Function search_content

atomic-repository/src/content_search.rs:153–239

Search the content index. Returns matching lines with file paths, line numbers, and content. Supports path filtering, file type filtering, and case-insensitive search. Results are ranked: source files (`src/`, `lib/`, `pkg/`, `internal/`) are prioritized over config, test, build, and generated files. Within each tier, results preserve filesystem order from the index. The returned [`ContentSearc

(
    repo_root: &Path,
    pattern: &str,
    opts: ContentSearchOptions,
)

/// (capped to `max_results`) and the `total_matches` count so callers
/// know how many were found before truncation.
pub fn search_content(
    repo_root: &Path,
    pattern: &str,
    opts: ContentSearchOptions,
) -> Result<ContentSearchResult, ContentSearchError> {
    let config = content_config(repo_root);
    let index = Index::open(config)?;

    let display_limit = opts.max_results.unwrap_or(50);

    // Fetch a large candidate set so we can rank before truncating.
    // We ask for 10x the display limit (capped at 2000) to get good
    // coverage across the repo before applying our own ranking.
    let fetch_limit = (display_limit * 10).min(2000);

    let search_opts = SearchOptions {
        path_filter: opts.path_filter,
        file_type: opts.file_type,
        exclude_type: opts.exclude_type,
        max_results: Some(fetch_limit),
        case_insensitive: opts.case_insensitive,
    };

    let raw_matches = index.search(pattern, &search_opts)?;

    // Drop matches in ignored or Atomic-internal paths. syntext's index walk
    // respects `.gitignore`/`.ignore` but NOT `.atomicignore`, and it descends
    // into hidden dirs like `.atomic`/`.vault`, so build artifacts and
    // dependencies excluded only by `.atomicignore` (plus Atomic internals)
    // would otherwise surface here and, via the KG search's content-only file
    // promotion, as `file:` results. Filter at this shared consumption point
    // so both `query search` and `query code` stay clean regardless of what
    // the index contains.
    let ignore_rules = crate::ignore::IgnoreRules::load_for_enrichment(repo_root);
    let raw_matches: Vec<syntext::SearchMatch> = raw_matches
        .into_iter()
        .filter(|m| {
            !crate::ignore::is_enrichment_internal(&m.path)
                && !ignore_rules.is_ignored(&m.path, false)
        })
        .collect();
    let total_matches = raw_matches.len();

    // Build per-directory match counts for the facets summary.
    let mut dir_counts: std::collections::HashMap<String, usize> = std::collections::HashMap::new();
    for m in &raw_matches {
        let path_str = m.path.to_string_lossy();
        // Use the first two path components as the directory bucket.
        let dir = path_str
            .splitn(3, '/')
            .take(2)
            .collect::<Vec<_>>()
            .join("/");
        *dir_counts.entry(dir).or_insert(0) += 1;
    }

    // Sort directory facets by count (descending).
    let mut dir_facets: Vec<(String, usize)> = dir_counts.into_iter().collect();