Milvus

require 'milvus'

client = Milvus::Client.new(
    url: 'http://localhost:9091'
)

# Data types: "boolean", "int8", "int16", "int32", "int64", "float", "double", "string", "varchar", "binary_vector", "float_vector"

# Creating a new collection schema
client.collections.create(
  collection_name: "book",
  description: "Test book search",
  auto_id: false,
  fields: [
    {
      "name": "book_id",
      "description": "book id",
      "is_primary_key": true,
      "autoID": false,
      "data_type": Milvus::DATA_TYPES["int64"]
    },
    {
      "name": "word_count",
      "description": "count of words",
      "is_primary_key": false,
      "data_type": Milvus::DATA_TYPES["int64"]
    },
    {
      "name": "book_intro",
      "description": "embedded vector of book introduction",
      "data_type": Milvus::DATA_TYPES["binary_vector"],
      "is_primary_key": false,
      "type_params": [
        {
          "key": "dim",
          "value": "2"
        }
      ]
    }
  ]
)

# Get the collection info
client.collections.get(collection_name: "book")

# Delete the collection
client.collections.delete(collection_name: "book")

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

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

client.entities.insert(
  collection_name: "book",
  num_rows: 5, # Number of rows to be inserted. The number should be the same as the length of each field array.
  fields_data: [
    {
      "field_name": "book_id",
      "type": Milvus::DATA_TYPES["int64"],
      "field": [1,2,3,4,5]
    },
    {
      "field_name": "word_count",
      "type": Milvus::DATA_TYPES["int64"],
      "field": [1000,2000,3000,4000,5000]
    },
    {
      "field_name": "book_intro",
      "type": 101,
      "field": [ [1,1],[2,1],[3,1],[4,1],[5,1] ]
    }
  ]  
)

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

# Compact data manually
client.entities.compact!(
  collection_id: "book"
)
# => {"status"=>{}, "compactionID"=>440928616022809499}

# Check compaction status
client.entities.compact_status(
  compaction_id: 440928616022809499
)
# => {"status"=>{}, "state"=>2}

client.indices.create(
  collection_name: "book",
  field_name: "book_intro",
  extra_params: [
    { key: "metric_type", "value": "L2" },
    { key: "index_type", "value": "IVF_FLAT" },
    { key: "params", "value": "{\"nlist\":1024}" }
  ]
)

collection.indices.create(
  field_name: "book_name", 
  index_name: "scalar_index",
)

client.indices.delete(
  collection_name: "book",
  field_name: "book_intro"
)

client.search(
  collection_name: "book",
  output_fields: ["book_id"], # optional
  anns_field: "book_intro",
  top_k: "2",
  params: "{\"nprobe\": 10}",
  metric_type: "L2",
  round_decimal: "-1",
  vectors: [ [0.1,0.2] ],
  dsl_type: 1
)

client.query(
  collection_name: "book",
  output_fields: ["book_id", "book_intro"],
  expr: "book_id in [2,4,6,8]"
)

client.partitions.create(
  "collection_name": "book",
  "partition_name": "novel"
)

client.partitions.get(
  "collection_name": "book",
  "partition_name": "novel"
)

client.partitions.delete(
  "collection_name": "book",
  "partition_name": "novel"
)

client.partitions.load(
  "collection_name": "book",
  "partition_names": ["novel"],
  "replica_number": 1
)

client.partitions.release(
  "collection_name": "book",
  "partition_names": ["novel"],
  "replica_number": 1
)

# Live determines whether the application is alive. It can be used for Kubernetes liveness probe.
client.health