extractor",
153
+ stage_name="pdf_extraction",
154
154
  )
155
155
 
156
156
  # Process results: iterate over each image's inference output.
nv_ingest_api/util/pdf/pdfium.py CHANGED
@@ -119,7 +119,7 @@ def pdfium_try_get_bitmap_as_numpy(image_obj) -> np.ndarray:
119
119
  return img_array
120
120
 
121
121
 
122
- @traceable_func(trace_name="pdf_content_extractor::pdfium_pages_to_numpy")
122
+ @traceable_func(trace_name="pdf_extraction::pdfium_pages_to_numpy")
123
123
  def pdfium_pages_to_numpy(
124
124
  pages: List[pdfium.PdfPage],
125
125
  render_dpi: int = 300,
{nv_ingest_api-2025.5.13.dev20250513.dist-info → nv_ingest_api-2025.5.14.dev20250514.dist-info}/METADATA RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: nv-ingest-api
3
- Version: 2025.5.13.dev20250513
3
+ Version: 2025.5.14.dev20250514
4
4
  Summary: Python module with core document ingestion functions.
5
5
  Author-email: Jeremy Dyer <jdyer@nvidia.com>
6
6
  License: Apache License
{nv_ingest_api-2025.5.13.dev20250513.dist-info → nv_ingest_api-2025.5.14.dev20250514.dist-info}/RECORD RENAMED
@@ -18,19 +18,19 @@ nv_ingest_api/internal/extract/docx/engines/docxreader_helpers/__init__.py,sha25
18
18
  nv_ingest_api/internal/extract/docx/engines/docxreader_helpers/docx_helper.py,sha256=1wkciAxu8lz9WuPuoleJFy2s09ieSzXl1S71F9r0BWA,4385
19
19
  nv_ingest_api/internal/extract/docx/engines/docxreader_helpers/docxreader.py,sha256=CM2yV8lfEw1F1ORAjupD4gyIKX0PDDJrL3nsZ5Mnrgg,31539
20
20
  nv_ingest_api/internal/extract/image/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
21
- nv_ingest_api/internal/extract/image/chart_extractor.py,sha256=Jy_fNmDbZcdni55Fq7vT6NdbYnCyGoyw0J7QjpK-KPc,13315
21
+ nv_ingest_api/internal/extract/image/chart_extractor.py,sha256=CkaW8ihPmGMQGrZh0ih14gtEpWuGOJ8InPQfZwpsP2g,13300
22
22
  nv_ingest_api/internal/extract/image/image_extractor.py,sha256=4tUWinuFMN3ukWa2tZa2_LtzRiTyUAUCBF6BDkUEvm0,8705
23
- nv_ingest_api/internal/extract/image/infographic_extractor.py,sha256=k4Z6JwsoNKsyfmpaQkN_dxJpAv9-RVsRL1BfSWUtXTM,8908
24
- nv_ingest_api/internal/extract/image/table_extractor.py,sha256=80FQef4Dsn6__MNIRCQzFf32s4wUyTOzBFgmA84JZJk,13133
23
+ nv_ingest_api/internal/extract/image/infographic_extractor.py,sha256=yc9b2q_Ea08CEVclZ47UkpU4F7qlakPuU3UV9P013W0,8903
24
+ nv_ingest_api/internal/extract/image/table_extractor.py,sha256=ivHaJxYjeHvFM1PZIpxVabPadxtcTsu51j398ZjMhD4,13123
25
25
  nv_ingest_api/internal/extract/image/image_helpers/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
26
- nv_ingest_api/internal/extract/image/image_helpers/common.py,sha256=NU8TEU9p2aIL_KppyhtTgRUPqD4MsanxATG19rKhGjw,15032
26
+ nv_ingest_api/internal/extract/image/image_helpers/common.py,sha256=P8rcl4YPyeWeMJg7u1yejD3k9EnDVEbJgfYEnJ4WO5c,15025
27
27
  nv_ingest_api/internal/extract/pdf/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
28
28
  nv_ingest_api/internal/extract/pdf/pdf_extractor.py,sha256=CxtWaD6mql9MEqSdk2CfSQ9T-Bn87beBkCOuGGjxGt8,2934
29
29
  nv_ingest_api/internal/extract/pdf/engines/__init__.py,sha256=u4GnAZmDKRl0RwYGIRiozIRw70Kybw3A72-lcKFeoTI,582
30
30
  nv_ingest_api/internal/extract/pdf/engines/adobe.py,sha256=VT0dEqkU-y2uGkaCqxtKYov_Q8R1028UQVBchgMLca4,17466
31
31
  nv_ingest_api/internal/extract/pdf/engines/llama.py,sha256=PpKTqS8jGHBV6mKLGZWwjpfT8ga6Fy8ffrvL-gPAf2c,8182
32
- nv_ingest_api/internal/extract/pdf/engines/nemoretriever.py,sha256=6xI8RKceIUqZvByLm433nWzdJUoAOWvqRpTX0Fy-8lg,22933
33
- nv_ingest_api/internal/extract/pdf/engines/pdfium.py,sha256=jUcquCWbyQPNCHZLaV-XnVqUFsajX4YxVFCiWWwD4QQ,22367
32
+ nv_ingest_api/internal/extract/pdf/engines/nemoretriever.py,sha256=Uqj1NH7yWga9P6_vCzgny1WKALfF--UdAaGHUF8K_aQ,22926
33
+ nv_ingest_api/internal/extract/pdf/engines/pdfium.py,sha256=fDbrZwJ-lgeHYOq107WXehzdSvyF8zEDza_9UkDm5aE,22360
34
34
  nv_ingest_api/internal/extract/pdf/engines/tika.py,sha256=6GyR2l6EsgNZl9jnYDXLeKNK9Fj2Mw9y2UWDq-eSkOc,3169
35
35
  nv_ingest_api/internal/extract/pdf/engines/unstructured_io.py,sha256=jrv2B4VZAH4PevAQrFz965qz8UyXq3rViiOTbGLejec,14908
36
36
  nv_ingest_api/internal/extract/pdf/engines/pdf_helpers/__init__.py,sha256=Jk3wrQ2CZs167juvEZ-uV6qXWQjR08hhIu8otk2MWj4,4931
@@ -118,7 +118,7 @@ nv_ingest_api/util/exception_handlers/pdf.py,sha256=FUC41QJKDCfiTv-1c1_8Isxwt1xM
118
118
  nv_ingest_api/util/exception_handlers/schemas.py,sha256=NJngVNf9sk5Uz6CFFfkNO_LBAMt2QZUcMYGxX64oYRk,2179
119
119
  nv_ingest_api/util/image_processing/__init__.py,sha256=Jiy8C1ZuSrNb_eBM1ZTV9IKFIsnjhZi6Ku3JJhVLimA,104
120
120
  nv_ingest_api/util/image_processing/clustering.py,sha256=sUGlZI4cx1q8h4Pns1N9JVpdfSM2BOH8zRmn9QFCtzI,9236
121
- nv_ingest_api/util/image_processing/processing.py,sha256=dHyoxoI2btKT04ODJK0ChB8MR6eCnZ0ZLpbEQowCb5A,6561
121
+ nv_ingest_api/util/image_processing/processing.py,sha256=LSoDDEmahr7a-qSS12McVcowRe3dOrAZwa1h-PD_JPQ,6554
122
122
  nv_ingest_api/util/image_processing/table_and_chart.py,sha256=bxOu9PZYkG_WFCDGw_JLaO60S2pDSN8EOWK3xkIwr2A,14376
123
123
  nv_ingest_api/util/image_processing/transforms.py,sha256=Kz9hrizV314Hy7cRCYK9ZmhmBbVUOZ_z0HEpzZYcslQ,14081
124
124
  nv_ingest_api/util/logging/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -134,7 +134,7 @@ nv_ingest_api/util/multi_processing/__init__.py,sha256=4fojP8Rp_5Hu1YAkqGylqTyEZ
134
134
  nv_ingest_api/util/multi_processing/mp_pool_singleton.py,sha256=dTfP82DgGPaXEJH3jywTO8rNlLZUniD4FFzwv84_giE,7372
135
135
  nv_ingest_api/util/nim/__init__.py,sha256=UqbiXFCqjWcjNvoduXd_0gOUOGBT8JvppiYHOmMyneA,1775
136
136
  nv_ingest_api/util/pdf/__init__.py,sha256=uLsBITo_XfgbwpzqXUm1IYX6XlZrTfx6T1cIhdILwG8,140
137
- nv_ingest_api/util/pdf/pdfium.py,sha256=1Py9qj-dWHDisuSc50iHGRSaY4QlyfRKYTwz99nskZI,15768
137
+ nv_ingest_api/util/pdf/pdfium.py,sha256=Ch9Gh5jRLcBr3stjCckqWwTUL-T0sI50PlQnZHo_9NA,15761
138
138
  nv_ingest_api/util/schema/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
139
139
  nv_ingest_api/util/schema/schema_validator.py,sha256=H0yZ_i_HZaiBRUCGmTBfRB9-hURhVqyd10aS_ynM1_0,321
140
140
  nv_ingest_api/util/service_clients/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
@@ -147,8 +147,8 @@ nv_ingest_api/util/service_clients/rest/rest_client.py,sha256=dZ-jrk7IK7oNtHoXFS
147
147
  nv_ingest_api/util/string_processing/__init__.py,sha256=mkwHthyS-IILcLcL1tJYeF6mpqX3pxEw5aUzDGjTSeU,1411
148
148
  nv_ingest_api/util/system/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
149
149
  nv_ingest_api/util/system/hardware_info.py,sha256=JGxBbF3kvgYbwhhWvtjNzPxVZQV_npmsordAioBrglo,19252
150
- nv_ingest_api-2025.5.13.dev20250513.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
151
- nv_ingest_api-2025.5.13.dev20250513.dist-info/METADATA,sha256=SnzH1EvOcXSwFdwNfRNQsNfEUbYekGLBHgCugs9UNQ8,13889
152
- nv_ingest_api-2025.5.13.dev20250513.dist-info/WHEEL,sha256=DnLRTWE75wApRYVsjgc6wsVswC54sMSJhAEd4xhDpBk,91
153
- nv_ingest_api-2025.5.13.dev20250513.dist-info/top_level.txt,sha256=abjYMlTJGoG5tOdfIB-IWvLyKclw6HLaRSc8MxX4X6I,14
154
- nv_ingest_api-2025.5.13.dev20250513.dist-info/RECORD,,
150
+ nv_ingest_api-2025.5.14.dev20250514.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
151
+ nv_ingest_api-2025.5.14.dev20250514.dist-info/METADATA,sha256=ovvmVcwxqjyt5DQyPur4gOwFTmMRNiB-0S0lVN9v9OU,13889
152
+ nv_ingest_api-2025.5.14.dev20250514.dist-info/WHEEL,sha256=QZxptf4Y1BKFRCEDxD4h2V0mBFQOVFLFEpvxHmIs52A,91
153
+ nv_ingest_api-2025.5.14.dev20250514.dist-info/top_level.txt,sha256=abjYMlTJGoG5tOdfIB-IWvLyKclw6HLaRSc8MxX4X6I,14
154
+ nv_ingest_api-2025.5.14.dev20250514.dist-info/RECORD,,
{nv_ingest_api-2025.5.13.dev20250513.dist-info → nv_ingest_api-2025.5.14.dev20250514.dist-info}/WHEEL RENAMED
@@ -1,5 +1,5 @@
1
1
  Wheel-Version: 1.0
2
- Generator: setuptools (80.4.0)
2
+ Generator: setuptools (80.6.0)
3
3
  Root-Is-Purelib: true
4
4
  Tag: py3-none-any
5
5