Dllb

dllb

Elixir client for the dllb multi-model NoSQL database

Hex.pmHex Docs

Dllb provides a high-level Elixir API for communicating with the dllb database over TCP. It manages a NimblePool-based connection pool, speaks the dllb line-based wire protocol, and exposes a query builder plus result parsing so your application can focus on data rather than sockets.

Features

Installation

Add dllb to your list of dependencies in mix.exs:

def deps do

  [

    {:dllb, "~> 0.1.0"}

  ]

end

Configuration

# config/config.exs

config :dllb,

  enabled: true,

  host: "127.0.0.1",

  port: 3009,

  pool_size: 5,

  outcome: :json,

  timeout: 30_000

Setting enabled: false (the default) starts the application without the connection pool, which is useful for compile-time or test environments where no dllb server is available.

Options

Usage

Basic queries

{:ok, %Dllb.Result.Rows{count: 3, data: rows}} = Dllb.query("SELECT * FROM users")



result = Dllb.query!("SELECT * FROM users WHERE age > 25")

Query builder

Dllb.Query.create("user", %{name: "Alice", age: 30})

# => "CREATE user SET age = 30, name = 'Alice'"



Dllb.Query.select("user", where: "age > 25", limit: 10)

# => "SELECT * FROM user WHERE age > 25 LIMIT 10"



Dllb.Query.relate("user:a", "follows", "user:b", %{since: "2024"})

# => "RELATE user:a->follows->user:b SET since = '2024'"

Secondary indexes

# Single-field secondary index.

Dllb.Query.define_index("user", "by_age", ["age"])

# => "DEFINE INDEX by_age ON TABLE user FIELDS age"



# Composite index (leftmost-prefix planning: list the leading field first).

Dllb.Query.define_index("ast_node", "idx_file_kind", ["file_path", "kind"])

# => "DEFINE INDEX idx_file_kind ON TABLE ast_node FIELDS file_path, kind"



# Unique constraint over the full indexed tuple.

Dllb.Query.define_index("user", "by_email", ["email"], unique: true)

# => "DEFINE INDEX by_email ON TABLE user FIELDS email UNIQUE"



# Drop an index (queries then fall back to full scans).

Dllb.Query.remove_index("user", "by_age")

# => "REMOVE INDEX by_age ON TABLE user"

Once an index exists, no query changes are required: SELECT, COUNT, and UPDATE statements whose WHERE clause has equality or range predicates on indexed fields are accelerated automatically.

Full-text and vector search

Full-text (BM25) and vector (HNSW) indexes are created over the wire and queried with the SEARCH, VECTOR SEARCH, and HYBRID SEARCH verbs, each accepting an optional server-side WHERE scope. All require a dllb server with search services enabled (the default server build).

# Define a full-text index (optionally with a language analyzer).

Dllb.Query.define_fulltext_index("article", "ft_body", "body", analyzer: "english")

# => "DEFINE FULLTEXT INDEX ft_body ON TABLE article FIELDS body ANALYZER english"



# BM25 search; each row carries a "score" field, best-first.

Dllb.Query.search("article", "body", "graph database", limit: 5)

# => "SEARCH article body 'graph database' LIMIT 5"



# Define a vector (HNSW) index over a dense embedding field.

Dllb.Query.define_vector_index("ast_node", "vec_src", "source_embedding", 768, metric: "cosine")

# => "DEFINE VECTOR INDEX vec_src ON TABLE ast_node FIELDS source_embedding DIMENSION 768 METRIC cosine"



# Approximate nearest-neighbour search; each row carries a "distance" field.

Dllb.Query.vector_search("ast_node", "source_embedding", [0.12, 0.07, 0.91], k: 10)

# => "VECTOR SEARCH ast_node source_embedding [0.12, 0.07, 0.91] K 10"



# Scope results server-side (multi-project isolation, kind/language filters).

Dllb.Query.vector_search("ast_node", "source_embedding", [0.12, 0.07],

  where: "project_path = '/app'",

  k: 10

)

# => "VECTOR SEARCH ast_node source_embedding [0.12, 0.07] WHERE project_path = '/app' K 10"



# Hybrid search fuses BM25 and HNSW; rows carry score, text_score, vector_score.

Dllb.Query.hybrid_search("ast_node", "source_text", "parse tokens", "source_embedding", [0.12, 0.07],

  alpha: 0.6,

  limit: 10

)

# => "HYBRID SEARCH ast_node TEXT source_text 'parse tokens' VECTOR source_embedding [0.12, 0.07] ALPHA 0.6 LIMIT 10"

Valid analyzers: default, simple, english, spanish, french, german, italian, portuguese, russian. Valid metrics: cosine, euclidean (alias l2), dot (alias dotproduct/dot_product).

Aggregation, deletion, and graph analytics

# Grouped COUNT: one row per kind, each with a count (best-first).

Dllb.Query.count("ast_node", group_by: "kind")

# => "COUNT ast_node GROUP BY kind"



# Server-side delete-by-predicate (engine maintains all indexes).

Dllb.Query.delete_where("ast_node", "file_path = '/app/lib/old.ex'")

# => "DELETE ast_node WHERE file_path = '/app/lib/old.ex'"



# Weighted PageRank over an edge table, top-N by score.

Dllb.Query.graph_pagerank("calls", damping: 0.85, limit: 20)

# => "GRAPH PAGERANK calls DAMPING 0.85 LIMIT 20"



# Degree centrality (also :indegree / :outdegree).

Dllb.Query.graph_centrality("calls", mode: :indegree, limit: 20)

# => "GRAPH CENTRALITY calls INDEGREE LIMIT 20"



# Shortest directed path between two vertices.

Dllb.Query.graph_path("a", "b", "calls", max_depth: 6)

# => "GRAPH PATH a -> b ON calls MAX_DEPTH 6"



# List edges with their stored weights (default 1.0).

Dllb.Query.graph_edges("calls", where: "weight > 0.5")

# => "GRAPH EDGES calls WHERE weight > 0.5"

Upserts

# Insert, or merge the same fields on conflict.

Dllb.Query.upsert("user", "u1", %{name: "Alice", age: 30})

# => "CREATE user:u1 SET age = 30, name = 'Alice' ON CONFLICT UPDATE"



# Insert, or apply explicit fields on conflict.

Dllb.Query.upsert("user", "u1", %{name: "Alice", age: 30}, %{age: 31})

# => "CREATE user:u1 SET age = 30, name = 'Alice' ON CONFLICT UPDATE SET age = 31"

Schema bootstrap

{:ok, :bootstrapped} = Dllb.Schema.bootstrap(&Dllb.query/1)

MetaAST ingestion

context = %{language: :elixir, file_path: "/app/lib/parser.ex"}

{:ok, %{nodes: 42, edges: 17}} = Dllb.MetaAST.ingest_tree(ast, context, &Dllb.query/1)

Modules

Documentation

hexdocs.pm/dllb

Credits

Created as part of the Oeditus code quality tooling ecosystem.

License

MIT