chDB for Go

How to install and use chDB with Go

chDB-go provides Go bindings for chDB, enabling you to run ClickHouse queries directly in your Go applications with zero external dependencies.

Installation

Step 1: Install libchdb

First, install the chDB library:

curl -sL https://lib.chdb.io | bash

Step 2: Install chdb-go

Install the Go package:

go install github.com/chdb-io/chdb-go@latest

Or add it to your go.mod:

go get github.com/chdb-io/chdb-go

Usage

Command line interface

chDB-go includes a CLI for quick queries:

# Simple query
./chdb-go "SELECT 123"

# Interactive mode
./chdb-go

# Interactive mode with persistent storage
./chdb-go --path /tmp/chdb

Go Library - quick start

Stateless queries

For simple, one-off queries:

package main

import (
    "fmt"
    "github.com/chdb-io/chdb-go/chdb"
)

func main() {
    // Execute a simple query
    result, err := chdb.Query("SELECT version()", "CSV")
    if err != nil {
        panic(err)
    }
    fmt.Println(result)
}

Stateful queries with session

For complex queries with persistent state:

package main

import (
    "fmt"
    "github.com/chdb-io/chdb-go/chdb"
)

func main() {
    // Create a session with persistent storage
    session, err := chdb.NewSession("/tmp/chdb-data")
    if err != nil {
        panic(err)
    }
    defer session.Cleanup()

    // Create database and table
    _, err = session.Query(`
        CREATE DATABASE IF NOT EXISTS testdb;
        CREATE TABLE IF NOT EXISTS testdb.test_table (
            id UInt32,
            name String
        ) ENGINE = MergeTree() ORDER BY id
    `, "")
    
    if err != nil {
        panic(err)
    }

    // Insert data
    _, err = session.Query(`
        INSERT INTO testdb.test_table VALUES 
        (1, 'Alice'), (2, 'Bob'), (3, 'Charlie')
    `, "")
    
    if err != nil {
        panic(err)
    }

    // Query data
    result, err := session.Query("SELECT * FROM testdb.test_table ORDER BY id", "Pretty")
    if err != nil {
        panic(err)
    }
    
    fmt.Println(result)
}

SQL driver interface

chDB-go implements Go’s database/sql interface:

package main

import (
    "database/sql"
    "fmt"
    _ "github.com/chdb-io/chdb-go/chdb/driver"
)

func main() {
    // Open database connection
    db, err := sql.Open("chdb", "")
    if err != nil {
        panic(err)
    }
    defer db.Close()

    // Query with standard database/sql interface
    rows, err := db.Query("SELECT COUNT(*) FROM url('https://datasets.clickhouse.com/hits/hits.parquet')")
    if err != nil {
        panic(err)
    }
    defer rows.Close()

    for rows.Next() {
        var count int
        err := rows.Scan(&count)
        if err != nil {
            panic(err)
        }
        fmt.Printf("Count: %d\n", count)
    }
}

Query streaming for large datasets

For processing large datasets that don’t fit in memory, use streaming queries:

package main

import (
    "fmt"
    "log"
    "github.com/chdb-io/chdb-go/chdb"
)

func main() {
    // Create a session for streaming queries
    session, err := chdb.NewSession("/tmp/chdb-stream")
    if err != nil {
        log.Fatal(err)
    }
    defer session.Cleanup()

    // Execute a streaming query for large dataset
    streamResult, err := session.QueryStreaming(
        "SELECT number, number * 2 as double FROM system.numbers LIMIT 1000000", 
        "CSV",
    )
    if err != nil {
        log.Fatal(err)
    }
    defer streamResult.Free()

    rowCount := 0
    
    // Process data in chunks
    for {
        chunk := streamResult.GetNext()
        if chunk == nil {
            // No more data
            break
        }
        
        // Check for streaming errors
        if err := streamResult.Error(); err != nil {
            log.Printf("Streaming error: %v", err)
            break
        }
        
        rowsRead := chunk.RowsRead()
        // You can process the chunk data here
        // For example, write to file, send over network, etc.
        fmt.Printf("Processed chunk with %d rows\n", rowsRead)
        rowCount += int(rowsRead)
        if rowCount%100000 == 0 {
            fmt.Printf("Processed %d rows so far...\n", rowCount)
        }
    }
    
    fmt.Printf("Total rows processed: %d\n", rowCount)
}

Benefits of query streaming:

• Memory efficient - Process large datasets without loading everything into memory
• Real-time processing - Start processing data as soon as first chunk arrives
• Cancellation support - Can cancel long-running queries with Cancel()
• Error handling - Check for errors during streaming with Error()

API documentation

chDB-go provides both high-level and low-level APIs:

• High-Level API Documentation - Recommended for most use cases
• Low-Level API Documentation - For advanced use cases requiring fine-grained control

System requirements

• Go 1.21 or later
• Compatible with Linux, macOS