alint_rules/
line_max_width.rs

1//! `line_max_width` — cap on characters per line.
2//!
3//! Counts Unicode scalar values (chars) per line, not bytes or
4//! display cells. CJK, combining marks, and emoji that occupy
5//! two terminal columns count as one char — if you want real
6//! display-width accounting, use a formatter (Biome, prettier);
7//! that's out of alint's byte/structure scope.
8//!
9//! Check-only: truncation isn't a safe auto-fix. Users either
10//! refactor the line or widen the limit.
11
12use std::path::Path;
13
14use alint_core::{Context, Error, Level, PerFileRule, Result, Rule, RuleSpec, Scope, Violation};
15use serde::Deserialize;
16
17#[derive(Debug, Deserialize)]
18#[serde(deny_unknown_fields)]
19struct Options {
20    max_width: usize,
21}
22
23#[derive(Debug)]
24pub struct LineMaxWidthRule {
25    id: String,
26    level: Level,
27    policy_url: Option<String>,
28    message: Option<String>,
29    scope: Scope,
30    max_width: usize,
31}
32
33impl Rule for LineMaxWidthRule {
34    fn id(&self) -> &str {
35        &self.id
36    }
37    fn level(&self) -> Level {
38        self.level
39    }
40    fn policy_url(&self) -> Option<&str> {
41        self.policy_url.as_deref()
42    }
43
44    fn evaluate(&self, ctx: &Context<'_>) -> Result<Vec<Violation>> {
45        let mut violations = Vec::new();
46        for entry in ctx.index.files() {
47            if !self.scope.matches(&entry.path, ctx.index) {
48                continue;
49            }
50            let full = ctx.root.join(&entry.path);
51            let Ok(bytes) = std::fs::read(&full) else {
52                continue;
53            };
54            violations.extend(self.evaluate_file(ctx, &entry.path, &bytes)?);
55        }
56        Ok(violations)
57    }
58
59    fn as_per_file(&self) -> Option<&dyn PerFileRule> {
60        Some(self)
61    }
62}
63
64impl PerFileRule for LineMaxWidthRule {
65    fn path_scope(&self) -> &Scope {
66        &self.scope
67    }
68
69    fn evaluate_file(
70        &self,
71        _ctx: &Context<'_>,
72        path: &Path,
73        bytes: &[u8],
74    ) -> Result<Vec<Violation>> {
75        // `chars().count()` requires a UTF-8-validated `&str` —
76        // line widths count Unicode scalars, not bytes. Non-UTF-8
77        // files silently skip, matching the rule-major path.
78        let Ok(text) = std::str::from_utf8(bytes) else {
79            return Ok(Vec::new());
80        };
81        let Some((line_no, width)) = first_overlong_line(text, self.max_width) else {
82            return Ok(Vec::new());
83        };
84        let msg = self.message.clone().unwrap_or_else(|| {
85            format!(
86                "line {line_no} is {width} chars wide; max is {}",
87                self.max_width
88            )
89        });
90        Ok(vec![
91            Violation::new(msg)
92                .with_path(std::sync::Arc::<Path>::from(path))
93                .with_location(line_no, self.max_width + 1),
94        ])
95    }
96}
97
98fn first_overlong_line(text: &str, max_width: usize) -> Option<(usize, usize)> {
99    for (idx, line) in text.split('\n').enumerate() {
100        let trimmed = line.strip_suffix('\r').unwrap_or(line);
101        let width = trimmed.chars().count();
102        if width > max_width {
103            return Some((idx + 1, width));
104        }
105    }
106    None
107}
108
109pub fn build(spec: &RuleSpec) -> Result<Box<dyn Rule>> {
110    let _paths = spec
111        .paths
112        .as_ref()
113        .ok_or_else(|| Error::rule_config(&spec.id, "line_max_width requires a `paths` field"))?;
114    let opts: Options = spec
115        .deserialize_options()
116        .map_err(|e| Error::rule_config(&spec.id, format!("invalid options: {e}")))?;
117    if opts.max_width == 0 {
118        return Err(Error::rule_config(
119            &spec.id,
120            "line_max_width `max_width` must be > 0",
121        ));
122    }
123    if spec.fix.is_some() {
124        return Err(Error::rule_config(
125            &spec.id,
126            "line_max_width has no fix op — truncation is unsafe",
127        ));
128    }
129    Ok(Box::new(LineMaxWidthRule {
130        id: spec.id.clone(),
131        level: spec.level,
132        policy_url: spec.policy_url.clone(),
133        message: spec.message.clone(),
134        scope: Scope::from_spec(spec)?,
135        max_width: opts.max_width,
136    }))
137}
138
139#[cfg(test)]
140mod tests {
141    use super::*;
142
143    #[test]
144    fn short_file_is_ok() {
145        assert_eq!(first_overlong_line("hi\nthere\n", 10), None);
146    }
147
148    #[test]
149    fn flags_first_overlong_line() {
150        let txt = "short\nway too looooong for ten\nok\n";
151        // "way too looooong for ten" is 24 chars.
152        assert_eq!(first_overlong_line(txt, 10), Some((2, 24)));
153    }
154
155    #[test]
156    fn width_exactly_at_limit_is_ok() {
157        assert_eq!(first_overlong_line("0123456789\n", 10), None);
158    }
159
160    #[test]
161    fn crlf_is_stripped_before_counting() {
162        // "hi\r\n" should count as 2 chars ("hi"), not 3.
163        assert_eq!(first_overlong_line("hi\r\n", 2), None);
164    }
165
166    #[test]
167    fn counts_unicode_scalar_values_not_bytes() {
168        // "☃☃☃" is 3 scalars / 9 bytes. Under `max_width: 3` it's fine.
169        assert_eq!(first_overlong_line("☃☃☃\n", 3), None);
170        // Under max_width: 2 it's flagged.
171        assert_eq!(first_overlong_line("☃☃☃\n", 2), Some((1, 3)));
172    }
173}
alint_rules/line_max_width.rs

alint_rules/
line_max_width.rs
