rand: add PRNG.fill_buffer_from_set/2 (#21037)

vlib/rand/rand.c.v

@@ -93,6 +93,7 @@ fn internal_ulid_at_millisecond(mut rng PRNG, unix_time_milli u64) string {
 	}
 }
 
+@[direct_array_access]
 fn internal_string_from_set(mut rng PRNG, charset string, len int) string {
 	if len == 0 {
 		return ''
@@ -100,7 +101,7 @@ fn internal_string_from_set(mut rng PRNG, charset string, len int) string {
 	mut buf := unsafe { malloc_noscan(len + 1) }
 	for i in 0 .. len {
 		unsafe {
-			buf[i] = charset[intn(charset.len) or { 0 }]
+			buf[i] = charset[rng.u32() % charset.len]
 		}
 	}
 	unsafe {
@@ -109,6 +110,19 @@ fn internal_string_from_set(mut rng PRNG, charset string, len int) string {
 	return unsafe { buf.vstring_with_len(len) }
 }
 
+@[direct_array_access]
+fn internal_fill_buffer_from_set(mut rng PRNG, charset string, mut buf []u8) {
+	if buf.len == 0 {
+		return
+	}
+	blen := buf.len
+	for i in 0 .. blen {
+		unsafe {
+			buf[i] = charset[rng.u32() % charset.len]
+		}
+	}
+}
+
 fn deinit() {
 	unsafe {
 		default_rng.free() // free the implementation

vlib/rand/rand.js.v

@@ -5,6 +5,10 @@ fn init() {
 	default_rng = new_default()
 }
 
+fn internal_fill_buffer_from_set(mut rng PRNG, charset string, mut buf []u8) {
+	panic('todo')
+}
+
 fn internal_string_from_set(mut rng PRNG, charset string, len int) string {
 	result := ''
 	#

vlib/rand/rand.v

@@ -351,6 +351,12 @@ pub fn (mut rng PRNG) ascii(len int) string {
 	return internal_string_from_set(mut rng, rand.ascii_chars, len)
 }
 
+// fill_buffer_from_set fills the mutable `buf` with random characters from the given `charset`
+@[inline]
+pub fn (mut rng PRNG) fill_buffer_from_set(charset string, mut buf []u8) {
+	internal_fill_buffer_from_set(mut rng, charset, mut buf)
+}
+
 // bernoulli returns true with a probability p. Note that 0 <= p <= 1.
 pub fn (mut rng PRNG) bernoulli(p f64) !bool {
 	if p < 0 || p > 1 {
@@ -675,6 +681,12 @@ pub fn string_from_set(charset string, len int) string {
 	return default_rng.string_from_set(charset, len)
 }
 
+// fill_buffer_from_set fills the array `buf` with random characters sampled from the given `charset`
+@[inline]
+pub fn fill_buffer_from_set(charset string, mut buf []u8) {
+	default_rng.fill_buffer_from_set(charset, mut buf)
+}
+
 // string returns a string of length `len` containing random characters in range `[a-zA-Z]`.
 pub fn string(len int) string {
 	return string_from_set(rand.english_letters, len)

vlib/rand/random_numbers_test.v

@@ -236,19 +236,44 @@ fn test_rand_string_from_set() {
 	}
 }
 
+fn test_rand_fill_buffer_from_set() {
+	rand.seed([u32(0), 1])
+	outputs := [
+		[u8(52), 48, 55, 57, 50, 49, 53, 49, 53, 53],
+		[u8(57), 51, 56, 53, 56, 55, 56, 52, 56, 51],
+		[u8(57), 54, 52, 53, 57, 56, 57, 57, 48, 57],
+		[u8(57), 54, 50, 50, 52, 57, 53, 55, 50, 57],
+		[u8(51), 48, 55, 54, 49, 55, 53, 54, 52, 57],
+		[u8(57), 50, 48, 50, 48, 49, 52, 54, 50, 48],
+		[u8(55), 54, 51, 48, 51, 54, 49, 55, 56, 52],
+		[u8(52), 56, 52, 54, 50, 50, 50, 56, 54, 53],
+		[u8(53), 53, 55, 52, 51, 54, 55, 56, 51, 51],
+		[u8(52), 50, 51, 57, 54, 52, 50, 48, 49, 53],
+		[u8(49), 51, 54, 57, 55, 51, 48, 51, 51, 50],
+		[u8(56), 54, 50, 54, 51, 54, 49, 55, 57, 49],
+	]
+	for output in outputs {
+		mut buf := []u8{len: 10}
+		rand.fill_buffer_from_set('0123456789', mut buf)
+		assert buf == output
+	}
+}
+
 fn test_rand_string() {
 	rand.seed([u32(0), 1])
 	outputs := [
-		'rzJfVBJgvAyCNpEdXIteDQezg',
-		'AJOeswgoelDOCfcrSUWzVPjeL',
-		'NQfKauQqsXYXSUMFPGnXXPJIn',
-		'vfBGUKbpLoBMQVYXfkvRplWih',
-		'aYHLjMJqvUJmJJHGxEnrEmQGl',
-		'rBJXkQZcembAteaRFoxXmECJo',
-		'HYVLfHmDOCTlSbiSzHrsAIaBH',
-		'zgOiwyISjLSdLGhLzJsSKHVBi',
-		'UiAtobWXGcHsEtgzuNatxfkoI',
-		'NisnYlffJgFEcIdcgzWcGjnHy',
+		'oIfPOHLBZTlvGhYtCMolfssbZ',
+		'yHFGzDYeWIRldsBzMtkDhzQqF',
+		'vwoeerAKsEZiludKtRKoCoiuE',
+		'EQAaJDRZkvKTKNLkEPhWeEKFX',
+		'rDIhxzIbDUIusiTuzLHRslfzu',
+		'KCUoAEugYvUwzXcKRrAiwMzXH',
+		'NIOXerfCpEwbfhLmbbWKjoxbL',
+		'baJWQWarRRRmXCvMKcEjxQBpk',
+		'CkVLxbJEPhviBTohEVBnMAFHZ',
+		'ZdnGGhYShqzwnDXqHncLgLcdo',
+		'zRiSLsgnApmvtlIVrQQaBzOJD',
+		'VeeBcztImGquJnzEsXCdUaUed',
 	]
 	for output in outputs {
 		assert rand.string(25) == output
@@ -258,16 +283,20 @@ fn test_rand_string() {
 fn test_rand_hex() {
 	rand.seed([u32(0), 1])
 	outputs := [
-		'fc30e495deee09e008e15ffc3',
-		'4320efa837788397fb59b28f4',
-		'4995210abf33b6765c240ce62',
-		'f3d20dbe0a8aa6b9c88cd1f6f',
-		'8d7d58b256ab00213dd519cf7',
-		'fa2251284bc20a21eff48127c',
-		'5fef90cdc0c37143117599092',
-		'2a6170531c76dfb50c54126bc',
-		'a686dfd536042d1c1a9afdaf4',
-		'7f12013f6e1177e2d63726de3',
+		'847b633d9f9765c1a84d38035',
+		'efdef342641958db89cfdb4e1',
+		'704ee34204d29e9e99aca0ae0',
+		'0c8e1fd5472f65fc4b9668adf',
+		'3349538378c2023ef7f14dfbe',
+		'ae4080a0cb4cbb0693c68037b',
+		'90e3a7be588b3dfeb3663c97f',
+		'f25a82eb559ab6f0288bd8590',
+		'649f579cb93e9f414d9f40539',
+		'553a210a52bcbfbafb0783850',
+		'3daef80b45ef518d30c6db6db',
+		'56a187106e6e5fb88761024a5',
+		'b5cd8b7a24054d7dc66e62f88',
+		'306eed0c4207d8db185f04afd',
 	]
 	for output in outputs {
 		assert rand.hex(25) == output
@@ -277,16 +306,15 @@ fn test_rand_hex() {
 fn test_rand_ascii() {
 	rand.seed([u32(0), 1])
 	outputs := [
-		"2Z:&PeD'V;9=mn\$C>yKg'DIr%",
-		'Ub7ix,}>I=&#2QJki{%FHKv&K',
-		'1WStRylMO|p.R~qqRtr&AOEsd',
-		'yka<GPZ&m+r0^Zi!ShB*1dU~W',
-		'uDA?.zU2X,<DkKT#_-halW\\ki',
-		'fsx!@uRc?re/fSPXj`Y&\\BU}p',
-		'fI_qM"):2;CUno!<dX:Yv*FX$',
-		'FnA(Fr|D`WZVWEzp<k)O;auub',
-		"QRkxH!kjXh&/j{)uSe&{D'v?|",
-		"_CyaU\$z':#}At*v2|xDu6w=;1",
+		r"KqdNI|*bDh42kn'z-}}nhmKd~",
+		r'IZ4wVRC-Q3@TviD>G4#Z(2}s4',
+		r"l7'1Ute)i?4Efo$sX^sOk;s%m",
+		r"3}3s^l(PeNY>I8&'a>$)AW14*",
+		r'V.a^b>GN"\\9e-Vs"&.vS0"F_',
+		r"U-;S}OY+e>Ca>p'UD|7{}?6`x",
+		r'$/EN5*2w@/KdN~pU||c=*yn6|',
+		r'FsLkK{gFrPn)>EVW53uJLa<8?',
+		r'1#PB<"P}pLtY@F}^\TfNyCDB$',
 	]
 	for output in outputs {
 		assert rand.ascii(25) == output

vlib/v/tests/bench/bench_rand_fill_buffer_from_set.v

@@ -0,0 +1,54 @@
+import os
+import rand
+import time
+import rand.pcg32
+
+const buf_len = os.getenv_opt('BUF_LEN') or { '100_000_000' }.int()
+const nthreads = os.getenv_opt('VJOBS') or { '2' }.int()
+const max_iterations = os.getenv_opt('MAX_ITERATIONS') or { '4' }.int()
+
+fn main() {
+	mut buf := []u8{len: buf_len}
+	mut arr := []thread{}
+	sw := time.new_stopwatch()
+	for i in 0 .. nthreads {
+		part_len := buf.len / nthreads
+		start := i * part_len
+		mut chunk := &WorkChunk{
+			thread_id: i
+			// make the last thread fill the remaining characters too:
+			part: unsafe { buf[start..if i == nthreads - 1 { buf.len } else { start + part_len }] }
+		}
+		arr << spawn worker(mut chunk)
+	}
+	arr.wait()
+	elapsed := sw.elapsed().milliseconds()
+	mut histogram := []u64{len: 58}
+	for b in buf {
+		histogram[b]++
+	}
+	println('            buf: ${buf#[..10]} ... ${buf#[-10..]}')
+	println('      histogram: ${histogram#[48..]}')
+	println('Total took ${elapsed:6}ms, VJOBS: ${nthreads:2}, MAX_ITERATIONS: ${max_iterations:5}, BUF_LEN: ${buf_len:6}')
+	println('')
+}
+
+struct WorkChunk {
+	thread_id int
+mut:
+	part []u8
+}
+
+const charset = '0123456789'
+
+fn worker(mut chunk WorkChunk) {
+	mut rng := rand.PRNG(pcg32.PCG32RNG{})
+	sw := time.new_stopwatch()
+	for _ in 0 .. max_iterations {
+		rng.fill_buffer_from_set(charset, mut chunk.part)
+	}
+	elapsed_time_ns := sw.elapsed().nanoseconds()
+	fill_ms := u64(f64(elapsed_time_ns) / f64(max_iterations * 1_000_000))
+	rand_character_ns := f64(elapsed_time_ns) / f64(max_iterations * chunk.part.len)
+	println('      thread ${chunk.thread_id:2}, took ${elapsed_time_ns / 1_000_000:6}ms, per fill: ${fill_ms:6}ms, per character: ${rand_character_ns:5.0}ns, part: ${chunk.part#[..3]}...${chunk.part#[-3..]}, part.len: ${chunk.part.len:6}')
+}