Data Structures

Reference documentation for Memista's data structures.

Rust Types

AppState

Shared application state passed to all handlers.

pub struct AppState {
    pub db_pool: Pool,
}

Field	Type	Description
db_pool	Pool	Async SQLite connection pool

Config

Application configuration.

pub struct Config {
    pub database_path: String,
    pub server_host: String,
    pub server_port: u16,
    pub log_level: String,
}

Field	Type	Default	Description
database_path	String	memista.db	SQLite database file
server_host	String	127.0.0.1	Server bind address
server_port	u16	8083	Server port
log_level	String	info	Logging level

Methods

impl Config {
    /// Load configuration from environment variables
    pub fn from_env() -> Result<Self, ConfigError>;
}

ChunkData

Individual text chunk with embedding.

#[derive(Clone, Serialize, Deserialize, JsonSchema)]
pub struct ChunkData {
    pub embedding: Vec<f32>,
    pub text: String,
    pub metadata: String,
}

Field	Type	Description
embedding	Vec<f32>	Vector embedding
text	String	Text content
metadata	String	JSON metadata string

InsertChunkRequest

Request body for insert operations.

#[derive(Deserialize, JsonSchema)]
pub struct InsertChunkRequest {
    pub database_id: String,
    pub chunks: Vec<ChunkData>,
}

Field	Type	Description
database_id	String	Target database identifier
chunks	Vec<ChunkData>	Chunks to insert

SearchRequest

Request body for search operations.

#[derive(Deserialize, JsonSchema)]
pub struct SearchRequest {
    pub database_id: String,
    pub embeddings: Vec<Vec<f32>>,
    pub num_results: usize,
}

Field	Type	Description
database_id	String	Database to search
embeddings	Vec<Vec<f32>>	Query embeddings
num_results	usize	Max results per query

SearchResult

Individual search result.

#[derive(Serialize, JsonSchema)]
pub struct SearchResult {
    pub chunk_id: i64,
    pub text: String,
    pub metadata: Option<String>,
    pub score: f32,
}

Field	Type	Description
chunk_id	i64	Chunk identifier
text	String	Chunk text content
metadata	Option<String>	JSON metadata
score	f32	Similarity score

DropTableRequest

Request body for drop operations.

#[derive(Deserialize, JsonSchema)]
pub struct DropTableRequest {
    pub database_id: String,
}

Field	Type	Description
database_id	String	Database to drop

Database Schema

SQLite Table Structure

Each database_id creates a table named chunks_{database_id}:

CREATE TABLE IF NOT EXISTS chunks_{database_id} (
    chunk_id INTEGER PRIMARY KEY AUTOINCREMENT,
    text TEXT NOT NULL,
    metadata TEXT
);

Column	Type	Description
chunk_id	INTEGER	Auto-incrementing primary key
text	TEXT	Chunk text content
metadata	TEXT	Optional JSON metadata

USearch Index

Each database has a corresponding index file: {database_id}.usearch

The index maps chunk_id (u64) to embedding vectors.

JSON Schemas

Insert Request

{
  "type": "object",
  "required": ["database_id", "chunks"],
  "properties": {
    "database_id": { "type": "string" },
    "chunks": {
      "type": "array",
      "items": {
        "type": "object",
        "required": ["embedding", "text", "metadata"],
        "properties": {
          "embedding": {
            "type": "array",
            "items": { "type": "number" }
          },
          "text": { "type": "string" },
          "metadata": { "type": "string" }
        }
      }
    }
  }
}

Search Request

{
  "type": "object",
  "required": ["database_id", "embeddings", "num_results"],
  "properties": {
    "database_id": { "type": "string" },
    "embeddings": {
      "type": "array",
      "items": {
        "type": "array",
        "items": { "type": "number" }
      }
    },
    "num_results": { "type": "integer", "minimum": 1 }
  }
}

Search Response

{
  "type": "object",
  "properties": {
    "results": {
      "type": "array",
      "items": {
        "type": "array",
        "items": {
          "type": "object",
          "properties": {
            "chunk_id": { "type": "integer" },
            "text": { "type": "string" },
            "metadata": { "type": "string" },
            "score": { "type": "number" }
          }
        }
      }
    }
  }
}