rand: add rand.ulid() (#5979)

* removed debug println

* added newline to the end of the file

* time: add .unix_time_milli() method; rand,time: add tests

* rand: add more ulid tests; move tests to a separate file random_identifiers_test.v

* run vfmt over vlib/rand/random_identifiers_test.v

* speed up time.unix_time_milli

* simplify and speedup time.unix_time/0 and time.new_time/1

* update comment about rand.ulid()

* fix terminating 0 off by 1 issue in rand.ulid()

* optimize time.new_time()

* restore the master version of vlib/time/parse.v

* make test_unix_time more robust

Co-authored-by: Delyan Angelov <delian66@gmail.com>

vlib/rand/rand.v

@@ -6,6 +6,8 @@ module rand
 import rand.util
 import rand.wyrand
 
+import time
+
 // Configuration struct for creating a new instance of the default RNG.
 pub struct PRNGConfigStruct {
 	seed []u32 = util.time_seed_array(2)
@@ -182,3 +184,51 @@ pub fn uuid_v4() string {
 	}
 	return string(buf, buflen)
 }
+
+const(
+	ulid_encoding = "0123456789ABCDEFGHJKMNPQRSTVWXYZ"
+)
+
+// rand.ulid generates an Unique Lexicographically sortable IDentifier.
+// See https://github.com/ulid/spec .
+// NB: ULIDs can leak timing information, if you make them public, because
+// you can infer the rate at which some resource is being created, like 
+// users or business transactions.
+// (https://news.ycombinator.com/item?id=14526173)
+pub fn ulid() string {
+	buflen := 26
+	mut buf := malloc(27)
+	// time section
+	mut t := time.utc().unix_time_milli()
+	mut i := 9
+	for i >= 0 {
+		unsafe{
+			buf[i] = ulid_encoding[t & 0x1F]
+		}
+		t = t >> 5
+		i--
+	}
+	// first rand set
+	mut x := default_rng.u64()
+	i = 10
+	for i < 19 {
+		unsafe{
+			buf[i] = ulid_encoding[x & 0x1F]
+		}
+		x = x >> 5
+		i++
+	}
+	// second rand set
+	x = default_rng.u64()
+	for i < 26 {
+		unsafe{
+			buf[i] = ulid_encoding[x & 0x1F]
+		}
+		x = x >> 5
+		i++
+	}
+	unsafe{
+		buf[26] = 0
+	}
+	return string(buf,buflen)
+}