rag-rat 0.8.0

//! `rag-rat claude-hook`: the Claude Code hook client (PreToolUse + SessionStart).
//!
//! Reads the hook JSON from stdin and branches on `hook_event_name`:
//! - `"SessionStart"`: injects a read-only repo orientation digest into the model context.
//! - anything else (or absent): the PreToolUse grep-augmentation path — asks the elected listener
//!   or falls back to a direct read-only query, prints `additionalContext` JSON.
//!
//! Exit 0 on every path — the hook must never block a tool call or session start.

use std::io::Read as _;
use std::path::{Path, PathBuf};
use std::time::Duration;

use rag_rat_core::config::Config;
use rag_rat_core::locks;
use rag_rat_core::query::grep_augment;
use rag_rat_core::query::orientation::Orientation;
use rag_rat_core::storage::IndexConnection;
use serde::Deserialize;

const SOCKET_BUDGET: Duration = Duration::from_millis(250);

/// Parsed hook input.  Fields absent on `SessionStart` (`tool_name`, `tool_input`) are
/// `#[serde(default)]` so deserialization succeeds for every event type.
#[derive(Debug, Default, Deserialize)]
pub struct HookInput {
    #[serde(default)]
    pub session_id: String,
    #[serde(default)]
    pub cwd: String,
    /// `"PreToolUse"`, `"SessionStart"`, etc. — absent on some older hook payloads.
    pub hook_event_name: Option<String>,
    /// `"startup"` | `"resume"` | `"clear"` | `"compact"` (SessionStart only).
    pub source: Option<String>,
    /// Present on PreToolUse only.
    #[serde(default)]
    pub tool_name: String,
    /// Present on PreToolUse only.
    #[serde(default)]
    pub tool_input: serde_json::Value,
}

pub struct Search {
    pub pattern: String,
    pub search_path: Option<String>,
    pub source: &'static str,
}

/// Pull a search intent out of the hook input; `None` means "not a grep, stay silent".
pub fn extract_search(input: &HookInput) -> Option<Search> {
    match input.tool_name.as_str() {
        "Grep" => {
            let pattern = input.tool_input.get("pattern")?.as_str()?.to_string();
            let search_path =
                input.tool_input.get("path").and_then(|v| v.as_str()).map(str::to_string);
            Some(Search { pattern, search_path, source: "grep_tool" })
        },
        "Bash" => {
            let command = input.tool_input.get("command")?.as_str()?;
            let (pattern, search_path) = parse_bash_search(command)?;
            Some(Search { pattern, search_path, source: "bash" })
        },
        _ => None,
    }
}

const SEARCH_COMMANDS: &[&str] = &["grep", "rg", "ag"];
/// Flags whose *next* token is a value, not the pattern. Conservative superset across the
/// three tools — a missed flag only costs a wrong-pattern no-op downstream, never a block.
const ARG_FLAGS: &[&str] = &[
    "-A",
    "-B",
    "-C",
    "-m",
    "-g",
    "-t",
    "-T",
    "-f",
    "-M",
    "--glob",
    "--type",
    "--type-not",
    "--include",
    "--exclude",
    "--exclude-dir",
    "--max-count",
    "--max-depth",
    "--context",
    "--after-context",
    "--before-context",
    "--file",
    "--ignore-file",
    "--threads",
    "--colors",
];

/// Extract (pattern, path) from a shell command that runs grep/rg/ag, or `None` when the
/// command doesn't or parsing would have to guess. False negatives are fine; false
/// positives are not (spec: Bash command parsing).
pub fn parse_bash_search(command: &str) -> Option<(String, Option<String>)> {
    if command.contains('`') || command.contains("$(") {
        return None; // substitution: ambiguous
    }
    // Split into pipeline/sequence segments; examine each for a search command.
    for (piped, segment) in split_top_level(command) {
        let tokens = shell_tokens(&segment)?;
        let mut tokens = tokens.as_slice();
        // Skip env-var prefixes (FOO=bar) before the command word.
        while tokens.first().is_some_and(|t| t.contains('=') && !t.starts_with('-')) {
            tokens = &tokens[1..];
        }
        let Some(command_word) = tokens.first() else { continue };
        let base = command_word.rsplit('/').next().unwrap_or(command_word);
        if base == "xargs" || base == "find" {
            return None; // grep as an argument of these is ambiguous
        }
        if !SEARCH_COMMANDS.contains(&base) {
            continue;
        }
        // A search command DOWNSTREAM of a pipe is filtering another tool's output
        // (`cargo test | grep …`, `gh … | rg …`), not a code search — skip it so augmentation
        // doesn't fire on incidental greps (#138). A grep that is the pipeline HEAD (`grep … |
        // head`) or a sequenced command (`cd x && rg …`) is a real search and still
        // matches.
        if piped {
            continue;
        }
        let mut pattern: Option<String> = None;
        let mut path: Option<String> = None;
        let mut rest = tokens[1..].iter();
        while let Some(token) = rest.next() {
            if let Some(value) = token.strip_prefix("--regexp=") {
                pattern.get_or_insert_with(|| value.to_string());
            } else if token == "-e" || token == "--regexp" {
                if let Some(value) = rest.next() {
                    pattern.get_or_insert_with(|| value.to_string());
                }
            } else if ARG_FLAGS.contains(&token.as_str()) {
                rest.next(); // consume the flag's value
            } else if token.starts_with('-') && token.len() > 1 {
                // value-less flag (or unknown): skip
            } else if pattern.is_none() {
                pattern = Some(token.to_string());
            } else if path.is_none() {
                path = Some(token.to_string());
            }
        }
        return pattern.map(|p| (p, path));
    }
    None
}

/// Split on top-level `|`, `&&`, `||`, `;` (quote-aware); also drop a leading `cd …` segment.
/// Each returned segment carries a `piped` flag: true when it was preceded by a single `|` (i.e. it
/// consumes the previous command's output). `||`, `&&`, `;`, `&`, and the first segment are not
/// piped — they're independent commands. This lets the caller tell a real grep (pipeline head or
/// sequenced) from an incidental output filter (#138).
///
/// Quote characters are preserved verbatim into the segment so that [`shell_tokens`] can
/// strip them itself — a top-level separator inside quotes must not split, and a quoted
/// pattern (`rg "quoted pattern" src`) must survive intact for re-tokenization.
fn split_top_level(command: &str) -> Vec<(bool, String)> {
    let mut segments = Vec::new();
    let mut current = String::new();
    let mut quote: Option<char> = None;
    // Whether the segment currently being accumulated was preceded by a single `|`.
    let mut piped = false;
    let mut chars = command.chars().peekable();
    while let Some(ch) = chars.next() {
        match (quote, ch) {
            // Closing quote: keep the quote char so shell_tokens sees a balanced pair.
            (Some(q), c) if c == q => {
                quote = None;
                current.push(c);
            },
            (Some(_), c) => current.push(c),
            // Opening quote: keep the quote char verbatim.
            (None, '\'' | '"') => {
                quote = Some(ch);
                current.push(ch);
            },
            (None, '|') => {
                // `|` pipes into the next segment; `|&` (bash shorthand for `2>&1 |`) also pipes
                // (stdout+stderr) so the next segment is still a downstream filter; `||` (logical
                // or) does NOT pipe — it's an independent command.
                let next_piped = match chars.peek() {
                    Some('|') => {
                        chars.next();
                        false
                    },
                    Some('&') => {
                        chars.next();
                        true
                    },
                    _ => true,
                };
                segments.push((piped, std::mem::take(&mut current)));
                piped = next_piped;
            },
            (None, ';') => {
                segments.push((piped, std::mem::take(&mut current)));
                piped = false;
            },
            (None, '&') => {
                if chars.peek() == Some(&'&') {
                    chars.next();
                }
                segments.push((piped, std::mem::take(&mut current)));
                piped = false;
            },
            (None, c) => current.push(c),
        }
    }
    segments.push((piped, current));
    segments
        .into_iter()
        .map(|(piped, s)| (piped, s.trim().to_string()))
        .filter(|(_, s)| !s.is_empty() && !s.starts_with("cd ") && *s != "cd")
        .collect()
}

/// Quote-aware tokenization of one segment. `None` on unbalanced quotes (ambiguous).
fn shell_tokens(segment: &str) -> Option<Vec<String>> {
    let mut tokens = Vec::new();
    let mut current = String::new();
    let mut quote: Option<char> = None;
    let mut quoted = false;
    for ch in segment.chars() {
        match (quote, ch) {
            (Some(q), c) if c == q => quote = None,
            (Some(_), c) => current.push(c),
            (None, '\'' | '"') => {
                quote = Some(ch);
                quoted = true;
            },
            (None, c) if c.is_whitespace() =>
                if !current.is_empty() || quoted {
                    tokens.push(std::mem::take(&mut current));
                    quoted = false;
                },
            (None, c) => current.push(c),
        }
    }
    if quote.is_some() {
        return None;
    }
    if !current.is_empty() || quoted {
        tokens.push(current);
    }
    Some(tokens)
}

/// Entry point for `rag-rat claude-hook`. Every failure path prints nothing and returns
/// Ok(()) — the hook must never block a grep (spec: error posture).
pub fn run() -> anyhow::Result<()> {
    let _ = run_inner(); // swallow: silence is the contract
    Ok(())
}

fn run_inner() -> anyhow::Result<()> {
    let mut raw = String::new();
    std::io::stdin().read_to_string(&mut raw)?;
    // Tolerant parse: missing fields use Default so both PreToolUse and SessionStart succeed.
    let input: HookInput = serde_json::from_str(&raw).unwrap_or_default();
    match input.hook_event_name.as_deref() {
        Some("SessionStart") => session_start(&input),
        _ => pretooluse(&input),
    }
}

/// SessionStart path: inject a read-only repo orientation digest as plain stdout.
/// Every error path prints nothing and returns Ok — never block session start.
fn session_start(input: &HookInput) -> anyhow::Result<()> {
    // Allowlist: only fire for meaningful session triggers, not resume.
    match input.source.as_deref() {
        Some("startup") | Some("clear") | Some("compact") => {},
        _ => return Ok(()),
    }
    let Some(config) = find_config(Path::new(&input.cwd)) else { return Ok(()) };
    // If the DB file does not exist, print a minimal nudge and exit — do NOT open/create it.
    if !config.database.is_file() {
        print!("{}", db_absent_notice());
        return Ok(());
    }
    // Open read-only; compose the orientation; print the digest.
    // Any error (locked, corrupt, etc.) propagates via `?` to run_inner, which run() swallows
    // (`let _ = run_inner()`) — so the hook stays silent and never blocks session start. Do NOT
    // add error prints here: this branch's stdout is injected as model context.
    let conn = IndexConnection::open_read_only(&config.database)?;
    // Scope orientation to the session's worktree: `config.root` (anchored to the main worktree) is
    // the base index, `input.cwd` is where the session is — a linked worktree gets its branch
    // overlay (#219). find_config already anchored config.root to the main worktree, so the two
    // together resolve the right overlay even when the session is launched from a linked
    // checkout.
    let o = rag_rat_core::query::orientation::orientation(
        conn.connection(),
        &config.root,
        Path::new(&input.cwd),
    )?;
    let (live, enabled) = watcher_state(&config);
    print!("{}", format_digest(&o, live, enabled));
    if let Some(line) = version_check_line(&config) {
        print!("{line}");
    }
    Ok(())
}

/// One digest line stating the running version vs the latest published on crates.io, with the
/// update command when behind. Reads the cached check only (no network — the MCP server refreshes
/// it out of band); `None` when version checking is disabled or no check has been cached yet, so a
/// fresh repo or an opted-out user sees nothing.
fn version_check_line(config: &Config) -> Option<String> {
    let status =
        rag_rat_core::version_check::cached_status(config.version_check.enabled, &config.database)?;
    version_line(&status)
}

/// Format the digest version line from a status (pure, so it's testable without a config/cache).
/// `None` when the latest version is unknown (no successful check cached yet) — stay quiet rather
/// than print a half-answer.
fn version_line(status: &rag_rat_core::version_check::VersionStatus) -> Option<String> {
    let latest = status.latest_version.as_deref()?;
    if status.update_available {
        Some(format!(
            "\n⚠ rag-rat update available: {} → {} — run `{}`\n",
            status.current_version, latest, status.update_command
        ))
    } else if latest == status.current_version {
        Some(format!("\nrag-rat {} (latest on crates.io)\n", status.current_version))
    } else {
        // Local build ahead of the published latest (dev / pre-release after a version bump) —
        // don't call it the crates.io latest.
        Some(format!("\nrag-rat {} (ahead of crates.io latest {latest})\n", status.current_version))
    }
}

/// PreToolUse path (grep augmentation).
fn pretooluse(input: &HookInput) -> anyhow::Result<()> {
    let Some(search) = extract_search(input) else { return Ok(()) };
    let Some(config) = find_config(Path::new(&input.cwd)) else { return Ok(()) };

    // Pass the session cwd through both paths so the grep-augmentation is scoped to the worktree
    // the session is in (a linked worktree gets its branch overlay), not just the base index
    // (#219).
    let context = ask_listener(&config, &input.session_id, &input.cwd, &search)
        .unwrap_or_else(|| fallback_compose(&config, &input.cwd, &search));
    if let Some(context) = context {
        // PreToolUse contract: allow + additionalContext; plain stdout is debug-only.
        println!(
            "{}",
            serde_json::json!({
                "hookSpecificOutput": {
                    "hookEventName": "PreToolUse",
                    "permissionDecision": "allow",
                    "additionalContext": context,
                }
            })
        );
    }
    Ok(())
}

/// One-liner shown when the DB file is absent (no config directory walk — `find_config` already
/// succeeded, so we know we're in a rag-rat repo; the DB just hasn't been built yet).
fn db_absent_notice() -> String {
    format!("{}\nindex not built — run 'rag-rat index'\n", ATTRIBUTION_HEADER.trim_end())
}

/// Probe whether the per-worktree watcher election lock is currently held (i.e. a watcher is live).
///
/// Algorithm: try to acquire the election lock non-blocking.
/// - `Ok(None)` → lock is held by another process → watcher is live.
/// - `Ok(Some(_))` → we acquired it (no holder); release immediately → not live.
/// - `Err(_)` → treat as not live (conservative).
pub fn watcher_state(config: &Config) -> (bool /* live */, bool /* enabled */) {
    let enabled = config.watch.enabled && std::env::var_os("RAG_RAT_NO_WATCH").is_none();
    let base_dir =
        config.database.parent().map(Path::to_path_buf).unwrap_or_else(|| config.root.clone());
    let election_path = locks::election_lock_path(&base_dir, &config.root);
    // try_acquire: Ok(None) means the lock is held (watcher is live).
    let live = matches!(locks::FileLock::try_acquire(&election_path), Ok(None));
    (live, enabled)
}

// ─── Attribution header ───────────────────────────────────────────────────────

const ATTRIBUTION_HEADER: &str = "\
▶ rag-rat repo intelligence — injected by the rag-rat MCP server (prefer it over grep/cat)
  concept → semantic_search · callers/callees → find_callers/trace_callees
  before editing a symbol → impact_surface · exact symbol → symbol_lookup
  why/rationale → repo memories ride along; memory_search to dig
";

// ─── Digest formatting ────────────────────────────────────────────────────────

/// Strip a leading `crates/<crate>/src/` prefix from a repo-relative path.
///
/// Converts e.g. `crates/rag-rat-core/src/index/mod.rs` → `index/mod.rs`.
/// Paths that do not match the three-segment `crates/<anything>/src/` prefix are returned
/// unchanged.
pub fn short_path(p: &str) -> String {
    let parts: Vec<&str> = p.splitn(4, '/').collect();
    if parts.len() == 4 && parts[0] == "crates" && parts[2] == "src" {
        parts[3].to_string()
    } else {
        p.to_string()
    }
}

/// Render the full orientation digest as a plain-text string.
///
/// `live` = watcher election lock is currently held; `enabled` = watch is configured on.
pub fn format_digest(o: &Orientation, live: bool, enabled: bool) -> String {
    let mut out = String::with_capacity(2048);

    // Attribution + capability nudge.
    out.push_str(ATTRIBUTION_HEADER);
    out.push('\n');

    // Purpose line (root dir memory title) — omit entirely if absent.
    if let Some(ref title) = o.tree.root_memory_title {
        out.push_str(title);
        out.push('\n');
    }

    // LAYOUT — directory tree.
    out.push_str(&format!("LAYOUT  ({} files · ‹…› = directory memory)\n", o.total_files));
    for node in &o.tree.nodes {
        let indent = "  ".repeat(node.depth as usize);
        if let Some(ref title) = node.memory_title {
            out.push_str(&format!("{}{}  ‹{}›\n", indent, node.label, title));
        } else {
            out.push_str(&format!("{}{}\n", indent, node.label));
        }
    }
    if o.tree.truncated > 0 {
        out.push_str(&format!("  … (+{} more)\n", o.tree.truncated));
    }

    // Load-bearing files (paths shortened: crates/<crate>/src/X → X).
    if !o.load_bearing.is_empty() {
        let parts: Vec<String> = o
            .load_bearing
            .iter()
            .map(|(p, fi)| format!("{} (fan_in {})", short_path(p), fi))
            .collect();
        out.push_str(&format!("load-bearing: {}\n", parts.join(" · ")));
    }

    // Recent activity.
    {
        let mut line_parts: Vec<String> = Vec::new();
        if !o.recent_commits.is_empty() {
            line_parts.push(format!("recent: {}", o.recent_commits.join(" · ")));
        }
        if !o.hot_files.is_empty() {
            let short_hot: Vec<String> = o.hot_files.iter().map(|p| short_path(p)).collect();
            line_parts.push(format!("hot: {}", short_hot.join(", ")));
        }
        if !line_parts.is_empty() {
            out.push_str(&format!("{}\n", line_parts.join(" · ")));
        }
    }

    // Active non-dir memory titles — the list is already truncated to the display cap by
    // the query; the overflow note reflects the TRUE total, not the truncated list length.
    if !o.active_memory_titles.is_empty() {
        let mut mem_line = o.active_memory_titles.join(" · ");
        let extra = (o.active_memory_total as usize).saturating_sub(o.active_memory_titles.len());
        if extra > 0 {
            mem_line.push_str(&format!(" (+{extra} more)"));
        }
        out.push_str(&format!("memories: {mem_line}\n"));
    }

    // Watcher-aware health line.
    let fresh = o.head == o.indexed_head || o.head.is_empty() || o.indexed_head.is_empty();
    let health_status = match (live, enabled, fresh) {
        (true, _, true) => "index fresh (watcher live)".to_string(),
        (true, _, false) => "index syncing (watcher live)".to_string(),
        (false, true, false) => "index stale — start the rag-rat MCP server".to_string(),
        (false, false, false) => "watcher off; index stale — run 'rag-rat index'".to_string(),
        _ => "index fresh".to_string(),
    };
    let active = o.anchor.current + o.anchor.relocated;
    let mut health = format!("health: {} · memories {} active", health_status, active);
    if o.anchor.stale > 0 {
        health.push_str(&format!("/{} stale", o.anchor.stale));
    }
    if o.anchor.gone > 0 {
        health.push_str(&format!(" · {} gone → run 'rag-rat memory doctor'", o.anchor.gone));
    }
    if o.parser_failures > 0 {
        health.push_str(&format!(" · parser failures: {}", o.parser_failures));
    }
    out.push_str(&health);
    out.push('\n');

    out
}

/// Walk up from the hook's cwd to the nearest rag-rat.toml. `None` ⇒ not a rag-rat repo ⇒
/// silent no-op (what makes `--global` install safe).
fn find_config(start: &Path) -> Option<Config> {
    let mut dir = Some(start);
    while let Some(current) = dir {
        let candidate = current.join("rag-rat.toml");
        if candidate.is_file() {
            return Config::load(&candidate).ok();
        }
        dir = current.parent();
    }
    None
}

/// Outer Option: did the listener answer at all (None ⇒ fall back). Inner Option: did it
/// have anything new to say.
fn ask_listener(
    config: &Config,
    session_id: &str,
    cwd: &str,
    search: &Search,
) -> Option<Option<String>> {
    #[cfg(unix)]
    {
        use std::io::{BufRead, BufReader, Write as _};
        use std::os::unix::net::UnixStream;
        let socket = socket_path(config);
        // SOCKET_BUDGET covers both read and write. Unix-domain connect() completes into
        // the listener's backlog immediately (no network round-trip), so no separate connect
        // timeout is needed.
        let stream = UnixStream::connect(&socket).ok()?;
        stream.set_read_timeout(Some(SOCKET_BUDGET)).ok()?;
        stream.set_write_timeout(Some(SOCKET_BUDGET)).ok()?;
        // `cwd` lets the listener scope the augmentation to the session's worktree overlay (#219);
        // an older listener without the field just ignores it (lenient deserialize) → base scope.
        let request = serde_json::json!({
            "v": 1, "kind": "grep_augment", "session_id": session_id,
            "cwd": cwd,
            "pattern": search.pattern, "search_path": search.search_path,
            "source": search.source,
        });
        let mut writer = stream.try_clone().ok()?;
        writeln!(writer, "{request}").ok()?;
        let mut line = String::new();
        BufReader::new(stream).read_line(&mut line).ok()?;
        let reply: serde_json::Value = serde_json::from_str(&line).ok()?;
        if reply.get("v")?.as_u64()? != 1 {
            return None;
        }
        Some(reply.get("context")?.as_str().map(str::to_string))
    }
    #[cfg(not(unix))]
    {
        let _ = (config, session_id, cwd, search);
        None
    }
}

/// Single source of truth via `locks::hook_socket_path_for`; same computation as the MCP
/// listener's `socket_path_for`, guaranteed not to diverge.
fn socket_path(config: &Config) -> PathBuf {
    locks::hook_socket_path_for(config)
}

/// Stateless direct read (no dedupe — spec: fallback path). Any error ⇒ silence.
fn fallback_compose(config: &Config, cwd: &str, search: &Search) -> Option<String> {
    let conn = IndexConnection::open_read_only(&config.database).ok()?;
    // Scope to the session's worktree overlay before composing — `compose` queries the `files`
    // view, so without this it would read raw (unscoped) rows. config.root is the anchored main
    // worktree; cwd is the session dir (a linked worktree → its overlay, else base) (#219).
    rag_rat_core::index::install_worktree_scope_view(
        conn.connection(),
        &config.root,
        Path::new(cwd),
    )
    .ok()?;
    grep_augment::compose(
        conn.connection(),
        &search.pattern,
        search.search_path.as_deref(),
        &grep_augment::DedupeFilter::default(),
    )
    .ok()
    .flatten()
    .map(|out| out.context)
}

#[cfg(test)]
mod tests;