feat: Add native_embedding option to allow Weaviate to generate embeddings during data ingestion and querying.

[selvamanigovindaraj]

Description

This feature enables the use of Weaviate's native embedding capabilities, which are otherwise not directly accessible via standard LlamaIndex workflows that assume client-side embedding generation.

Fixes #18666

New Package?

Did I fill in the tool.llamahub section in the pyproject.toml and provide a detailed README.md for my new integration or package?

• Yes
• No

Version Bump?

Did I bump the version in the pyproject.toml file of the package I am updating? (Except for the llama-index-core package)

• Yes
• No

Type of Change

Please delete options that are not relevant.

• New feature (non-breaking change which adds functionality)

How Has This Been Tested?

Your pull-request will likely not be merged unless it is covered by some form of impactful unit testing.

• I added new unit tests to cover this change
• I believe this change is already covered by existing unit tests

###Test Details:

Unit Tests: Added TestWeaviateEmbedding class to tests/test_vector_stores_weaviate.py covering:

• Add method with native embedding enabled (verifies vector is not sent)
• Add method with native embedding disabled (verifies vector is passed)
• Query method with native embedding enabled (verifies vector is not sent)
• Query method with native embedding disabled (verifies vector is passed)
• Async add method with native embedding enabled (verifies vector is not sent)
• Async add method with native embedding disabled (verifies vector is passed)
• Async query method with native embedding enabled (verifies vector is not sent)

E2E Verification: Ran comprehensive E2E tests against Weaviate Cloud and Embedded Weaviate, verifying:

• Server-Side Embedding: Native embedding enabled successfully adds nodes without client-side vectors.
• Client-Side Embedding: Native embedding disabled successfully adds nodes with provided vectors.
• Retrieval: Verified semantic search works in all scenarios.
• Async Support: Verified async add works correctly.

Suggested Checklist:

• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• I have added Google Colab support for the newly added notebooks.
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I ran uv run make format; uv run make lint to appease the lint gods

[Copilot]

Pull request overview

This PR fixes a bug where the embed_on_weaviate parameter was not properly preventing client-side embedding generation in the WeaviateVectorStore. The fix ensures that when embed_on_weaviate=True, vectors are not retrieved from nodes during data ingestion and querying, allowing Weaviate to generate embeddings server-side instead.

Key changes:

• Added a use_vector parameter to the get_data_object utility function to conditionally retrieve embeddings from nodes
• Updated add, async_add, and query methods to pass use_vector=False when embed_on_weaviate=True
• Added comprehensive unit tests covering both sync and async operations with the new parameter

Reviewed changes

Copilot reviewed 4 out of 5 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
llama_index/vector_stores/weaviate/utils.py	Added use_vector parameter to get_data_object function to conditionally retrieve node embeddings
llama_index/vector_stores/weaviate/base.py	Added _embed_on_weaviate private attribute and updated add, async_add, and get_query_parameters methods to respect the embedding mode
tests/test_vector_stores_weaviate.py	Added TestWeaviateEmbedding class with unit tests for add/async_add/query operations with both embedding modes, plus fixed async test method signature
pyproject.toml	Bumped version from 1.4.1 to 1.4.2
uv.lock	Updated version lock file to reflect version bump

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review for a chance to win a $100 gift card. Take the survey.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 5 changed files in this pull request and generated 3 comments.

Comments suppressed due to low confidence (1)

llama-index-integrations/vector_stores/llama-index-vector-stores-weaviate/llama_index/vector_stores/weaviate/base.py:491

• When native_embedding=True, the vector is set to None on line 486, which breaks the alpha calculation logic for hybrid queries. On line 490, the condition if vector is not None and query.query_str: will always be False when using native embeddings, causing alpha to remain 1.0 instead of using the configured value (default 0.5). Consider changing line 490 to: if (not self._native_embedding or query.query_embedding is not None) and query.query_str: to correctly handle both native and client-side embedding scenarios.

        vector = query.query_embedding if not self._native_embedding else None
        alpha = 1
        if query.mode == VectorStoreQueryMode.HYBRID:
            _logger.debug(f"Using hybrid search with alpha {query.alpha}")
            if vector is not None and query.query_str:
                alpha = query.alpha or 0.5

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review for a chance to win a $100 gift card. Take the survey.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 5 changed files in this pull request and generated 4 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review for a chance to win a $100 gift card. Take the survey.

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 5 changed files in this pull request and generated no new comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review for a chance to win a $100 gift card. Take the survey.

[logan-markewich]

Can we avoid class based tests? None of the other tests follow this pattern

[logan-markewich]

it'd be nice to update the example notebook in docs/examples/vector_stores/WeaviateIndexDemo.ipynb to mention this