OXYGENHUB
◆ v0.9.8 oxygenhub / crates / hub / src / repos / search.rs
2.9 KiB · Rust · last changed Milestone 0.6: releases, Git mirror, code search, anchored API comments 1 month ago source blame history raw
1 
//! Code search (hub spec §11): substring or regular expression over the
2 
//! text files of one state, with a time budget and a result cap. Results
3 
//! are cached by state hash and query, since a state never changes.
4
5 
use super::cache::Loaded;
6 
use oxy_core::state::FileContent;
7 
use std::collections::HashMap;
8 
use std::sync::Mutex;
9 
use std::time::{Duration, Instant};
10
11 
#[derive(Clone, Debug, PartialEq, Eq)]
12 
pub struct Hit {
13 
    pub path: String,
14 
    /// 1-based; 0 for a match on the path itself.
15 
    pub line: usize,
16 
    pub text: String,
17 
}
18
19 
#[derive(Clone, Debug, Default)]
20 
pub struct Results {
21 
    pub hits: Vec<Hit>,
22 
    pub files: usize,
23 
    /// The budget or the cap stopped the search early.
24 
    pub truncated: bool,
25 
}
26
27 
pub struct Query<'a> {
28 
    pub text: &'a str,
29 
    pub regex: bool,
30 
    pub case_sensitive: bool,
31 
    /// Only paths starting with this.
32 
    pub path_prefix: &'a str,
33 
}
34
35 
pub const MAX_HITS: usize = 300;
36 
pub const BUDGET: Duration = Duration::from_secs(2);
37
38 
pub fn run(loaded: &Loaded, q: &Query<'_>) -> Result<Results, String> {
39 
    let pattern = if q.regex { q.text.to_string() } else { regex::escape(q.text) };
40 
    let re = regex::RegexBuilder::new(&pattern)
41 
        .case_insensitive(!q.case_sensitive)
42 
        .size_limit(1 << 20)
43 
        .build()
44 
        .map_err(|e| format!("not a valid regular expression: {e}"))?;
45 
    let started = Instant::now();
46 
    let mut out = Results::default();
47 
    for (path, f) in &loaded.snapshot.files {
48 
        if !path.starts_with(q.path_prefix) {
49 
            continue;
50 
        }
51 
        if started.elapsed() > BUDGET || out.hits.len() >= MAX_HITS {
52 
            out.truncated = true;
53 
            break;
54 
        }
55 
        let mut found = false;
56 
        if re.is_match(path) {
57 
            out.hits.push(Hit { path: path.clone(), line: 0, text: String::new() });
58 
            found = true;
59 
        }
60 
        let FileContent::Text { bytes, .. } = &f.content else { continue };
61 
        if bytes.len() > 1 << 20 || bytes.contains(&0) {
62 
            continue;
63 
        }
64 
        let text = String::from_utf8_lossy(bytes);
65 
        for (i, line) in text.lines().enumerate() {
66 
            if re.is_match(line) {
67 
                out.hits.push(Hit { path: path.clone(), line: i + 1, text: line.chars().take(300).collect() });
68 
                found = true;
69 
                if out.hits.len() >= MAX_HITS {
70 
                    out.truncated = true;
71 
                    break;
72 
                }
73 
            }
74 
        }
75 
        if found {
76 
            out.files += 1;
77 
        }
78 
    }
79 
    Ok(out)
80 
}
81
82 
/// A small cache of results by (state, query).
83 
#[derive(Default)]
84 
pub struct SearchCache {
85 
    inner: Mutex<HashMap<String, Results>>,
86 
}
87
88 
impl SearchCache {
89 
    pub fn get(&self, key: &str) -> Option<Results> {
90 
        self.inner.lock().unwrap().get(key).cloned()
91 
    }
92
93 
    pub fn put(&self, key: String, r: Results) {
94 
        let mut m = self.inner.lock().unwrap();
95 
        if m.len() > 256 {
96 
            m.clear();
97 
        }
98 
        m.insert(key, r);
99 
    }
100 
}