A high-performance, cross-platform Kotlin library for OpenLDAP's Lightning Memory-Mapped Database (LMDB).

kotlin-lmdb provides a type-safe, idiomatic Kotlin API for LMDB, one of the fastest and most reliable key-value stores available. This library enables direct integration with LMDB across JVM and Native platforms from a single, consistent codebase.

• Cross-Platform Compatibility: Works seamlessly on JVM, Android, Native platforms (Linux, macOS, Windows, iOS, watchOS, tvOS), and WebAssembly (wasmJs).
• Performance: Direct bindings to native LMDB without additional layers for maximum performance
• Memory Efficiency: Leverages LMDB's memory-mapped architecture for efficient memory usage
• Transactional: Fully ACID-compliant with robust transaction support
• Consistent API: Unified Kotlin API regardless of platform
• Type Safety: Leverages Kotlin's type system for safer database operations

kotlin {
    sourceSets {
        val commonMain by getting {
            dependencies {
                implementation("io.github.crowded-libs:kotlin-lmdb:0.3.1")
            }
        }
    }
}

<dependency>
  <groupId>io.github.crowded-libs</groupId>
  <artifactId>kotlin-lmdb</artifactId>
  <version>0.3.1</version>
</dependency>

// Create and open an environment
val env = Env()
env.open("/path/to/database")

// Perform database operations within a transaction
env.beginTxn { txn ->
    val dbi = txn.dbiOpen()

    // Write data
    txn.put(dbi, "key1".encodeToByteArray(), "value1".encodeToByteArray())

    // Read data
    val (resultCode, k, v) = txn.get(dbi, "key1".encodeToByteArray())
    if (resultCode == 0) {
        val value = v.toByteArray()!!.decodeToString()
        println(value)  // Outputs: value1
    }

    txn.commit()
}

// Always close the environment when done
env.close()

LMDB supports storing multiple values for a single key using DupSort:

// Create environment and open database with dupsort flag
val env = Env()
env.open("/path/to/database")

env.beginTxn { txn ->
    // Open database with DupSort option enabled
    val dbi = txn.dbiOpen(null, DbiOption.DupSort, DbiOption.Create)

    // Store multiple values for the same key
    val key = "category".encodeToByteArray()
    txn.put(dbi, key, "item1".encodeToByteArray())
    txn.put(dbi, key, "item2".encodeToByteArray())
    txn.put(dbi, key, "item3".encodeToByteArray())

    // Retrieve all values for a key using a cursor
    val cursor = txn.openCursor(dbi)
    cursor.use {
        // Position cursor at the first value for this key
        val result = cursor.set(key)

        if (result.first == 0) {
            // Process the first value
            println("Value: ${result.third.toByteArray()!!.decodeToString()}")

            // Iterate through remaining values for this key
            var moreValues = true
            while (moreValues) {
                val nextResult = cursor.nextDuplicate()
                if (nextResult.first == 0) {
                    println("Value: ${nextResult.third.toByteArray()!!.decodeToString()}")
                } else {
                    moreValues = false
                }
            }
        }
    }

    txn.commit()
}

env.close()

The library comes with several pre-built comparers that you can use without writing custom code:

val env = Env()
env.open("/path/to/database")

env.beginTxn { txn ->
    // Use pre-built INTEGER_KEY comparer for numeric sorting
    val config = DbiConfig(keyComparer = ValComparer.INTEGER_KEY)
    val intDb = txn.dbiOpen("integer-keys", config, DbiOption.Create)

    // Now keys will be sorted as integers
    txn.put(intDb, byteArrayOf(10), "value-10".encodeToByteArray())
    txn.put(intDb, byteArrayOf(5), "value-5".encodeToByteArray()) 
    txn.put(intDb, byteArrayOf(20), "value-20".encodeToByteArray())

    // When iterating, they will be in order: 5, 10, 20
    // Open a different database with lexicographic string sorting
    val stringConfig = DbiConfig(keyComparer = ValComparer.LEXICOGRAPHIC_STRING)
    val stringDb = txn.dbiOpen("string-keys", stringConfig, DbiOption.Create)

    txn.commit()
}

env.close()

You can also define your own custom sorting logic for keys and values:

// Register a custom first-byte comparator
val firstByteComparer: ValCompare = { a, b -> 
    val aBytes = a.toByteArray() ?: byteArrayOf()
    val bBytes = b.toByteArray() ?: byteArrayOf()

    val aFirstByte = if (aBytes.isNotEmpty()) aBytes[0].toInt() else 0
    val bFirstByte = if (bBytes.isNotEmpty()) bBytes[0].toInt() else 0

    aFirstByte - bFirstByte
}

// Clear any existing custom comparers
clearCustomComparers()

// Register the custom comparer
registerCustomComparer(ValComparer.CUSTOM_1, firstByteComparer)

val env = Env()
env.open("/path/to/database")

env.beginTxn { txn ->
    // Create database config that uses the custom comparer
    val config = DbiConfig(keyComparer = ValComparer.CUSTOM_1)
    val dbi = txn.dbiOpen("custom-comparer-db", config, DbiOption.Create)

    // Now keys will be sorted by their first byte
    // Add some test data
    txn.put(dbi, "abc".encodeToByteArray(), "value1".encodeToByteArray())
    txn.put(dbi, "bcd".encodeToByteArray(), "value2".encodeToByteArray())
    txn.put(dbi, "zde".encodeToByteArray(), "value3".encodeToByteArray())

    txn.commit()
}

env.close()

The library provides these pre-built comparers through the ValComparer enum:

• BITWISE - Default byte-by-byte comparison (ascending)
• REVERSE_BITWISE - Byte-by-byte comparison in reverse order
• LEXICOGRAPHIC_STRING - Compare as UTF-8 strings (ascending)
• REVERSE_LEXICOGRAPHIC_STRING - Compare as UTF-8 strings (descending)
• INTEGER_KEY - Compare as integers (ascending)
• REVERSE_INTEGER_KEY - Compare as integers (descending)
• LENGTH - Compare by length first, then by content
• REVERSE_LENGTH - Compare by length (longer first), then by content
• LENGTH_ONLY - Compare only by length, ignoring content
• HASH_CODE - Compare by hash code (for faster large value comparisons)

• LMDB is memory-mapped, providing exceptional read performance
• Consider using a single transaction for multiple operations when possible
• Memory usage scales with database size as pages are mapped into memory

• JVM: Linux, macOS, Windows
• Android: All supported Android devices
• Kotlin/Native:
  • Desktop: macOS (x64, Arm64), Linux (x64, Arm64), Windows (x64)
  • iOS: iOS (Arm64, x64), iOS Simulator (Arm64)
  • watchOS: watchOS (Arm32, Arm64), watchOS Simulator (Arm64)
  • tvOS: tvOS (Arm64), tvOS Simulator (Arm64)
• WebAssembly: wasmJs (browser and Node.js)

The library includes pre-built native libraries for all common Android architectures:

• arm64-v8a (64-bit ARM)
• armeabi-v7a (32-bit ARM)
• x86_64 (64-bit x86)
• x86 (32-bit x86)

No additional configuration is required - the native libraries are automatically included and loaded when you add the dependency.

The wasmJs target provides LMDB functionality in web browsers and Node.js environments, but has some differences from other platforms:

Implementation Details:

• Uses a compiled WASM version of LMDB with Emscripten
• File system operations are handled through browser/Node.js filesystem abstraction layers
• Database files are stored in virtual filesystems (IDBFS for browsers, NODEFS for Node.js)

Behavioral Differences:

• The MDB_WRITEMAP flag is always enabled when opening environments, regardless of whether it's explicitly specified
• This is due to Emscripten's memory mapping implementation which doesn't support MAP_SHARED without write access
• Write operations may have different performance characteristics compared to native platforms

Limitations:

• File system persistence varies by environment (IndexedDB in browsers, real filesystem in Node.js)
• Memory constraints are typically lower than native platforms
• Some advanced LMDB features may have different behavior due to WASM limitations
• Performance characteristics differ from native implementations

Usage Notes:

• Must call WasmUtils.mountBestFilesystem(...) before using.
• In browsers, databases persist in IndexedDB between sessions
• In Node.js, databases behave similarly to other platforms
• Always test wasmJs-specific functionality thoroughly in your target environment
• You will need to setup a copy task for the js assets with your own project to ensure the modules load successfully. I haven't figured out how to make the packaging work seamlessly for consumption of the maven package yet. (pull request?)

• Official LMDB Documentation
• LMDB GitHub Repository