$mol_bloom - Bloom filter with automatic optimal parameters.

bloom/bloom.test.ts

@@ -0,0 +1,30 @@
+namespace $ {
+	$mol_test({
+
+		'Add and check strings'() {
+
+			const bloom = new $mol_bloom( 5 )
+			$mol_assert_equal( bloom.has_str( 'Hello' ), 0 )
+			$mol_assert_equal( bloom.has_str( 'World' ), 0 )
+			$mol_assert_equal( bloom.has_str( 'Hello World' ), 0 )
+			$mol_assert_equal( bloom.has_str( 'hello' ), 0 )
+			$mol_assert_equal( bloom.has_str( 'world' ), 0 )
+
+			bloom.add_str( 'Hello World' )
+			$mol_assert_equal( bloom.has_str( 'Hello' ), 0 )
+			$mol_assert_equal( bloom.has_str( 'World' ), 0 )
+			$mol_assert_equal( bloom.has_str( 'Hello World' ), 1 )
+			$mol_assert_equal( bloom.has_str( 'hello' ), 0 )
+			$mol_assert_equal( bloom.has_str( 'world' ), 0 )
+
+			bloom.add_str( 'Hello' )
+			bloom.add_str( 'World' )
+			$mol_assert_equal( bloom.has_str( 'Hello' ), 1 )
+			$mol_assert_equal( bloom.has_str( 'World' ), 1 )
+			$mol_assert_equal( bloom.has_str( 'hello' ), 0 )
+			$mol_assert_equal( bloom.has_str( 'world' ), 0 )
+
+		},
+
+	})
+}

bloom/bloom.ts

@@ -0,0 +1,74 @@
+namespace $ {
+
+	/**
+	 * Bloom filter with automatic optimal parameters.
+	 * False negative is impossible.
+	 * False positive is controlled.
+	 */
+	export class $mol_bloom extends Object {
+
+		bitmap: Uint32Array
+		hashes: number
+
+		constructor(
+			/** Max count of stored values. */
+			count: number,
+			/** Chance of false positive. 1e-6 by default */
+			public risk = 1e-6,
+		) {
+			super()
+			this.hashes = Math.ceil( -1.44 * Math.log( risk ) )
+
+			const length = Math.ceil( -.065 * count * Math.log( risk ) )
+			this.bitmap = new Uint32Array( 2 ** Math.ceil( Math.log2( length ) ) )
+		}
+
+		add_str( word: string ) {
+			this.add_bin( $mol_charset_encode( word ) )
+		}
+
+		has_str( word: string ) {
+			return this.has_bin( $mol_charset_encode( word ) )
+		}
+
+		add_bin( bin: Uint8Array ) {
+			for( const index of this.hash( bin ) ) {
+				this.add_bit( index )
+			}
+		}
+
+		has_bin( bin: Uint8Array ) {
+			let res = 1
+			for( const index of this.hash( bin ) ) {
+				res &= this.has_bit( index )
+			}
+			return res
+		}
+
+		hash( data: Uint8Array ) {
+			const res = [] as number[]
+			fill: while( true ) {
+				data = $mol_crypto_hash( data )
+				for( const index of new Uint32Array( data.buffer ) ) {
+					res.push( index )
+					if( res.length >= this.hashes ) break fill
+				}
+			}
+			return res
+		}
+
+		add_bit( index: number ) {
+			const int = Math.ceil( index / 32 ) % this.bitmap.length
+			const bit = index & 0b11111
+			this.bitmap[ int ] |= 1 << bit
+		}
+
+		has_bit( index: number ) {
+			const int = Math.ceil( index / 32 ) % this.bitmap.length
+			const bit = index & 0b11111
+			return 1&( this.bitmap[ int ] >> bit )
+		}
+
+	}
+
+}

crypto/hash/hash.ts

@@ -2,7 +2,7 @@ namespace $ {
 
 	let sponge = new Uint32Array(80)
 
-	/** Fast small sync SHA-1 */
+	/** Fast small sync SHA-1 (20 bytes, 160 bits) */
 	export function $mol_crypto_hash( data: Uint8Array ) {
 
 		const bits = data.byteLength << 3