search.rs 86.5%

45 of 52 lines run, 4 of 8 functions · run #510 on main · run missed

//! Code search (hub spec §11): substring or regular expression over the
//! text files of one state, with a time budget and a result cap. Results
//! are cached by state hash and query, since a state never changes.

use super::cache::Loaded;
use oxy_core::state::FileContent;
use std::collections::HashMap;
use std::sync::Mutex;
use std::time::{Duration, Instant};

#[derive(Clone, Debug, PartialEq, Eq)]
pub struct Hit {
    pub path: String,
    /// 1-based; 0 for a match on the path itself.
    pub line: usize,
    pub text: String,
}

#[derive(Clone, Debug, Default)]
pub struct Results {
    pub hits: Vec<Hit>,
    pub files: usize,
    /// The budget or the cap stopped the search early.
    pub truncated: bool,
}

pub struct Query<'a> {
    pub text: &'a str,
    pub regex: bool,
    pub case_sensitive: bool,
    /// Only paths starting with this.
    pub path_prefix: &'a str,
}

pub const MAX_HITS: usize = 300;
pub const BUDGET: Duration = Duration::from_secs(2);

pub fn run(loaded: &Loaded, q: &Query<'_>) -> Result<Results, String> {
    let pattern = if q.regex { q.text.to_string() } else { regex::escape(q.text) };
    let re = regex::RegexBuilder::new(&pattern)
        .case_insensitive(!q.case_sensitive)
        .size_limit(1 << 20)
        .build()
        .map_err(|e| format!("not a valid regular expression: {e}"))?;
    let started = Instant::now();
    let mut out = Results::default();
    for (path, f) in &loaded.snapshot.files {
        if !path.starts_with(q.path_prefix) {
            continue;
        }
        if started.elapsed() > BUDGET || out.hits.len() >= MAX_HITS {
            out.truncated = true;
            break;
        }
        let mut found = false;
        if re.is_match(path) {
            out.hits.push(Hit { path: path.clone(), line: 0, text: String::new() });
            found = true;
        }
        let FileContent::Text { bytes, .. } = &f.content else { continue };
        if bytes.len() > 1 << 20 || bytes.contains(&0) {
            continue;
        }
        let text = String::from_utf8_lossy(bytes);
        for (i, line) in text.lines().enumerate() {
            if re.is_match(line) {
                out.hits.push(Hit { path: path.clone(), line: i + 1, text: line.chars().take(300).collect() });
                found = true;
                if out.hits.len() >= MAX_HITS {
                    out.truncated = true;
                    break;
                }
            }
        }
        if found {
            out.files += 1;
        }
    }
    Ok(out)
}

/// A small cache of results by (state, query).
#[derive(Default)]
pub struct SearchCache {
    inner: Mutex<HashMap<String, Results>>,
}

impl SearchCache {
    pub fn get(&self, key: &str) -> Option<Results> {
        self.inner.lock().unwrap().get(key).cloned()
    }

    pub fn put(&self, key: String, r: Results) {
        let mut m = self.inner.lock().unwrap();
        if m.len() > 256 {
            m.clear();
        }
        m.insert(key, r);
    }
}

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	4×	pub fn run(loaded: &Loaded, q: &Query<'_>) -> Result<Results, String> {
39	4×	let pattern = if q.regex { q.text.to_string() } else { regex::escape(q.text) };
40	4×	let re = regex::RegexBuilder::new(&pattern)
41	4×	.case_insensitive(!q.case_sensitive)
42	4×	.size_limit(1 << 20)
43	4×	.build()
44	4×	.map_err(\|e\| format!("not a valid regular expression: {e}"))?;
45	3×	let started = Instant::now();
46	3×	let mut out = Results::default();
47	12×	for (path, f) in &loaded.snapshot.files {
48	12×	if !path.starts_with(q.path_prefix) {
49	0×	continue;
50	12×	}
51	12×	if started.elapsed() > BUDGET \|\| out.hits.len() >= MAX_HITS {
52	0×	out.truncated = true;
53	0×	break;
54	12×	}
55	12×	let mut found = false;
56	12×	if re.is_match(path) {
57	1×	out.hits.push(Hit { path: path.clone(), line: 0, text: String::new() });
58	1×	found = true;
59	11×	}
60	12×	let FileContent::Text { bytes, .. } = &f.content else { continue };
61	9×	if bytes.len() > 1 << 20 \|\| bytes.contains(&0) {
62	0×	continue;
63	9×	}
64	9×	let text = String::from_utf8_lossy(bytes);
65	18×	for (i, line) in text.lines().enumerate() {
66	18×	if re.is_match(line) {
67	2×	out.hits.push(Hit { path: path.clone(), line: i + 1, text: line.chars().take(300).collect() });
68	2×	found = true;
69	2×	if out.hits.len() >= MAX_HITS {
70	0×	out.truncated = true;
71	0×	break;
72	2×	}
73	16×	}
74		}
75	9×	if found {
76	3×	out.files += 1;
77	6×	}
78		}
79	3×	Ok(out)
80	4×	}
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	4×	pub fn get(&self, key: &str) -> Option<Results> {
90	4×	self.inner.lock().unwrap().get(key).cloned()
91	4×	}
92
93	3×	pub fn put(&self, key: String, r: Results) {
94	3×	let mut m = self.inner.lock().unwrap();
95	3×	if m.len() > 256 {
96	0×	m.clear();
97	3×	}
98	3×	m.insert(key, r);
99	3×	}
100		}

`/`	Search
`g` `c`	Code
`g` `h`	History
`g` `b`	Bookmarks
`g` `t`	Trust
`y`	Permalink to this state (files and trees)
`?`	This sheet

Coverage / crates/hub/src/repos/search.rs 86.5%

Keyboard