alint_rules/
file_min_lines.rs

1//! `file_min_lines` — files in scope must have at least
2//! `min_lines` lines.
3//!
4//! Catches the "README is a title plus two sentences" case
5//! where the file exists, isn't empty, but is far too thin to
6//! actually document anything. Pairs well with `file_exists`
7//! on README / CHANGELOG / SECURITY.md in governance rulesets.
8//!
9//! A **line** is any run of bytes terminated by `\n`. The
10//! trailing segment after the last newline (or the whole file
11//! when there is no newline) counts as one additional line
12//! only when it is non-empty — so `"a\nb\n"` and `"a\nb"` both
13//! report 2 lines, while `"a\nb\n\n"` reports 3 (the empty
14//! line between the two newlines counts). This matches the
15//! usual `wc -l` semantics closely enough for policy use;
16//! pedantic counting differences aren't worth the surprise.
17
18use std::path::Path;
19
20use alint_core::{Context, Error, Level, PerFileRule, Result, Rule, RuleSpec, Scope, Violation};
21use serde::Deserialize;
22
23#[derive(Debug, Deserialize)]
24struct Options {
25    min_lines: u64,
26}
27
28#[derive(Debug)]
29pub struct FileMinLinesRule {
30    id: String,
31    level: Level,
32    policy_url: Option<String>,
33    message: Option<String>,
34    scope: Scope,
35    min_lines: u64,
36}
37
38impl Rule for FileMinLinesRule {
39    fn id(&self) -> &str {
40        &self.id
41    }
42    fn level(&self) -> Level {
43        self.level
44    }
45    fn policy_url(&self) -> Option<&str> {
46        self.policy_url.as_deref()
47    }
48
49    fn evaluate(&self, ctx: &Context<'_>) -> Result<Vec<Violation>> {
50        let mut violations = Vec::new();
51        for entry in ctx.index.files() {
52            if !self.scope.matches(&entry.path, ctx.index) {
53                continue;
54            }
55            let full = ctx.root.join(&entry.path);
56            let Ok(bytes) = std::fs::read(&full) else {
57                // Unreadable (permission, race with a remove, …)
58                // — mirror the rest of the content-family rules
59                // and skip silently rather than blowing up a
60                // whole check run.
61                continue;
62            };
63            violations.extend(self.evaluate_file(ctx, &entry.path, &bytes)?);
64        }
65        Ok(violations)
66    }
67
68    fn as_per_file(&self) -> Option<&dyn PerFileRule> {
69        Some(self)
70    }
71}
72
73impl PerFileRule for FileMinLinesRule {
74    fn path_scope(&self) -> &Scope {
75        &self.scope
76    }
77
78    fn evaluate_file(
79        &self,
80        _ctx: &Context<'_>,
81        path: &Path,
82        bytes: &[u8],
83    ) -> Result<Vec<Violation>> {
84        let lines = count_lines(bytes);
85        if lines >= self.min_lines {
86            return Ok(Vec::new());
87        }
88        let msg = self.message.clone().unwrap_or_else(|| {
89            format!(
90                "file has {} line(s); at least {} required",
91                lines, self.min_lines,
92            )
93        });
94        Ok(vec![
95            Violation::new(msg).with_path(std::sync::Arc::<Path>::from(path)),
96        ])
97    }
98}
99
100/// Count lines with `wc -l`-style semantics: every `\n` is a
101/// line terminator, plus one more line when the file doesn't
102/// end with `\n` but has content after the last `\n`. Empty
103/// file → 0 lines.
104fn count_lines(bytes: &[u8]) -> u64 {
105    if bytes.is_empty() {
106        return 0;
107    }
108    // `bytecount` would be faster, but line-count files are
109    // typically READMEs / CHANGELOGs (small). Not worth a
110    // dep for a hot loop that isn't.
111    #[allow(clippy::naive_bytecount)]
112    let newlines = bytes.iter().filter(|&&b| b == b'\n').count() as u64;
113    let trailing_unterminated = u64::from(!bytes.ends_with(b"\n"));
114    newlines + trailing_unterminated
115}
116
117pub fn build(spec: &RuleSpec) -> Result<Box<dyn Rule>> {
118    let Some(_paths) = &spec.paths else {
119        return Err(Error::rule_config(
120            &spec.id,
121            "file_min_lines requires a `paths` field",
122        ));
123    };
124    let opts: Options = spec
125        .deserialize_options()
126        .map_err(|e| Error::rule_config(&spec.id, format!("invalid options: {e}")))?;
127    Ok(Box::new(FileMinLinesRule {
128        id: spec.id.clone(),
129        level: spec.level,
130        policy_url: spec.policy_url.clone(),
131        message: spec.message.clone(),
132        scope: Scope::from_spec(spec)?,
133        min_lines: opts.min_lines,
134    }))
135}
136
137#[cfg(test)]
138mod tests {
139    use super::count_lines;
140
141    #[test]
142    fn empty_file_has_zero_lines() {
143        assert_eq!(count_lines(b""), 0);
144    }
145
146    #[test]
147    fn content_with_trailing_newline_counts_each_line() {
148        assert_eq!(count_lines(b"a\n"), 1);
149        assert_eq!(count_lines(b"a\nb\n"), 2);
150        assert_eq!(count_lines(b"a\nb\nc\n"), 3);
151    }
152
153    #[test]
154    fn content_without_trailing_newline_adds_one_for_tail() {
155        assert_eq!(count_lines(b"a"), 1);
156        assert_eq!(count_lines(b"a\nb"), 2);
157    }
158
159    #[test]
160    fn blank_lines_count() {
161        assert_eq!(count_lines(b"a\n\nb\n"), 3);
162        assert_eq!(count_lines(b"\n\n"), 2);
163    }
164}
alint_rules/file_min_lines.rs

alint_rules/
file_min_lines.rs
