nv-ingest-client 2025.8.19.dev20250819__py3-none-any.whl → 2025.8.20.dev20250820__py3-none-any.whl

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/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.dist-info → nv_ingest_client-2025.8.20.dev20250820.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest-client
-Version: 2025.8.19.dev20250819
+Version: 2025.8.20.dev20250820
 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.dist-info → nv_ingest_client-2025.8.20.dev20250820.dist-info}/RECORD

@@ -30,7 +30,7 @@ nv_ingest_client/primitives/tasks/task_factory.py,sha256=x8FXrhlgRYTxM0rLvsUvM8w
 nv_ingest_client/primitives/tasks/udf.py,sha256=5e_WJVgocnK-z0EGCEwPO_zG8WJEhuIsOUTjPmr8REY,12833
 nv_ingest_client/primitives/tasks/vdb_upload.py,sha256=mXOyQJfQfaoN96nntzevd0sKUs60-AHi8lc1jxG3DAw,1765
 nv_ingest_client/util/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-nv_ingest_client/util/dataset.py,sha256=b6if_hM15iUJC4rvSHS0cmGBsSuZ3W-NoKDMTulx4b8,3316
+nv_ingest_client/util/dataset.py,sha256=2yDPs47HNj8AOdOAfJL4XVji0BMRJq_NH8CG4s4xT-Q,3701
 nv_ingest_client/util/milvus.py,sha256=MwBix_UBg54i7xONBIwjcqeKSBkqunxBJBK2f0bPMoo,61
 nv_ingest_client/util/process_json_files.py,sha256=YKR-fGT4kM8zO2p8r5tpo5-vvFywkcLuNieozvPWvo0,3785
 nv_ingest_client/util/processing.py,sha256=bAy8it-OUgGFO3pcy6D3ezpyZ6p2DfmoQUGhx3QmVf8,8989
@@ -42,11 +42,11 @@ nv_ingest_client/util/file_processing/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCe
 nv_ingest_client/util/file_processing/extract.py,sha256=uXEATBYZXjxdymGTNQvvzDD2eHgpuq4PdU6HsMl0Lp0,4662
 nv_ingest_client/util/vdb/__init__.py,sha256=ZmoEzeM9LzwwrVvu_DVUnjRNx-x8ahkNeIrSfSKzbAk,513
 nv_ingest_client/util/vdb/adt_vdb.py,sha256=UubzAMSfyrqqpD-OQErpBs25hC2Mw8zGZ4waenGXPOk,515
-nv_ingest_client/util/vdb/milvus.py,sha256=PC3qXjrdTab2xVS3FZkhj_28T5R9DNaHZ8a7D721Pik,77269
+nv_ingest_client/util/vdb/milvus.py,sha256=5yjn9uZ0fB10RrJml0WdImsfvfcowDtwrPrl_oYnnF0,77436
 nv_ingest_client/util/vdb/opensearch.py,sha256=I4FzF95VWCOkyzhfm-szdfK1Zd9ugUc8AxxpAdEMWGE,7538
-nv_ingest_client-2025.8.19.dev20250819.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-nv_ingest_client-2025.8.19.dev20250819.dist-info/METADATA,sha256=QYd4COuKD4YRil1snCzCeQpUBEuSzpnb5wGVGW__VMk,30737
-nv_ingest_client-2025.8.19.dev20250819.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nv_ingest_client-2025.8.19.dev20250819.dist-info/entry_points.txt,sha256=3uQVZkTZIjO08_bjTV-g0CDF5H1nrP1zWXU9gJOweuI,137
-nv_ingest_client-2025.8.19.dev20250819.dist-info/top_level.txt,sha256=1eMhBFD3SiWmpXnod2LM66C1HrSLSk96ninZi5XX-cE,17
-nv_ingest_client-2025.8.19.dev20250819.dist-info/RECORD,,
+nv_ingest_client-2025.8.20.dev20250820.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+nv_ingest_client-2025.8.20.dev20250820.dist-info/METADATA,sha256=54Czy3ATSEasGQ0SwUpgxKw1wErWaryTKkHw9-LYpcE,30737
+nv_ingest_client-2025.8.20.dev20250820.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nv_ingest_client-2025.8.20.dev20250820.dist-info/entry_points.txt,sha256=3uQVZkTZIjO08_bjTV-g0CDF5H1nrP1zWXU9gJOweuI,137
+nv_ingest_client-2025.8.20.dev20250820.dist-info/top_level.txt,sha256=1eMhBFD3SiWmpXnod2LM66C1HrSLSk96ninZi5XX-cE,17
+nv_ingest_client-2025.8.20.dev20250820.dist-info/RECORD,,