Milvus

Milvus logo Ruby logo

Ruby wrapper for the Milvus vector search database API.

Part of the Langchain.rb stack.

Available for paid consulting engagements! Email me.

API Docs

https://docs.zilliz.com/reference/restful/data-plane-v2

Installation

Install the gem and add to the application’s Gemfile by executing:

$ bundle add milvus

If bundler is not being used to manage dependencies, install the gem by executing:

$ gem install milvus

Usage

Instantiating API client

require 'milvus'

client = Milvus::Client.new(
    url: 'http://localhost:19530'
)

Using the Collections endpoints

# Check if the collection exists.
client.collections.has(collection_name: "example_collection")

# Rename a collection.
client.collections.rename(collection_name: "example_collection", new_collection_name: "example_collection")

# Get collection stats
client.collections.get_stats(collection_name: "example_collection")

# Data types: https://github.com/patterns-ai-core/milvus/blob/main/lib/milvus/constants.rb

# Creating a new collection schema
client.collections.create(
  collection_name: "example_collection",
  auto_id: true,
  fields: [
    {
      fieldName: "book_id",
      isPrimary: true,
      autoID: false,
      dataType: "Int64"
    },
    {
      fieldName: "content",
      dataType: "VarChar",
      elementTypeParams: {
        max_length: "512"
      }
    },
    {
      fieldName: "vector",
      dataType: "FloatVector",
      elementTypeParams: {
        dim: 1536
      }
    }
  ]
)

# Descrbie the collection
client.collections.describe(collection_name: "example_collection")

# Drop the collection
client.collections.drop(collection_name: "example_collection")

# Load the collection to memory before a search or a query
client.collections.load(collection_name: "example_collection")

# Load status of a specific collection.
client.collections.get_load_state(collection_name: "example_collection")

# List all collections in the specified database.
client.collections.list

# Release a collection from memory after a search or a query to reduce memory usage
client.collections.release(collection_name: "example_collection")

Inserting Data

client.entities.insert(
  collection_name: "example_collection",
  data: [
    { id: 1, content: "The quick brown fox jumps over the lazy dog", vector: ([0.1]*1536) },
    { id: 2, content: "Lorem ipsum dolor sit amet, consectetur adipiscing elit", vector: ([0.2]*1536) },
    { id: 3, content: "Sed do eiusmod tempor incididunt ut labore et dolore magna aliqua", vector: ([0.3]*1536) }
  ]  
)

# Delete the entities with the boolean expression you created
client.entities.delete(
  collection_name: "example_collection",
  expression: "book_id in [0,1]"
)

# Inserts new records into the database or updates existing ones.
client.entities.upsert()

# Get specific entities by their IDs
client.entities.get()

Indexes

# Create an index
index_params = [
  {
    metricType: "L2",
    fieldName: "vector",
    indexName: "vector_idx",
    indexConfig: {
      index_type: "AUTOINDEX"
    }
  }
]

client.indexes.create(
  collection_name: "example_collection",
  index_params: index_params
)

# Describe an index
client.indexes.describe(
  collection_name: "example_collection",
  index_name: "example_index"
)

# List indexes
client.indexes.list(
  collection_name: "example_collection"
)

# Drop an index
client.indexes.drop(
  collection_name: "example_collection",
  index_name: "example_index"
)

Search, Querying & Hybrid Search

client.entities.search(
  collection_name: "example_collection",
  anns_field: "vectors",
  data: [embedding],
  filter: "id in [450847466900987454]"
)

client.entities.query(
  collection_name: "example_collection",
  filter: "id in [450847466900987455, 450847466900987454]"
)

client.entities.hybrid_search(
  collection_name: "example_collection",
  search: [{
    filter: "id in [450847466900987455]",
    data: [embedding],
    annsField: "vectors",
    limit: 10,
    outputFields: ["content", "id"]
  }],
  rerank: {
    "strategy": "rrf",
    "params": {
      "k": 10
    }
  },
  limit: 10,
  output_fields: ["content", "id"]
)

Partitions

# List partitions
client.partitions.list(
  collection_name: "example_collection"
)

# Create a partition
client.partitions.create(
  collection_name: "example_collection",
  partition_name: "example_partition"
)

# Check if a partition exists
client.partitions.has(
  collection_name: "example_collection",
  partition_name: "example_partition"
)

# Load partition data into memory
client.partitions.load(
  collection_name: "example_collection",
  partition_names: ["example_partition"]
)

# Release partition data from memory
client.partitions.release(
  collection_name: "example_collection",
  partition_names: ["example_partition"]
)

# Get statistics of a partition
client.partitions.get_stats(
  collection_name: "example_collection",
  partition_name: "example_partition"
)

# Drop a partition
client.partitions.drop(
  collection_name: "example_collection",
  partition_name: "example_partition"
)

Roles

# List roles available on the server
client.roles.list

# Describe the role
client.roles.describe(role_name: 'public')

Users

# Create new user
client.users.create(user_name: 'user_name', password: 'password')

# List of roles assigned to the user
client.users.describe(user_name: 'user_name')

# List all users in the specified database.
client.users.list

# Drop existing user
client.users.drop(user_name: 'user_name')

# Update password for the user
client.users.update_password(user_name: 'user_name', password: 'old_password', new_password: 'new_password')

# Grant role to the user
client.users.grant_role(user_name: 'user_name', role_name: 'admin')

# Revoke role from the user 
client.users.revoke_role(user_name: 'user_name', role_name: 'admin')

Aliases

# Lists all existing collection aliases in the specified database
client.aliases.list

# Describes the details of a specific alias
client.aliases.describe

# Reassigns the alias of one collection to another.
client.aliases.alter

# Drops a specified alias
client.aliases.drop

# Creates an alias for an existing collection
client.aliases.create

Development

After checking out the repo, run bin/setup to install dependencies. Then, run rake spec to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and the created tag, and push the .gem file to rubygems.org.

Development with Docker

Run docker compose run --rm ruby_app bash and install required gems (bundle install). It will give you a fully working development environment with Milvus services and gem’s code.

For example inside docker container run bin/console and inside the ruby console:

client = Milvus::Client.new(url: ENV["MILVUS_URL"])
client.collections.list

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/patterns-ai-core/milvus.

License

milvus is licensed under the Apache License, Version 2.0. View a copy of the License file.

Modules

Classes