nv-ingest-api 2025.7.23.dev20250723__py3-none-any.whl → 2025.7.25.dev20250725__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 (8) hide show
  1. nv_ingest_api/internal/extract/pdf/engines/nemoretriever.py +0 -3
  2. nv_ingest_api/internal/schemas/transform/transform_image_caption_schema.py +1 -1
  3. nv_ingest_api/util/pdf/pdfium.py +1 -1
  4. {nv_ingest_api-2025.7.23.dev20250723.dist-info → nv_ingest_api-2025.7.25.dev20250725.dist-info}/METADATA +1 -1
  5. {nv_ingest_api-2025.7.23.dev20250723.dist-info → nv_ingest_api-2025.7.25.dev20250725.dist-info}/RECORD +8 -8
  6. {nv_ingest_api-2025.7.23.dev20250723.dist-info → nv_ingest_api-2025.7.25.dev20250725.dist-info}/WHEEL +0 -0
  7. {nv_ingest_api-2025.7.23.dev20250723.dist-info → nv_ingest_api-2025.7.25.dev20250725.dist-info}/licenses/LICENSE +0 -0
  8. {nv_ingest_api-2025.7.23.dev20250723.dist-info → nv_ingest_api-2025.7.25.dev20250725.dist-info}/top_level.txt +0 -0
nv_ingest_api/internal/extract/pdf/engines/nemoretriever.py CHANGED
@@ -258,9 +258,6 @@ def nemoretriever_parse_extractor(
258
258
  nemoretriever_parse_config.yolox_endpoints,
259
259
  nemoretriever_parse_config.yolox_infer_protocol,
260
260
  nemoretriever_parse_config.auth_token,
261
- input_names=["INPUT_IMAGES", "THRESHOLDS"],
262
- dtypes=["BYTES", "FP32"],
263
- output_names=["OUTPUT"],
264
261
  execution_trace_log=execution_trace_log,
265
262
  )
266
263
  futures.append(future_yolox)
nv_ingest_api/internal/schemas/transform/transform_image_caption_schema.py CHANGED
@@ -8,7 +8,7 @@ from pydantic import ConfigDict, BaseModel
8
8
 
9
9
  class ImageCaptionExtractionSchema(BaseModel):
10
10
  api_key: str = "api_key"
11
- endpoint_url: str = "https://ai.api.nvidia.com/v1/gr/nvidia/llama-3.1-nemotron-nano-vl-8b-v1/chat/completions"
11
+ endpoint_url: str = "https://integrate.api.nvidia.com/v1/chat/completions"
12
12
  prompt: str = "Caption the content of this image:"
13
13
  model_name: str = "nvidia/llama-3.1-nemotron-nano-vl-8b-v1"
14
14
  raise_on_failure: bool = False
nv_ingest_api/util/pdf/pdfium.py CHANGED
@@ -386,7 +386,7 @@ def extract_image_like_objects_from_pdfium_page(page, merge=True, **kwargs):
386
386
  try:
387
387
  original_images, _ = pdfium_pages_to_numpy(
388
388
  [page], # A batch with a single image.
389
- render_dpi=300, # dpi = 72 is equivalent to scale = 1.
389
+ render_dpi=72, # dpi = 72 is equivalent to scale = 1.
390
390
  rotation=rotation, # Without rotation, coordinates from page.get_pos() will not match.
391
391
  )
392
392
  image_bboxes = extract_merged_images_from_pdfium_page(page, merge=merge, **kwargs)
{nv_ingest_api-2025.7.23.dev20250723.dist-info → nv_ingest_api-2025.7.25.dev20250725.dist-info}/METADATA RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: nv-ingest-api
3
- Version: 2025.7.23.dev20250723
3
+ Version: 2025.7.25.dev20250725
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-2025.7.23.dev20250723.dist-info → nv_ingest_api-2025.7.25.dev20250725.dist-info}/RECORD RENAMED
@@ -31,7 +31,7 @@ nv_ingest_api/internal/extract/pdf/pdf_extractor.py,sha256=CxtWaD6mql9MEqSdk2CfS
31
31
  nv_ingest_api/internal/extract/pdf/engines/__init__.py,sha256=u4GnAZmDKRl0RwYGIRiozIRw70Kybw3A72-lcKFeoTI,582
32
32
  nv_ingest_api/internal/extract/pdf/engines/adobe.py,sha256=VT0dEqkU-y2uGkaCqxtKYov_Q8R1028UQVBchgMLca4,17466
33
33
  nv_ingest_api/internal/extract/pdf/engines/llama.py,sha256=PpKTqS8jGHBV6mKLGZWwjpfT8ga6Fy8ffrvL-gPAf2c,8182
34
- nv_ingest_api/internal/extract/pdf/engines/nemoretriever.py,sha256=VrIfhDXzYVOwvuoQ9dkiYX-y0CHYhy_Ludadq0YpHqY,23132
34
+ nv_ingest_api/internal/extract/pdf/engines/nemoretriever.py,sha256=IVbNcH_phMiRSxnkZ04pGfQrPJ-x1zVR3hXyhxv7juc,22977
35
35
  nv_ingest_api/internal/extract/pdf/engines/pdfium.py,sha256=SKmias2iZmAE6Q8WXxmFEjvLOZy-vXRoaRIPpi7Tuhs,22962
36
36
  nv_ingest_api/internal/extract/pdf/engines/tika.py,sha256=6GyR2l6EsgNZl9jnYDXLeKNK9Fj2Mw9y2UWDq-eSkOc,3169
37
37
  nv_ingest_api/internal/extract/pdf/engines/unstructured_io.py,sha256=jrv2B4VZAH4PevAQrFz965qz8UyXq3rViiOTbGLejec,14908
@@ -90,7 +90,7 @@ nv_ingest_api/internal/schemas/store/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQu
90
90
  nv_ingest_api/internal/schemas/store/store_embedding_schema.py,sha256=tdKeiraim9CDL9htgp4oUSCoPMoO5PrHBnlXqDyCpMw,956
91
91
  nv_ingest_api/internal/schemas/store/store_image_schema.py,sha256=p2LGij9i6sG6RYmsfdiQOiWIc2j-POjxYrNuMrp3ELU,1010
92
92
  nv_ingest_api/internal/schemas/transform/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
93
- nv_ingest_api/internal/schemas/transform/transform_image_caption_schema.py,sha256=OtM1iPw26uioC3mghbOJQurKGg641uQfhASH462VqOY,578
93
+ nv_ingest_api/internal/schemas/transform/transform_image_caption_schema.py,sha256=gRJCfEGrJXErMF_GsZDjyDR8HOrLrUNxz2BVS6LSPY8,542
94
94
  nv_ingest_api/internal/schemas/transform/transform_image_filter_schema.py,sha256=31ThI5fr0yyENeJeE1xMAA-pxk1QVJLwM842zMate_k,429
95
95
  nv_ingest_api/internal/schemas/transform/transform_text_embedding_schema.py,sha256=RZCISA8CUqKiY8eJuk4uWxzo4PZ-fuYdzMO7_LYFkoM,1117
96
96
  nv_ingest_api/internal/schemas/transform/transform_text_splitter_schema.py,sha256=D9K8tvu-tkEBQkZo7uuRzgrHdGyM3ZcNycHbHy5HV2E,791
@@ -140,7 +140,7 @@ nv_ingest_api/util/multi_processing/__init__.py,sha256=4fojP8Rp_5Hu1YAkqGylqTyEZ
140
140
  nv_ingest_api/util/multi_processing/mp_pool_singleton.py,sha256=dTfP82DgGPaXEJH3jywTO8rNlLZUniD4FFzwv84_giE,7372
141
141
  nv_ingest_api/util/nim/__init__.py,sha256=UqbiXFCqjWcjNvoduXd_0gOUOGBT8JvppiYHOmMyneA,1775
142
142
  nv_ingest_api/util/pdf/__init__.py,sha256=uLsBITo_XfgbwpzqXUm1IYX6XlZrTfx6T1cIhdILwG8,140
143
- nv_ingest_api/util/pdf/pdfium.py,sha256=qTiTlSaiCk_rxm_eoQBoAFKq_5OQrioHVSbPbGDxVkE,15668
143
+ nv_ingest_api/util/pdf/pdfium.py,sha256=1aPCnPKXHWnncYoMO8HllYjrhODSXIeRBIsSLDevpYs,15667
144
144
  nv_ingest_api/util/schema/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
145
145
  nv_ingest_api/util/schema/schema_validator.py,sha256=H0yZ_i_HZaiBRUCGmTBfRB9-hURhVqyd10aS_ynM1_0,321
146
146
  nv_ingest_api/util/service_clients/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
@@ -153,8 +153,8 @@ nv_ingest_api/util/service_clients/rest/rest_client.py,sha256=dZ-jrk7IK7oNtHoXFS
153
153
  nv_ingest_api/util/string_processing/__init__.py,sha256=mkwHthyS-IILcLcL1tJYeF6mpqX3pxEw5aUzDGjTSeU,1411
154
154
  nv_ingest_api/util/system/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
155
155
  nv_ingest_api/util/system/hardware_info.py,sha256=ORZeKpH9kSGU_vuPhyBwkIiMyCViKUX2CP__MCjrfbU,19463
156
- nv_ingest_api-2025.7.23.dev20250723.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
157
- nv_ingest_api-2025.7.23.dev20250723.dist-info/METADATA,sha256=f0DDwSsXN9EyR1RZ9-8NUUG_MgGRtu5B2suV_l9xW7Q,13947
158
- nv_ingest_api-2025.7.23.dev20250723.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
159
- nv_ingest_api-2025.7.23.dev20250723.dist-info/top_level.txt,sha256=abjYMlTJGoG5tOdfIB-IWvLyKclw6HLaRSc8MxX4X6I,14
160
- nv_ingest_api-2025.7.23.dev20250723.dist-info/RECORD,,
156
+ nv_ingest_api-2025.7.25.dev20250725.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
157
+ nv_ingest_api-2025.7.25.dev20250725.dist-info/METADATA,sha256=C6sRMPmRDImC5iZaOQr3tyvxvJ-DZSC5NPMeEzk4hM0,13947
158
+ nv_ingest_api-2025.7.25.dev20250725.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
159
+ nv_ingest_api-2025.7.25.dev20250725.dist-info/top_level.txt,sha256=abjYMlTJGoG5tOdfIB-IWvLyKclw6HLaRSc8MxX4X6I,14
160
+ nv_ingest_api-2025.7.25.dev20250725.dist-info/RECORD,,