_diffctx/
candidate_files.rs

1use std::path::{Path, PathBuf};
2
3use rayon::prelude::*;
4use rustc_hash::FxHashSet;
5use walkdir::WalkDir;
6
7use crate::config::graph_filtering::GRAPH_FILTERING;
8use crate::config::limits::LIMITS;
9use crate::git;
10use crate::languages::get_language_for_file;
11
12fn is_allowed_file(path: &Path) -> bool {
13    get_language_for_file(&path.to_string_lossy()).is_some()
14}
15
16fn is_candidate_file(
17    file_path: &Path,
18    _root_dir: &Path,
19    included_set: &FxHashSet<PathBuf>,
20) -> bool {
21    if !file_path.is_file() {
22        return false;
23    }
24    if !is_allowed_file(file_path) {
25        return false;
26    }
27    if included_set.contains(file_path) {
28        return false;
29    }
30    match file_path.metadata() {
31        Ok(meta) if meta.len() as usize > LIMITS.max_file_size => return false,
32        Err(_) => return false,
33        _ => {}
34    }
35    true
36}
37
38pub fn collect_candidate_files(root_dir: &Path, included_set: &FxHashSet<PathBuf>) -> Vec<PathBuf> {
39    if let Ok(parts) = git::run_git_z(root_dir, &["ls-files", "-z"]) {
40        let all_paths: Vec<PathBuf> = parts.into_iter().map(|f| root_dir.join(f)).collect();
41        let files: Vec<PathBuf> = all_paths
42            .into_par_iter()
43            .filter(|f| is_candidate_file(f, root_dir, included_set))
44            .collect();
45        return filter_ignored_and_secret(root_dir, files);
46    }
47
48    let mut fallback: Vec<PathBuf> = Vec::new();
49    for entry in WalkDir::new(root_dir)
50        .sort_by_file_name()
51        .into_iter()
52        .filter_entry(|e| {
53            if e.depth() == 0 || !e.file_type().is_dir() {
54                return true;
55            }
56            match e.file_name().to_str() {
57                Some(name) => {
58                    !name.starts_with('.') && name != "node_modules" && name != "__pycache__"
59                }
60                None => true,
61            }
62        })
63        .filter_map(|e| e.ok())
64    {
65        if !entry.file_type().is_file() {
66            continue;
67        }
68        if fallback.len() >= GRAPH_FILTERING.fallback_max_files {
69            break;
70        }
71        let path = entry.into_path();
72        if is_candidate_file(&path, root_dir, included_set) {
73            fallback.push(path);
74        }
75    }
76    filter_ignored_and_secret(root_dir, fallback)
77}
78
79/// Discovery's candidate universe otherwise skips straight from a language
80/// check to the graph: an unchanged file a changed file imports would render
81/// as neighbour context even when `.diffctx/ignore` explicitly excludes it,
82/// or when it is secret-like (`id_rsa`, `*.pem`, ...) — `changed_files` is
83/// filtered this way already (`pipeline::compute_scored_state`), the
84/// discovery universe was not. One batched `git check-ignore` call covers
85/// every survivor of the language filter, so cost stays O(1) subprocess
86/// invocations regardless of repo size, and the final filter still runs in
87/// parallel via rayon.
88fn filter_ignored_and_secret(root_dir: &Path, files: Vec<PathBuf>) -> Vec<PathBuf> {
89    let rel_paths: Vec<String> = files
90        .iter()
91        .filter_map(|f| crate::pipeline::rel_path_string(root_dir, f))
92        .collect();
93    let ignored_rel_paths = git::find_ignored_paths(root_dir, &rel_paths);
94    files
95        .into_par_iter()
96        .filter(|f| {
97            !crate::pipeline::is_secret_path(f)
98                && !crate::pipeline::is_ignored_path(root_dir, f, &ignored_rel_paths)
99        })
100        .collect()
101}
102
103pub fn normalize_path(path: &Path, root_dir: &Path) -> PathBuf {
104    if path.is_absolute() {
105        path.canonicalize().unwrap_or_else(|_| path.to_path_buf())
106    } else {
107        let joined = root_dir.join(path);
108        joined.canonicalize().unwrap_or(joined)
109    }
110}
_diffctx/candidate_files.rs

_diffctx/
candidate_files.rs
