Skip to main content

reference_query/cli/
mod.rs

1//! Command-line surface. Search is the default action: `rq <query>`.
2
3use std::collections::HashSet;
4use std::io::{IsTerminal, Write};
5use std::path::PathBuf;
6use std::process::ExitCode;
7use std::time::Duration;
8
9use clap::{CommandFactory, Parser};
10use clap_complete::Shell;
11
12use crate::store::Store;
13
14/// Search is the default action (`rq <query>`). Operations are flags rather
15/// than subcommands so no word is reserved — `rq index`, `rq status`, and
16/// `rq record` all search for those symbols. This also matches the rg/fd feel.
17#[derive(Parser)]
18#[command(
19    name = "rq",
20    version,
21    about = "Ranked definition lookup — the one place a symbol is defined, first.",
22    long_about = "rq finds where a symbol is defined and ranks the one you most \
23likely meant to the top — not every match.\n\n\
24Search is the default action; operations are flags, not subcommands, so every \
25word (including \"index\", \"status\", \"record\") stays searchable. Ranking favors \
26your current repo and recently-active files, and learns from the results you open \
27(see RECORDING below). Run `rq <query> --explain` to see the score behind each result.",
28    after_help = "EXAMPLES:\n  \
29rq thing                  search for a definition named or like \"thing\"\n  \
30rq wibble --explain       same, plus the score behind each result\n  \
31rq thing --json           machine-readable results (for editors/agents)\n  \
32rq thing --no-record      search without recording it (speculative/agent queries)\n  \
33rq thing --no-wait        answer now from the committed index; don't block on a rebuild\n  \
34rq thing --wait 2s        ...or wait up to a bounded time for the index to warm\n  \
35rq thing app/web          restrict to a directory (rg-style)\n  \
36rq perform -k method      restrict to a symbol kind (c/mod/m/f/s/e/t)\n  \
37rq class Widget           a leading kind keyword is shorthand for -k\n  \
38rq --symbols FILE         outline a file's definitions, in line order\n  \
39rq thing -x rust          restrict to a language (ruby/rust/go/python/ts/js)\n  \
40rq -o thing               open the best match in your editor (and record it)\n  \
41rq --index                index the current repository\n  \
42rq --status               show indexing coverage\n  \
43rq --usage                show how rq has been called (by caller and flags)\n  \
44rq --drop                 remove this repo's index (opposite of --index)\n\n\
45SHORT FLAGS (easy to misread):\n  \
46-j = --json (not jobs; --jobs is long-only)   -l = --limit (not lang)   -x = --lang\n\n\
47RECORDING (editor/shell hook):\n  \
48rq --record --file <path> --line <n> <query>\n  \
49Tells rq which result you opened for a query, so ranking learns. Pass --no-record \
50to a search to skip this. Editors and the script/rq-open wrapper call --record for you.\n\n\
51The index is a SQLite file at $RQ_DB (default ~/.local/share/rq/rq.db); it warms \
52automatically on the first search in a git repo. On a large, cold repo a search \
53keeps indexing until it can answer rather than reporting a premature \"no \
54matches\" (an interactive run shows progress and stops on Ctrl-C). Exit codes: 0 \
55= matched, 1 = no match, 2 = no match yet (index still warming — try again)."
56)]
57struct Cli {
58    /// Search query. With --drop, the repo path/identity to drop; with --record,
59    /// the query the selection was made for.
60    //
61    // `Other` keeps shells from offering filenames here: a search query isn't a
62    // path. The path-valued operations (--index, --symbols) carry their own
63    // value with a path hint instead, so completion is scoped to them.
64    #[arg(value_name = "TARGET", value_hint = clap::ValueHint::Other)]
65    target: Option<String>,
66
67    /// Directories to restrict results to (rg-style; same as repeated --path).
68    #[arg(value_name = "PATH")]
69    dirs: Vec<String>,
70
71    /// Show the score breakdown for each result.
72    #[arg(short = 'e', long)]
73    explain: bool,
74
75    /// Don't let this invocation teach ranking or count as usage — suppresses
76    /// recording the result you open, select, or `--show`, and keeps the call
77    /// out of `--usage`. For benchmark and CI loops, whose repeated queries
78    /// would otherwise dominate both.
79    #[arg(long)]
80    no_record: bool,
81
82    /// Answer immediately from the committed index — never block waiting on a
83    /// background (re)index. For agents/scripts: a query issued mid-rebuild
84    /// returns at once (a miss reports `warming`, exit 2, so a caller can retry)
85    /// instead of blocking up to the wait budget. Shorthand for `--wait 0`;
86    /// leftover warming still detaches to a background child.
87    #[arg(long = "no-wait")]
88    no_wait: bool,
89
90    /// How long a query may wait for the index to warm before answering with
91    /// whatever's committed: a duration like `50ms`, `2s`, `1m`, or a bare number
92    /// of seconds. `0` doesn't wait at all (same as `--no-wait`). Overrides
93    /// `RQ_WAIT_BUDGET_MS` for this call (default 1 minute).
94    #[arg(long, value_name = "DUR", value_parser = parse_wait, conflicts_with = "no_wait")]
95    wait: Option<Duration>,
96
97    /// Open the best match in your editor and record the pick, so ranking learns.
98    /// On a terminal with several matches, prompts to choose. Launcher: `RQ_OPEN`
99    /// (a template with `{file}`/`{line}`/`{}` = path:line), else VS Code
100    /// (`code`), else `$VISUAL`/`$EDITOR`, else prints the resolved path:line.
101    #[arg(short = 'o', long, conflicts_with_all = ["index", "status", "record", "json", "ndjson"])]
102    open: bool,
103
104    /// Print the definition's source, not just its location — but only when the
105    /// top match is confident; otherwise falls back to the ranked list. Pipe to a
106    /// pager (`rq --show foo | less`). JSON adds a `body` field.
107    #[arg(long, conflicts_with_all = ["open", "index", "status", "record", "symbols", "drop"])]
108    show: bool,
109
110    /// Emit results as a JSON array (for editors and scripts).
111    #[arg(short = 'j', long)]
112    json: bool,
113
114    /// Emit results as newline-delimited JSON, one object per line.
115    #[arg(short = 'J', long, conflicts_with = "json")]
116    ndjson: bool,
117
118    /// Restrict results to files under this repo-relative directory (repeatable).
119    #[arg(short = 'p', long, value_name = "DIR")]
120    path: Vec<String>,
121
122    /// Maximum number of results to show; `0` shows every match.
123    #[arg(short = 'l', long, value_name = "N", default_value_t = DEFAULT_LIMIT)]
124    limit: usize,
125
126    /// Restrict to symbol kinds: class, module, method, function, struct, enum,
127    /// trait (shortcuts: c, mod, m, f, s, e, t; `interface` = trait, `type` =
128    /// struct). Repeatable or comma-separated.
129    #[arg(short = 'k', long, value_name = "KIND", value_delimiter = ',')]
130    kind: Vec<String>,
131
132    /// Restrict to languages: ruby, rust, go, python, typescript, javascript.
133    /// Prefix-matched, so `r` means ruby+rust and `p` means python; aliases rb,
134    /// rs, golang, ts, tsx, js, jsx. Repeatable or comma-separated.
135    #[arg(short = 'x', long = "lang", value_name = "LANG", value_delimiter = ',')]
136    lang: Vec<String>,
137
138    /// Search every indexed repository, not just the current one. By default a
139    /// search inside a repo returns only that repo's definitions.
140    #[arg(short = 'a', long = "all-repos")]
141    all_repos: bool,
142
143    /// Index a repository (PATH, or the current directory).
144    #[arg(long, value_name = "PATH", num_args = 0..=1, value_hint = clap::ValueHint::AnyPath, conflicts_with_all = ["status", "record"])]
145    index: Option<Option<String>>,
146
147    /// Show indexing coverage per known repository.
148    #[arg(long, conflicts_with_all = ["index", "record"])]
149    status: bool,
150
151    /// Show how rq has been used: searches per day, by caller and flags.
152    #[arg(long, conflicts_with_all = ["index", "record", "status"])]
153    usage: bool,
154
155    /// List the symbols defined in FILE, in line order — a structural outline,
156    /// not a ranked search. Honors -k/-x to filter by kind/language.
157    #[arg(long, value_name = "FILE", value_hint = clap::ValueHint::FilePath, conflicts_with_all = ["index", "status", "record", "drop", "open"])]
158    symbols: Option<String>,
159
160    /// Drop a repository's index — the opposite of --index. Removes its symbols,
161    /// files, coverage, and learned ranking. TARGET is the repo's path (or the
162    /// current repo); a known identity string (as shown by --status) also works.
163    #[arg(long, conflicts_with_all = ["index", "status", "record", "open"])]
164    drop: bool,
165
166    /// Record an interaction (editor/shell hook): the result opened for a query.
167    /// Requires --file.
168    #[arg(long, requires = "file", conflicts_with_all = ["index", "status"])]
169    record: bool,
170
171    /// (--record) File that was opened/selected.
172    #[arg(long)]
173    file: Option<String>,
174
175    /// (--record) Line landed on (attributes the selection to a definition).
176    #[arg(long)]
177    line: Option<i64>,
178
179    /// (--record) Event kind (select or open).
180    #[arg(long, default_value = "select")]
181    event: String,
182
183    /// Finish warming a repository's index in the background — the target a
184    /// search re-execs after printing results, detached, so the shell never
185    /// waits on it. Single-flighted per repo; safe to run by hand.
186    #[arg(long, hide = true, value_name = "PATH", num_args = 0..=1, value_hint = clap::ValueHint::AnyPath, conflicts_with_all = ["index", "status", "record", "drop", "symbols", "open", "show"])]
187    warm: Option<Option<String>>,
188
189    /// Print a shell completion script (bash, zsh, fish, elvish, powershell).
190    #[arg(long, value_name = "SHELL")]
191    completions: Option<Shell>,
192
193    /// Trace what rq decides (root, coverage, warming, reconcile) to stderr —
194    /// for debugging. `RQ_LOG=1` does the same for an installed binary.
195    #[arg(short = 'v', long)]
196    verbose: bool,
197
198    /// Report where a search spent its time, phase by phase, to stderr — as
199    /// JSON alongside --json, so a baseline can be stored and diffed.
200    /// `RQ_PROFILE=1` does the same for an installed binary.
201    #[arg(long)]
202    profile: bool,
203
204    /// Parse worker threads the background indexer uses (0 = auto). (`-j` is
205    /// taken by `--json`, so this is `--jobs` only.) `RQ_JOBS` works too.
206    #[arg(long, value_name = "N", default_value_t = 0)]
207    jobs: usize,
208}
209
210/// Parse arguments and dispatch. Returns the process exit code.
211pub fn run() -> ExitCode {
212    let cli = Cli::parse();
213    crate::trace::enable_from(cli.verbose);
214    crate::profile::enable_from(cli.profile);
215    crate::index::set_parse_jobs(cli.jobs);
216
217    if let Some(shell) = cli.completions {
218        clap_complete::generate(shell, &mut Cli::command(), "rq", &mut std::io::stdout());
219        return ExitCode::SUCCESS;
220    }
221    if let Some(path) = &cli.index {
222        // index PATH (else cwd); with --path, seed only those subtrees
223        let out = output_format(&cli);
224        return cmd_index(path.as_deref().map(PathBuf::from), &cli.path, out);
225    }
226    if let Some(path) = &cli.warm {
227        return cmd_warm(path.as_deref());
228    }
229    if cli.status {
230        return cmd_status(output_format(&cli));
231    }
232    if cli.usage {
233        return cmd_usage(output_format(&cli));
234    }
235    if cli.drop {
236        let out = output_format(&cli);
237        return cmd_drop(cli.target, out);
238    }
239    if cli.record {
240        // a typo'd --event would otherwise record silently and never roll up
241        if !matches!(cli.event.as_str(), "select" | "open") {
242            return fail(format_args!(
243                "rq --record: unknown --event {:?} (expected select or open)",
244                cli.event
245            ));
246        }
247        // clap guarantees --file is present via `requires`
248        let file = cli.file.expect("--record requires --file");
249        return cmd_record(&cli.event, cli.target.as_deref(), &file, cli.line);
250    }
251    let out = output_format(&cli);
252    let mut kinds: Vec<String> = cli.kind.iter().map(|k| canonical_kind(k)).collect();
253    // a language token can expand to several tags (`r` → ruby + rust)
254    let langs: Vec<String> = cli.lang.iter().flat_map(|x| canonical_langs(x)).collect();
255    if let Some(file) = &cli.symbols {
256        return cmd_symbols(file, &kinds, &langs, out);
257    }
258    // path filters: trailing positionals (rg-style) plus any --path flags
259    let mut paths = cli.path.clone();
260    match cli.target {
261        Some(target) => {
262            // A leading kind keyword (`rq class Foo`) is shorthand for `-k`; skip
263            // it when the user gave an explicit `-k`, so the two never conflict.
264            let query = if cli.kind.is_empty() {
265                let (kw, query, dirs) = split_kind_keyword(target, cli.dirs.clone());
266                if let Some(k) = kw {
267                    kinds.push(k.to_string());
268                }
269                paths.extend(dirs);
270                query
271            } else {
272                paths.extend(cli.dirs.clone());
273                target
274            };
275            let mut session = match Session::open() {
276                Ok(s) => s,
277                Err(code) => return code,
278            };
279            cmd_search(
280                &mut session,
281                &SearchArgs {
282                    query: &query,
283                    explain: cli.explain,
284                    out,
285                    paths: &paths,
286                    kinds: &kinds,
287                    langs: &langs,
288                    want: requested_limit(cli.limit),
289                    no_record: cli.no_record,
290                    no_wait: cli.no_wait,
291                    wait: cli.wait,
292                    open: cli.open,
293                    all_repos: cli.all_repos,
294                    show: cli.show,
295                    batch: false,
296                },
297            )
298        }
299        // No query, but a pipe on stdin: each line is one, all sharing this
300        // run's store, repo resolution and warm.
301        None if !std::io::stdin().is_terminal() => cmd_batch(&cli, out, &paths, &kinds, &langs),
302        // bare `rq` (or just flags like --explain with no query): show help
303        None => {
304            let _ = Cli::command().print_long_help();
305            ExitCode::SUCCESS
306        }
307    }
308}
309
310/// How results are rendered.
311#[derive(Clone, Copy, PartialEq)]
312enum Output {
313    Text,
314    Json,
315    Ndjson,
316}
317
318fn output_format(cli: &Cli) -> Output {
319    if cli.ndjson {
320        Output::Ndjson
321    } else if cli.json {
322        Output::Json
323    } else {
324        Output::Text
325    }
326}
327
328/// Results shown when `--limit` isn't given.
329const DEFAULT_LIMIT: usize = 10;
330
331/// Minimum headroom to rank before a `--path` filter (so filtered-in results
332/// aren't lost to the cutoff).
333const PATH_HEADROOM: usize = 200;
334
335/// `--limit 0` means unlimited: every ranked hit, bounded only by how many
336/// candidates recall returned.
337fn requested_limit(limit: usize) -> usize {
338    if limit == 0 { usize::MAX } else { limit }
339}
340
341/// Count one search for `--usage`. Observability only: `search` rows are
342/// invisible to the rollup that feeds ranking, so this can never move a result.
343fn record_usage(
344    store: &mut Store,
345    args: &SearchArgs,
346    repository_id: Option<i64>,
347    results: usize,
348    status: &str,
349    coverage: Option<&str>,
350) {
351    if args.no_record {
352        return;
353    }
354    let _ = store.record_search(&crate::store::SearchRecord {
355        query: &args.query.to_ascii_lowercase(),
356        repository_id,
357        results,
358        source: &crate::origin::detect(),
359        flags: &flag_summary(args),
360        status,
361        coverage: coverage.unwrap_or("none"),
362    });
363}
364
365/// The call's flags as a canonical, comma-joined string, for usage counts.
366/// A fixed vocabulary in a fixed order, so the same call always produces the
367/// same string and the counter table stays small — values are never included,
368/// only which knobs were reached for.
369fn flag_summary(args: &SearchArgs) -> String {
370    let mut on: Vec<&str> = Vec::new();
371    match args.out {
372        Output::Json => on.push("json"),
373        Output::Ndjson => on.push("ndjson"),
374        Output::Text => {}
375    }
376    for (present, name) in [
377        (args.explain, "explain"),
378        (args.show, "show"),
379        (args.open, "open"),
380        (args.all_repos, "all-repos"),
381        (args.no_wait, "no-wait"),
382        (args.batch, "batch"),
383        (!args.paths.is_empty(), "path"),
384        (!args.kinds.is_empty(), "kind"),
385        (!args.langs.is_empty(), "lang"),
386        (args.want != DEFAULT_LIMIT, "limit"),
387    ] {
388        if present {
389            on.push(name);
390        }
391    }
392    on.join(",")
393}
394
395/// How often the search re-checks the index while a cold repo warms on the
396/// background thread. Each poll runs a full read query against the DB the
397/// indexer is actively writing, so polling too fast steals CPU and read-lock
398/// churn from the warm; 100 ms keeps that pressure low while staying
399/// imperceptible (an early answer or completion appears within a frame, and the
400/// progress line only redraws every `PROGRESS_REDRAW` anyway).
401const POLL_INTERVAL: Duration = Duration::from_millis(100);
402
403/// How long a cold-repo query may wait silently before we tell the user we're
404/// indexing — short enough to explain the pause, long enough that a repo which
405/// indexes quickly never flashes a message.
406const HEADS_UP_DELAY: Duration = Duration::from_millis(500);
407
408/// Minimum gap between progress-line redraws once the heads-up is showing — keeps
409/// the line from flickering (and the count query off the hot path) while still
410/// feeling live.
411const PROGRESS_REDRAW: Duration = Duration::from_millis(120);
412
413/// Everything `rq <query>` needs, bundled from the parsed CLI flags.
414struct SearchArgs<'a> {
415    query: &'a str,
416    explain: bool,
417    out: Output,
418    paths: &'a [String],
419    kinds: &'a [String],
420    langs: &'a [String],
421    /// Number of results to show (`--limit`).
422    want: usize,
423    no_record: bool,
424    /// Answer from the committed index without blocking on a (re)index (`--no-wait`).
425    no_wait: bool,
426    /// Cap on how long to wait for the index to warm (`--wait`); `None` = the
427    /// default/`RQ_WAIT_BUDGET_MS` budget.
428    wait: Option<Duration>,
429    open: bool,
430    all_repos: bool,
431    /// One of several queries sharing a run, so each row says which query it
432    /// answers — a single stream serving many questions is otherwise
433    /// unattributable.
434    batch: bool,
435    show: bool,
436}
437
438/// Default action: search the index and print ranked results.
439/// Everything a search needs that doesn't depend on the query: the open store,
440/// the repo it's rooted in, the branch's changed files, and who that repo is.
441///
442/// Split out because it's the expensive half — opening the store, resolving the
443/// root, reading branch files, resolving identity — and none of it varies per
444/// query. One search builds one and drops it; a caller answering many can build
445/// it once. Deliberately *not* holding the warm decision: that one is entangled
446/// with the query (the indexer path-prioritises toward it) and belongs to a
447/// single search.
448struct Session {
449    store: Store,
450    cwd: Option<PathBuf>,
451    cwd_is_git: bool,
452    root: Option<PathBuf>,
453    active_paths: Vec<String>,
454    branch_refresh: Option<BranchRefresh>,
455    identity: Option<String>,
456    coverage: Option<String>,
457}
458
459impl Session {
460    /// Resolve the search context, or the exit code to fail with.
461    fn open() -> std::result::Result<Session, ExitCode> {
462        let open_span = crate::profile::span("store open");
463        let store = match open_store() {
464            Ok(s) => s,
465            Err(e) => return Err(fail(format_args!("rq: cannot open database: {e}"))),
466        };
467        drop(open_span);
468        let git_span = crate::profile::span("setup: git root");
469        let cwd = std::env::current_dir().ok();
470        let cwd_is_git = cwd.as_deref().is_some_and(crate::index::is_git_repo);
471
472        // Index relative to the repo ROOT, not wherever the search happens to run.
473        // Paths and the stored checkout root must be repo-root-relative and stable, or
474        // a search from a subdirectory would re-key the same repo under subdir-relative
475        // paths — and the deletion reconcile / staleness revalidation would then forget
476        // everything indexed from the root. Outside git, the root is just the cwd.
477        let root = cwd
478            .as_deref()
479            .map(|c| crate::index::repo_root(c).unwrap_or_else(|| c.to_path_buf()));
480        drop(git_span);
481
482        // Files you're changing on this feature branch (and their directory
483        // neighbors): the branch ranking boost, and the warm pass's priority set.
484        let mut branch_span = crate::profile::span("setup: branch files");
485        let (active_paths, branch_refresh) = match &root {
486            Some(c) if cwd_is_git => cached_branch_files(&store, c),
487            _ => (Vec::new(), None),
488        };
489        branch_span.note(|| {
490            let how = if branch_refresh.is_some() {
491                "cached, refreshing alongside"
492            } else {
493                "cached"
494            };
495            format!("{} changed, {how}", active_paths.len())
496        });
497        drop(branch_span);
498
499        // Resolve identity from the repo root, cache-first: looked up by checkout root
500        // (no `git remote` fork), falling back to git only the first time we see a
501        // repo. Computed even for non-git dirs so an explicitly `--index`ed one is
502        // still recognized as the current repo below.
503        let mut identity_span = crate::profile::span("setup: identity");
504        let identity = root.as_deref().map(|c| resolve_identity(&store, c));
505        let coverage = identity
506            .as_deref()
507            .and_then(|id| store.coverage_status(id).ok())
508            .flatten();
509        identity_span.note(|| coverage.as_deref().unwrap_or("unknown").to_string());
510        drop(identity_span);
511        Ok(Session {
512            store,
513            cwd,
514            cwd_is_git,
515            root,
516            active_paths,
517            branch_refresh,
518            identity,
519            coverage,
520        })
521    }
522}
523
524/// Answer a stream of queries, one per line on stdin, in a single run.
525///
526/// Everything a query doesn't vary — the store, the repo, its identity, the
527/// branch's changed files — is resolved once and reused, which on a large repo
528/// is most of what a single lookup costs. Agents and scripts do runs of
529/// lookups; this is that shape.
530///
531/// A cold repo warms **once, up front, until complete** rather than answering
532/// each line from whatever happens to be indexed. Block-until-*answered*
533/// doesn't generalise to queries we haven't read yet — you can't prioritise
534/// toward them — so block-until-*complete* is the batch-shaped equivalent, and
535/// it keeps this file's own rule that correctness beats the first query's
536/// latency. `--no-wait` opts out, exactly as it does for one query, and any
537/// line the index can't yet answer says so with `status: "warming"`.
538fn cmd_batch(
539    cli: &Cli,
540    out: Output,
541    paths: &[String],
542    kinds: &[String],
543    langs: &[String],
544) -> ExitCode {
545    if out == Output::Json {
546        return fail(format_args!(
547            "rq: --json can't frame a stream of queries — use --ndjson (-J), \
548             where each line carries the query it answers"
549        ));
550    }
551    if cli.open || cli.show {
552        return fail(format_args!(
553            "rq: --open and --show act on a single result, not a stream of queries"
554        ));
555    }
556
557    use std::io::BufRead;
558    let queries: Vec<String> = std::io::stdin()
559        .lock()
560        .lines()
561        .map_while(std::result::Result::ok)
562        .map(|l| l.trim().to_string())
563        .filter(|l| !l.is_empty())
564        .collect();
565    // Nothing on stdin isn't a batch — it's a bare invocation that happens to
566    // run without a terminal (a script, a test harness, stdin from /dev/null).
567    // Treat it the way `rq` with no arguments is always treated.
568    if queries.is_empty() {
569        let _ = Cli::command().print_long_help();
570        return ExitCode::SUCCESS;
571    }
572
573    let mut session = match Session::open() {
574        Ok(s) => s,
575        Err(code) => return code,
576    };
577
578    // Warm to completion before answering anything, so a cold repo doesn't
579    // return a page of misses that only mean "not indexed yet".
580    if !cli.no_wait
581        && session.coverage.as_deref() != Some("complete")
582        && let Some(root) = session.root.clone()
583    {
584        {
585            let budget = cli.wait.unwrap_or_else(wait_budget);
586            crate::trace!(
587                "batch: warming {} queries' worth of index first",
588                queries.len()
589            );
590            let active = session.active_paths.clone();
591            let _ = crate::index::index_budgeted(&mut session.store, &root, &active, budget, None);
592            session.coverage = session
593                .identity
594                .as_deref()
595                .and_then(|id| session.store.coverage_status(id).ok())
596                .flatten();
597        }
598    }
599
600    let mut worst = ExitCode::SUCCESS;
601    let mut any_hit = false;
602    for query in &queries {
603        let code = cmd_search(
604            &mut session,
605            &SearchArgs {
606                query,
607                explain: cli.explain,
608                out,
609                paths,
610                kinds,
611                langs,
612                want: requested_limit(cli.limit),
613                no_record: cli.no_record,
614                // The warm happened above, once. Per-query warming would undo
615                // the point of batching, and block-until-answered is meaningless
616                // when the queries were all read up front.
617                no_wait: true,
618                wait: cli.wait,
619                open: false,
620                all_repos: cli.all_repos,
621                show: false,
622                batch: true,
623            },
624        );
625        if code == ExitCode::SUCCESS {
626            any_hit = true;
627        } else {
628            worst = code;
629        }
630    }
631    // The batch ran; per-line `status` carries each query's outcome. Only a
632    // wholly fruitless batch reports failure, mirroring one query's contract.
633    if any_hit { ExitCode::SUCCESS } else { worst }
634}
635
636fn cmd_search(session: &mut Session, args: &SearchArgs) -> ExitCode {
637    let &SearchArgs {
638        query,
639        out,
640        want,
641        no_record,
642        no_wait,
643        wait,
644        open,
645        all_repos,
646        show,
647        ..
648    } = args;
649    // `--wait DUR` overrides the wait budget for this call; `--wait 0` (or
650    // `--no-wait`) means don't block or warm in-process at all.
651    let wait_budget = wait.unwrap_or_else(wait_budget);
652    let no_wait = no_wait || wait_budget.is_zero();
653    // post-filters (--path, --kind, --lang) need headroom before the cutoff so a
654    // filtered-in result isn't lost to the top-N truncation
655    let limit = if args.paths.is_empty() && args.kinds.is_empty() && args.langs.is_empty() {
656        want
657    } else {
658        want.saturating_mul(20).max(PATH_HEADROOM)
659    };
660    let _timer = crate::trace::Timer::start("search done");
661    let profile_started = std::time::Instant::now();
662    let t_setup = std::time::Instant::now();
663    // Brackets the warm decision as well as the session, so it outlives both.
664    let setup_span = crate::profile::span("setup");
665    // Borrowed field-by-field so the body reads the same as when it owned them,
666    // while the session itself outlives this call and can answer again.
667    let Session {
668        store,
669        cwd,
670        cwd_is_git,
671        root,
672        active_paths,
673        branch_refresh,
674        identity,
675        coverage,
676    } = session;
677    let cwd_is_git = *cwd_is_git;
678
679    // Opportunistic indexing (Layer 5), time-bounded so the first query in a
680    // large repo never blocks on a full walk. We may warm a git work tree (safe
681    // to auto-discover) *or* any dir we already track — one earns tracking by
682    // being explicitly `--index`ed, which opts a non-git dir in. We never warm
683    // an unknown non-git dir (don't walk a random directory). A subtree index
684    // (`--index --path …`) is a seed, not a fence: coverage stays `warming`, so
685    // warming continues over the rest of the repo from here.
686    let known = coverage.is_some();
687    let warming_ok = cwd_is_git || known;
688    if crate::trace::enabled() {
689        crate::trace!(
690            "query {query:?}: root={} identity={} coverage={} warming_ok={warming_ok} active={}",
691            root.as_deref().map_or("?".into(), crate::trace::abbrev),
692            identity.as_deref().unwrap_or("none"),
693            coverage.as_deref().unwrap_or("none"),
694            active_paths.len(),
695        );
696    }
697    let repo_span = crate::profile::span("setup: repo state");
698    let current = identity
699        .as_deref()
700        .and_then(|id| store.repository_id(id).ok().flatten());
701    // Default: scope results to the current repo (when it's indexed) so a search
702    // never leaks another repo's definitions. `--all-repos` searches everything.
703    let only_repo = if all_repos { None } else { current };
704    let active = crate::search::ActiveFiles::new(active_paths.clone());
705
706    drop(repo_span);
707    let warm_span = crate::profile::span("setup: warm decision");
708
709    // Warm the index on a background thread (its own connection — WAL lets it
710    // write while we read) whenever there's work: a not-yet-complete repo, or a
711    // complete one changed since it was indexed. The search below reads whatever
712    // it has committed so far. With detach on (the default), this in-process
713    // warm only serves *this* answer — leftover work goes to a detached child
714    // after results print, so the shell never waits on it.
715    let warm_budget = if warm_detach_enabled() {
716        answer_warm_budget()
717    } else {
718        answer_warm_budget() + deferred_warm_budget()
719    };
720    let was_warming = coverage.as_deref() != Some("complete");
721
722    // On a complete repo the only question left is whether the worktree moved
723    // since it was indexed — and answering it forks `git status`, which on a
724    // large worktree is most of a query's cost. It decides nothing this answer
725    // depends on: with `was_warming` false, `block` and `polling` below are
726    // false too, the search reads the committed index, and `revalidate_top`
727    // guarantees the freshness of what we print. So start it alongside the
728    // search and collect it in `settle_warm` once results are out.
729    //
730    // A still-warming repo never ran this check at all — the `||` short-circuit
731    // saw to that — so its path here is unchanged.
732    let indexed_head = (!was_warming)
733        .then(|| current.and_then(|id| store.indexed_head(id).ok().flatten()))
734        .flatten();
735    // Whether the worktree moved is a property of the repo, not of the query,
736    // so a batch asks once (up front) instead of forking `git status` per line.
737    let staleness = (!was_warming && warming_ok && !args.batch)
738        .then(|| root.clone())
739        .flatten()
740        .map(|c| std::thread::spawn(move || worktree_changed(&c, indexed_head.as_deref())));
741    // Only a repo that's still warming warms *before* the answer now; a
742    // complete-but-edited one is reindexed by `settle_warm` afterwards.
743    let want_warm = warming_ok && was_warming && root.is_some();
744
745    // Block-until-answered on a cold/partial repo. A bounded warm exists so a
746    // query never hangs, but on a *huge, cold* repo it can expire before the
747    // symbol is indexed — turning a real hit into a false "no matches". Since
748    // correctness beats the first query's latency (and once warm the repo answers
749    // fast), we keep indexing until the answer appears or the repo is fully
750    // indexed — for humans *and* programs alike. Small/medium repos finish inside
751    // the normal budget and are unaffected; only a genuinely large cold repo
752    // waits, and only once.
753    // `--no-wait`: a scripted/agent caller that would rather answer from the
754    // committed index right now than block up to the wait budget while a
755    // background rebuild rewrites the index. It suppresses the block-until-answered
756    // escalation *and* the in-process warm (no lock contention, no join) — leftover
757    // warming still detaches below, so the index keeps improving for next time.
758    let block = want_warm && was_warming && !no_wait;
759    // A human at a plain-text terminal also gets a live progress heads-up and a
760    // graceful Ctrl-C; piped/`--json` callers (agents, scripts) block silently and
761    // are bounded by a wait budget instead, since there's nothing to draw to and
762    // no one to interrupt.
763    let progress_ui = block && show_progress(out, stderr_interactive());
764    let indexer_budget = if block { wait_budget } else { warm_budget };
765    if progress_ui {
766        install_interrupt_handler();
767    }
768
769    // `warm_done` lets the poll stop the instant the indexer finishes — so a miss
770    // on a small repo returns as soon as it's indexed, not at the deadline.
771    let warm_done = std::sync::Arc::new(std::sync::atomic::AtomicBool::new(false));
772    let indexer = (want_warm && root.is_some() && !no_wait).then(|| {
773        crate::trace!(
774            "background warm ({indexer_budget:?}, block={block}, progress_ui={progress_ui}, {} jobs)",
775            crate::index::parse_jobs()
776        );
777        let root = root.clone().expect("checked");
778        let active = active_paths.clone();
779        let q = query.to_string();
780        let warm_done = std::sync::Arc::clone(&warm_done);
781        std::thread::spawn(move || {
782            if let Ok(mut idx) = open_store() {
783                // path-prioritize toward the query so the relevant file indexes first
784                let _ = if block {
785                    // the abort flag (`INTERRUPTED`) lets a Ctrl-C, a wait timeout,
786                    // or an early answer stop the pass without losing committed work
787                    crate::index::index_budgeted_cancellable(
788                        &mut idx,
789                        &root,
790                        &active,
791                        indexer_budget,
792                        Some(&q),
793                        &INTERRUPTED,
794                    )
795                } else {
796                    crate::index::index_budgeted(&mut idx, &root, &active, indexer_budget, Some(&q))
797                };
798            }
799            warm_done.store(true, std::sync::atomic::Ordering::Relaxed);
800        })
801    });
802
803    // Poll while a cold/partial repo warms. Don't print the first hit off a sparse
804    // index — a fuzzy or path match can be wrong once more is indexed. Hold until a
805    // *high-confidence* (exact or prefix name) match appears; otherwise keep
806    // building until the index is complete (a "no matches" is then trustworthy), a
807    // wait deadline passes, or — interactively — Ctrl-C. A human sees a progress
808    // line once the pause is noticeable.
809    crate::trace!(
810        "setup (open + repo detect + warm decision): {} ms",
811        t_setup.elapsed().as_millis()
812    );
813    let poll_start = std::time::Instant::now();
814    // Deadline: an interactive block waits unbounded (Ctrl-C escapes); a
815    // programmatic block waits out the wait budget; a non-block (complete repo)
816    // keeps the original fast answer budget.
817    let deadline = if progress_ui {
818        None
819    } else if block {
820        Some(poll_start + wait_budget)
821    } else {
822        Some(poll_start + answer_warm_budget())
823    };
824    drop(warm_span);
825    let polling = indexer.is_some() && was_warming;
826    // Everything before the first search: resolving the repo root, checking
827    // coverage, deciding whether to warm. It runs on every query, so it counts
828    // toward the first-answer budget even though no searching happened yet.
829    drop(setup_span);
830    let mut query_span = crate::profile::span("query");
831    let label = repo_label(root.as_deref());
832    let mut drew_progress = false;
833    let mut last_draw = poll_start;
834    let mut hits = loop {
835        match crate::search::search(store, query, current, only_repo, &active, limit) {
836            Ok(h) => {
837                let confident = h.first().is_some_and(|hit| {
838                    hit.features
839                        .iter()
840                        .any(|f| matches!(f.name, "exact" | "prefix"))
841                });
842                let warm_finished = warm_done.load(std::sync::atomic::Ordering::Relaxed);
843                let stopped = INTERRUPTED.load(std::sync::atomic::Ordering::Relaxed);
844                let timed_out = deadline.is_some_and(|d| std::time::Instant::now() >= d);
845                if !polling || confident || warm_finished || stopped || timed_out {
846                    break h;
847                }
848                if progress_ui
849                    && poll_start.elapsed() >= HEADS_UP_DELAY
850                    && last_draw.elapsed() >= PROGRESS_REDRAW
851                {
852                    draw_progress(store, identity.as_deref(), &label);
853                    drew_progress = true;
854                    last_draw = std::time::Instant::now();
855                }
856            }
857            Err(e) => {
858                if let Some(h) = indexer {
859                    let _ = h.join();
860                }
861                return fail(format_args!("rq: {e}"));
862            }
863        }
864        std::thread::sleep(POLL_INTERVAL);
865    };
866    query_span.note(|| {
867        if polling {
868            "polled a warming index".to_string()
869        } else {
870            String::new()
871        }
872    });
873    drop(query_span);
874    if drew_progress {
875        clear_progress();
876    }
877    // Captured before we self-cancel below, so it reflects only a *user's* Ctrl-C.
878    let interrupted = INTERRUPTED.load(std::sync::atomic::Ordering::Relaxed);
879
880    // Staleness: revalidate the files behind the top hits; re-rank once if changed.
881    if !hits.is_empty() && revalidate_top(store, &hits) {
882        hits = crate::search::search(store, query, current, only_repo, &active, limit)
883            .unwrap_or_default();
884    }
885
886    // Untracked non-git dir — nothing persisted, no warmer running — so scan it
887    // live in-memory (substring, then fuzzy) and blend with whatever the index
888    // gave. The only non-persisting scan left.
889    if !hits.iter().any(strong)
890        && indexer.is_none()
891        && coverage.is_none()
892        && let Some(root) = &root
893    {
894        let tail = live_fallback(root, query, limit);
895        hits = crate::search::merge(hits, tail, limit);
896    }
897
898    apply_gates(query, &mut hits);
899    apply_post_filters(args, cwd.as_deref(), root.as_deref(), &mut hits);
900
901    if hits.is_empty() {
902        // Stop a still-running block so the join is prompt, then settle coverage.
903        if block {
904            INTERRUPTED.store(true, std::sync::atomic::Ordering::Relaxed);
905        }
906        if let Some(h) = indexer {
907            let _ = h.join();
908        }
909        // A miss against a *complete* index is definitive (the symbol isn't
910        // there); against a still-warming one it's only "not yet". Distinguish
911        // them so a caller — agent or script — isn't misled into thinking the
912        // symbol is absent when the index simply hasn't reached it. `--no-wait`
913        // returns without blocking, so its miss is judged the same way — an
914        // incomplete index yields `warming` (exit 2, "retry"), not a false absence.
915        let mut incomplete = (block || no_wait)
916            && identity
917                .as_deref()
918                .and_then(|id| store.coverage_status(id).ok().flatten())
919                .as_deref()
920                != Some("complete");
921        // a "not yet" miss leaves work behind — reindex an edited worktree and
922        // let a detached child keep warming, so the retry lands on a better
923        // index. This is the path a just-added symbol takes, so it has to do
924        // the same settling the render path does.
925        // A worktree that has moved since we indexed it makes this miss
926        // provisional, not definitive: the symbol may be in an edit the
927        // detached warm hasn't caught up with. Say "warming" (exit 2, retry)
928        // rather than "no match" (exit 1, absent) — a just-added symbol is
929        // exactly this case, and a confident no is the wrong answer to it.
930        incomplete |= settle_warm(
931            store,
932            staleness,
933            was_warming,
934            warming_ok,
935            root.as_deref(),
936            active_paths,
937            query,
938            warm_budget,
939            no_wait,
940            identity.as_deref(),
941        );
942        // Recorded here rather than above the branch: whether this was a
943        // definitive miss or a not-ready one is only known now, and counting
944        // them as one number overstates how often rq truly finds nothing.
945        record_usage(
946            store,
947            args,
948            current,
949            0,
950            if incomplete { "warming" } else { "miss" },
951            coverage.as_deref(),
952        );
953        return no_match_code(out, query, interrupted, incomplete);
954    }
955
956    // The hit path's single count, above the --show/--open/list forks so it
957    // covers all three.
958    record_usage(store, args, current, hits.len(), "hit", coverage.as_deref());
959
960    // Attach each result's definition line (e.g. `def perform(refund)`) — shown
961    // in text output and carried in JSON. Cheap: only the displayed results.
962    for hit in &mut hits {
963        hit.signature = read_signature(
964            store,
965            &hit.repo_identity,
966            &hit.file,
967            hit.line,
968            cwd.as_deref(),
969        );
970    }
971    attach_confidence(&mut hits);
972
973    // --show: print the top hit's full source when confident; otherwise fall
974    // through to the normal ranked list (rq won't dump a body it isn't sure of).
975    if show
976        && let Some(code) = show_top_definition(
977            store,
978            &mut hits,
979            query,
980            out,
981            cwd.as_deref(),
982            current,
983            no_record,
984        )
985    {
986        return code;
987    }
988
989    // --open: pick the best match (prompting on a TTY with several), record the
990    // pick so ranking learns, and hand off to the editor. Returns before the
991    // normal print / warm-join — opening should be snappy, and a launcher `exec`s.
992    if open {
993        return finish_open(store, &hits, query, current, root.as_deref(), no_record);
994    }
995
996    if let Some(code) = render_hits(args, &hits) {
997        return code;
998    }
999
1000    // Report before the deferred maintenance below, so the total covers
1001    // getting answers out rather than the bookkeeping that follows them.
1002    if crate::profile::enabled() {
1003        let total = profile_started.elapsed();
1004        if args.out == Output::Text {
1005            for line in crate::profile::report(total) {
1006                eprintln!("{line}");
1007            }
1008        } else {
1009            // stdout stays exactly the results, so the profile can be captured
1010            // separately and diffed.
1011            eprintln!("{}", crate::profile::json(total));
1012        }
1013    }
1014
1015    // Collect the refresh started back at setup. It ran alongside the search
1016    // rather than after it, so by now it has usually finished — and it only
1017    // ever feeds the *next* query, never this one's ranking, so waiting on it
1018    // can't reorder what was just printed.
1019    // Taken, not borrowed: the refresh is one-shot, and a session answering
1020    // several queries must not re-store a result it already consumed.
1021    if let Some(refresh) = branch_refresh.take() {
1022        refresh.store(store);
1023    }
1024
1025    // Results are out — now the cheap deferred work, amortized across
1026    // interactions: roll the `open`/`select` picks that teach ranking into
1027    // `selection_stats`, and prune the raw log. The `search` row written above
1028    // is skipped by the rollup — it counts usage, it doesn't teach.
1029    deferred_maintenance(store);
1030
1031    // Results are out; stop the in-process warm (it persists as it goes, so a
1032    // cut pass keeps everything parsed) and join it — then hand whatever's left
1033    // to a detached child, which finishes coverage with a budget no foreground
1034    // query could afford. The shell only ever waits on the answer.
1035    if block {
1036        INTERRUPTED.store(true, std::sync::atomic::Ordering::Relaxed);
1037    }
1038    if let Some(h) = indexer {
1039        let _ = h.join();
1040    }
1041    let _ = settle_warm(
1042        store,
1043        staleness,
1044        was_warming,
1045        warming_ok,
1046        root.as_deref(),
1047        active_paths,
1048        query,
1049        warm_budget,
1050        no_wait,
1051        identity.as_deref(),
1052    );
1053
1054    ExitCode::SUCCESS
1055}
1056
1057/// Re-exec a detached warm child when this query's warming didn't finish the
1058/// job. No-op when detach is off, nothing was warming, or coverage completed.
1059fn maybe_detach_warm(
1060    store: &Store,
1061    want_warm: bool,
1062    changed: bool,
1063    root: Option<&std::path::Path>,
1064    identity: Option<&str>,
1065) {
1066    if !warm_detach_enabled() || !want_warm {
1067        return;
1068    }
1069    let (Some(root), Some(id)) = (root, identity) else {
1070        return;
1071    };
1072    // Coverage measures breadth, not freshness — an edit never demotes it. So
1073    // "complete" alone isn't done; it's done only if the worktree also hasn't
1074    // moved since we indexed it.
1075    if !changed && store.coverage_status(id).ok().flatten().as_deref() == Some("complete") {
1076        return; // the in-process pass finished the job
1077    }
1078    spawn_detached_warm(root);
1079}
1080
1081/// Spawn `rq --warm <root>` fully detached: null stdio and its own process
1082/// group, so it survives this process and a later Ctrl-C in the terminal
1083/// can't reach it. The child nices itself and is single-flighted per repo.
1084fn spawn_detached_warm(root: &std::path::Path) {
1085    use std::os::unix::process::CommandExt;
1086    let Ok(exe) = std::env::current_exe() else {
1087        return;
1088    };
1089    let mut cmd = std::process::Command::new(exe);
1090    cmd.arg("--warm")
1091        .arg(root)
1092        .stdin(std::process::Stdio::null())
1093        .stdout(std::process::Stdio::null())
1094        .stderr(std::process::Stdio::null())
1095        .process_group(0);
1096    match cmd.spawn() {
1097        Ok(child) => crate::trace!(
1098            "background warm (detached): pid {} for {}",
1099            child.id(),
1100            crate::trace::abbrev(root)
1101        ),
1102        Err(e) => crate::trace!("detached warm failed to spawn: {e}"),
1103    }
1104}
1105
1106/// How long a warm lock is trusted without a liveness hit — past this, a
1107/// stamp is a crashed warmer's leftover and a new child takes over.
1108const WARM_LOCK_TTL_SECS: i64 = 600;
1109
1110/// `rq --warm [PATH]`: the detached child a search re-execs after printing —
1111/// finishes warming the repo's index in the background. Niced so it stays out
1112/// of the foreground's way; single-flighted per repo so a burst of queries
1113/// runs at most one warmer. Safe (and boring) to run by hand.
1114fn cmd_warm(path: Option<&str>) -> ExitCode {
1115    // Stay out of the way: drop scheduling priority, and throttle disk I/O on
1116    // macOS. Best-effort — a failure just means a less-polite warm.
1117    #[cfg(target_os = "macos")]
1118    unsafe extern "C" {
1119        // <sys/resource.h>; not in the libc crate. Args below:
1120        // IOPOL_TYPE_DISK=0, IOPOL_SCOPE_PROCESS=0, IOPOL_THROTTLE=3.
1121        fn setiopolicy_np(
1122            iotype: libc::c_int,
1123            scope: libc::c_int,
1124            policy: libc::c_int,
1125        ) -> libc::c_int;
1126    }
1127    unsafe {
1128        libc::nice(10);
1129        #[cfg(target_os = "macos")]
1130        setiopolicy_np(0, 0, 3);
1131    }
1132    let mut store = match open_store() {
1133        Ok(s) => s,
1134        Err(_) => return ExitCode::FAILURE,
1135    };
1136    let start = path
1137        .map(PathBuf::from)
1138        .or_else(|| std::env::current_dir().ok())
1139        .unwrap_or_else(|| PathBuf::from("."));
1140    let root = crate::index::repo_root(&start).unwrap_or(start);
1141    let identity = resolve_identity(&store, &root);
1142
1143    // Single-flight: if another live rq is already warming this repo, bow out.
1144    // A dead pid or a stale stamp is a crashed warmer — take over.
1145    if let Ok(Some((pid, ts))) = store.warm_lock(&identity)
1146        && pid != std::process::id()
1147        && unsafe { libc::kill(pid as libc::pid_t, 0) } == 0
1148        && now_secs() - ts < WARM_LOCK_TTL_SECS
1149    {
1150        return ExitCode::SUCCESS;
1151    }
1152    let _ = store.set_warm_lock(&identity, std::process::id());
1153
1154    // Sweep until coverage completes, the budget runs out, or a pass stops
1155    // making progress (each pass converges — mtime-skips what's done).
1156    let deadline = std::time::Instant::now() + warm_bg_budget();
1157    let active = crate::index::branch_changed_files(&root);
1158    loop {
1159        let remaining = deadline.saturating_duration_since(std::time::Instant::now());
1160        if remaining.is_zero() {
1161            break;
1162        }
1163        let stats = match crate::index::index_budgeted(&mut store, &root, &active, remaining, None)
1164        {
1165            Ok(s) => s,
1166            Err(_) => break,
1167        };
1168        if store.coverage_status(&identity).ok().flatten().as_deref() == Some("complete")
1169            || stats.files_indexed == 0
1170        {
1171            break;
1172        }
1173    }
1174    let _ = store.clear_warm_lock(&identity);
1175    ExitCode::SUCCESS
1176}
1177
1178fn now_secs() -> i64 {
1179    std::time::SystemTime::now()
1180        .duration_since(std::time::UNIX_EPOCH)
1181        .map(|d| d.as_secs() as i64)
1182        .unwrap_or(0)
1183}
1184
1185/// Live in-memory scan of an untracked (non-git, never-indexed) dir: substring
1186/// pre-filtered first, then the unfiltered fuzzy retry. Persists nothing.
1187fn live_fallback(root: &std::path::Path, query: &str, limit: usize) -> Vec<crate::search::Hit> {
1188    crate::trace!("empty → live (in-memory) scan of an untracked dir");
1189    let deadline = std::time::Instant::now() + live_fallback_budget();
1190    let h = crate::search::live_search(root, query, limit, &HashSet::new(), Some(deadline), true);
1191    if !h.is_empty() {
1192        return h;
1193    }
1194    crate::search::live_search(root, query, limit, &HashSet::new(), Some(deadline), false)
1195}
1196
1197/// A high-confidence name match: exact or prefix (not fuzzy/path-only).
1198fn strong(h: &crate::search::Hit) -> bool {
1199    h.features
1200        .iter()
1201        .any(|f| matches!(f.name, "exact" | "prefix"))
1202}
1203
1204/// The result-quality gates, in order:
1205/// - relevance: when the query lands a real name match (exact or prefix), drop
1206///   the scattered fuzzy / path-only near-matches — they're noise next to a
1207///   solid hit, and rq favors fewer, better results. A purely-fuzzy query (no
1208///   exact/prefix anywhere) keeps its matches.
1209/// - scope: a qualified query (`Foo::Bar#baz`) that lands inside the named
1210///   scope keeps only the in-scope results; if none match, the others stay
1211///   (the definition may live elsewhere).
1212fn apply_gates(query: &str, hits: &mut Vec<crate::search::Hit>) {
1213    if hits.iter().any(strong) {
1214        hits.retain(strong);
1215    }
1216    crate::search::apply_scope_gate(query, hits);
1217}
1218
1219/// Post-filters: keep only results under a `--path` dir, of a `--kind`, and/or
1220/// in a `--lang`, then trim to the requested count.
1221fn apply_post_filters(
1222    args: &SearchArgs,
1223    cwd: Option<&std::path::Path>,
1224    root: Option<&std::path::Path>,
1225    hits: &mut Vec<crate::search::Hit>,
1226) {
1227    if !args.paths.is_empty() {
1228        // --path values may be absolute or cwd-relative; stored files are
1229        // repo-root-relative, so normalize before prefix-matching or an
1230        // absolute path would silently filter everything out.
1231        let here = cwd.map_or_else(|| PathBuf::from("."), PathBuf::from);
1232        let base = root.map_or_else(|| here.clone(), PathBuf::from);
1233        let norm: Vec<String> = args
1234            .paths
1235            .iter()
1236            .map(|p| repo_relative(&base, &here, p))
1237            .collect();
1238        hits.retain(|h| under_any(&h.file, &norm));
1239    }
1240    if !args.kinds.is_empty() {
1241        hits.retain(|h| args.kinds.iter().any(|k| k == &h.kind));
1242    }
1243    if !args.langs.is_empty() {
1244        hits.retain(|h| args.langs.iter().any(|l| l == &h.language));
1245    }
1246    if !args.paths.is_empty() || !args.kinds.is_empty() || !args.langs.is_empty() {
1247        hits.truncate(args.want);
1248    }
1249}
1250
1251/// Report a miss and pick its exit code. Structured callers get a reason, not
1252/// a bare `[]`/empty: `warming` (retry — index incomplete), `interrupted` (a
1253/// stopped block), or `no_match` (definitive). Text keeps its human message.
1254/// Exit 2 = indeterminate (index incomplete), 1 = a definitive miss — both
1255/// non-zero, so `rq … && …` still reads as "found something".
1256fn no_match_code(out: Output, query: &str, interrupted: bool, incomplete: bool) -> ExitCode {
1257    let status = if interrupted {
1258        "interrupted"
1259    } else if incomplete {
1260        "warming"
1261    } else {
1262        "no_match"
1263    };
1264    match out {
1265        Output::Json | Output::Ndjson => {
1266            let obj = serde_json::json!({ "status": status, "query": query });
1267            let _ = emit_json(out, &obj); // the exit code below carries the miss
1268        }
1269        Output::Text if interrupted => {
1270            eprintln!("rq: indexing interrupted — run again to finish")
1271        }
1272        Output::Text if incomplete => eprintln!(
1273            "rq: still indexing — no match for {query:?} yet (run again, or `rq --index` to finish)"
1274        ),
1275        Output::Text => eprintln!("no matches for {query:?}"),
1276    }
1277    if incomplete {
1278        ExitCode::from(2)
1279    } else {
1280        ExitCode::FAILURE
1281    }
1282}
1283
1284/// Normalized confidence per hit: match quality scaled by dominance over the
1285/// other results (needs the whole ranked set). "Best other" is the top score —
1286/// or the runner-up, for the top hit itself.
1287fn attach_confidence(hits: &mut [crate::search::Hit]) {
1288    let (top, second) = hits.iter().fold((None::<f64>, None::<f64>), |(t, s), h| {
1289        if t.is_none_or(|t| h.score > t) {
1290            (Some(h.score), t)
1291        } else if s.is_none_or(|s| h.score > s) {
1292            (t, Some(h.score))
1293        } else {
1294            (t, s)
1295        }
1296    });
1297    for hit in hits.iter_mut() {
1298        let best_other = if Some(hit.score) == top { second } else { top };
1299        hit.confidence = crate::search::confidence(
1300            hit.score,
1301            crate::search::match_quality(&hit.features),
1302            best_other,
1303        );
1304    }
1305}
1306
1307/// Print the ranked results (JSON array, NDJSON lines, or highlighted text).
1308/// `Some(exit)` on a serialization failure, `None` on success.
1309fn render_hits(args: &SearchArgs, hits: &[crate::search::Hit]) -> Option<ExitCode> {
1310    // Time to the first printed result, not to the last: rq streams, and the
1311    // sub-50 ms budget is about the first answer. A change that speeds the
1312    // total while delaying this one is a regression.
1313    let render_span = crate::profile::span("render");
1314    if args.batch {
1315        // One stream, many questions: tag each row with the query it answers,
1316        // the same way `no_match_code` already tags a miss.
1317        #[derive(serde::Serialize)]
1318        struct Tagged<'a> {
1319            query: &'a str,
1320            #[serde(flatten)]
1321            hit: &'a crate::search::Hit,
1322        }
1323        let rows: Vec<Tagged> = hits
1324            .iter()
1325            .map(|hit| Tagged {
1326                query: args.query,
1327                hit,
1328            })
1329            .collect();
1330        if let Some(code) = emit_rows(args.out, &rows) {
1331            return Some(code);
1332        }
1333    } else if let Some(code) = emit_rows(args.out, hits) {
1334        return Some(code);
1335    }
1336    if args.out != Output::Text {
1337        return None;
1338    }
1339    drop(render_span);
1340    let color = match_color();
1341    let c = color.as_deref();
1342    let query = args.query;
1343    if args.show {
1344        // fell through from --show: no single confident match to print
1345        eprintln!(
1346            "rq: no single confident match for {query:?} — {} candidates below; narrow the query to --show one",
1347            hits.len()
1348        );
1349    }
1350    for hit in hits {
1351        // highlight the chars the query matched — in the name, the
1352        // filename, and the definition line (great for fuzzy matches)
1353        let name = hl(&hit.name, query, c);
1354        let qualified = match &hit.parent {
1355            Some(p) => format!("{name} · {p}"),
1356            None => name,
1357        };
1358        println!(
1359            "{}:{}  {} {}",
1360            hl_path(&hit.file, query, c),
1361            hit.line,
1362            hit.kind,
1363            qualified
1364        );
1365        if let Some(sig) = &hit.signature {
1366            println!("    {}", hl(sig, query, c));
1367        }
1368        if args.explain {
1369            let parts: Vec<String> = hit
1370                .features
1371                .iter()
1372                .map(|f| format!("{} {:.0}", f.name, f.value))
1373                .collect();
1374            println!(
1375                "    confidence {:.2} · score {:.0} = {}",
1376                hit.confidence,
1377                hit.score,
1378                parts.join(" + ")
1379            );
1380        }
1381    }
1382    None
1383}
1384
1385/// Pick a hit for `--open`: the top match, unless we're on an interactive
1386/// terminal with several — then print a short numbered menu and read a choice
1387/// (empty = the top match). `None` means abort (EOF or unparseable input).
1388fn choose_hit(hits: &[crate::search::Hit]) -> Option<&crate::search::Hit> {
1389    use std::io::{IsTerminal, Write};
1390    if hits.len() == 1 || !std::io::stdin().is_terminal() || !std::io::stderr().is_terminal() {
1391        return hits.first();
1392    }
1393    let mut err = std::io::stderr();
1394    let _ = writeln!(err, "rq: {} matches — pick one (enter = 1):", hits.len());
1395    for (i, h) in hits.iter().enumerate() {
1396        let _ = writeln!(
1397            err,
1398            "  {}. {}:{}  {} {}",
1399            i + 1,
1400            h.file,
1401            h.line,
1402            h.kind,
1403            h.name
1404        );
1405    }
1406    let _ = write!(err, "rq> ");
1407    let _ = err.flush();
1408    let mut line = String::new();
1409    if std::io::stdin().read_line(&mut line).unwrap_or(0) == 0 {
1410        return None; // Ctrl-D
1411    }
1412    parse_choice(&line, hits.len()).and_then(|i| hits.get(i))
1413}
1414
1415/// Resolve a menu reply to a 0-based index: blank → 0 (the top match), `N` → N-1
1416/// when in range, anything else → `None` (abort). Pure, so it's unit-tested.
1417fn parse_choice(input: &str, n: usize) -> Option<usize> {
1418    let s = input.trim();
1419    if s.is_empty() {
1420        return Some(0);
1421    }
1422    let i = s.parse::<usize>().ok()?.checked_sub(1)?;
1423    (i < n).then_some(i)
1424}
1425
1426/// `--open`: choose a hit, record it as a selection so ranking learns, then hand
1427/// off to the editor. The launcher `exec`s (replacing this process), so the shell
1428/// waits on the editor — not on rq's background warm.
1429fn finish_open(
1430    store: &mut Store,
1431    hits: &[crate::search::Hit],
1432    query: &str,
1433    current: Option<i64>,
1434    root: Option<&std::path::Path>,
1435    no_record: bool,
1436) -> ExitCode {
1437    let Some(hit) = choose_hit(hits) else {
1438        return ExitCode::SUCCESS; // aborted at the prompt
1439    };
1440
1441    // Record the pick — same signal as `rq --record`. The hit's path is already
1442    // repo-relative, which is what the selection rollup keys off.
1443    if !no_record {
1444        let _ = store.record_event(
1445            "select",
1446            Some(&query.to_ascii_lowercase()),
1447            current,
1448            Some(&hit.file),
1449            Some(hit.line),
1450            None,
1451        );
1452        deferred_maintenance(store);
1453    }
1454
1455    // Results are repo-root-relative, so resolve against the root — the bare path
1456    // wouldn't open from a subdirectory.
1457    let target = match root {
1458        Some(r) => r.join(&hit.file),
1459        None => PathBuf::from(&hit.file),
1460    };
1461    launch_editor(&target, hit.line)
1462}
1463
1464/// Launch the editor on `file:line`, resolving the command in order: `RQ_OPEN`
1465/// template → VS Code (`code`) → `$VISUAL`/`$EDITOR` → print the location. The
1466/// chosen command replaces this process via `exec`.
1467fn launch_editor(file: &std::path::Path, line: i64) -> ExitCode {
1468    use std::os::unix::process::CommandExt;
1469    let loc = format!("{}:{}", file.display(), line);
1470    match open_command(file, line, &loc) {
1471        Some((prog, args)) => {
1472            // exec returns only on failure
1473            let err = std::process::Command::new(&prog).args(&args).exec();
1474            fail(format_args!("rq --open: cannot run {prog}: {err}"))
1475        }
1476        None => {
1477            println!("{loc}");
1478            ExitCode::SUCCESS
1479        }
1480    }
1481}
1482
1483/// Resolve the editor command + args. `None` → no launcher configured (the
1484/// caller prints the location). `RQ_OPEN` is split on whitespace (no shell) with
1485/// `{file}` / `{line}` / `{}` (= `path:line`) substituted per token.
1486fn open_command(file: &std::path::Path, line: i64, loc: &str) -> Option<(String, Vec<String>)> {
1487    let fstr = file.to_string_lossy().into_owned();
1488
1489    if let Some(t) = std::env::var_os("RQ_OPEN") {
1490        let t = t.to_string_lossy();
1491        let mut parts = t.split_whitespace().map(|p| {
1492            p.replace("{file}", &fstr)
1493                .replace("{line}", &line.to_string())
1494                .replace("{}", loc)
1495        });
1496        if let Some(prog) = parts.next() {
1497            return Some((prog, parts.collect()));
1498        }
1499    }
1500
1501    if on_path("code") {
1502        return Some(("code".into(), vec!["--goto".into(), loc.into()]));
1503    }
1504
1505    if let Some(ed) = std::env::var_os("VISUAL").or_else(|| std::env::var_os("EDITOR")) {
1506        let ed = ed.to_string_lossy().into_owned();
1507        let l = ed.to_ascii_lowercase();
1508        // line-aware launch for the common terminal editors; others just get the file
1509        if ["vim", "nvim", "vi", "nano", "emacs", "kak", "micro"]
1510            .iter()
1511            .any(|e| l.contains(e))
1512        {
1513            return Some((ed, vec![format!("+{line}"), fstr]));
1514        }
1515        return Some((ed, vec![fstr]));
1516    }
1517
1518    None
1519}
1520
1521/// Whether `prog` resolves on `PATH` (a regular file; symlinks followed).
1522fn on_path(prog: &str) -> bool {
1523    std::env::var_os("PATH")
1524        .is_some_and(|paths| std::env::split_paths(&paths).any(|dir| dir.join(prog).is_file()))
1525}
1526
1527/// Whether a complete repo is provably unchanged since its last index — same
1528/// HEAD and a clean work tree — so the deferred re-walk can be skipped. The git
1529/// HEAD + dirty check is cheap (~tens of ms) and authoritative at any size, so
1530/// it gates warming for small and large repos alike: a clean, fully-indexed repo
1531/// has nothing to warm, and re-walking it on every query just to discover that
1532/// wasted a full sweep (~hundreds of ms) per search. Conservative: any
1533/// uncertainty (not complete, non-git / no recorded head, git hiccup) returns
1534/// false, so we warm.
1535/// Seconds since the epoch.
1536fn unix_now() -> i64 {
1537    std::time::SystemTime::now()
1538        .duration_since(std::time::UNIX_EPOCH)
1539        .map(|d| d.as_secs() as i64)
1540        .unwrap_or(0)
1541}
1542
1543/// How long a branch-file list is served before it's refreshed. A commit or a
1544/// checkout is caught by the stamp; a bare working-tree edit touches neither
1545/// `.git/HEAD` nor `.git/index`, so only elapsed time catches that — short
1546/// enough that a burst of searches shares one computation and the edit you just
1547/// made is reflected on the next search.
1548const BRANCH_FILES_TTL_SECS: i64 = 15;
1549
1550/// A branch-file recomputation running alongside the search. The git work
1551/// happens on the thread; the store write waits for the main thread, since a
1552/// SQLite connection isn't shared.
1553struct BranchRefresh {
1554    handle: std::thread::JoinHandle<Vec<String>>,
1555    identity: String,
1556    stamp: String,
1557}
1558
1559impl BranchRefresh {
1560    /// Wait for the recomputation and store it for the next query.
1561    fn store(self, store: &Store) {
1562        let Ok(files) = self.handle.join() else {
1563            return;
1564        };
1565        let _ = store.branch_files_set(&self.identity, &self.stamp, unix_now(), &files);
1566    }
1567}
1568
1569/// The branch-changed file list, served from the store when it's still good.
1570/// Returns the list, plus a recomputation to collect after results print when
1571/// the stored one has aged out.
1572///
1573/// The list feeds a *ranking boost*, so serving a slightly old one costs a
1574/// little ranking quality, while recomputing it first would cost every search
1575/// the git diff behind it — which is O(tracked files). So the stored list is
1576/// served immediately and the refresh runs concurrently with the search rather
1577/// than after it, which usually hides its cost entirely. It feeds only the next
1578/// query, so nothing about this run's ranking depends on how the race lands.
1579///
1580/// The first search in a repo has nothing to serve and computes inline; that's
1581/// once per repo, like the first index.
1582fn cached_branch_files(
1583    store: &Store,
1584    root: &std::path::Path,
1585) -> (Vec<String>, Option<BranchRefresh>) {
1586    let identity = resolve_identity(store, root);
1587    let stamp = crate::index::branch_files_stamp(root);
1588    let cached = store.branch_files_get(&identity).ok().flatten();
1589    let now = unix_now();
1590
1591    if let (Some((cached_stamp, at, files)), Some(stamp)) = (&cached, &stamp) {
1592        if cached_stamp == stamp && now.saturating_sub(*at) < BRANCH_FILES_TTL_SECS {
1593            return (files.clone(), None);
1594        }
1595        let owned_root = root.to_path_buf();
1596        let refresh = BranchRefresh {
1597            handle: std::thread::spawn(move || crate::index::branch_changed_files(&owned_root)),
1598            identity,
1599            stamp: stamp.clone(),
1600        };
1601        return (files.clone(), Some(refresh));
1602    }
1603
1604    // Nothing cached (or nowhere to cache it, e.g. a worktree): compute inline.
1605    let files = crate::index::branch_changed_files(root);
1606    if let Some(stamp) = stamp {
1607        let _ = store.branch_files_set(&identity, &stamp, now, &files);
1608    }
1609    (files, None)
1610}
1611
1612/// Whether the worktree has moved since it was indexed — a different HEAD, or
1613/// uncommitted edits. Split out from the store read so this half can run on its
1614/// own thread: `is_dirty` forks `git status`, which on a large worktree costs
1615/// more than the search it was gating (measured: 12.6ms of a 16.8ms query on a
1616/// 6k-file repo, against 0.1ms on a 54-file one).
1617///
1618/// `None` for `indexed_head` means we never recorded one, which counts as
1619/// changed — there's nothing to compare against, so assume work is due.
1620fn worktree_changed(cwd: &std::path::Path, indexed_head: Option<&str>) -> bool {
1621    let Some(head) = indexed_head else {
1622        return true;
1623    };
1624    crate::index::git_head(cwd).as_deref() != Some(head) || crate::index::is_dirty(cwd)
1625}
1626
1627/// Settle warming once the answer is out: collect the staleness check started
1628/// back at setup, reindex if the worktree moved, and hand any remainder to a
1629/// detached child.
1630///
1631/// Called from *both* exits. The miss path matters as much as the render one —
1632/// a symbol added a moment ago is precisely a miss, and reindexing before we
1633/// exit is what makes the immediate retry hit.
1634#[allow(clippy::too_many_arguments)]
1635fn settle_warm(
1636    store: &Store,
1637    staleness: Option<std::thread::JoinHandle<bool>>,
1638    was_warming: bool,
1639    warming_ok: bool,
1640    root: Option<&std::path::Path>,
1641    active: &[String],
1642    query: &str,
1643    budget: Duration,
1644    no_wait: bool,
1645    identity: Option<&str>,
1646) -> bool {
1647    // A panicked check counts as changed: warming needlessly costs a little
1648    // time, skipping it wrongly serves a stale index.
1649    let changed = staleness.is_some_and(|h| h.join().unwrap_or(true));
1650    // Reindexing an edited worktree means sweeping every file to find the few
1651    // that moved — ~32ms on a 3000-file repo, and it was paid on *every* query
1652    // for as long as anything stayed uncommitted, which is exactly while you're
1653    // working. The shell shouldn't wait for that: hand it to the detached
1654    // child, which is what "the shell never waits on it" already promises
1655    // everywhere else.
1656    //
1657    // With detach off (the harness pins it so no child races a test's cleanup)
1658    // there's nobody to hand it to, so do it here as before.
1659    if changed
1660        && !no_wait
1661        && !warm_detach_enabled()
1662        && let Some(r) = root
1663        && let Ok(mut idx) = open_store()
1664    {
1665        crate::trace!("background warm (deferred, {budget:?}): worktree changed since index");
1666        let _ = crate::index::index_budgeted(&mut idx, r, active, budget, Some(query));
1667    }
1668    maybe_detach_warm(
1669        store,
1670        warming_ok && (was_warming || changed),
1671        changed,
1672        root,
1673        identity,
1674    );
1675    // Report only that work was *deferred*, which is what makes a miss
1676    // provisional. When the reindex ran inline just above (detach off), the
1677    // index is as current as we can make it and a miss is definitive.
1678    changed && warm_detach_enabled()
1679}
1680
1681/// Inline warm budget on the search path. A *cap*, not a fixed delay:
1682/// `index_budgeted` returns the moment a full sweep finishes, so small/medium
1683/// repos index completely and pay only their real cost. The cap only bites a
1684/// genuinely huge, never-indexed repo — where a bigger budget buys a much better
1685/// first answer (a tiny budget can return nothing, since a git repo has no
1686/// live-scan fallback). 500 ms is a one-time cold-cache cost, trivial next to
1687/// scanning a large tree from scratch; the deferred pass and later queries fill
1688/// in the rest.
1689fn answer_warm_budget() -> Duration {
1690    env_budget("RQ_ANSWER_BUDGET_MS", 500)
1691}
1692
1693/// Deferred warm budget, spent after results are printed: larger, to make real
1694/// progress on coverage per query while keeping each invocation snappy.
1695fn deferred_warm_budget() -> Duration {
1696    env_budget("RQ_DEFERRED_BUDGET_MS", 250)
1697}
1698
1699/// Bound for the git-repo live-scan fallback (index empty, still warming): enough
1700/// to surface a result the warm hasn't reached, without an unbounded walk.
1701fn live_fallback_budget() -> Duration {
1702    env_budget("RQ_FALLBACK_BUDGET_MS", 250)
1703}
1704
1705/// Budget for the *detached* warm child — generous, because nothing waits on
1706/// it: the shell got its results and the child runs niced in the background.
1707fn warm_bg_budget() -> Duration {
1708    env_budget("RQ_WARM_BUDGET_MS", 20_000)
1709}
1710
1711/// Whether a search hands leftover warming to a detached child (default) or
1712/// finishes it in-process before exiting (`RQ_WARM_DETACH=0` — used by the
1713/// test harness for hermetic runs, and handy for debugging).
1714fn warm_detach_enabled() -> bool {
1715    std::env::var("RQ_WARM_DETACH").map_or(true, |v| v != "0")
1716}
1717
1718/// How long a query may block indexing a cold repo before giving up with an
1719/// honest "still indexing" rather than a false miss. A generous backstop, not the
1720/// real cost: `index_budgeted` returns the moment the sweep completes, so any
1721/// normal repo finishes well under it, and an interactive run isn't bounded by it
1722/// at all (Ctrl-C escapes). It mainly bounds a programmatic caller on a
1723/// pathologically huge repo — where the partial index still persists for the next
1724/// query. `RQ_WAIT_BUDGET_MS=0` makes a programmatic caller non-blocking again —
1725/// it answers immediately from whatever's already indexed.
1726fn wait_budget() -> Duration {
1727    env_budget("RQ_WAIT_BUDGET_MS", 60_000)
1728}
1729
1730/// Parse a `--wait` value into a duration: `<n>ms`, `<n>s`, `<n>m`, or a bare
1731/// `<n>` (seconds). Fractions are allowed (`1.5s`); `0` (any unit) means "don't
1732/// wait". A `clap` value parser, so an invalid duration is rejected at parse
1733/// time with a usage error.
1734fn parse_wait(s: &str) -> std::result::Result<Duration, String> {
1735    let s = s.trim();
1736    let bad = || format!("invalid duration {s:?} — use e.g. 50ms, 2s, 1m, or 0");
1737    // check "ms" before "s" so the "s" arm doesn't swallow it
1738    let (num, unit_ms) = if let Some(n) = s.strip_suffix("ms") {
1739        (n, 1.0)
1740    } else if let Some(n) = s.strip_suffix('s') {
1741        (n, 1_000.0)
1742    } else if let Some(n) = s.strip_suffix('m') {
1743        (n, 60_000.0)
1744    } else {
1745        // a bare number is seconds
1746        (s, 1_000.0)
1747    };
1748    let val: f64 = num.trim().parse().map_err(|_| bad())?;
1749    if !val.is_finite() || val < 0.0 {
1750        return Err(bad());
1751    }
1752    Ok(Duration::from_millis((val * unit_ms).round() as u64))
1753}
1754
1755/// Set by the SIGINT handler during an interactive cold-start escalation. The
1756/// poll loop and the running index pass watch it, so Ctrl-C stops the wait
1757/// promptly and prints the best partial results instead of killing the process.
1758static INTERRUPTED: std::sync::atomic::AtomicBool = std::sync::atomic::AtomicBool::new(false);
1759
1760extern "C" fn on_sigint(_: libc::c_int) {
1761    // Async-signal-safe: a lone relaxed atomic store — no allocation, no locks.
1762    INTERRUPTED.store(true, std::sync::atomic::Ordering::Relaxed);
1763}
1764
1765/// Install the SIGINT handler once. Scoped to the escalation path: a normal fast
1766/// query keeps the default behavior (Ctrl-C kills it outright).
1767fn install_interrupt_handler() {
1768    static ONCE: std::sync::Once = std::sync::Once::new();
1769    ONCE.call_once(|| unsafe {
1770        let mut action: libc::sigaction = std::mem::zeroed();
1771        action.sa_sigaction = on_sigint as *const () as usize;
1772        libc::sigemptyset(&mut action.sa_mask);
1773        libc::sigaction(libc::SIGINT, &action, std::ptr::null_mut());
1774    });
1775}
1776
1777/// Is a human watching stderr? True for a real terminal; `RQ_ASSUME_INTERACTIVE`
1778/// forces it on so the progress/Ctrl-C path is exercisable under test (where
1779/// stderr is a pipe), mirroring the `RQ_*_BUDGET_MS` testing knobs.
1780fn stderr_interactive() -> bool {
1781    std::io::stderr().is_terminal() || std::env::var_os("RQ_ASSUME_INTERACTIVE").is_some()
1782}
1783
1784/// Whether to show the live "indexing…" progress heads-up and handle Ctrl-C
1785/// gracefully while a cold repo blocks — a human watching a plain-text terminal.
1786/// Piped / `--json` / `--ndjson` callers block silently instead (no line to draw,
1787/// no one to interrupt); the *decision to block* is the same for both.
1788fn show_progress(out: Output, interactive: bool) -> bool {
1789    interactive && matches!(out, Output::Text)
1790}
1791
1792/// A short, friendly name for the repo being indexed — its directory name, for
1793/// the progress line.
1794fn repo_label(root: Option<&std::path::Path>) -> String {
1795    root.and_then(|r| r.file_name())
1796        .map(|n| n.to_string_lossy().into_owned())
1797        .unwrap_or_else(|| "repo".into())
1798}
1799
1800/// Redraw the in-place "indexing…" progress line on stderr (kept off stdout so
1801/// piped/`--json` output stays clean). The file count comes from the index the
1802/// background pass is filling, so it climbs as warming proceeds.
1803fn draw_progress(store: &Store, identity: Option<&str>, label: &str) {
1804    let files = identity
1805        .and_then(|id| store.repository_id(id).ok().flatten())
1806        .and_then(|rid| store.repo_totals(rid).ok())
1807        .map_or(0, |(f, _)| f);
1808    eprint!("\r\x1b[Krq: indexing {label}… {files} files");
1809    let _ = std::io::stderr().flush();
1810}
1811
1812/// Erase the progress line so results print to a clean terminal.
1813fn clear_progress() {
1814    eprint!("\r\x1b[K");
1815    let _ = std::io::stderr().flush();
1816}
1817
1818/// Read a budget (milliseconds) from an env var, else the default. The env knobs
1819/// exist mainly for testing — a tiny budget reproduces large-repo warming
1820/// behavior on a small repo.
1821fn env_budget(var: &str, default_ms: u64) -> Duration {
1822    let ms = std::env::var(var)
1823        .ok()
1824        .and_then(|v| v.parse().ok())
1825        .unwrap_or(default_ms);
1826    Duration::from_millis(ms)
1827}
1828
1829/// How many events to roll up per interaction. Bounded so the deferred pass
1830/// after a command stays quick.
1831const AGGREGATE_BATCH: usize = 256;
1832
1833/// Recent raw events to retain after rollup (enough for repeat detection); the
1834/// rest, once aggregated, are pruned to keep the log from growing unbounded.
1835const KEEP_RECENT_EVENTS: i64 = 200;
1836
1837/// The bounded background work run after a user interaction, once results are
1838/// out: roll new events into the learning rollup, then prune the raw log.
1839fn deferred_maintenance(store: &mut Store) {
1840    let _ = store.aggregate_events(AGGREGATE_BATCH);
1841    let _ = store.prune_events(KEEP_RECENT_EVENTS);
1842}
1843
1844/// Hook entry point: record that `file` was opened/selected for `query`, then
1845/// amortize a chunk of event aggregation.
1846fn cmd_record(kind: &str, query: Option<&str>, file: &str, line: Option<i64>) -> ExitCode {
1847    let mut store = match open_store() {
1848        Ok(s) => s,
1849        Err(e) => return fail(format_args!("rq: cannot open database: {e}")),
1850    };
1851    let cwd = std::env::current_dir().unwrap_or_else(|_| PathBuf::from("."));
1852    let identity = crate::index::detect_identity(&cwd).to_string();
1853    let repo_id = store.repository_id(&identity).ok().flatten();
1854
1855    // Store the path repo-relative so the rollup can resolve it against indexed
1856    // files.
1857    let rel = match repo_id.and_then(|id| store.checkout_root(id).ok().flatten()) {
1858        Some(root) => repo_relative(std::path::Path::new(&root), &cwd, file),
1859        None => file.to_string(),
1860    };
1861    let query_norm = query.map(|q| q.to_ascii_lowercase());
1862
1863    if let Err(e) = store.record_event(kind, query_norm.as_deref(), repo_id, Some(&rel), line, None)
1864    {
1865        return fail(format_args!("rq record: {e}"));
1866    }
1867    deferred_maintenance(&mut store);
1868    ExitCode::SUCCESS
1869}
1870
1871/// Candidate on-disk roots that may hold a hit's file, most-current first: every
1872/// checkout root recorded for the repo (newest first), then the cwd (for live
1873/// results, and as a fallback when the stored root is stale — a moved repo keeps
1874/// its old checkout row, and reading from that path fails). Callers read from the
1875/// first candidate that actually has the file.
1876fn hit_file_roots(
1877    store: &Store,
1878    repo_identity: &str,
1879    cwd: Option<&std::path::Path>,
1880) -> Vec<PathBuf> {
1881    let mut roots: Vec<PathBuf> = store
1882        .repository_id(repo_identity)
1883        .ok()
1884        .flatten()
1885        .map(|id| store.checkout_roots(id).unwrap_or_default())
1886        .unwrap_or_default()
1887        .into_iter()
1888        .map(PathBuf::from)
1889        .collect();
1890    if let Some(c) = cwd {
1891        let c = c.to_path_buf();
1892        if !roots.contains(&c) {
1893            roots.push(c);
1894        }
1895    }
1896    roots
1897}
1898
1899/// The definition's source line (trimmed) for a hit — read from the first
1900/// candidate root that has the file (see [`hit_file_roots`]). Best-effort.
1901fn read_signature(
1902    store: &Store,
1903    repo_identity: &str,
1904    file: &str,
1905    line: i64,
1906    cwd: Option<&std::path::Path>,
1907) -> Option<String> {
1908    hit_file_roots(store, repo_identity, cwd)
1909        .into_iter()
1910        .find_map(|root| signature_in(&std::fs::read_to_string(root.join(file)).ok()?, line))
1911}
1912
1913/// Confidence at or above which `--show` prints a body instead of a list. Exact
1914/// (1.0) and a unique prefix (0.9) clear it; a fuzzy or tied match does not — so
1915/// `--show` never prints a definition it isn't sure about.
1916const SHOW_CONFIDENCE: f64 = 0.85;
1917
1918/// `--show`: if the top hit is confident, read and print its full source span
1919/// and return the exit code; otherwise return `None` to fall through to the
1920/// ranked list. Emits a single object in JSON/NDJSON (with a `body` field).
1921///
1922/// Printing the body *is* the selection — the caller asked for one definition
1923/// and consumed exactly this one — so it records the same signal `--open` does,
1924/// no follow-up call needed. Unlike a ranked list, there was no choice left to
1925/// the caller, and unlike a bare search, rq observed what was taken.
1926fn show_top_definition(
1927    store: &mut Store,
1928    hits: &mut [crate::search::Hit],
1929    query: &str,
1930    out: Output,
1931    cwd: Option<&std::path::Path>,
1932    current: Option<i64>,
1933    no_record: bool,
1934) -> Option<ExitCode> {
1935    let top = hits.first()?;
1936    if top.confidence < SHOW_CONFIDENCE {
1937        return None; // ambiguous / weak — let the caller list candidates
1938    }
1939    let end = top.end_line.unwrap_or(top.line);
1940    let body = read_span(store, &top.repo_identity, &top.file, top.line, end, cwd);
1941    hits[0].body = body;
1942    let top = &hits[0];
1943    let shown = (top.file.clone(), top.line);
1944    let code = match out {
1945        Output::Json | Output::Ndjson => {
1946            // fail loudly on a serialize error, like every other JSON path
1947            emit_json(out, top)
1948        }
1949        Output::Text => {
1950            let color = match_color();
1951            let c = color.as_deref();
1952            let name = hl(&top.name, query, c);
1953            let qualified = match &top.parent {
1954                Some(p) => format!("{name} · {p}"),
1955                None => name,
1956            };
1957            println!(
1958                "{}:{}  {} {}",
1959                hl_path(&top.file, query, c),
1960                top.line,
1961                top.kind,
1962                qualified
1963            );
1964            match (&top.body, &top.signature) {
1965                (Some(body), _) => println!("{body}"),
1966                // end_line unknown (pre-v4 row) → at least the definition line
1967                (None, Some(sig)) => println!("{sig}"),
1968                (None, None) => {}
1969            }
1970            ExitCode::SUCCESS
1971        }
1972    };
1973
1974    // After the output, like every other post-interaction write.
1975    if !no_record {
1976        let (file, line) = shown;
1977        let _ = store.record_event(
1978            "select",
1979            Some(&query.to_ascii_lowercase()),
1980            current,
1981            Some(&file),
1982            Some(line),
1983            None,
1984        );
1985        deferred_maintenance(store);
1986    }
1987    Some(code)
1988}
1989
1990/// The source span `start..=end` (1-based, inclusive) of a hit — the full
1991/// definition body for `--show`. Best-effort, mirroring [`read_signature`].
1992fn read_span(
1993    store: &Store,
1994    repo_identity: &str,
1995    file: &str,
1996    start: i64,
1997    end: i64,
1998    cwd: Option<&std::path::Path>,
1999) -> Option<String> {
2000    hit_file_roots(store, repo_identity, cwd)
2001        .into_iter()
2002        .find_map(|root| span_in(&std::fs::read_to_string(root.join(file)).ok()?, start, end))
2003}
2004
2005/// Lines `start..=end` (1-based, inclusive) of already-read `content`, joined —
2006/// clamped to the file's bounds. `None` if `start` is past the end.
2007fn span_in(content: &str, start: i64, end: i64) -> Option<String> {
2008    let s = usize::try_from(start).ok()?.checked_sub(1)?;
2009    let lines: Vec<&str> = content.lines().collect();
2010    if s >= lines.len() {
2011        return None;
2012    }
2013    let e = usize::try_from(end).ok()?.clamp(s + 1, lines.len());
2014    Some(lines[s..e].join("\n"))
2015}
2016
2017/// The trimmed source line `line` (1-based) of already-read `content`, if
2018/// non-empty — a symbol's definition line. Splitting this out lets `--symbols`
2019/// read one file once instead of re-reading it per symbol.
2020fn signature_in(content: &str, line: i64) -> Option<String> {
2021    let idx = usize::try_from(line).ok()?.checked_sub(1)?;
2022    let l = content.lines().nth(idx)?.trim();
2023    (!l.is_empty()).then(|| l.to_string())
2024}
2025
2026/// One symbol in `rq --symbols` output. Same field names as a search hit
2027/// (`repo`, `signature`) for agent consistency, but no score/features — an
2028/// outline is structural, not ranked.
2029#[derive(serde::Serialize)]
2030struct SymbolOut {
2031    name: String,
2032    kind: String,
2033    language: String,
2034    file: String,
2035    line: i64,
2036    #[serde(skip_serializing_if = "Option::is_none")]
2037    end_line: Option<i64>,
2038    #[serde(skip_serializing_if = "Option::is_none")]
2039    parent: Option<String>,
2040    #[serde(skip_serializing_if = "Option::is_none")]
2041    visibility: Option<String>,
2042    repo: String,
2043    #[serde(skip_serializing_if = "Option::is_none")]
2044    signature: Option<String>,
2045}
2046
2047/// `rq --symbols <file>`: list a file's symbols in line order — a structural
2048/// outline, not a ranked search. Warms the file's repo if it's cold/incomplete or
2049/// changed (same gate as search), then reads straight from the index. Honors
2050/// --kind/--lang filters and --json/--ndjson.
2051fn cmd_symbols(file_arg: &str, kinds: &[String], langs: &[String], out: Output) -> ExitCode {
2052    let mut store = match open_store() {
2053        Ok(s) => s,
2054        Err(e) => return fail(format_args!("rq: cannot open database: {e}")),
2055    };
2056    let cwd = std::env::current_dir().unwrap_or_else(|_| PathBuf::from("."));
2057    let root = crate::index::repo_root(&cwd).unwrap_or_else(|| cwd.clone());
2058    let rel = repo_relative(&root, &cwd, file_arg);
2059
2060    let identity = resolve_identity(&store, &root);
2061    let coverage = store.coverage_status(&identity).ok().flatten();
2062    let warming_ok = crate::index::is_git_repo(&root) || coverage.is_some();
2063    let current = store.repository_id(&identity).ok().flatten();
2064    // Listing a file's symbols warms synchronously — there's no answer to get
2065    // out of the way of here, so the staleness fork is paid inline as before.
2066    let indexed_head = current.and_then(|id| store.indexed_head(id).ok().flatten());
2067    let needs_warm = warming_ok
2068        && (coverage.as_deref() != Some("complete")
2069            || worktree_changed(&root, indexed_head.as_deref()));
2070    if needs_warm {
2071        // Path-prioritize the warm toward the requested file so it indexes first.
2072        let budget = answer_warm_budget() + deferred_warm_budget();
2073        let _ = crate::index::index_budgeted(&mut store, &root, &[], budget, Some(&rel));
2074    }
2075
2076    let Some(repo_id) = store.repository_id(&identity).ok().flatten() else {
2077        return emit_symbols(out, &[]); // unknown / un-indexed repo → nothing
2078    };
2079    let mut rows = match store.symbols_in_file(repo_id, &rel) {
2080        Ok(r) => r,
2081        Err(e) => return fail(format_args!("rq: {e}")),
2082    };
2083    if !kinds.is_empty() {
2084        rows.retain(|r| kinds.iter().any(|k| k == &r.kind));
2085    }
2086    if !langs.is_empty() {
2087        rows.retain(|r| langs.iter().any(|l| l == &r.language));
2088    }
2089
2090    // Read the source once for signatures (every row is the same file), from
2091    // the first root that actually has it (see `hit_file_roots` — a moved repo
2092    // keeps a stale checkout row, so the first-recorded root can be dead).
2093    let content = hit_file_roots(&store, &identity, Some(&root))
2094        .iter()
2095        .find_map(|r| std::fs::read_to_string(r.join(&rel)).ok());
2096    let syms: Vec<SymbolOut> = rows
2097        .into_iter()
2098        .map(|r| SymbolOut {
2099            signature: content.as_deref().and_then(|c| signature_in(c, r.line)),
2100            name: r.name,
2101            kind: r.kind,
2102            language: r.language,
2103            file: r.file,
2104            line: r.line,
2105            end_line: r.end_line,
2106            parent: r.parent,
2107            visibility: r.visibility,
2108            repo: r.repo_identity,
2109        })
2110        .collect();
2111    emit_symbols(out, &syms)
2112}
2113
2114/// Render the outline. Exit 0 if any symbols, non-zero if none — rq's exit-code
2115/// convention, matching how search reports an empty result per format.
2116fn emit_symbols(out: Output, syms: &[SymbolOut]) -> ExitCode {
2117    if syms.is_empty() {
2118        match out {
2119            Output::Json | Output::Ndjson => {
2120                let obj = serde_json::json!({ "status": "no_match" });
2121                let _ = emit_json(out, &obj); // exit code below carries the miss
2122            }
2123            Output::Text => eprintln!("no symbols"),
2124        }
2125        return ExitCode::FAILURE;
2126    }
2127    if let Some(code) = emit_rows(out, syms) {
2128        return code;
2129    }
2130    match out {
2131        Output::Json | Output::Ndjson => {}
2132        Output::Text => {
2133            for s in syms {
2134                let qualified = match &s.parent {
2135                    Some(p) => format!("{} · {p}", s.name),
2136                    None => s.name.clone(),
2137                };
2138                println!("{}:{}  {} {}", s.file, s.line, s.kind, qualified);
2139                if let Some(sig) = &s.signature {
2140                    println!("    {sig}");
2141                }
2142            }
2143        }
2144    }
2145    ExitCode::SUCCESS
2146}
2147
2148/// A leading positional that names a symbol kind — the shorthand behind
2149/// `rq class Foo` and `rq method zoom`. Only the full, unambiguous keyword forms
2150/// count (never the single-letter `-k` shortcuts, which are far likelier to be a
2151/// real query). Returns the canonical kind, so it filters exactly like `--kind`.
2152fn keyword_kind(token: &str) -> Option<&'static str> {
2153    match token.to_ascii_lowercase().as_str() {
2154        "class" => Some("class"),
2155        "module" => Some("module"),
2156        "method" => Some("method"),
2157        "function" | "fn" => Some("function"),
2158        "struct" | "type" => Some("struct"),
2159        "enum" => Some("enum"),
2160        "trait" | "interface" => Some("trait"),
2161        _ => None,
2162    }
2163}
2164
2165/// Peel a leading kind keyword off the query, so `rq class Foo` (or the quoted
2166/// `rq 'class Foo'`) means `-k class` + query `Foo`. The keyword must be followed
2167/// by a real query token — a bare `rq class` stays a search for a symbol literally
2168/// named `class`. Returns `(kind, query, trailing_path_dirs)`; the trailing dirs
2169/// are the rg-style positionals left after the query is consumed.
2170fn split_kind_keyword(
2171    target: String,
2172    dirs: Vec<String>,
2173) -> (Option<&'static str>, String, Vec<String>) {
2174    // Quoted form: the whole thing is one arg (`"class Foo"`), so peel the first
2175    // whitespace-separated word and keep the remainder as the query.
2176    if let Some((head, rest)) = target.split_once(char::is_whitespace) {
2177        let rest = rest.trim();
2178        if let Some(k) = keyword_kind(head)
2179            && !rest.is_empty()
2180        {
2181            return (Some(k), rest.to_string(), dirs);
2182        }
2183    } else if let Some(k) = keyword_kind(&target)
2184        && let Some((query, extra)) = dirs.split_first()
2185    {
2186        // Unquoted form: `rq class Foo` — the next positional is the query.
2187        return (Some(k), query.clone(), extra.to_vec());
2188    }
2189    (None, target, dirs)
2190}
2191
2192/// Normalize a `--kind` value (name or shortcut) to a canonical symbol kind.
2193/// Unknown values pass through lowercased (so they simply match nothing).
2194fn canonical_kind(s: &str) -> String {
2195    match s.to_ascii_lowercase().as_str() {
2196        "c" | "class" => "class",
2197        "m" | "method" => "method",
2198        "f" | "fn" | "func" | "function" => "function",
2199        "mod" | "module" => "module",
2200        "s" | "struct" | "type" => "struct",
2201        "e" | "enum" => "enum",
2202        "t" | "trait" | "interface" => "trait",
2203        other => return other.to_string(),
2204    }
2205    .to_string()
2206}
2207
2208/// Expand a `--lang` value to the language tag(s) it selects: a **prefix** of any
2209/// known language name (so `r` → ruby+rust, `p`/`py` → python, `g` → go,
2210/// `t` → typescript, `j` → javascript), plus a few non-prefix aliases
2211/// (`rb`→ruby, `rs`→rust, `golang`→go, `ts`/`tsx`→typescript,
2212/// `js`/`jsx`→javascript). An unknown value passes through lowercased so it
2213/// simply matches nothing.
2214fn canonical_langs(s: &str) -> Vec<String> {
2215    let t = s.to_ascii_lowercase();
2216    let alias = match t.as_str() {
2217        "rb" => Some("ruby"),
2218        "rs" => Some("rust"),
2219        "golang" => Some("go"),
2220        "ts" | "tsx" => Some("typescript"),
2221        "js" | "jsx" => Some("javascript"),
2222        _ => None,
2223    };
2224    let matched: Vec<String> = crate::lang::languages()
2225        .into_iter()
2226        .filter(|lang| alias == Some(*lang) || lang.starts_with(&t))
2227        .map(str::to_string)
2228        .collect();
2229    if matched.is_empty() { vec![t] } else { matched }
2230}
2231
2232/// The ANSI SGR code for highlighting matches, or `None` to disable color.
2233/// Off unless stdout is a terminal; honors `NO_COLOR`; takes the match style
2234/// from `GREP_COLORS` (`mt`/`ms`) when set, else grep's default bold red.
2235fn match_color() -> Option<String> {
2236    if std::env::var_os("NO_COLOR").is_some() || !std::io::stdout().is_terminal() {
2237        return None;
2238    }
2239    let style = std::env::var("GREP_COLORS").ok().and_then(|gc| {
2240        gc.split(':').find_map(|e| {
2241            e.strip_prefix("mt=")
2242                .or_else(|| e.strip_prefix("ms="))
2243                .filter(|v| !v.is_empty())
2244                .map(str::to_string)
2245        })
2246    });
2247    Some(style.unwrap_or_else(|| "1;31".to_string()))
2248}
2249
2250/// Highlight the chars of `text` that `query` matched (no-op when `color` is
2251/// `None`, e.g. piped output).
2252fn hl(text: &str, query: &str, color: Option<&str>) -> String {
2253    match color {
2254        Some(c) => highlight(text, &crate::search::match_positions(query, text), c),
2255        None => text.to_string(),
2256    }
2257}
2258
2259/// Like [`hl`], but only over a path's filename — so matched chars light up in
2260/// `payrolls_controller.rb`, not scattered across the directory parts.
2261fn hl_path(path: &str, query: &str, color: Option<&str>) -> String {
2262    let Some(c) = color else {
2263        return path.to_string();
2264    };
2265    let base_byte = path.rfind('/').map(|b| b + 1).unwrap_or(0);
2266    let base_start = path[..base_byte].chars().count();
2267    // align on the filename *stem* (drop the extension), the same string the
2268    // scorer matched — so the query can't straggle into `.rb` instead of lighting
2269    // up the logical name (`employees_controller`)
2270    let stem = crate::search::path_stem(path);
2271    let positions: Vec<usize> = crate::search::match_positions(query, stem)
2272        .into_iter()
2273        .map(|p| p + base_start)
2274        .collect();
2275    highlight(path, &positions, c)
2276}
2277
2278/// Wrap the matched character positions of `text` in an ANSI color run.
2279/// Consecutive matched chars share one escape sequence.
2280fn highlight(text: &str, positions: &[usize], color: &str) -> String {
2281    if positions.is_empty() {
2282        return text.to_string();
2283    }
2284    let matched: std::collections::HashSet<usize> = positions.iter().copied().collect();
2285    let mut out = String::new();
2286    let mut on = false;
2287    for (i, c) in text.chars().enumerate() {
2288        match (matched.contains(&i), on) {
2289            (true, false) => {
2290                out.push_str("\x1b[");
2291                out.push_str(color);
2292                out.push('m');
2293                on = true;
2294            }
2295            (false, true) => {
2296                out.push_str("\x1b[0m");
2297                on = false;
2298            }
2299            _ => {}
2300        }
2301        out.push(c);
2302    }
2303    if on {
2304        out.push_str("\x1b[0m");
2305    }
2306    out
2307}
2308
2309/// Whether a repo-relative `file` sits under one of the `--path` directories
2310/// (prefix match on a path boundary). `app/services` matches
2311/// `app/services/refund.rb` but not `app/services_old/x.rb`.
2312fn under_any(file: &str, paths: &[String]) -> bool {
2313    paths.iter().any(|p| {
2314        let p = p.trim_start_matches("./").trim_end_matches('/');
2315        p.is_empty() || file == p || file.starts_with(&format!("{p}/"))
2316    })
2317}
2318
2319/// Resolve a possibly-absolute or cwd-relative path to a repo-relative one.
2320fn repo_relative(root: &std::path::Path, cwd: &std::path::Path, file: &str) -> String {
2321    let p = std::path::Path::new(file);
2322    let abs = if p.is_absolute() {
2323        p.to_path_buf()
2324    } else {
2325        cwd.join(p)
2326    };
2327    let abs = abs.canonicalize().unwrap_or(abs);
2328    abs.strip_prefix(root)
2329        .map(|r| r.to_string_lossy().into_owned())
2330        .unwrap_or_else(|_| file.to_string())
2331}
2332
2333/// Revalidate the files behind the top hits against disk, refreshing any that
2334/// changed and forgetting any that were deleted. Returns true if anything
2335/// changed (so the caller re-runs the search).
2336fn revalidate_top(store: &mut Store, hits: &[crate::search::Hit]) -> bool {
2337    use std::collections::HashSet;
2338    let mut seen = HashSet::new();
2339    let mut changed = false;
2340    for hit in hits {
2341        if !seen.insert((hit.repo_identity.clone(), hit.file.clone())) {
2342            continue;
2343        }
2344        let Some(repo_id) = store.repository_id(&hit.repo_identity).ok().flatten() else {
2345            continue;
2346        };
2347        let Some(root) = store.checkout_root(repo_id).ok().flatten() else {
2348            continue;
2349        };
2350        if let Ok(crate::index::Refresh::Updated) =
2351            crate::index::refresh_file(store, repo_id, std::path::Path::new(&root), &hit.file)
2352        {
2353            changed = true;
2354        }
2355    }
2356    changed
2357}
2358
2359/// The repository's normalized identity for `cwd`, cache-first: look it up by
2360/// the canonical cwd (the checkout root indexing records), so a known repo (git
2361/// or explicitly `--index`ed) costs no `git` fork. On a cache miss, a non-git
2362/// dir resolves to its `local:` path directly (still no fork); only a git work
2363/// tree we haven't seen yet pays a `git remote` call.
2364fn resolve_identity(store: &Store, cwd: &std::path::Path) -> String {
2365    if let Ok(canon) = cwd.canonicalize() {
2366        if let Ok(Some(identity)) = store.identity_for_root(&canon.to_string_lossy()) {
2367            return identity;
2368        }
2369        if crate::index::repo_root(cwd).is_none() {
2370            return crate::core::RepoIdentity::local(&canon.to_string_lossy()).to_string();
2371        }
2372    }
2373    crate::index::detect_identity(cwd).to_string()
2374}
2375
2376fn cmd_index(path: Option<PathBuf>, subdirs: &[String], out: Output) -> ExitCode {
2377    let explicit = path.is_some();
2378    let target = path.unwrap_or_else(|| PathBuf::from("."));
2379    // Normalize to the repo root: the index is repo-root-relative, so indexing
2380    // from a subdirectory must still key off the root (a subdir-relative index
2381    // would mismatch a later search and get reconciled away). `--path` scopes a
2382    // subset; outside git the target is used as-is.
2383    let root = crate::index::repo_root(&target).unwrap_or_else(|| target.clone());
2384    // An explicit TARGET *inside* the repo scopes the index to that subtree — the
2385    // user pointed at a subdir, not the whole repo, and shouldn't pay to walk
2386    // everything. Folded in alongside any `--path` subdirs. (A bare `rq --index`
2387    // with no target still walks the whole repo.)
2388    let mut subdirs = subdirs.to_vec();
2389    if explicit
2390        && let (Ok(t), Ok(r)) = (target.canonicalize(), root.canonicalize())
2391        && t != r
2392        && let Ok(rel) = t.strip_prefix(&r)
2393        && !rel.as_os_str().is_empty()
2394    {
2395        subdirs.push(rel.to_string_lossy().into_owned());
2396    }
2397    let mut store = match open_store() {
2398        Ok(s) => s,
2399        Err(e) => return fail(format_args!("rq: cannot open database: {e}")),
2400    };
2401    let identity = crate::index::detect_identity(&root).to_string();
2402    match crate::index::index_under(&mut store, &root, &subdirs) {
2403        Ok(stats) => {
2404            let subtree = !subdirs.is_empty();
2405            // distinguish this run's incremental work from the index totals
2406            let totals = store
2407                .repository_id(&identity)
2408                .ok()
2409                .flatten()
2410                .and_then(|id| store.repo_totals(id).ok());
2411            match out {
2412                Output::Json | Output::Ndjson => {
2413                    let (files, symbols) = match totals {
2414                        Some((f, s)) => (Some(f), Some(s)),
2415                        None => (None, None),
2416                    };
2417                    return emit_json(
2418                        out,
2419                        &serde_json::json!({
2420                            "repo": identity,
2421                            "scope": if subtree { "subtree" } else { "full" },
2422                            "files_added": stats.files_indexed,
2423                            "symbols_added": stats.symbols,
2424                            "files": files,
2425                            "symbols": symbols,
2426                        }),
2427                    );
2428                }
2429                Output::Text => {
2430                    let scope = if subtree { " (subtree seed)" } else { "" };
2431                    match totals {
2432                        Some((files, symbols)) => println!(
2433                            "{} file(s)/{} symbol(s) added this run; index{scope} now {files} files, {symbols} symbols",
2434                            stats.files_indexed, stats.symbols
2435                        ),
2436                        None => println!(
2437                            "{} file(s)/{} symbol(s) added this run{scope}",
2438                            stats.files_indexed, stats.symbols
2439                        ),
2440                    }
2441                }
2442            }
2443            ExitCode::SUCCESS
2444        }
2445        Err(e) => fail(format_args!("rq --index: {e}")),
2446    }
2447}
2448
2449fn cmd_drop(target: Option<String>, out: Output) -> ExitCode {
2450    let mut store = match open_store() {
2451        Ok(s) => s,
2452        Err(e) => return fail(format_args!("rq: cannot open database: {e}")),
2453    };
2454
2455    // Resolve the repo to drop: TARGET as a path (→ repo root → identity, like
2456    // --index), falling back to TARGET as a literal identity string — so cruft
2457    // shown by --status can be dropped by name even if the checkout is gone.
2458    let path = PathBuf::from(target.clone().unwrap_or_else(|| ".".to_string()));
2459    let root = crate::index::repo_root(&path).unwrap_or(path);
2460    let from_path = crate::index::detect_identity(&root).to_string();
2461    let resolved = match store.repository_id(&from_path) {
2462        Ok(Some(id)) => Some((from_path.clone(), id)),
2463        Ok(None) => target.as_deref().and_then(|s| {
2464            store
2465                .repository_id(s)
2466                .ok()
2467                .flatten()
2468                .map(|id| (s.to_string(), id))
2469        }),
2470        Err(e) => return fail(format_args!("rq --drop: {e}")),
2471    };
2472
2473    let Some((identity, repo_id)) = resolved else {
2474        // nothing to drop — idempotent. `dropped: false` lets a script tell.
2475        return match out {
2476            Output::Text => {
2477                println!("not indexed: {from_path}");
2478                ExitCode::SUCCESS
2479            }
2480            _ => emit_json(
2481                out,
2482                &serde_json::json!({"repo": from_path, "files": 0, "symbols": 0, "dropped": false}),
2483            ),
2484        };
2485    };
2486
2487    let (files, symbols) = store.repo_totals(repo_id).unwrap_or((0, 0));
2488    match store.drop_repository(repo_id) {
2489        Ok(()) => match out {
2490            Output::Text => {
2491                println!("dropped {identity} ({files} file(s), {symbols} symbol(s))");
2492                ExitCode::SUCCESS
2493            }
2494            _ => emit_json(
2495                out,
2496                &serde_json::json!({"repo": identity, "files": files, "symbols": symbols, "dropped": true}),
2497            ),
2498        },
2499        Err(e) => fail(format_args!("rq --drop: {e}")),
2500    }
2501}
2502
2503/// Print a single value as JSON: `--json` pretty, `--ndjson` compact one-liner.
2504/// Used by the single-object operations (`--index`, `--drop`) and the
2505/// no-match status objects; [`emit_rows`] is the multi-row twin.
2506fn emit_json<T: serde::Serialize>(out: Output, value: &T) -> ExitCode {
2507    let rendered = if out == Output::Json {
2508        serde_json::to_string_pretty(value)
2509    } else {
2510        serde_json::to_string(value)
2511    };
2512    match rendered {
2513        Ok(s) => {
2514            println!("{s}");
2515            ExitCode::SUCCESS
2516        }
2517        Err(e) => fail(format_args!("rq: {e}")),
2518    }
2519}
2520
2521/// Print a row set as structured output: `--json` one pretty array, `--ndjson`
2522/// one compact object per line. Returns `Some(exit)` on a serialization
2523/// failure, `None` on success (Text output is the caller's business).
2524fn emit_rows<T: serde::Serialize>(out: Output, rows: &[T]) -> Option<ExitCode> {
2525    match out {
2526        Output::Json => match serde_json::to_string_pretty(rows) {
2527            Ok(s) => println!("{s}"),
2528            Err(e) => return Some(fail(format_args!("rq: {e}"))),
2529        },
2530        Output::Ndjson => {
2531            for r in rows {
2532                match serde_json::to_string(r) {
2533                    Ok(line) => println!("{line}"),
2534                    Err(e) => return Some(fail(format_args!("rq: {e}"))),
2535                }
2536            }
2537        }
2538        Output::Text => {}
2539    }
2540    None
2541}
2542
2543fn cmd_status(out: Output) -> ExitCode {
2544    let store = match open_store() {
2545        Ok(s) => s,
2546        Err(e) => return fail(format_args!("rq: cannot open database: {e}")),
2547    };
2548    let rows = match store.coverage_overview() {
2549        Ok(rows) => rows,
2550        Err(e) => return fail(format_args!("rq --status: {e}")),
2551    };
2552    if let Some(code) = emit_rows(out, &rows) {
2553        return code;
2554    }
2555    match out {
2556        Output::Json | Output::Ndjson => {}
2557        Output::Text if rows.is_empty() => {
2558            println!("no repositories indexed yet (try `rq --index`)");
2559        }
2560        Output::Text => {
2561            for r in &rows {
2562                println!(
2563                    "{:<10} {:>6} files  {:>7} symbols  {}",
2564                    r.status, r.files, r.symbols, r.identity
2565                );
2566            }
2567        }
2568    }
2569    ExitCode::SUCCESS
2570}
2571
2572/// `--usage`: how rq has actually been called, by day, caller, and flag set.
2573/// Reads `usage_daily`, which outlives the pruned raw event log.
2574fn cmd_usage(out: Output) -> ExitCode {
2575    let store = match open_store() {
2576        Ok(s) => s,
2577        Err(e) => return fail(format_args!("rq: cannot open database: {e}")),
2578    };
2579    let rows = match store.usage_overview() {
2580        Ok(rows) => rows,
2581        Err(e) => return fail(format_args!("rq --usage: {e}")),
2582    };
2583    if let Some(code) = emit_rows(out, &rows) {
2584        return code;
2585    }
2586    match out {
2587        Output::Json | Output::Ndjson => {}
2588        Output::Text if rows.is_empty() => {
2589            println!("no usage recorded yet");
2590        }
2591        Output::Text => {
2592            // Columns of bare numbers need naming; `--status` gets away without
2593            // a header because its columns carry their own units.
2594            println!(
2595                "{:<10}  {:<16} {:>6} {:>7} {:>8}  flags",
2596                "day", "caller", "found", "missed", "warming"
2597            );
2598            for r in &rows {
2599                let flags = if r.flags.is_empty() { "-" } else { &r.flags };
2600                println!(
2601                    "{:<10}  {:<16} {:>6} {:>7} {:>8}  {}",
2602                    r.day,
2603                    r.source,
2604                    r.searches - r.misses - r.warming,
2605                    r.misses,
2606                    r.warming,
2607                    flags
2608                );
2609            }
2610            let searches: i64 = rows.iter().map(|r| r.searches).sum();
2611            let misses: i64 = rows.iter().map(|r| r.misses).sum();
2612            let warming: i64 = rows.iter().map(|r| r.warming).sum();
2613            let complete: i64 = rows.iter().map(|r| r.on_complete).sum();
2614            let plural = if searches == 1 { "search" } else { "searches" };
2615            // Counts, not a percentage: these totals are often small enough
2616            // that a percentage would read as more evidence than there is.
2617            println!(
2618                "{searches} {plural} · {misses} missed · {warming} asked too early · {complete} on a complete index"
2619            );
2620        }
2621    }
2622    // Nothing recorded is the "nothing happened" case, like an empty --status.
2623    if rows.is_empty() {
2624        return ExitCode::from(1);
2625    }
2626    ExitCode::SUCCESS
2627}
2628
2629/// Open the rq database, honoring `RQ_DB` and creating parent dirs.
2630fn open_store() -> Result<Store, Box<dyn std::error::Error>> {
2631    let path = db_path()?;
2632    if let Some(parent) = path.parent() {
2633        std::fs::create_dir_all(parent)?;
2634    }
2635    Ok(Store::open(&path)?)
2636}
2637
2638/// Resolve the database path: `$RQ_DB`, else `$HOME/.local/share/rq/rq.db`.
2639fn db_path() -> Result<PathBuf, Box<dyn std::error::Error>> {
2640    if let Ok(p) = std::env::var("RQ_DB") {
2641        return Ok(PathBuf::from(p));
2642    }
2643    let home = std::env::var("HOME")?;
2644    Ok(PathBuf::from(home).join(".local/share/rq/rq.db"))
2645}
2646
2647fn fail(args: std::fmt::Arguments) -> ExitCode {
2648    eprintln!("{args}");
2649    ExitCode::FAILURE
2650}
2651
2652#[cfg(test)]
2653mod tests {
2654    use super::*;
2655
2656    #[test]
2657    fn open_menu_choice_parsing() {
2658        // blank reply takes the top match; a valid number maps to its index
2659        assert_eq!(parse_choice("\n", 5), Some(0));
2660        assert_eq!(parse_choice("  ", 5), Some(0));
2661        assert_eq!(parse_choice("3", 5), Some(2));
2662        assert_eq!(parse_choice("5", 5), Some(4));
2663        // out of range, zero, or non-numeric aborts
2664        assert_eq!(parse_choice("6", 5), None);
2665        assert_eq!(parse_choice("0", 5), None);
2666        assert_eq!(parse_choice("q", 5), None);
2667    }
2668
2669    #[test]
2670    fn wait_duration_parsing() {
2671        use std::time::Duration;
2672        // units: ms / s / m, and a bare number is seconds
2673        assert_eq!(parse_wait("50ms"), Ok(Duration::from_millis(50)));
2674        assert_eq!(parse_wait("2s"), Ok(Duration::from_secs(2)));
2675        assert_eq!(parse_wait("1m"), Ok(Duration::from_secs(60)));
2676        assert_eq!(parse_wait("250"), Ok(Duration::from_secs(250)));
2677        // fractions and zero
2678        assert_eq!(parse_wait("1.5s"), Ok(Duration::from_millis(1500)));
2679        assert_eq!(parse_wait("0"), Ok(Duration::ZERO));
2680        assert!(parse_wait("0s").unwrap().is_zero());
2681        // surrounding whitespace is tolerated
2682        assert_eq!(parse_wait(" 2s "), Ok(Duration::from_secs(2)));
2683        // garbage, empty, and negatives are rejected (a usage error at parse time)
2684        assert!(parse_wait("2x").is_err());
2685        assert!(parse_wait("").is_err());
2686        assert!(parse_wait("s").is_err());
2687        assert!(parse_wait("-1s").is_err());
2688    }
2689
2690    #[test]
2691    fn leading_kind_keyword_becomes_a_kind_filter() {
2692        let d = |s: &[&str]| s.iter().map(|x| x.to_string()).collect::<Vec<_>>();
2693        // unquoted: `rq class Widget` — keyword + next positional is the query
2694        assert_eq!(
2695            split_kind_keyword("class".into(), d(&["Widget"])),
2696            (Some("class"), "Widget".into(), vec![])
2697        );
2698        // quoted: `rq 'method zoom'` — one arg, peel the first word
2699        assert_eq!(
2700            split_kind_keyword("method zoom".into(), vec![]),
2701            (Some("method"), "zoom".into(), vec![])
2702        );
2703        // `fn` is an alias for function; composes with a qualifier tail
2704        assert_eq!(
2705            split_kind_keyword("fn".into(), d(&["Foo::run"])),
2706            (Some("function"), "Foo::run".into(), vec![])
2707        );
2708        // extra positionals after the query stay as rg-style path dirs
2709        assert_eq!(
2710            split_kind_keyword("struct".into(), d(&["Gadget", "src"])),
2711            (Some("struct"), "Gadget".into(), d(&["src"]))
2712        );
2713    }
2714
2715    #[test]
2716    fn a_bare_or_non_keyword_query_is_left_alone() {
2717        let d = |s: &[&str]| s.iter().map(|x| x.to_string()).collect::<Vec<_>>();
2718        // a keyword with no following query token is a search for that literal name
2719        assert_eq!(
2720            split_kind_keyword("class".into(), vec![]),
2721            (None, "class".into(), vec![])
2722        );
2723        // an ordinary query is untouched, trailing dirs preserved
2724        assert_eq!(
2725            split_kind_keyword("Widget".into(), d(&["app"])),
2726            (None, "Widget".into(), d(&["app"]))
2727        );
2728        // single-letter `-k` shortcuts are NOT keywords here (too query-like)
2729        assert_eq!(
2730            split_kind_keyword("c".into(), d(&["Foo"])),
2731            (None, "c".into(), d(&["Foo"]))
2732        );
2733    }
2734
2735    #[test]
2736    fn a_language_selects_by_prefix_or_alias() {
2737        // a prefix can name more than one language
2738        assert_eq!(canonical_langs("r"), ["ruby", "rust"]);
2739        assert_eq!(canonical_langs("t"), ["typescript"]);
2740        // the names people actually type aren't prefixes of the tag
2741        assert_eq!(canonical_langs("ts"), ["typescript"]);
2742        assert_eq!(canonical_langs("jsx"), ["javascript"]);
2743        assert_eq!(canonical_langs("rb"), ["ruby"]);
2744        // an unknown value passes through and simply matches nothing
2745        assert_eq!(canonical_langs("COBOL"), ["cobol"]);
2746    }
2747
2748    #[test]
2749    fn a_kind_normalizes_language_specific_spellings() {
2750        assert_eq!(canonical_kind("f"), "function");
2751        // TypeScript's spellings land on the shared model's kinds
2752        assert_eq!(canonical_kind("interface"), "trait");
2753        assert_eq!(canonical_kind("type"), "struct");
2754        // …and work as the leading-keyword shorthand too
2755        let d = |s: &[&str]| s.iter().map(|x| x.to_string()).collect::<Vec<_>>();
2756        assert_eq!(
2757            split_kind_keyword("interface".into(), d(&["Renderer"])),
2758            (Some("trait"), "Renderer".into(), vec![])
2759        );
2760    }
2761
2762    #[test]
2763    fn highlight_wraps_matched_runs() {
2764        assert_eq!(
2765            highlight("FooThing", &[0, 1, 2], "1;31"),
2766            "\u{1b}[1;31mFoo\u{1b}[0mThing"
2767        );
2768        // scattered matches get separate runs
2769        assert_eq!(
2770            highlight("FooThing", &[0, 3], "1"),
2771            "\u{1b}[1mF\u{1b}[0moo\u{1b}[1mT\u{1b}[0mhing"
2772        );
2773        // nothing matched → unchanged
2774        assert_eq!(highlight("FooThing", &[], "1;31"), "FooThing");
2775    }
2776
2777    #[test]
2778    fn progress_ui_only_for_an_interactive_text_terminal() {
2779        // a person at a terminal, plain text → live progress + graceful Ctrl-C
2780        assert!(show_progress(Output::Text, true));
2781
2782        // machine-readable output blocks silently (no progress line to corrupt it)
2783        assert!(!show_progress(Output::Json, true));
2784        assert!(!show_progress(Output::Ndjson, true));
2785
2786        // not a terminal (a script/agent/pipe) — block, but without the UI
2787        assert!(!show_progress(Output::Text, false));
2788    }
2789
2790    #[test]
2791    fn repo_label_uses_the_directory_name() {
2792        assert_eq!(
2793            repo_label(Some(std::path::Path::new("/src/widgets"))),
2794            "widgets"
2795        );
2796        assert_eq!(repo_label(None), "repo");
2797    }
2798
2799    #[test]
2800    fn hl_path_highlights_the_stem_not_the_extension() {
2801        // matching `employeescontroller`, the highlight covers the logical name in
2802        // the stem and never straggles into `.rb`
2803        let out = hl_path(
2804            "app/employees_controller.rb",
2805            "employeescontroller",
2806            Some("1;31"),
2807        );
2808        assert!(
2809            out.starts_with("app/\u{1b}[1;31memployees"),
2810            "stem highlighted: {out:?}"
2811        );
2812        assert!(
2813            out.ends_with("controller\u{1b}[0m.rb"),
2814            "`.rb` left un-highlighted: {out:?}"
2815        );
2816    }
2817}