nv-ingest-api 25.7.7.dev20250707__py3-none-any.whl → 25.8.0rc2__py3-none-any.whl

nv_ingest_api/util/message_brokers/simple_message_broker/simple_client.py

@@ -14,7 +14,7 @@ import logging
 from typing import Optional, Tuple, Union
 
 from nv_ingest_api.internal.schemas.message_brokers.response_schema import ResponseSchema
-from nv_ingest_api.util.service_clients.client_base import MessageBrokerClientBase, FetchMode
+from nv_ingest_api.util.service_clients.client_base import MessageBrokerClientBase
 
 logger = logging.getLogger(__name__)
 
@@ -108,29 +108,23 @@ class SimpleClient(MessageBrokerClientBase):
         return self._handle_push(queue_name, message, timeout, for_nv_ingest)
 
     def fetch_message(
-        self,
-        queue_name: str,
-        timeout: Optional[Tuple[int, Union[float]]] = (100, None),
-        override_fetch_mode: FetchMode = None,
+        self, queue_name: str, timeout: Optional[Tuple[int, Union[float, None]]] = (1200, None)
     ) -> ResponseSchema:
         """
-        Fetch a message from the specified queue.
+        Fetch a message from a specified queue.
 
         Parameters
         ----------
         queue_name : str
             The name of the queue.
-        timeout : float, optional
-            Timeout in seconds for the operation.
+        timeout : tuple, optional
+            A tuple containing the timeout value and an unused second element.
 
         Returns
         -------
         ResponseSchema
-            The response containing the fetched message.
+            The response from the broker.
         """
-        if isinstance(timeout, int):
-            timeout = (timeout, None)
-
         return self._handle_pop(queue_name, timeout)
 
     def ping(self) -> ResponseSchema:
@@ -208,6 +202,7 @@ class SimpleClient(MessageBrokerClientBase):
 
             try:
                 with socket.create_connection((self._host, self._port), timeout=self._connection_timeout) as sock:
+                    sock.settimeout(self._connection_timeout)
                     self._send(sock, json.dumps(command).encode("utf-8"))
                     # Receive initial response with transaction ID
                     response_data = self._recv(sock)
@@ -241,8 +236,9 @@ class SimpleClient(MessageBrokerClientBase):
 
                     return ResponseSchema(**final_response)
 
-            except (ConnectionError, socket.error, BrokenPipeError):
-                pass
+            except (ConnectionError, socket.error, BrokenPipeError, socket.timeout) as e:
+                logger.debug(f"Connection error during PUSH: {e}")
+                pass  # Will be retried
             except json.JSONDecodeError:
                 return ResponseSchema(response_code=1, response_reason="Invalid JSON response from server.")
             except Exception as e:
@@ -272,61 +268,67 @@ class SimpleClient(MessageBrokerClientBase):
 
         command = {"command": "POP", "queue_name": queue_name}
 
-        timeout = int(timeout[0])
+        timeout_val = timeout[0] if isinstance(timeout, tuple) else timeout
 
-        if timeout is not None:
-            command["timeout"] = timeout
+        if timeout_val is not None:
+            command["timeout"] = timeout_val
 
         start_time = time.time()
+        backoff_delay = 1  # Start with a 1-second backoff
+
         while True:
             elapsed = time.time() - start_time
-            remaining_timeout = timeout - elapsed if timeout else None
-            if remaining_timeout is not None and remaining_timeout <= 0:
-                return ResponseSchema(response_code=1, response_reason="POP operation timed out.")
+            if timeout_val is not None and elapsed >= timeout_val:
+                return ResponseSchema(response_code=2, response_reason="Job not ready.")
 
             try:
                 with socket.create_connection((self._host, self._port), timeout=self._connection_timeout) as sock:
+                    sock.settimeout(self._connection_timeout)
                     self._send(sock, json.dumps(command).encode("utf-8"))
                     # Receive initial response with transaction ID and message
                     response_data = self._recv(sock)
                     response = json.loads(response_data)
 
-                    if response.get("response_code") != 0:
-                        if response.get("response_reason") == "Queue is empty":
-                            time.sleep(0.1)
-                            continue
-                        else:
-                            return ResponseSchema(**response)
-
-                    if "transaction_id" not in response:
-                        error_msg = "No transaction_id in response."
+                    # The broker now returns a response_code of 2 for a timeout, which the high-level
+                    # client should handle as a retryable event.
+                    if response.get("response_code") == 2:
+                        # Queue is empty or job not ready, continue to backoff and retry
+                        pass
+                    elif response.get("response_code") != 0:
+                        return ResponseSchema(**response)
+                    else:
+                        # Success case: we received a message.
+                        if "transaction_id" not in response:
+                            return ResponseSchema(response_code=1, response_reason="No transaction_id in response.")
 
-                        return ResponseSchema(response_code=1, response_reason=error_msg)
+                        transaction_id = response["transaction_id"]
+                        message = response.get("response")
 
-                    transaction_id = response["transaction_id"]
-                    message = response.get("response")
+                        # Send ACK
+                        ack_data = json.dumps({"transaction_id": transaction_id, "ack": True}).encode("utf-8")
+                        self._send(sock, ack_data)
 
-                    # Send ACK
-                    ack_data = json.dumps({"transaction_id": transaction_id, "ack": True}).encode("utf-8")
-                    self._send(sock, ack_data)
+                        # Receive final response
+                        final_response_data = self._recv(sock)
+                        final_response = json.loads(final_response_data)
 
-                    # Receive final response
-                    final_response_data = self._recv(sock)
-                    final_response = json.loads(final_response_data)
-
-                    if final_response.get("response_code") == 0:
-                        return ResponseSchema(response_code=0, response=message, transaction_id=transaction_id)
-                    else:
-                        return ResponseSchema(**final_response)
+                        if final_response.get("response_code") == 0:
+                            return ResponseSchema(response_code=0, response=message, transaction_id=transaction_id)
+                        else:
+                            return ResponseSchema(**final_response)
 
-            except (ConnectionError, socket.error, BrokenPipeError):
-                pass
+            except (ConnectionError, socket.error, BrokenPipeError, socket.timeout) as e:
+                # Let the high-level client handle connection errors as retryable.
+                logger.debug(f"Connection error during POP: {e}, will retry after backoff.")
+                pass  # Fall through to backoff and retry
             except json.JSONDecodeError:
                 return ResponseSchema(response_code=1, response_reason="Invalid JSON response from server.")
             except Exception as e:
                 return ResponseSchema(response_code=1, response_reason=str(e))
 
-            time.sleep(0.1)  # Backoff delay before retry
+            # Exponential backoff
+            time.sleep(backoff_delay)
+            backoff_delay = min(backoff_delay * 2, self._max_backoff)
 
     def _execute_simple_command(self, command: dict) -> ResponseSchema:
         """
@@ -350,12 +352,13 @@ class SimpleClient(MessageBrokerClientBase):
 
         try:
             with socket.create_connection((self._host, self._port), timeout=self._connection_timeout) as sock:
+                sock.settimeout(self._connection_timeout)
                 self._send(sock, data)
                 response_data = self._recv(sock)
                 response = json.loads(response_data)
                 return ResponseSchema(**response)
-        except (ConnectionError, socket.error, BrokenPipeError) as e:
-            return ResponseSchema(response_code=1, response_reason=f"Connection error: {e}")
+        except (ConnectionError, socket.error, BrokenPipeError, socket.timeout) as e:
+            return ResponseSchema(response_code=2, response_reason=f"Connection error: {e}")
         except json.JSONDecodeError:
             return ResponseSchema(response_code=1, response_reason="Invalid JSON response from server.")
         except Exception as e:

nv_ingest_api/util/metadata/aggregators.py

@@ -201,6 +201,8 @@ def construct_image_metadata_from_base64(
     page_count: int,
     source_metadata: Dict[str, Any],
     base_unified_metadata: Dict[str, Any],
+    subtype: None | ContentTypeEnum | str = "",
+    text: str = "",
 ) -> List[Any]:
     """
     Extracts image data from a base64-encoded image string, decodes the image to get
@@ -252,6 +254,7 @@ def construct_image_metadata_from_base64(
             "line": -1,
             "span": -1,
         },
+        "subtype": subtype or "",
     }
 
     # Construct image metadata
@@ -259,7 +262,7 @@ def construct_image_metadata_from_base64(
         "image_type": DocumentTypeEnum.PNG,
         "structured_image_type": ContentTypeEnum.UNKNOWN,
         "caption": "",
-        "text": "",
+        "text": text,
         "image_location": bbox,
         "image_location_max_dimensions": (width, height),
         "height": height,

nv_ingest_api/util/pdf/pdfium.py

@@ -7,7 +7,6 @@ from typing import List, Any
 from typing import Optional
 from typing import Tuple
 
-import PIL
 import numpy as np
 import pypdfium2 as pdfium
 import pypdfium2.raw as pdfium_c
@@ -20,8 +19,9 @@ from nv_ingest_api.util.image_processing.clustering import (
     combine_groups_into_bboxes,
     remove_superset_bboxes,
 )
-from nv_ingest_api.util.image_processing.transforms import pad_image, numpy_to_base64, crop_image
+from nv_ingest_api.util.image_processing.transforms import pad_image, numpy_to_base64, crop_image, scale_numpy_image
 from nv_ingest_api.util.metadata.aggregators import Base64Image
+from nv_ingest_api.internal.primitives.nim.model_interface.yolox import YOLOX_PAGE_IMAGE_FORMAT
 
 logger = logging.getLogger(__name__)
 
@@ -176,18 +176,10 @@ def pdfium_pages_to_numpy(
     for idx, page in enumerate(pages):
         # Render the page as a bitmap with the specified scale and rotation
         page_bitmap = page.render(scale=scale, rotation=rotation)
-
-        # Convert the bitmap to a PIL image
-        pil_image = page_bitmap.to_pil()
-
+        img_arr = convert_bitmap_to_corrected_numpy(page_bitmap)
         # Apply scaling using the thumbnail approach if specified
         if scale_tuple:
-            pil_image.thumbnail(scale_tuple, PIL.Image.LANCZOS)
-
-        # Convert the PIL image to a NumPy array and force a full copy,
-        # ensuring the returned array is entirely independent of the original buffer.
-        img_arr = np.array(pil_image).copy()
-
+            img_arr = scale_numpy_image(img_arr, scale_tuple)
         # Apply padding if specified
         if padding_tuple:
             img_arr, (pad_width, pad_height) = pad_image(
@@ -250,7 +242,7 @@ def extract_simple_images_from_pdfium_page(page, max_depth):
         try:
             # Attempt to retrieve the image bitmap
             image_numpy: np.ndarray = pdfium_try_get_bitmap_as_numpy(obj)  # noqa
-            image_base64: str = numpy_to_base64(image_numpy)
+            image_base64: str = numpy_to_base64(image_numpy, format=YOLOX_PAGE_IMAGE_FORMAT)
             image_bbox = obj.get_pos()
             image_size = obj.get_size()
             if image_size[0] < 10 and image_size[1] < 10:
@@ -394,7 +386,7 @@ def extract_image_like_objects_from_pdfium_page(page, merge=True, **kwargs):
     try:
         original_images, _ = pdfium_pages_to_numpy(
             [page],  # A batch with a single image.
-            render_dpi=300,  # dpi = 72 is equivalent to scale = 1.
+            render_dpi=72,  # dpi = 72 is equivalent to scale = 1.
             rotation=rotation,  # Without rotation, coordinates from page.get_pos() will not match.
         )
         image_bboxes = extract_merged_images_from_pdfium_page(page, merge=merge, **kwargs)

{nv_ingest_api-25.7.7.dev20250707.dist-info → nv_ingest_api-25.8.0rc2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest-api
-Version: 25.7.7.dev20250707
+Version: 25.8.0rc2
 Summary: Python module with core document ingestion functions.
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License
@@ -217,6 +217,7 @@ Requires-Dist: backoff==2.2.1
 Requires-Dist: pandas>=2.0
 Requires-Dist: pydantic>2.0.0
 Requires-Dist: pydantic-settings>2.0.0
+Requires-Dist: tritonclient
 Dynamic: license-file
 
 # nv-ingest-api

{nv_ingest_api-25.7.7.dev20250707.dist-info → nv_ingest_api-25.8.0rc2.dist-info}/RECORD

@@ -1,13 +1,13 @@
 nv_ingest_api/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/interface/__init__.py,sha256=ltWlfmtCewHSRK4B7DF__QvlSUPuliz58JEcEIeIgI0,10134
-nv_ingest_api/interface/extract.py,sha256=GyBfXKKTGwSb-y0k0nMiTf4HcCT2E-lxLY4aMYAPeOI,38815
+nv_ingest_api/interface/extract.py,sha256=o9OdoWxYsj-O4HsDe6wWbyd69OAueb2rlMtKSzOrKZo,38743
 nv_ingest_api/interface/mutate.py,sha256=eZkd3sbHEJQiEPJyMbhewlPxQNMnL_Xur15icclnb-U,5934
 nv_ingest_api/interface/store.py,sha256=aR3Cf19lq9Yo9AHlAy1VVcrOP2dgyN01yYhwxyTprkQ,8207
 nv_ingest_api/interface/transform.py,sha256=g6YnFR7TpEU0xNtzCvv6kqnFbuCwQ6vRMjjBxz3G4n4,15815
 nv_ingest_api/interface/utility.py,sha256=AL4l0cJNvTjG1MAe1YNTk1jbbPED3g4HCewzx6Ffcio,7296
 nv_ingest_api/internal/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nv_ingest_api/internal/enums/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest_api/internal/enums/common.py,sha256=HSj7qqNr6KXu_FIyK_Wvel24R-r8lV7dLA173z5XFBc,12321
+nv_ingest_api/internal/enums/common.py,sha256=lzDJ35VWfIwlL_Lx_q0dfHUuwEB7CXudHIQAilpjoRw,12611
 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=_jf_UC_FTqZr-xEpwG8edwBzdDjM01gGhqm9ulOsDcY,6973
@@ -20,22 +20,22 @@ nv_ingest_api/internal/extract/docx/engines/docxreader_helpers/docxreader.py,sha
 nv_ingest_api/internal/extract/html/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/extract/html/html_extractor.py,sha256=I9oWfj6_As4898GDDh0zsSuKxO3lBsvyYzhvUotjzJI,3282
 nv_ingest_api/internal/extract/image/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest_api/internal/extract/image/chart_extractor.py,sha256=CkaW8ihPmGMQGrZh0ih14gtEpWuGOJ8InPQfZwpsP2g,13300
+nv_ingest_api/internal/extract/image/chart_extractor.py,sha256=gk-O-9wjZBoaLVE_6Erb4gMwsSFk4UtPQ2QLpMCW4H4,13212
 nv_ingest_api/internal/extract/image/image_extractor.py,sha256=4tUWinuFMN3ukWa2tZa2_LtzRiTyUAUCBF6BDkUEvm0,8705
-nv_ingest_api/internal/extract/image/infographic_extractor.py,sha256=yc9b2q_Ea08CEVclZ47UkpU4F7qlakPuU3UV9P013W0,8903
-nv_ingest_api/internal/extract/image/table_extractor.py,sha256=ivHaJxYjeHvFM1PZIpxVabPadxtcTsu51j398ZjMhD4,13123
+nv_ingest_api/internal/extract/image/infographic_extractor.py,sha256=i7zt_ow1gytU4hK2JCRg7T1wlbokaeuUpXX69LIQkzY,9687
+nv_ingest_api/internal/extract/image/table_extractor.py,sha256=O0m3N2Tz9W6X7TBI4o-rbBXc8dFOf9zSZq1v9qC1U4M,13780
 nv_ingest_api/internal/extract/image/image_helpers/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest_api/internal/extract/image/image_helpers/common.py,sha256=P8rcl4YPyeWeMJg7u1yejD3k9EnDVEbJgfYEnJ4WO5c,15025
+nv_ingest_api/internal/extract/image/image_helpers/common.py,sha256=80jRhGzisHvQ9Ky3MKUMM7soKUmvZ5LqRVzwNYjgdPY,14988
 nv_ingest_api/internal/extract/pdf/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/extract/pdf/pdf_extractor.py,sha256=CxtWaD6mql9MEqSdk2CfSQ9T-Bn87beBkCOuGGjxGt8,2934
 nv_ingest_api/internal/extract/pdf/engines/__init__.py,sha256=u4GnAZmDKRl0RwYGIRiozIRw70Kybw3A72-lcKFeoTI,582
 nv_ingest_api/internal/extract/pdf/engines/adobe.py,sha256=VT0dEqkU-y2uGkaCqxtKYov_Q8R1028UQVBchgMLca4,17466
 nv_ingest_api/internal/extract/pdf/engines/llama.py,sha256=PpKTqS8jGHBV6mKLGZWwjpfT8ga6Fy8ffrvL-gPAf2c,8182
-nv_ingest_api/internal/extract/pdf/engines/nemoretriever.py,sha256=Uqj1NH7yWga9P6_vCzgny1WKALfF--UdAaGHUF8K_aQ,22926
-nv_ingest_api/internal/extract/pdf/engines/pdfium.py,sha256=fDbrZwJ-lgeHYOq107WXehzdSvyF8zEDza_9UkDm5aE,22360
+nv_ingest_api/internal/extract/pdf/engines/nemoretriever.py,sha256=IVbNcH_phMiRSxnkZ04pGfQrPJ-x1zVR3hXyhxv7juc,22977
+nv_ingest_api/internal/extract/pdf/engines/pdfium.py,sha256=SKmias2iZmAE6Q8WXxmFEjvLOZy-vXRoaRIPpi7Tuhs,22962
 nv_ingest_api/internal/extract/pdf/engines/tika.py,sha256=6GyR2l6EsgNZl9jnYDXLeKNK9Fj2Mw9y2UWDq-eSkOc,3169
 nv_ingest_api/internal/extract/pdf/engines/unstructured_io.py,sha256=jrv2B4VZAH4PevAQrFz965qz8UyXq3rViiOTbGLejec,14908
-nv_ingest_api/internal/extract/pdf/engines/pdf_helpers/__init__.py,sha256=Jk3wrQ2CZs167juvEZ-uV6qXWQjR08hhIu8otk2MWj4,4931
+nv_ingest_api/internal/extract/pdf/engines/pdf_helpers/__init__.py,sha256=4bvN6LsPksLicI6jM0JqbJFiOZNHEcuc8MVVW4XfgV8,5875
 nv_ingest_api/internal/extract/pptx/__init__.py,sha256=HIHfzSig66GT0Uk8qsGBm_f13fKYcPtItBicRUWOOVA,183
 nv_ingest_api/internal/extract/pptx/pptx_extractor.py,sha256=o-0P2dDyRFW37uQi_lKk6-eFozTcZvbq-2Y4I0EBMIY,7749
 nv_ingest_api/internal/extract/pptx/engines/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -48,19 +48,19 @@ nv_ingest_api/internal/primitives/control_message_task.py,sha256=nWVB3QsP6p8BKwH
 nv_ingest_api/internal/primitives/ingest_control_message.py,sha256=rvipBiiUaHuRhupFCFDCG8rv0PylSJibCiJ7rDeb98A,8514
 nv_ingest_api/internal/primitives/nim/__init__.py,sha256=i_i_fBR2EcRCh2Y19DF6GM3s_Q0VPgo_thPnhEIJUyg,266
 nv_ingest_api/internal/primitives/nim/default_values.py,sha256=W92XjfyeC6uuVxut6J7p00x1kpNsnXIDb97gSVytZJk,380
-nv_ingest_api/internal/primitives/nim/nim_client.py,sha256=pvNxwMgNS7lFjXvLgRVUrL6FdXDGkfGnOGQh9WrpYzs,14858
+nv_ingest_api/internal/primitives/nim/nim_client.py,sha256=m8_whtmfXFGSw4-Efu938NKG97nzhygyuHB8Tq-a0Ec,16570
 nv_ingest_api/internal/primitives/nim/nim_model_interface.py,sha256=wMEgoi79YQn_4338MVemkeZgM1J-vnz0aZWpvqDhib4,2392
 nv_ingest_api/internal/primitives/nim/model_interface/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/primitives/nim/model_interface/cached.py,sha256=b1HX-PY1ExW5V6pXC1ZiHdobeG_BmbPr3rBbVJef13s,11003
 nv_ingest_api/internal/primitives/nim/model_interface/decorators.py,sha256=qwubkHs4WjnexM6rI0wkjWCsrVNEbA4Wjk2oKL9OYCU,1499
 nv_ingest_api/internal/primitives/nim/model_interface/deplot.py,sha256=TvKdk6PTuI1WNhRmNNrvygaI_DIutkJkDL-XdtLZQac,10787
-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/helpers.py,sha256=iyGxAr4tG2UZ7LtXXoWO_kF-KsObhPrmZ46Nl0Mi-Ag,11592
+nv_ingest_api/internal/primitives/nim/model_interface/nemoretriever_parse.py,sha256=WysjDZeegclO3mZgVcGOwzWbr8wSI4pWRiYD4iC2EXo,7098
+nv_ingest_api/internal/primitives/nim/model_interface/ocr.py,sha256=Vhim3py_rc5jA0BoKubwfekEqOwxUUePzcmc59pRuOk,21458
 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
+nv_ingest_api/internal/primitives/nim/model_interface/yolox.py,sha256=zpfEZIPctWhNfREnP6e77zffU8vs_RfnMprBj-2jXXk,42847
 nv_ingest_api/internal/primitives/tracing/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nv_ingest_api/internal/primitives/tracing/latency.py,sha256=5kVTeYRbRdTlT_aI4MeS20N_S7mqCcLqZR6YHtxhXkY,2215
 nv_ingest_api/internal/primitives/tracing/logging.py,sha256=SSzIgS7afLH-e1C7VagYDmkkA6rTXmQ-bmtLjoEguhg,3851
@@ -68,21 +68,21 @@ nv_ingest_api/internal/primitives/tracing/tagging.py,sha256=O5dD7Z7j43nrjqn0Axhx
 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=W-nEBriqiNkjpaQ5AT_8LhtVXlW8AhlcftmoeQQtKAs,3812
-nv_ingest_api/internal/schemas/extract/extract_chart_schema.py,sha256=iu8lHQC0zbBB9VRK7PZisAVzpeSpFqjcXRAnwZ9OzoM,4301
+nv_ingest_api/internal/schemas/extract/extract_chart_schema.py,sha256=wDcvQ5XtOjIBGSWtNjQPiPtVKrSOYqbf2mnLrhfhue4,4283
 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
 nv_ingest_api/internal/schemas/extract/extract_image_schema.py,sha256=GC4xV8Z9TPLOuxlEtf2fbklSSp8ETGMrDpZgMQ02UwA,3766
-nv_ingest_api/internal/schemas/extract/extract_infographic_schema.py,sha256=rl_hFDoJaJLTKbtnEpDSBj-73KQL9aUEVKGiW0IdXiU,3991
+nv_ingest_api/internal/schemas/extract/extract_infographic_schema.py,sha256=z42cs7w-U-IUCMGByp5e_iBUZ7KCl5vTIXkP64ty6gY,3973
 nv_ingest_api/internal/schemas/extract/extract_pdf_schema.py,sha256=G9g1lEORmryUWTzDyZ0vHAuPnVMK7VaRx0E4xzmAw3Q,6589
 nv_ingest_api/internal/schemas/extract/extract_pptx_schema.py,sha256=5dT0kv-Mmpe5KW-BZc1JOW3rUlgzVZI0rpB79NWytmw,3761
-nv_ingest_api/internal/schemas/extract/extract_table_schema.py,sha256=sbt3TvQrLsXc8-muKnsyOs4MfpA4VzrprYHdu1IrY8M,3950
+nv_ingest_api/internal/schemas/extract/extract_table_schema.py,sha256=vd_1mf_LmQGvSTpQCuWr6ubsiav4TMhp_SpKGO-6RLc,3935
 nv_ingest_api/internal/schemas/message_brokers/__init__.py,sha256=uLsBITo_XfgbwpzqXUm1IYX6XlZrTfx6T1cIhdILwG8,140
 nv_ingest_api/internal/schemas/message_brokers/message_broker_client_schema.py,sha256=4xTSFE_vH7yZE9RRJRflFAG9hNXIaF6K020M_xA7ylw,1351
 nv_ingest_api/internal/schemas/message_brokers/request_schema.py,sha256=LZX_wXDxTamVFqTQs2Yd8uvWyPE5mddHAWSU4PtfEIQ,966
 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=szDvgc2A_JetD2Jyewyl4ac4lwpy3NiLxD9dOYz42sM,8116
+nv_ingest_api/internal/schemas/meta/ingest_job_schema.py,sha256=ceYQjRjhBSDbbZ6q-Db7Y6GHVOvWPdGAMb3TX1vMWfY,8321
 nv_ingest_api/internal/schemas/meta/metadata_schema.py,sha256=VnAzkSFat_ckI19mlwQTlFrvP6EZVCwyNl9bt51b8oU,7193
 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
@@ -90,17 +90,17 @@ nv_ingest_api/internal/schemas/store/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQu
 nv_ingest_api/internal/schemas/store/store_embedding_schema.py,sha256=tdKeiraim9CDL9htgp4oUSCoPMoO5PrHBnlXqDyCpMw,956
 nv_ingest_api/internal/schemas/store/store_image_schema.py,sha256=p2LGij9i6sG6RYmsfdiQOiWIc2j-POjxYrNuMrp3ELU,1010
 nv_ingest_api/internal/schemas/transform/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest_api/internal/schemas/transform/transform_image_caption_schema.py,sha256=OtM1iPw26uioC3mghbOJQurKGg641uQfhASH462VqOY,578
+nv_ingest_api/internal/schemas/transform/transform_image_caption_schema.py,sha256=gRJCfEGrJXErMF_GsZDjyDR8HOrLrUNxz2BVS6LSPY8,542
 nv_ingest_api/internal/schemas/transform/transform_image_filter_schema.py,sha256=31ThI5fr0yyENeJeE1xMAA-pxk1QVJLwM842zMate_k,429
-nv_ingest_api/internal/schemas/transform/transform_text_embedding_schema.py,sha256=ongmHkJA2953f9_RI7ZYzf5BUnFzVL6Al5E8WKyfgw4,885
+nv_ingest_api/internal/schemas/transform/transform_text_embedding_schema.py,sha256=RZCISA8CUqKiY8eJuk4uWxzo4PZ-fuYdzMO7_LYFkoM,1117
 nv_ingest_api/internal/schemas/transform/transform_text_splitter_schema.py,sha256=D9K8tvu-tkEBQkZo7uuRzgrHdGyM3ZcNycHbHy5HV2E,791
 nv_ingest_api/internal/store/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/store/embed_text_upload.py,sha256=maxb4FPsBvWgvlrjAPEBlRZEFdJX5NxPG-p8kUbzV7I,9898
 nv_ingest_api/internal/store/image_upload.py,sha256=GNlY4k3pfcHv3lzXxkbmGLeHFsf9PI25bkBn6Xn9h3I,9654
 nv_ingest_api/internal/transform/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/transform/caption_image.py,sha256=0ILCG2F8ESqKtZiPUM-6F1BHUflFZ76Dzi2GNzkE-lU,8517
-nv_ingest_api/internal/transform/embed_text.py,sha256=A8JMotTkC8KQ0pmz4AIJhaKebza6JzhQ0aEnHX2oHY8,16539
-nv_ingest_api/internal/transform/split_text.py,sha256=-kwpRWSVZrPldm1hn3-tVz_TkzuKM-kPvNU3HTp9zOY,7476
+nv_ingest_api/internal/transform/embed_text.py,sha256=pVM9gULOrVf3IJrpFDZH0X7lWQbr4BTbrLG3LXp_G7s,20095
+nv_ingest_api/internal/transform/split_text.py,sha256=LAtInGVuydH43UwjNMQWFVC1A6NdhXP_dZup2xX4qEo,7745
 nv_ingest_api/util/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/util/control_message/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nv_ingest_api/util/control_message/validators.py,sha256=KvvbyheJ5rbzvJbH9JKpMR9VfoI0b0uM6eTAZte1p44,1315
@@ -122,8 +122,8 @@ nv_ingest_api/util/exception_handlers/schemas.py,sha256=NJngVNf9sk5Uz6CFFfkNO_LB
 nv_ingest_api/util/image_processing/__init__.py,sha256=Jiy8C1ZuSrNb_eBM1ZTV9IKFIsnjhZi6Ku3JJhVLimA,104
 nv_ingest_api/util/image_processing/clustering.py,sha256=sUGlZI4cx1q8h4Pns1N9JVpdfSM2BOH8zRmn9QFCtzI,9236
 nv_ingest_api/util/image_processing/processing.py,sha256=LSoDDEmahr7a-qSS12McVcowRe3dOrAZwa1h-PD_JPQ,6554
-nv_ingest_api/util/image_processing/table_and_chart.py,sha256=bxOu9PZYkG_WFCDGw_JLaO60S2pDSN8EOWK3xkIwr2A,14376
-nv_ingest_api/util/image_processing/transforms.py,sha256=Kz9hrizV314Hy7cRCYK9ZmhmBbVUOZ_z0HEpzZYcslQ,14081
+nv_ingest_api/util/image_processing/table_and_chart.py,sha256=idCIjiLkY-usI2EARchg3omWLtIYmYA-1tdUUV2lbno,16338
+nv_ingest_api/util/image_processing/transforms.py,sha256=FBcORrvjimn3c1naaVxRMm6PMJ2Dt6Uy9AZRUxdbkR0,23829
 nv_ingest_api/util/imports/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/util/imports/callable_signatures.py,sha256=e2bJB1pmkN4Ee-Bf-VggOSBaQ4RXofWF5eKkWXgIj2U,1855
 nv_ingest_api/util/imports/dynamic_resolvers.py,sha256=7GByV_-8z2X0tnVoabCxVioxOP3sYMros3ZllVAW-wY,4343
@@ -131,16 +131,16 @@ nv_ingest_api/util/logging/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJ
 nv_ingest_api/util/logging/configuration.py,sha256=XUo7yON9V8IDPfN3x8RBwpZ3Gv4zrRq8QwsByf4dGNE,981
 nv_ingest_api/util/message_brokers/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/util/message_brokers/simple_message_broker/__init__.py,sha256=WaQ3CWIpIKWEivT5kL-bkmzcSQKLGFNFHdXHUJjqZFs,325
-nv_ingest_api/util/message_brokers/simple_message_broker/broker.py,sha256=h9Q4q_alXGxCLNlJUZPan46q8fJ7B72sQy2eBfHdk6I,17265
+nv_ingest_api/util/message_brokers/simple_message_broker/broker.py,sha256=PekxaxVcAa9k1wgUtozlr04SW3sAeqYJE-wdVBZf9eo,17264
 nv_ingest_api/util/message_brokers/simple_message_broker/ordered_message_queue.py,sha256=3p-LRqG8qLnsfEhBNf73_DG22C08JKahTqUvPLS2Apg,2554
-nv_ingest_api/util/message_brokers/simple_message_broker/simple_client.py,sha256=fh7Q0wO5H_FtrHV1VdT6V66aZNqglOh_2XdkfLt8hgg,15722
+nv_ingest_api/util/message_brokers/simple_message_broker/simple_client.py,sha256=CCRAbq2EBH2quX9UTfuBbz3tTMDnWqhEF33roFwbyuk,16484
 nv_ingest_api/util/metadata/__init__.py,sha256=HIHfzSig66GT0Uk8qsGBm_f13fKYcPtItBicRUWOOVA,183
-nv_ingest_api/util/metadata/aggregators.py,sha256=Y5JSKuLhhk_ldpzT3eRIcVg7QM7cTNhfQZn4g5bcbq4,15884
+nv_ingest_api/util/metadata/aggregators.py,sha256=YYdvJ1E04eGFZKKHUxXoH6mzLg8nor9Smvnv0qzqK5w,15988
 nv_ingest_api/util/multi_processing/__init__.py,sha256=4fojP8Rp_5Hu1YAkqGylqTyEZ-HBVVEunn5Z9I99swA,242
 nv_ingest_api/util/multi_processing/mp_pool_singleton.py,sha256=dTfP82DgGPaXEJH3jywTO8rNlLZUniD4FFzwv84_giE,7372
 nv_ingest_api/util/nim/__init__.py,sha256=UqbiXFCqjWcjNvoduXd_0gOUOGBT8JvppiYHOmMyneA,1775
 nv_ingest_api/util/pdf/__init__.py,sha256=uLsBITo_XfgbwpzqXUm1IYX6XlZrTfx6T1cIhdILwG8,140
-nv_ingest_api/util/pdf/pdfium.py,sha256=Ch9Gh5jRLcBr3stjCckqWwTUL-T0sI50PlQnZHo_9NA,15761
+nv_ingest_api/util/pdf/pdfium.py,sha256=1aPCnPKXHWnncYoMO8HllYjrhODSXIeRBIsSLDevpYs,15667
 nv_ingest_api/util/schema/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/util/schema/schema_validator.py,sha256=H0yZ_i_HZaiBRUCGmTBfRB9-hURhVqyd10aS_ynM1_0,321
 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
 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-25.7.7.dev20250707.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-nv_ingest_api-25.7.7.dev20250707.dist-info/METADATA,sha256=-ayjdRz79l_WfVI1kTdgFhIu0gHWZ4SLMBjhtuLfnXg,13916
-nv_ingest_api-25.7.7.dev20250707.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nv_ingest_api-25.7.7.dev20250707.dist-info/top_level.txt,sha256=abjYMlTJGoG5tOdfIB-IWvLyKclw6HLaRSc8MxX4X6I,14
-nv_ingest_api-25.7.7.dev20250707.dist-info/RECORD,,
+nv_ingest_api-25.8.0rc2.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+nv_ingest_api-25.8.0rc2.dist-info/METADATA,sha256=xEisyVAPeQGxKOHZ8gu9MuDrJYsGVaNxJXeEtCcEII8,13935
+nv_ingest_api-25.8.0rc2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nv_ingest_api-25.8.0rc2.dist-info/top_level.txt,sha256=abjYMlTJGoG5tOdfIB-IWvLyKclw6HLaRSc8MxX4X6I,14
+nv_ingest_api-25.8.0rc2.dist-info/RECORD,,