Skip to main content

thoughts_tool/git/
ref_key.rs

1use anyhow::{Result, bail};
2use sha2::{Digest, Sha256};
3
4const MAX_REF_KEY_LEN: usize = 120;
5const HASH_HEX_LEN: usize = 16;
6const PREFIX: &str = "r-";
7
8pub fn encode_ref_key(ref_name: &str) -> Result<String> {
9    let ref_name = ref_name.trim();
10    if ref_name.is_empty() {
11        bail!("Reference name cannot be empty");
12    }
13    if ref_name.contains('\0') {
14        bail!("Reference name cannot contain NUL bytes");
15    }
16    if looks_like_raw_git_oid(ref_name) {
17        bail!("Raw commit SHAs are not supported; provide a named ref instead");
18    }
19
20    let mut encoded = String::with_capacity(ref_name.len() + PREFIX.len());
21    encoded.push_str(PREFIX);
22    for byte in ref_name.bytes() {
23        match byte {
24            b'a'..=b'z' | b'0'..=b'9' | b'-' | b'_' | b'.' => encoded.push(byte as char),
25            _ => push_escape(&mut encoded, byte),
26        }
27    }
28
29    if encoded.len() <= MAX_REF_KEY_LEN {
30        return Ok(encoded);
31    }
32
33    let hash = short_hash_hex(ref_name.as_bytes());
34    let suffix = format!("--{hash}");
35    let keep = MAX_REF_KEY_LEN
36        .checked_sub(PREFIX.len() + suffix.len())
37        .ok_or_else(|| anyhow::anyhow!("Invalid ref key length configuration"))?;
38
39    let mut truncated = String::with_capacity(MAX_REF_KEY_LEN);
40    truncated.push_str(PREFIX);
41    truncated.push_str(&truncate_on_char_boundary(&encoded[PREFIX.len()..], keep));
42    truncated.push_str(&suffix);
43    Ok(truncated)
44}
45
46fn looks_like_raw_git_oid(value: &str) -> bool {
47    matches!(value.len(), 40 | 64) && value.bytes().all(|b| b.is_ascii_hexdigit())
48}
49
50fn push_escape(out: &mut String, byte: u8) {
51    const HEX: &[u8; 16] = b"0123456789abcdef";
52    out.push('~');
53    out.push(HEX[(byte >> 4) as usize] as char);
54    out.push(HEX[(byte & 0x0f) as usize] as char);
55}
56
57fn short_hash_hex(bytes: &[u8]) -> String {
58    let digest = Sha256::digest(bytes);
59    digest[..(HASH_HEX_LEN / 2)]
60        .iter()
61        .map(|byte| format!("{byte:02x}"))
62        .collect()
63}
64
65fn truncate_on_char_boundary(value: &str, max_len: usize) -> String {
66    if value.len() <= max_len {
67        return value.to_string();
68    }
69
70    let mut end = max_len;
71    while !value.is_char_boundary(end) {
72        end -= 1;
73    }
74    value[..end].to_string()
75}
76
77#[cfg(test)]
78mod tests {
79    use super::{MAX_REF_KEY_LEN, encode_ref_key};
80
81    #[test]
82    fn encodes_safe_ascii_directly() {
83        assert_eq!(encode_ref_key("main").unwrap(), "r-main");
84        assert_eq!(encode_ref_key("release-1.2_3").unwrap(), "r-release-1.2_3");
85    }
86
87    #[test]
88    fn escapes_slashes_and_uppercase_for_case_safe_segments() {
89        assert_eq!(encode_ref_key("feature/foo").unwrap(), "r-feature~2ffoo");
90        assert_eq!(encode_ref_key("Main").unwrap(), "r-~4dain");
91    }
92
93    #[test]
94    fn rejects_empty_nul_and_raw_sha_values() {
95        assert!(encode_ref_key("   ").is_err());
96        assert!(encode_ref_key("abc\0def").is_err());
97        assert!(encode_ref_key("0123456789abcdef0123456789abcdef01234567").is_err());
98    }
99
100    #[test]
101    fn truncates_long_values_with_stable_structure() {
102        let input = "refs/heads/".to_string() + &"very-long-".repeat(40);
103        let out = encode_ref_key(&input).unwrap();
104
105        assert!(out.starts_with("r-"), "must retain r- prefix");
106        assert!(
107            out.len() <= MAX_REF_KEY_LEN,
108            "must be bounded by MAX_REF_KEY_LEN"
109        );
110
111        let (_prefix_part, hash) = out.rsplit_once("--").expect("expected --{hash} suffix");
112        assert_eq!(hash.len(), 16, "expected 16 hex chars");
113        assert!(hash.chars().all(|c| c.is_ascii_hexdigit()));
114    }
115
116    #[test]
117    fn truncation_hash_makes_long_keys_unique() {
118        let base = "refs/heads/".to_string() + &"a".repeat(300);
119        let a = format!("{}-x", base);
120        let b = format!("{}-y", base);
121
122        let ka = encode_ref_key(&a).unwrap();
123        let kb = encode_ref_key(&b).unwrap();
124        assert_ne!(ka, kb, "distinct long refs must not collide");
125    }
126}