Adds multi vector query class

[justin-cechmanek]

No description provided.

[jit-ci]

Hi, I’m Jit, a friendly security platform designed to help developers build secure applications from day zero with an MVS (Minimal viable security) mindset.

In case there are security findings, they will be communicated to you as a comment inside the PR.

Hope you’ll enjoy using Jit.

Questions? Comments? Want to learn more? Get in touch with us.

[tylerhutcherson]

Very nice. I do think it could be interesting if this vector class also handled the conversion to bytes (and thus any compression based on choice of dtype)

I think we handle this within the query class today but some of that logic could be pulled into here as a true data container for vectors -- might be a cleaner and more testable pattern? @bsbodden @abrookins

[tylerhutcherson]

Add new query class to the sphinx docs!

[Copilot]

Pull Request Overview

This PR adds multi-vector query functionality to the Redis Vector Library, enabling simultaneous search across multiple vector fields in a document with weighted score combination.

• Introduces new MultiVectorQuery class and Vector helper class for multi-vector search capabilities
• Adds comprehensive test coverage for the new multi-vector query functionality including unit tests and integration tests
• Updates documentation to include the new Vector and MultiVectorQuery classes

Reviewed Changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
redisvl/query/aggregate.py	Implements Vector class and MultiVectorQuery class with validation and query building logic
redisvl/query/init.py	Exports new MultiVectorQuery and Vector classes
tests/unit/test_aggregation_types.py	Adds unit tests for MultiVectorQuery and Vector classes with validation and initialization tests
tests/integration/test_aggregation.py	Adds integration tests for multi-vector queries with various scenarios including filters, weights, and datatypes
tests/conftest.py	Adds multi_vector_data fixture with sample data for testing
docs/api/vector.rst	Adds documentation for the Vector class
docs/api/query.rst	Adds documentation for the MultiVectorQuery class
docs/api/index.md	Updates API documentation index to include vector documentation

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[abrookins]

@tylerhutcherson Yes, I think it makes sense to have fields own converting themselves to bytes:

from redisvl.redis.utils import array_to_buffer


class Vector(BaseModel):
    # ...
    def as_buffer(self) -> bytes:
        if isinstance(self.vector, (bytes, bytearray)):
            return bytes(self.vector)
        return array_to_buffer(self.vector, dtype=self.dtype)

We should be able to touch that up later though, if we want to get in a first pass.

Another thing I was wondering. This PR hard-codes the distance at 2.0 (all items?), but wouldn't it also work to expose that as a parameter on Vector? Similarly, we have room to change this later, but I wondered if that was intentional.

[abrookins]

I agree with @tylerhutcherson that making Vector the owner of conversion logic makes sense and encapsulates that better.

I had a question (posted separately) about whether we'd want to expose distance threshold as a parameter on Vector.

Neither are blocking, so I'd say if you want to keep the momentum and get this merged, and then consider those other improvements, I say 🫡 to that.

Seconded on the Copilot stuff, as well (just spelling, AFAIK).