amiss_wire/
human.rs

1pub const ATOM_SCALAR_BOUND: usize = 200;
2
3/// `human-atom-v1`: every repository-derived scalar is rendered as a
4/// double-quoted ASCII JSON-style string. At most the first two hundred
5/// Unicode scalar values are kept, with a literal `...` appended inside the
6/// quotes when any were omitted. Quote and backslash escape as `\"` and
7/// `\\`, printable ASCII stays literal, and every other scalar becomes a
8/// lowercase `\uXXXX` escape, non-BMP scalars as a UTF-16 surrogate pair, so
9/// CR, LF, tab, ESC, bidi controls, and ANSI bytes are never active terminal
10/// syntax.
11#[must_use]
12pub fn atom(text: &str) -> String {
13    let mut out = String::with_capacity(text.len().saturating_add(2));
14    out.push('"');
15    let mut omitted = false;
16    for (index, scalar) in text.chars().enumerate() {
17        if index >= ATOM_SCALAR_BOUND {
18            omitted = true;
19            break;
20        }
21        match scalar {
22            '"' => out.push_str("\\\""),
23            '\\' => out.push_str("\\\\"),
24            ' '..='~' => out.push(scalar),
25            _ => {
26                let mut units = [0_u16; 2];
27                for unit in scalar.encode_utf16(&mut units) {
28                    out.push_str("\\u");
29                    for shift in [12_u32, 8, 4, 0] {
30                        let nibble = (u32::from(*unit) >> shift) & 0xf;
31                        out.push(char::from_digit(nibble, 16).unwrap_or('0'));
32                    }
33                }
34            }
35        }
36    }
37    if omitted {
38        out.push_str("...");
39    }
40    out.push('"');
41    out
42}
43
44/// `human-atom-bytes-v1`: a repository-derived value that is raw bytes, not
45/// text, rendered under the same law as [`atom`]. At most the first two
46/// hundred bytes are kept with a literal `...` appended inside the quotes
47/// when any were omitted; printable ASCII stays literal with quote and
48/// backslash escaped, and every other byte becomes the lowercase `\u00xx`
49/// escape of its value, so no byte is ever active terminal syntax and no
50/// Unicode scalar is invented for bytes that never encoded one.
51#[must_use]
52pub fn atom_bytes(bytes: &[u8]) -> String {
53    let mut out = String::with_capacity(bytes.len().saturating_add(2));
54    out.push('"');
55    let mut omitted = false;
56    for (index, byte) in bytes.iter().enumerate() {
57        if index >= ATOM_SCALAR_BOUND {
58            omitted = true;
59            break;
60        }
61        match byte {
62            b'"' => out.push_str("\\\""),
63            b'\\' => out.push_str("\\\\"),
64            b' '..=b'~' => out.push(char::from(*byte)),
65            _ => {
66                out.push_str("\\u00");
67                for shift in [4_u32, 0] {
68                    let nibble = (u32::from(*byte) >> shift) & 0xf;
69                    out.push(char::from_digit(nibble, 16).unwrap_or('0'));
70                }
71            }
72        }
73    }
74    if omitted {
75        out.push_str("...");
76    }
77    out.push('"');
78    out
79}
amiss_wire/human.rs

amiss_wire/
human.rs
