phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-04 02:03:44 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge remote-tracking branch 'upstream/main' into elasticsearch-integration

Browse source
This commit is contained in:
Enrico Zimuel 2025-10-31 18:23:42 +01:00
commit 2407115ee8
No known key found for this signature in database
GPG key ID: 6CB203F6934A69F1
1050 changed files with 65153 additions and 2821 deletions
Download patch file Download diff file Expand all files Collapse all files

56
tests/integration/vector_io/test_openai_vector_stores.py
View file

@ -86,23 +86,37 @@ def skip_if_provider_doesnt_support_openai_vector_stores_search(client_with_mode
@pytest.fixture(scope="session")
def sample_chunks():
from llama_stack.providers.utils.vector_io.vector_utils import generate_chunk_id
chunks_data = [
(
"Python is a high-level programming language that emphasizes code readability and allows programmers to express concepts in fewer lines of code than would be possible in languages such as C++ or Java.",
"doc1",
"programming",
),
(
"Machine learning is a subset of artificial intelligence that enables systems to automatically learn and improve from experience without being explicitly programmed, using statistical techniques to give computer systems the ability to progressively improve performance on a specific task.",
"doc2",
"ai",
),
(
"Data structures are fundamental to computer science because they provide organized ways to store and access data efficiently, enable faster processing of data through optimized algorithms, and form the building blocks for more complex software systems.",
"doc3",
"computer_science",
),
(
"Neural networks are inspired by biological neural networks found in animal brains, using interconnected nodes called artificial neurons to process information through weighted connections that can be trained to recognize patterns and solve complex problems through iterative learning.",
"doc4",
"ai",
),
]
return [
Chunk(
content="Python is a high-level programming language that emphasizes code readability and allows programmers to express concepts in fewer lines of code than would be possible in languages such as C++ or Java.",
metadata={"document_id": "doc1", "topic": "programming"},
),
Chunk(
content="Machine learning is a subset of artificial intelligence that enables systems to automatically learn and improve from experience without being explicitly programmed, using statistical techniques to give computer systems the ability to progressively improve performance on a specific task.",
metadata={"document_id": "doc2", "topic": "ai"},
),
Chunk(
content="Data structures are fundamental to computer science because they provide organized ways to store and access data efficiently, enable faster processing of data through optimized algorithms, and form the building blocks for more complex software systems.",
metadata={"document_id": "doc3", "topic": "computer_science"},
),
Chunk(
content="Neural networks are inspired by biological neural networks found in animal brains, using interconnected nodes called artificial neurons to process information through weighted connections that can be trained to recognize patterns and solve complex problems through iterative learning.",
metadata={"document_id": "doc4", "topic": "ai"},
),
content=content,
chunk_id=generate_chunk_id(doc_id, content),
metadata={"document_id": doc_id, "topic": topic},
)
for content, doc_id, topic in chunks_data
]
@ -371,7 +385,7 @@ def test_openai_vector_store_with_chunks(
# Insert chunks using the native LlamaStack API (since OpenAI API doesn't have direct chunk insertion)
llama_client.vector_io.insert(
vector_db_id=vector_store.id,
vector_store_id=vector_store.id,
chunks=sample_chunks,
)
@ -438,7 +452,7 @@ def test_openai_vector_store_search_relevance(
# Insert chunks using native API
llama_client.vector_io.insert(
vector_db_id=vector_store.id,
vector_store_id=vector_store.id,
chunks=sample_chunks,
)
@ -488,7 +502,7 @@ def test_openai_vector_store_search_with_ranking_options(
# Insert chunks
llama_client.vector_io.insert(
vector_db_id=vector_store.id,
vector_store_id=vector_store.id,
chunks=sample_chunks,
)
@ -548,7 +562,7 @@ def test_openai_vector_store_search_with_high_score_filter(
# Insert chunks
llama_client.vector_io.insert(
vector_db_id=vector_store.id,
vector_store_id=vector_store.id,
chunks=sample_chunks,
)
@ -614,7 +628,7 @@ def test_openai_vector_store_search_with_max_num_results(
# Insert chunks
llama_client.vector_io.insert(
vector_db_id=vector_store.id,
vector_store_id=vector_store.id,
chunks=sample_chunks,
)
@ -1179,7 +1193,7 @@ def test_openai_vector_store_search_modes(
)
client_with_models.vector_io.insert(
vector_db_id=vector_store.id,
vector_store_id=vector_store.id,
chunks=sample_chunks,
)
query = "Python programming language"

58
tests/integration/vector_io/test_vector_io.py
View file

@ -13,23 +13,33 @@ from ..conftest import vector_provider_wrapper
@pytest.fixture(scope="session")
def sample_chunks():
from llama_stack.providers.utils.vector_io.vector_utils import generate_chunk_id
chunks_data = [
(
"Python is a high-level programming language that emphasizes code readability and allows programmers to express concepts in fewer lines of code than would be possible in languages such as C++ or Java.",
"doc1",
),
(
"Machine learning is a subset of artificial intelligence that enables systems to automatically learn and improve from experience without being explicitly programmed, using statistical techniques to give computer systems the ability to progressively improve performance on a specific task.",
"doc2",
),
(
"Data structures are fundamental to computer science because they provide organized ways to store and access data efficiently, enable faster processing of data through optimized algorithms, and form the building blocks for more complex software systems.",
"doc3",
),
(
"Neural networks are inspired by biological neural networks found in animal brains, using interconnected nodes called artificial neurons to process information through weighted connections that can be trained to recognize patterns and solve complex problems through iterative learning.",
"doc4",
),
]
return [
Chunk(
content="Python is a high-level programming language that emphasizes code readability and allows programmers to express concepts in fewer lines of code than would be possible in languages such as C++ or Java.",
metadata={"document_id": "doc1"},
),
Chunk(
content="Machine learning is a subset of artificial intelligence that enables systems to automatically learn and improve from experience without being explicitly programmed, using statistical techniques to give computer systems the ability to progressively improve performance on a specific task.",
metadata={"document_id": "doc2"},
),
Chunk(
content="Data structures are fundamental to computer science because they provide organized ways to store and access data efficiently, enable faster processing of data through optimized algorithms, and form the building blocks for more complex software systems.",
metadata={"document_id": "doc3"},
),
Chunk(
content="Neural networks are inspired by biological neural networks found in animal brains, using interconnected nodes called artificial neurons to process information through weighted connections that can be trained to recognize patterns and solve complex problems through iterative learning.",
metadata={"document_id": "doc4"},
),
content=content,
chunk_id=generate_chunk_id(doc_id, content),
metadata={"document_id": doc_id},
)
for content, doc_id in chunks_data
]
@ -123,12 +133,12 @@ def test_insert_chunks(
actual_vector_store_id = create_response.id
client_with_empty_registry.vector_io.insert(
vector_db_id=actual_vector_store_id,
vector_store_id=actual_vector_store_id,
chunks=sample_chunks,
)
response = client_with_empty_registry.vector_io.query(
vector_db_id=actual_vector_store_id,
vector_store_id=actual_vector_store_id,
query="What is the capital of France?",
)
assert response is not None
@ -137,7 +147,7 @@ def test_insert_chunks(
query, expected_doc_id = test_case
response = client_with_empty_registry.vector_io.query(
vector_db_id=actual_vector_store_id,
vector_store_id=actual_vector_store_id,
query=query,
)
assert response is not None
@ -169,19 +179,20 @@ def test_insert_chunks_with_precomputed_embeddings(
chunks_with_embeddings = [
Chunk(
content="This is a test chunk with precomputed embedding.",
chunk_id="chunk1",
metadata={"document_id": "doc1", "source": "precomputed", "chunk_id": "chunk1"},
embedding=[0.1] * int(embedding_dimension),
),
]
client_with_empty_registry.vector_io.insert(
vector_db_id=actual_vector_store_id,
vector_store_id=actual_vector_store_id,
chunks=chunks_with_embeddings,
)
provider = [p.provider_id for p in client_with_empty_registry.providers.list() if p.api == "vector_io"][0]
response = client_with_empty_registry.vector_io.query(
vector_db_id=actual_vector_store_id,
vector_store_id=actual_vector_store_id,
query="precomputed embedding test",
params=vector_io_provider_params_dict.get(provider, None),
)
@ -217,22 +228,25 @@ def test_query_returns_valid_object_when_identical_to_embedding_in_vdb(
actual_vector_store_id = register_response.id
from llama_stack.providers.utils.vector_io.vector_utils import generate_chunk_id
chunks_with_embeddings = [
Chunk(
content="duplicate",
chunk_id=generate_chunk_id("doc1", "duplicate"),
metadata={"document_id": "doc1", "source": "precomputed"},
embedding=[0.1] * int(embedding_dimension),
),
]
client_with_empty_registry.vector_io.insert(
vector_db_id=actual_vector_store_id,
vector_store_id=actual_vector_store_id,
chunks=chunks_with_embeddings,
)
provider = [p.provider_id for p in client_with_empty_registry.providers.list() if p.api == "vector_io"][0]
response = client_with_empty_registry.vector_io.query(
vector_db_id=actual_vector_store_id,
vector_store_id=actual_vector_store_id,
query="duplicate",
params=vector_io_provider_params_dict.get(provider, None),
)