// notevi — read code, take notes, read the trace back; every call is logged.
//
// Named for the bem-te-vi, the kiskadee that announces itself all day and keeps
// a bright crown patch hidden under plain feathers: the notes are there the
// whole time, tucked out of the project's way in a private jj sidecar.
//
// Each operation appends one JSON line to that sidecar recording who read what:
// session id, agent, model, reasoning effort, working dir, full jj change and
// commit ids, file and line range. Reads and greps are relative to a jj
// revision (default @, the working copy) — the tool assumes a jj repo.
// "notevi web" serves the same log as a browsable site.
//
// Agent identity comes from the caller's environment (Claude Code exports
// CLAUDE_CODE_SESSION_ID, AI_AGENT, CLAUDE_EFFORT into tool subprocesses); when
// absent, we walk the process tree with ps — the agent-id.py technique, macOS
// edition — matching harness names in ancestor cmdlines. Unlike /proc on
// Linux, macOS redacts other processes' env (KERN_PROCARGS2 is caller-only,
// ps -E shows nothing even for your own children), so the session id can only
// come from our own env, i.e. from harnesses that export it.
package main

import (
	"bytes"
	_ "embed"
	"encoding/json"
	"errors"
	"flag"
	"fmt"
	"os"
	"os/exec"
	"path/filepath"
	"regexp"
	"strconv"
	"strings"
	"time"

	"0x4200.cafe/notevi/sidecar"
	"0x4200.cafe/notevi/web"
)

type Entry struct {
	Time      string `json:"time"`
	Op        string `json:"op"`
	Session   string `json:"session,omitempty"`
	Agent     string `json:"agent,omitempty"`
	Model     string `json:"model,omitempty"`
	Reasoning string `json:"reasoning,omitempty"`
	Dir       string `json:"dir"`
	Rev       string `json:"rev,omitempty"`
	Change    string `json:"change,omitempty"`
	Commit    string `json:"commit,omitempty"`
	File      string `json:"file,omitempty"`
	Start     int    `json:"start,omitempty"`
	End       int    `json:"end,omitempty"`
	Pattern   string `json:"pattern,omitempty"`
	Matches   int    `json:"matches,omitempty"`
	NoteID    int    `json:"note_id,omitempty"`
	ReplyTo   int    `json:"reply_to,omitempty"`
	Type      string `json:"type,omitempty"`
	Text      string `json:"text,omitempty"`
}

var gSession, gAgent, gModel, gReasoning, gDir string
var gStore *sidecar.Store

//go:embed doc.txt
var doc string

//go:embed query-doc.txt
var queryDoc string

func main() {
	if len(os.Args) < 2 {
		usage()
	}
	gDir, _ = os.Getwd()
	switch strings.TrimLeft(os.Args[1], "-") {
	case "doc":
		fmt.Print(doc)
		return
	case "doc-query":
		fmt.Print(queryDoc)
		return
	}
	// web serves the log rather than appending to it, and brings its own flags.
	if os.Args[1] == "web" {
		web.Main(os.Args[2:])
		return
	}
	detect()
	switch os.Args[1] {
	case "read":
		cmdRead(os.Args[2:])
	case "grep":
		cmdGrep(os.Args[2:])
	case "note":
		cmdNote(os.Args[2:])
	case "query":
		cmdQuery(os.Args[2:])
	case "migrate":
		cmdMigrate(os.Args[2:])
	default:
		usage()
	}
}

func usage() {
	fmt.Fprint(os.Stderr, `notevi — read code and take notes; logs every call, hides them in a sidecar.

usage:
  notevi read  [-r REV] FILE[:START[-END]]
        print a file line-numbered, whole or a line range, at a jj revision
  notevi grep  [-r REV] PATTERN [PATH]
        search the working copy (rg) or another revision (jj file list/show);
        scope non-@ greps with PATH, they read every file through jj
  notevi note  [-r REV] [-f FILE[:START[-END]]] [-t KIND] [-reply NOTE] TEXT...
        record a note about a file range or a kind of thing (struct, enum,
        const, ...), optionally replying to a log-wide note number; ids count
        up from 1 per session
  notevi query [-session S] [-op OP] [-file SUBSTR] [-n N]
        print logged entries, oldest first
  notevi web   [-repo DIR] [-out DIR] ...
        browse the trace: repo tree, coverage, notes; "notevi web -h" for more
  notevi migrate [DIR...]
        rename a pre-notevi sidecar in place; run once per old repository

REV is any jj revset, default @ (working copy). The JSONL log lives in the
private jj sidecar named by the repo-local revset alias notevi_log.
"notevi -doc" prints the short agent-facing usage doc; "notevi -doc-query" the
query one.
`)
	os.Exit(2)
}

func fatal(format string, a ...any) {
	fmt.Fprintf(os.Stderr, "notevi: "+format+"\n", a...)
	os.Exit(1)
}

// ── agent / session / model detection ─────────────────────────────────────

func detect() {
	envClaude := os.Getenv("CLAUDECODE") == "1" || os.Getenv("CLAUDE_CODE_SESSION_ID") != ""
	envCodex := os.Getenv("CODEX_THREAD_ID") != "" || os.Getenv("CODEX_SANDBOX") != ""
	envHermes := os.Getenv("HERMES_SESSION_ID") != ""

	switch {
	case envClaude && !envCodex && !envHermes:
		gAgent = "claude"
	case envCodex && !envClaude && !envHermes:
		gAgent = "codex"
	case envHermes && !envClaude && !envCodex:
		gAgent = "hermes"
	default:
		// none, or several — nested harnesses inherit each other's env
		// (codex launched from a claude session carries both sets of vars),
		// so let the nearest ancestor in the process tree decide
		walkProcessTree()
	}

	switch {
	case strings.Contains(gAgent, "claude"):
		if v := os.Getenv("AI_AGENT"); v != "" {
			gAgent = v // claude-code_<version>_agent
		}
		gSession = os.Getenv("CLAUDE_CODE_SESSION_ID")
		gModel = os.Getenv("ANTHROPIC_MODEL")
		gReasoning = os.Getenv("CLAUDE_EFFORT")
		// Claude Code doesn't put the model in the env; it lives in settings.
		if gModel == "" || gReasoning == "" {
			home, _ := os.UserHomeDir()
			b, err := os.ReadFile(filepath.Join(home, ".claude", "settings.json"))
			if err == nil {
				var s struct {
					Model  string `json:"model"`
					Effort string `json:"effortLevel"`
				}
				json.Unmarshal(b, &s)
				if gModel == "" {
					gModel = s.Model
				}
				if gReasoning == "" {
					gReasoning = s.Effort
				}
			}
		}
	case gAgent == "codex":
		gSession = os.Getenv("CODEX_THREAD_ID")
		gModel, gReasoning = codexConfig()
	case gAgent == "hermes":
		gSession = os.Getenv("HERMES_SESSION_ID")
	}
}

// codexConfig reads the top-level model and model_reasoning_effort from
// ~/.codex/config.toml (keys before the first [section]; profiles ignored).
func codexConfig() (model, effort string) {
	home, _ := os.UserHomeDir()
	b, err := os.ReadFile(filepath.Join(home, ".codex", "config.toml"))
	if err != nil {
		return
	}
	modelRe := regexp.MustCompile(`^model\s*=\s*"([^"]+)"`)
	effortRe := regexp.MustCompile(`^model_reasoning_effort\s*=\s*"([^"]+)"`)
	for _, line := range strings.Split(string(b), "\n") {
		line = strings.TrimSpace(line)
		if strings.HasPrefix(line, "[") {
			break
		}
		if m := modelRe.FindStringSubmatch(line); m != nil {
			model = m[1]
		}
		if m := effortRe.FindStringSubmatch(line); m != nil {
			effort = m[1]
		}
	}
	return
}

var agentPatterns = []struct {
	name string
	re   *regexp.Regexp
}{
	{"hermes", regexp.MustCompile(`(?i)hermes[_-]?(cli|gateway)`)},
	{"codex", regexp.MustCompile(`(?i)codex`)},
	{"claude", regexp.MustCompile(`(?i)claude`)},
}

var shells = map[string]bool{"sh": true, "bash": true, "zsh": true, "fish": true, "dash": true, "ksh": true, "nu": true}

func walkProcessTree() {
	pid := os.Getppid()
	for depth := 0; pid > 1 && depth < 20; depth++ {
		// a shell's cmdline embeds arbitrary user command text (which can
		// mention any harness by name), so never pattern-match shells
		comm := strings.TrimPrefix(ps(pid, "-o", "comm="), "-")
		if !shells[filepath.Base(comm)] {
			cmdline := ps(pid, "-ww", "-o", "command=")
			for _, p := range agentPatterns {
				if p.re.MatchString(cmdline) {
					gAgent = p.name
					return
				}
			}
		}
		ppid, err := strconv.Atoi(ps(pid, "-o", "ppid="))
		if err != nil || ppid == pid {
			return
		}
		pid = ppid
	}
}

func ps(pid int, args ...string) string {
	out, _ := exec.Command("ps", append(args, "-p", strconv.Itoa(pid))...).Output()
	return strings.TrimSpace(string(out))
}

// ── jj helpers ────────────────────────────────────────────────────────────

func jjOut(args ...string) string {
	out, err := exec.Command("jj", args...).Output()
	if err != nil {
		if ee, ok := err.(*exec.ExitError); ok {
			os.Stderr.Write(ee.Stderr)
		}
		fatal("jj %s failed: %v", strings.Join(args, " "), err)
	}
	return string(out)
}

// --ignore-working-copy skips the snapshot: change ids are stable across
// snapshots, and this keeps `notevi note`/`notevi grep` working in sandboxes that
// block .git writes (codex workspace-write). read/grep -r snapshot anyway
// via their own jj file show/list calls.
func revisionIDs(rev string) (change, commit string) {
	out := jjOut("log", "-r", "exactly(("+rev+"), 1)", "--no-graph", "--ignore-working-copy",
		"-T", "change_id ++ \"\\n\" ++ commit_id ++ \"\\n\"")
	lines := strings.Fields(out)
	if len(lines) != 2 {
		fatal("revision %q did not resolve to one change and commit", rev)
	}
	return lines[0], lines[1]
}

// parseFileRange splits "file.go:10-20" / "file.go:10" / "file.go".
func parseFileRange(s string) (file string, start, end int) {
	i := strings.LastIndex(s, ":")
	if i < 0 {
		return s, 0, 0
	}
	m := regexp.MustCompile(`^(\d+)(?:-(\d+))?$`).FindStringSubmatch(s[i+1:])
	if m == nil {
		return s, 0, 0
	}
	start, _ = strconv.Atoi(m[1])
	end = start
	if m[2] != "" {
		end, _ = strconv.Atoi(m[2])
	}
	return s[:i], start, end
}

// ── subcommands ───────────────────────────────────────────────────────────

func cmdRead(args []string) {
	fs := flag.NewFlagSet("read", flag.ExitOnError)
	rev := fs.String("r", "@", "jj revision to read at")
	fs.Parse(args)
	if fs.NArg() != 1 {
		fatal("usage: notevi read [-r REV] FILE[:START[-END]]")
	}
	file, start, end := parseFileRange(fs.Arg(0))

	content := jjOut("file", "show", "-r", *rev, "--", file)
	lines := strings.Split(content, "\n")
	if n := len(lines); n > 0 && lines[n-1] == "" {
		lines = lines[:n-1]
	}
	if start == 0 {
		start, end = 1, len(lines)
	}
	if start > len(lines) {
		fatal("%s has only %d lines", file, len(lines))
	}
	if end > len(lines) {
		end = len(lines)
	}
	for i := start; i <= end; i++ {
		fmt.Printf("%d\t%s\n", i, lines[i-1])
	}
	change, commit := revisionIDs(*rev)
	writeLog(Entry{Op: "read", Rev: *rev, Change: change, Commit: commit, File: file, Start: start, End: end})
}

func cmdGrep(args []string) {
	fs := flag.NewFlagSet("grep", flag.ExitOnError)
	rev := fs.String("r", "@", "jj revision to search at")
	fs.Parse(args)
	if fs.NArg() < 1 || fs.NArg() > 2 {
		fatal("usage: notevi grep [-r REV] PATTERN [PATH]")
	}
	pattern, path := fs.Arg(0), fs.Arg(1)

	matches := 0
	if *rev == "@" {
		rgArgs := []string{"-n", "--no-heading", pattern}
		if path != "" {
			rgArgs = append(rgArgs, path)
		}
		out, err := exec.Command("rg", rgArgs...).Output()
		os.Stdout.Write(out)
		if ee, ok := err.(*exec.ExitError); ok && ee.ExitCode() != 1 { // 1 = no matches
			os.Stderr.Write(ee.Stderr)
			fatal("rg failed: %v", err)
		}
		matches = bytes.Count(out, []byte("\n"))
	} else {
		re, err := regexp.Compile(pattern)
		if err != nil {
			fatal("bad pattern: %v", err)
		}
		listArgs := []string{"file", "list", "-r", *rev}
		if path != "" {
			listArgs = append(listArgs, "--", path)
		}
		for _, f := range strings.Split(strings.TrimRight(jjOut(listArgs...), "\n"), "\n") {
			if f == "" {
				continue
			}
			content := strings.TrimSuffix(jjOut("file", "show", "-r", *rev, "--", f), "\n")
			for i, line := range strings.Split(content, "\n") {
				if re.MatchString(line) {
					fmt.Printf("%s:%d:%s\n", f, i+1, line)
					matches++
				}
			}
		}
	}
	change, commit := revisionIDs(*rev)
	writeLog(Entry{Op: "grep", Rev: *rev, Change: change, Commit: commit, Pattern: pattern, File: path, Matches: matches})
}

func cmdNote(args []string) {
	fs := flag.NewFlagSet("note", flag.ExitOnError)
	rev := fs.String("r", "@", "jj revision the note refers to")
	file := fs.String("f", "", "FILE[:START[-END]] the note is about")
	kind := fs.String("t", "", "kind of thing the note is about (struct, enum, const, ...)")
	reply := fs.Int("reply", 0, "log-wide note number this note replies to")
	fs.Parse(args)
	text := strings.Join(fs.Args(), " ")
	if text == "" {
		fatal("usage: notevi note [-r REV] [-f FILE[:START[-END]]] [-t KIND] [-reply NOTE] TEXT...")
	}
	if *reply < 0 {
		fatal("reply note must be positive")
	}
	change, commit := revisionIDs(*rev)
	e := Entry{Op: "note", Rev: *rev, Change: change, Commit: commit, ReplyTo: *reply, Type: *kind, Text: text}
	if *file != "" {
		e.File, e.Start, e.End = parseFileRange(*file)
	}
	writeLogWith(func(entries []Entry) (Entry, error) {
		if *reply > 0 && !hasNote(entries, *reply) {
			return Entry{}, fmt.Errorf("no log-wide note #%d in %s", *reply, store())
		}
		e.NoteID = nextNoteID(entries)
		return e, nil
	})
	fmt.Printf("note %d\n", e.NoteID)
}

// hasNote uses the same numbering as the web permalinks: note operations count
// from one over the append-only log, independent of their writers' sessions.
func hasNote(entries []Entry, id int) bool {
	n := 0
	for _, e := range entries {
		if e.Op == "note" {
			n++
			if n == id {
				return true
			}
		}
	}
	return false
}

func nextNoteID(entries []Entry) int {
	max := 0
	for _, e := range entries {
		if e.Op == "note" && e.Session == gSession && e.NoteID > max {
			max = e.NoteID
		}
	}
	return max + 1
}

func cmdQuery(args []string) {
	fs := flag.NewFlagSet("query", flag.ExitOnError)
	session := fs.String("session", "", "filter: session id substring")
	op := fs.String("op", "", "filter: read | grep | note")
	file := fs.String("file", "", "filter: file path substring")
	n := fs.Int("n", 0, "show only the last N entries")
	fs.Parse(args)

	entries := readLog()
	type result struct {
		Entry
		logNoteID int
	}
	var kept []result
	logNoteID := 0
	for _, e := range entries {
		if e.Op == "note" {
			logNoteID++
		}
		if *session != "" && !strings.Contains(e.Session, *session) {
			continue
		}
		if *op != "" && e.Op != *op {
			continue
		}
		if *file != "" && !strings.Contains(e.File, *file) {
			continue
		}
		kept = append(kept, result{Entry: e, logNoteID: logNoteID})
	}
	if *n > 0 && len(kept) > *n {
		kept = kept[len(kept)-*n:]
	}
	for _, r := range kept {
		e := r.Entry
		ts := e.Time
		if t, err := time.Parse(time.RFC3339, e.Time); err == nil {
			ts = t.Format("01-02 15:04:05")
		}
		head := fmt.Sprintf("%s  %-8.8s  %-5s %-8.8s", ts, e.Session, e.Op, e.Change)
		switch e.Op {
		case "read":
			fmt.Printf("%s  %s\n", head, fileRange(e))
		case "grep":
			loc := ""
			if e.File != "" {
				loc = " in " + e.File
			}
			fmt.Printf("%s  %q%s (%d matches)\n", head, e.Pattern, loc, e.Matches)
		case "note":
			tag := ""
			if e.Type != "" {
				tag = " (" + e.Type + ")"
			}
			loc := ""
			if e.File != "" {
				loc = " " + fileRange(e)
			}
			rel := ""
			if e.ReplyTo > 0 {
				rel = fmt.Sprintf(" reply to log #%d", e.ReplyTo)
			}
			fmt.Printf("%s  #%d [log #%d]%s%s%s — %s\n", head, e.NoteID, r.logNoteID, tag, loc, rel, e.Text)
		default:
			fmt.Printf("%s\n", head)
		}
	}
}

// cmdMigrate renames pre-notevi sidecars. It takes directories rather than only
// working on the current one so that a machine full of old repositories can be
// brought over in one run.
func cmdMigrate(args []string) {
	fs := flag.NewFlagSet("migrate", flag.ExitOnError)
	fs.Parse(args)
	dirs := fs.Args()
	if len(dirs) == 0 {
		dirs = []string{gDir}
	}
	failed := false
	for _, dir := range dirs {
		s, err := sidecar.Open(dir)
		if err != nil {
			fmt.Fprintf(os.Stderr, "notevi: %s: %v\n", dir, err)
			failed = true
			continue
		}
		migrated, err := s.Migrate()
		switch {
		case err != nil:
			fmt.Fprintf(os.Stderr, "notevi: %s: %v\n", s.Root(), err)
			failed = true
		case migrated:
			fmt.Printf("%s: renamed to %s\n", s.Root(), s)
		default:
			fmt.Printf("%s: nothing to rename\n", s.Root())
		}
	}
	if failed {
		os.Exit(1)
	}
}

func fileRange(e Entry) string {
	s := e.File
	if e.Start > 0 {
		s += ":" + strconv.Itoa(e.Start)
		if e.End > e.Start {
			s += "-" + strconv.Itoa(e.End)
		}
	}
	return s
}

// ── jj sidecar log ────────────────────────────────────────────────────────

func parseLog(b []byte) []Entry {
	var entries []Entry
	for _, line := range strings.Split(string(b), "\n") {
		if line == "" {
			continue
		}
		var e Entry
		if json.Unmarshal([]byte(line), &e) == nil {
			entries = append(entries, e)
		}
	}
	return entries
}

func store() *sidecar.Store {
	if gStore == nil {
		var err error
		gStore, err = sidecar.Open(gDir)
		if err != nil {
			fatal("open jj sidecar: %v", err)
		}
	}
	return gStore
}

func readLog() []Entry {
	b, err := store().Read()
	if errors.Is(err, sidecar.ErrNoLog) {
		return nil
	}
	if err != nil {
		fatal("read %s: %v", store(), err)
	}
	return parseLog(b)
}

func writeLog(e Entry) {
	writeLogWith(func([]Entry) (Entry, error) { return e, nil })
}

func writeLogWith(build func([]Entry) (Entry, error)) {
	err := store().Append(func(current []byte) ([]byte, error) {
		e, err := build(parseLog(current))
		if err != nil {
			return nil, err
		}
		e.Time = time.Now().Format(time.RFC3339)
		e.Session, e.Agent, e.Model, e.Reasoning, e.Dir = gSession, gAgent, gModel, gReasoning, gDir
		return json.Marshal(e)
	})
	if err != nil {
		fatal("write %s: %v", store(), err)
	}
}