huskies/server/src/chat/commands/loc.rs

//! Handler for the `loc` command — top source files by line count.

use super::CommandContext;
use walkdir::WalkDir;

const DEFAULT_TOP_N: usize = 10;

/// Directories to skip during traversal.
const SKIP_DIRS: &[&str] = &[
    "target",
    "node_modules",
    ".git",
    "dist",
    "build",
    ".next",
    "coverage",
    "test-results",
];

/// Path components that indicate a worktree path that should be skipped.
const SKIP_PATH_COMPONENTS: &[&str] = &[".huskies/worktrees"];

/// Known-huge or machine-generated files that are excluded from the loc count
/// even when they have a recognised source extension (e.g. `.json`, `.yaml`).
/// Add entries here to extend the exclusion list.
const EXCLUDED_FILENAMES: &[&str] = &[
    "package-lock.json",
    "yarn.lock",
    "pnpm-lock.yaml",
    "bun.lockb",
    "Cargo.lock",
    "composer.lock",
    "Gemfile.lock",
    "poetry.lock",
    "go.sum",
    "go.work.sum",
    "flake.lock",
];

pub(super) fn handle_loc(ctx: &CommandContext) -> Option<String> {
    let args = ctx.args.trim();

    if args.is_empty() {
        return Some(loc_top_n(ctx.project_root, DEFAULT_TOP_N));
    }

    let first_token = args.split_whitespace().next().unwrap_or("");
    Some(match first_token.parse::<usize>() {
        Ok(0) => format!(
            "Usage: `loc [N]` or `loc <filepath>` — show top N source files by line count (default {DEFAULT_TOP_N}), or line count for a specific file"
        ),
        Ok(n) => loc_top_n(ctx.project_root, n),
        Err(_) => loc_single_file(ctx.project_root, args),
    })
}

/// Count lines in a single file resolved relative to `project_root`.
pub(crate) fn loc_single_file(project_root: &std::path::Path, file_arg: &str) -> String {
    let path = if std::path::Path::new(file_arg).is_absolute() {
        std::path::PathBuf::from(file_arg)
    } else {
        project_root.join(file_arg)
    };

    match std::fs::read_to_string(&path) {
        Ok(content) => {
            let lines = content.lines().count();
            let display = path
                .strip_prefix(project_root)
                .unwrap_or(&path)
                .to_string_lossy();
            format!("`{display}` — {lines} lines")
        }
        Err(e) if e.kind() == std::io::ErrorKind::NotFound => {
            format!("File not found: `{file_arg}`")
        }
        Err(e) => format!("Error reading `{file_arg}`: {e}"),
    }
}

fn loc_top_n(project_root: &std::path::Path, top_n: usize) -> String {
    let mut files: Vec<(usize, String)> = WalkDir::new(project_root)
        .follow_links(false)
        .into_iter()
        .filter_entry(|e| {
            if e.file_type().is_dir() {
                let name = e.file_name().to_string_lossy();
                if SKIP_DIRS.iter().any(|s| *s == name.as_ref()) {
                    return false;
                }
                // Skip .huskies/worktrees — use relative path so the check
                // doesn't exclude the project root itself when running
                // from inside a worktree (where the absolute path contains
                // ".huskies/worktrees").
                let rel = e
                    .path()
                    .strip_prefix(project_root)
                    .map(|p| p.to_string_lossy().into_owned())
                    .unwrap_or_default();
                if SKIP_PATH_COMPONENTS.iter().any(|s| rel.contains(s)) {
                    return false;
                }
            }
            true
        })
        .filter_map(|entry| {
            let entry = entry.ok()?;
            if !entry.file_type().is_file() {
                return None;
            }
            let path = entry.path();
            // Skip known-huge or machine-generated files (lockfiles, etc.).
            let filename = path.file_name().and_then(|f| f.to_str()).unwrap_or("");
            if EXCLUDED_FILENAMES.contains(&filename) {
                return None;
            }
            // Skip binary/generated files without a recognisable text extension.
            let ext = path.extension().and_then(|e| e.to_str()).unwrap_or("");
            if !is_source_extension(ext) {
                return None;
            }
            let content = std::fs::read_to_string(path).ok()?;
            let line_count = content.lines().count();
            if line_count == 0 {
                return None;
            }
            // Make path relative to project_root for display.
            let rel = path
                .strip_prefix(project_root)
                .unwrap_or(path)
                .to_string_lossy()
                .into_owned();
            Some((line_count, rel))
        })
        .collect();

    files.sort_by(|a, b| b.0.cmp(&a.0));
    files.truncate(top_n);

    if files.is_empty() {
        return "No source files found.".to_string();
    }

    let mut out = format!("**Top {} files by line count**\n\n", files.len());
    for (rank, (lines, path)) in files.iter().enumerate() {
        out.push_str(&format!("{}. `{}` — {} lines\n", rank + 1, path, lines));
    }
    out
}

/// Returns true for file extensions considered source/text files.
fn is_source_extension(ext: &str) -> bool {
    matches!(
        ext,
        "rs" | "ts" | "tsx" | "js" | "jsx" | "py" | "go" | "java" | "c" | "cpp" | "h"
            | "hpp" | "cs" | "rb" | "swift" | "kt" | "scala" | "hs" | "ml" | "ex" | "exs"
            | "clj" | "lua" | "sh" | "bash" | "zsh" | "fish" | "ps1" | "toml" | "yaml"
            | "yml" | "json" | "md" | "html" | "css" | "scss" | "less" | "sql" | "graphql"
            | "proto" | "tf" | "hcl" | "nix" | "r" | "jl" | "dart" | "vue" | "svelte"
    )
}

// ---------------------------------------------------------------------------
// Tests
// ---------------------------------------------------------------------------

#[cfg(test)]
mod tests {
    use super::*;
    use crate::agents::AgentPool;
    use std::collections::HashSet;
    use std::sync::{Arc, Mutex};

    fn make_ctx<'a>(
        agents: &'a Arc<AgentPool>,
        ambient_rooms: &'a Arc<Mutex<HashSet<String>>>,
        project_root: &'a std::path::Path,
        args: &'a str,
    ) -> super::super::CommandContext<'a> {
        super::super::CommandContext {
            bot_name: "Timmy",
            args,
            project_root,
            agents,
            ambient_rooms,
            room_id: "!test:example.com",
        }
    }

    #[test]
    fn loc_command_is_registered() {
        use super::super::commands;
        let found = commands().iter().any(|c| c.name == "loc");
        assert!(found, "loc command must be in the registry");
    }

    #[test]
    fn loc_command_appears_in_help() {
        let result = super::super::tests::try_cmd_addressed(
            "Timmy",
            "@timmy:homeserver.local",
            "@timmy help",
        );
        let output = result.unwrap();
        assert!(output.contains("loc"), "help should list loc command: {output}");
    }

    #[test]
    fn loc_default_returns_top_10() {
        let agents = Arc::new(AgentPool::new_test(3000));
        let ambient_rooms = Arc::new(Mutex::new(HashSet::new()));
        let repo_root = std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
            .parent()
            .unwrap_or(std::path::Path::new("."));
        let ctx = make_ctx(&agents, &ambient_rooms, repo_root, "");
        let output = handle_loc(&ctx).unwrap();
        assert!(
            output.contains("Top"),
            "output should contain 'Top': {output}"
        );
        // At most 10 entries (numbered lines "1." through "10.")
        let count = output.lines().filter(|l| l.contains(". `")).count();
        assert!(count <= 10, "default should return at most 10 files, got {count}");
    }

    #[test]
    fn loc_with_arg_5_returns_at_most_5() {
        let agents = Arc::new(AgentPool::new_test(3000));
        let ambient_rooms = Arc::new(Mutex::new(HashSet::new()));
        let repo_root = std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
            .parent()
            .unwrap_or(std::path::Path::new("."));
        let ctx = make_ctx(&agents, &ambient_rooms, repo_root, "5");
        let output = handle_loc(&ctx).unwrap();
        let count = output.lines().filter(|l| l.contains(". `")).count();
        assert!(count <= 5, "loc 5 should return at most 5 files, got {count}");
    }

    #[test]
    fn loc_with_arg_20_returns_at_most_20() {
        let agents = Arc::new(AgentPool::new_test(3000));
        let ambient_rooms = Arc::new(Mutex::new(HashSet::new()));
        let repo_root = std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
            .parent()
            .unwrap_or(std::path::Path::new("."));
        let ctx = make_ctx(&agents, &ambient_rooms, repo_root, "20");
        let output = handle_loc(&ctx).unwrap();
        let count = output.lines().filter(|l| l.contains(". `")).count();
        assert!(count <= 20, "loc 20 should return at most 20 files, got {count}");
    }

    #[test]
    fn loc_output_contains_rank_and_line_count() {
        let agents = Arc::new(AgentPool::new_test(3000));
        let ambient_rooms = Arc::new(Mutex::new(HashSet::new()));
        let repo_root = std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
            .parent()
            .unwrap_or(std::path::Path::new("."));
        let ctx = make_ctx(&agents, &ambient_rooms, repo_root, "");
        let output = handle_loc(&ctx).unwrap();
        // Each entry should have "N. `path` — N lines"
        assert!(
            output.contains("1. `"),
            "first result should start with rank: {output}"
        );
        assert!(
            output.contains("lines"),
            "output should mention 'lines': {output}"
        );
    }

    #[test]
    fn loc_zero_arg_returns_usage() {
        let agents = Arc::new(AgentPool::new_test(3000));
        let ambient_rooms = Arc::new(Mutex::new(HashSet::new()));
        let repo_root = std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
            .parent()
            .unwrap_or(std::path::Path::new("."));
        let ctx = make_ctx(&agents, &ambient_rooms, repo_root, "0");
        let output = handle_loc(&ctx).unwrap();
        assert!(
            output.contains("Usage"),
            "loc 0 should show usage: {output}"
        );
    }

    #[test]
    fn loc_filepath_returns_line_count() {
        use std::io::Write as _;
        let dir = tempfile::tempdir().expect("tempdir");
        let src = dir.path().join("hello.rs");
        {
            let mut f = std::fs::File::create(&src).unwrap();
            for i in 0..42 {
                writeln!(f, "fn line_{i}() {{}}").unwrap();
            }
        }
        let agents = Arc::new(AgentPool::new_test(3000));
        let ambient_rooms = Arc::new(Mutex::new(HashSet::new()));
        let ctx = make_ctx(&agents, &ambient_rooms, dir.path(), "hello.rs");
        let output = handle_loc(&ctx).unwrap();
        assert!(
            output.contains("42"),
            "should report 42 lines for hello.rs: {output}"
        );
        assert!(
            output.contains("hello.rs"),
            "output should mention the filename: {output}"
        );
    }

    #[test]
    fn loc_filepath_nonexistent_returns_error() {
        let dir = tempfile::tempdir().expect("tempdir");
        let agents = Arc::new(AgentPool::new_test(3000));
        let ambient_rooms = Arc::new(Mutex::new(HashSet::new()));
        let ctx = make_ctx(&agents, &ambient_rooms, dir.path(), "does_not_exist.rs");
        let output = handle_loc(&ctx).unwrap();
        assert!(
            output.contains("not found") || output.contains("Error"),
            "nonexistent file should return a clear error: {output}"
        );
    }

    #[test]
    fn loc_skips_worktrees_directory() {
        let agents = Arc::new(AgentPool::new_test(3000));
        let ambient_rooms = Arc::new(Mutex::new(HashSet::new()));
        let repo_root = std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
            .parent()
            .unwrap_or(std::path::Path::new("."));
        let ctx = make_ctx(&agents, &ambient_rooms, repo_root, "");
        let output = handle_loc(&ctx).unwrap();
        assert!(
            !output.contains(".huskies/worktrees"),
            "output must not include paths inside worktrees: {output}"
        );
    }

    #[test]
    fn loc_excludes_lockfiles_from_results() {
        use std::io::Write as _;
        let dir = tempfile::tempdir().expect("tempdir");
        // Write a package-lock.json with many lines — it must NOT appear in output.
        let lockfile = dir.path().join("package-lock.json");
        {
            let mut f = std::fs::File::create(&lockfile).unwrap();
            for _ in 0..500 {
                writeln!(f, "  \"line\": true,").unwrap();
            }
        }
        // Write a real source file so the output is non-empty.
        let source = dir.path().join("main.rs");
        {
            let mut f = std::fs::File::create(&source).unwrap();
            for i in 0..20 {
                writeln!(f, "fn line_{i}() {{}}").unwrap();
            }
        }
        let agents = Arc::new(AgentPool::new_test(3000));
        let ambient_rooms = Arc::new(Mutex::new(HashSet::new()));
        let ctx = make_ctx(&agents, &ambient_rooms, dir.path(), "50");
        let output = handle_loc(&ctx).unwrap();
        assert!(
            !output.contains("package-lock.json"),
            "package-lock.json must be excluded from loc output: {output}"
        );
        assert!(
            output.contains("main.rs"),
            "main.rs should appear in loc output: {output}"
        );
    }

    #[test]
    fn loc_excludes_cargo_lock_from_results() {
        use std::io::Write as _;
        let dir = tempfile::tempdir().expect("tempdir");
        // Cargo.lock has no recognised source extension so it would be skipped
        // anyway — but we still verify EXCLUDED_FILENAMES contains it.
        assert!(
            EXCLUDED_FILENAMES.contains(&"Cargo.lock"),
            "EXCLUDED_FILENAMES must contain Cargo.lock"
        );
        // Write a Cargo.lock with many lines and verify it is excluded.
        let lockfile = dir.path().join("Cargo.lock");
        {
            let mut f = std::fs::File::create(&lockfile).unwrap();
            for _ in 0..500 {
                writeln!(f, "name = \"foo\"").unwrap();
            }
        }
        let source = dir.path().join("lib.rs");
        {
            let mut f = std::fs::File::create(&source).unwrap();
            for i in 0..10 {
                std::io::Write::write_all(&mut f, format!("fn f{i}() {{}}\n").as_bytes()).unwrap();
            }
        }
        let agents = Arc::new(AgentPool::new_test(3000));
        let ambient_rooms = Arc::new(Mutex::new(HashSet::new()));
        let ctx = make_ctx(&agents, &ambient_rooms, dir.path(), "50");
        let output = handle_loc(&ctx).unwrap();
        assert!(
            !output.contains("Cargo.lock"),
            "Cargo.lock must be excluded from loc output: {output}"
        );
    }

    #[test]
    fn excluded_filenames_constant_is_defined() {
        // The constant must contain the examples from the story.
        assert!(
            EXCLUDED_FILENAMES.contains(&"package-lock.json"),
            "EXCLUDED_FILENAMES must contain package-lock.json"
        );
        assert!(
            EXCLUDED_FILENAMES.contains(&"Cargo.lock"),
            "EXCLUDED_FILENAMES must contain Cargo.lock"
        );
    }

    #[test]
    fn loc_works_via_full_dispatch() {
        // Verifies the command is reachable through the same dispatch path used
        // by all transports (Matrix, WhatsApp, Slack).
        let result = super::super::tests::try_cmd_addressed(
            "Timmy",
            "@timmy:homeserver.local",
            "@timmy loc 1",
        );
        // /tmp has no source files, so we expect either "No source files found"
        // or a ranked result — either way the command must respond (not None).
        assert!(
            result.is_some(),
            "loc command must respond via dispatch (not fall through to LLM)"
        );
    }
}