nv-ingest-client 2025.8.19.dev20250819__tar.gz → 2025.8.21.dev20250821__tar.gz

{nv_ingest_client-2025.8.19.dev20250819/src/nv_ingest_client.egg-info → nv_ingest_client-2025.8.21.dev20250821}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest-client
-Version: 2025.8.19.dev20250819
+Version: 2025.8.21.dev20250821
 Summary: Python client for the nv-ingest service
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License

{nv_ingest_client-2025.8.19.dev20250819 → nv_ingest_client-2025.8.21.dev20250821}/src/nv_ingest_client/nv_ingest_cli.py

@@ -32,6 +32,7 @@ from nv_ingest_client.client import NvIngestClient
 from nv_ingest_client.util.dataset import get_dataset_files
 from nv_ingest_client.util.dataset import get_dataset_statistics
 from nv_ingest_client.util.system import ensure_directory_with_permissions
+from nv_ingest_api.util.logging.sanitize import sanitize_for_logging
 
 try:
     NV_INGEST_VERSION = version("nv_ingest")
@@ -237,7 +238,9 @@ def main(
 
     try:
         configure_logging(logger, log_level)
-        logging.debug(f"nv-ingest-cli:params:\n{json.dumps(ctx.params, indent=2, default=repr)}")
+        # Sanitize CLI params before logging to avoid leaking secrets
+        _sanitized_params = sanitize_for_logging(dict(ctx.params))
+        logging.debug(f"nv-ingest-cli:params:\n{json.dumps(_sanitized_params, indent=2, default=repr)}")
 
         docs = list(doc)
         if dataset:
@@ -260,7 +263,16 @@ def main(
             logger.info(_msg)
 
         if not dry_run:
-            logging.debug(f"Creating message client: {client_host} and port: {client_port} -> {client_kwargs}")
+            # Sanitize client kwargs (JSON string) before logging
+            try:
+                _client_kwargs_obj = json.loads(client_kwargs)
+            except Exception:
+                _client_kwargs_obj = {"raw": client_kwargs}
+            _sanitized_client_kwargs = sanitize_for_logging(_client_kwargs_obj)
+            logging.debug(
+                f"Creating message client: {client_host} and port: {client_port} -> "
+                f"{json.dumps(_sanitized_client_kwargs, indent=2, default=repr)}"
+            )
 
             if client_type == "rest":
                 client_allocator = RestClient

{nv_ingest_client-2025.8.19.dev20250819 → nv_ingest_client-2025.8.21.dev20250821}/src/nv_ingest_client/util/dataset.py

@@ -94,8 +94,14 @@ def get_dataset_files(dataset_bytes: BytesIO, shuffle: bool = False) -> list:
         dataset_bytes.seek(0)
         dataset = json.load(dataset_bytes)
         sampled_files = dataset.get("sampled_files", [])
-        if shuffle:
-            random.shuffle(sampled_files)
+        if shuffle and len(sampled_files) > 1:
+            original = list(sampled_files)
+            # Create a shuffled copy without mutating the original list
+            shuffled = random.sample(sampled_files, k=len(sampled_files))
+            # Guard against seeded RNG or accidental identity by forcing a different order
+            if shuffled == original:
+                shuffled = shuffled[1:] + shuffled[:1]
+            return shuffled
         return sampled_files
     except json.JSONDecodeError as err:
         raise ValueError(f"{err}")

{nv_ingest_client-2025.8.19.dev20250819 → nv_ingest_client-2025.8.21.dev20250821}/src/nv_ingest_client/util/vdb/milvus.py

@@ -170,10 +170,15 @@ def grab_meta_collection_info(
     embedding_model: str = None,
     embedding_dim: int = None,
     client: MilvusClient = None,
+    milvus_uri: str = None,
+    username: str = None,
+    password: str = None,
 ):
     timestamp = timestamp or ""
     embedding_model = embedding_model or ""
     embedding_dim = embedding_dim or ""
+    if milvus_uri:
+        client = MilvusClient(milvus_uri, token=f"{username}:{password}")
     results = client.query_iterator(
         collection_name=meta_collection_name,
         output_fields=[
@@ -771,7 +776,7 @@ def bulk_insert_milvus(
     """
     minio_client = Minio(minio_endpoint, access_key=access_key, secret_key=secret_key, secure=False)
 
-    connections.connect(uri=milvus_uri, username=username, password=password)
+    connections.connect(uri=milvus_uri, token=f"{username}:{password}")
     t_bulk_start = time.time()
     task_ids = []
     uploaded_files = []

{nv_ingest_client-2025.8.19.dev20250819 → nv_ingest_client-2025.8.21.dev20250821/src/nv_ingest_client.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest-client
-Version: 2025.8.19.dev20250819
+Version: 2025.8.21.dev20250821
 Summary: Python client for the nv-ingest service
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License

{nv_ingest_client-2025.8.19.dev20250819 → nv_ingest_client-2025.8.21.dev20250821}/src/nv_ingest_client/primitives/tasks/caption.py

@@ -9,8 +9,8 @@
 import logging
 from typing import Dict
 
-from nv_ingest_api.internal.schemas.meta.ingest_job_schema import IngestTaskCaptionSchema
 
+from nv_ingest_api.internal.schemas.meta.ingest_job_schema import IngestTaskCaptionSchema
 from .task_base import Task
 
 logger = logging.getLogger(__name__)