1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
//! Accurate token counting using tiktoken.
//!
//! Provides precise token counting for messages to enable better
//! context window management.
use once_cell::sync::Lazy;
use std::sync::Arc;
use tiktoken_rs::CoreBPE;
/// Global BPE encoder for token counting (cl100k_base for GPT-4/Claude).
static BPE: Lazy<Arc<CoreBPE>> = Lazy::new(|| {
Arc::new(tiktoken_rs::cl100k_base().expect("Failed to initialize tokenizer"))
});
/// Count tokens in a text string.
pub fn count_tokens(text: &str) -> u32 {
BPE.encode_with_special_tokens(text).len() as u32
}
/// Count tokens for a role prefix (e.g., "user: ", "assistant: ").
/// Each message has overhead for role markers and formatting.
pub fn message_overhead() -> u32 {
// Approximate overhead per message:
// - Role prefix: ~4 tokens
// - Message separators: ~2 tokens
// - Total overhead: ~6 tokens
6
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_count_tokens_simple() {
let text = "Hello, world!";
let count = count_tokens(text);
assert!(count > 0);
// "Hello, world!" is typically 4 tokens
assert!(count >= 3 && count <= 5);
}
#[test]
fn test_count_tokens_chinese() {
let text = "你好，世界！";
let count = count_tokens(text);
assert!(count > 0);
// Chinese characters typically use more tokens
assert!(count >= 5);
}
#[test]
fn test_count_tokens_code() {
let code = r#"
fn main() {
println!("Hello");
}
"#;
let count = count_tokens(code);
assert!(count > 0);
// Code typically uses more tokens due to symbols
// Actual count is around 13 tokens
assert!(count >= 10, "Code should use at least 10 tokens, got {}", count);
}
#[test]
fn test_message_overhead() {
let overhead = message_overhead();
assert_eq!(overhead, 6);
}
#[test]
fn test_token_counting_accuracy() {
// Compare with known token counts
// The phrase "Hello, world!" is 4 tokens in cl100k_base
assert_eq!(count_tokens("Hello, world!"), 4);
// Single word
assert_eq!(count_tokens("Hello"), 1);
// Numbers
assert_eq!(count_tokens("12345"), 2); // "123" + "45"
// Chinese characters (each typically 1-2 tokens)
let chinese = "你好世界";
let chinese_count = count_tokens(chinese);
assert!(chinese_count >= 4, "Chinese text should use at least 4 tokens, got {}", chinese_count);
// Empty string
assert_eq!(count_tokens(""), 0);
// Whitespace
assert_eq!(count_tokens(" "), 1);
}
}