// Package gitcmd runs the `git` binary for all repository access.
// It never links a git library. Every call goes through os/exec with a
// sanitized environment and an explicit context.
package gitcmd

import (
	"bytes"
	"context"
	"errors"
	"fmt"
	"os"
	"os/exec"
	"path/filepath"
	"regexp"
	"strconv"
	"strings"
	"sync"
	"time"

	"hearthforge/internal/config"
	"hearthforge/internal/util"
)

// Caps and cache settings for git command results.
const (
	MaxRefList     = 1000
	refCacheTTL    = 30 * time.Second
	maxBranchCache = 200
	maxTagCache    = 200
	staleLockAge   = 60 * time.Second
	maxPatchCache  = 100
	patchCacheTTL  = time.Hour
)

// Sentinel errors. Handlers map these to 404 / 400 / 409.
var (
	ErrInvalidName = errors.New("invalid repository name")
	ErrInvalidRef  = errors.New("invalid ref")
	ErrNotFound    = errors.New("not found")
	ErrExists      = errors.New("already exists")
	ErrBadRef      = errors.New("ref does not resolve")
	ErrConflict    = errors.New("patch does not apply")
	ErrRefChanged  = errors.New("ref changed concurrently")
)

var validRepoName = regexp.MustCompile(`^[a-zA-Z0-9._-]+$`)

// ValidRepoName mirrors VALID_REPO_NAME_RE plus the traversal guard.
func ValidRepoName(name string) bool {
	return name != "" && !strings.Contains(name, "..") && validRepoName.MatchString(name)
}

// ValidRef rejects names git would read as options or path traversal.
// `--end-of-options` covers the option case too. This is a second guard.
func ValidRef(ref string) bool {
	if ref == "" || strings.HasPrefix(ref, "-") || strings.Contains(ref, "..") {
		return false
	}
	// A colon would let a ref smuggle a path into `ref:path` forms.
	return !strings.ContainsAny(ref, " \t\n\r\x00:\\")
}

// ValidPath rejects paths that escape the tree or look like an option.
func ValidPath(p string) bool {
	if p == "" || strings.HasPrefix(p, "-") || strings.HasPrefix(p, "/") {
		return false
	}
	if strings.ContainsAny(p, "\x00\n") {
		return false
	}
	for _, seg := range strings.Split(p, "/") {
		if seg == ".." {
			return false
		}
	}
	return true
}

type Git struct {
	cfg *config.Config
	env []string

	mu       sync.Mutex
	locks    map[string]*sync.Mutex
	branches *util.Cache[string, []string]
	tags     *util.Cache[string, []string]

	archiveSem chan struct{}
}

func New(cfg *config.Config) *Git {
	return &Git{
		cfg:        cfg,
		env:        Env(),
		locks:      map[string]*sync.Mutex{},
		branches:   util.NewCache[string, []string](maxBranchCache, refCacheTTL),
		tags:       util.NewCache[string, []string](maxTagCache, refCacheTTL),
		archiveSem: make(chan struct{}, cfg.MaxConcurrentArchives),
	}
}

// Env is the sanitized environment every git subprocess runs with. A fixed
// environment keeps git output parseable and stops git from reading user or
// system config, or prompting for credentials. Callers that spawn git
// themselves (the transports, the CI runner) use it too.
func Env() []string {
	return append(os.Environ(),
		"LC_ALL=C",
		"LANG=C",
		"GIT_CONFIG_GLOBAL=/dev/null",
		"GIT_CONFIG_SYSTEM=/dev/null",
		"GIT_CONFIG_COUNT=0",
		"GIT_ASKPASS=echo",
		"GIT_TERMINAL_PROMPT=0",
	)
}

// RepoPath is the bare repo directory for a validated name.
func (g *Git) RepoPath(name string) string {
	return filepath.Join(g.cfg.ReposDir(), name+".git")
}

func (g *Git) repoDir(name string) (string, error) {
	if !ValidRepoName(name) {
		return "", fmt.Errorf("%q: %w", name, ErrInvalidName)
	}
	return g.RepoPath(name), nil
}

// lock serializes writes per repository. Two concurrent index writes in the
// same bare repo corrupt each other.
func (g *Git) lock(name string) *sync.Mutex {
	g.mu.Lock()
	defer g.mu.Unlock()
	m, ok := g.locks[name]
	if !ok {
		m = &sync.Mutex{}
		g.locks[name] = m
	}
	return m
}

type runOpts struct {
	extraEnv []string // appended to the sanitized env
	stdin    []byte
}

// run executes git and returns stdout. Stderr goes into the error.
func (g *Git) run(ctx context.Context, opt runOpts, args ...string) ([]byte, error) {
	cmd := exec.CommandContext(ctx, "git", args...)
	cmd.Env = g.env
	if len(opt.extraEnv) > 0 {
		cmd.Env = append(append([]string(nil), g.env...), opt.extraEnv...)
	}
	if opt.stdin != nil {
		cmd.Stdin = bytes.NewReader(opt.stdin)
	}
	var out, errBuf bytes.Buffer
	cmd.Stdout = &out
	cmd.Stderr = &errBuf
	if err := cmd.Run(); err != nil {
		return out.Bytes(), fmt.Errorf("git %s: %w: %s", args[0], err, strings.TrimSpace(errBuf.String()))
	}
	return out.Bytes(), nil
}

func (g *Git) text(ctx context.Context, args ...string) (string, error) {
	out, err := g.run(ctx, runOpts{}, args...)
	return string(out), err
}

func (g *Git) line(ctx context.Context, args ...string) (string, error) {
	s, err := g.text(ctx, args...)
	return strings.TrimSpace(s), err
}

// signArgs configure ssh commit signing with the server host key.
func (g *Git) signArgs() []string {
	return []string{"-c", "gpg.format=ssh", "-c", "user.signingKey=" + g.cfg.SSHHostKeyPath}
}

// verifyArgs configure signature verification against the allowed_signers file.
func (g *Git) verifyArgs() []string {
	return []string{"-c", "gpg.format=ssh", "-c", "gpg.ssh.allowedSignersFile=" + g.cfg.AllowedSignersPath()}
}

// SigStatus is the badge shown next to a commit.
type SigStatus string

const (
	SigGood SigStatus = "good"
	SigBad  SigStatus = "bad"
	SigNone SigStatus = "none"
)

// parseSigStatus maps git's %G? codes onto the three badges.
func parseSigStatus(code string) SigStatus {
	switch code {
	case "G", "X", "Y", "R":
		return SigGood
	case "B", "U", "E":
		return SigBad
	}
	return SigNone
}

type Commit struct {
	Hash      string
	Subject   string
	Author    string
	Date      string
	SigStatus SigStatus
}

type CommitMeta struct {
	Hash           string
	Subject        string
	Body           string
	Author         string
	Email          string
	Date           string
	Committer      string
	CommitterEmail string
	CommitterDate  string
	Parents        []string
	SigStatus      SigStatus
}

type TreeEntry struct {
	Mode string
	Type string // blob or tree
	Hash string
	Size string
	Name string
}

type BranchInfo struct {
	Name       string
	ShortHash  string
	Subject    string
	AuthorName string
	Date       string
}

type TagInfo struct {
	Name        string
	ShortHash   string
	Subject     string
	TaggerName  string
	Date        string
	IsAnnotated bool
}

// Ident is a git author or committer identity.
type Ident struct {
	Name  string
	Email string
}

func identEnv(author, committer Ident) []string {
	return []string{
		"GIT_AUTHOR_NAME=" + author.Name,
		"GIT_AUTHOR_EMAIL=" + author.Email,
		"GIT_COMMITTER_NAME=" + committer.Name,
		"GIT_COMMITTER_EMAIL=" + committer.Email,
	}
}

func splitLines(s string) []string {
	var out []string
	for _, l := range strings.Split(s, "\n") {
		if l != "" {
			out = append(out, l)
		}
	}
	return out
}

func field(parts []string, i int) string {
	if i < len(parts) {
		return parts[i]
	}
	return ""
}

// --- read operations ---

func (g *Git) Init(ctx context.Context, name, branch string) error {
	p, err := g.repoDir(name)
	if err != nil {
		return err
	}
	if branch == "" {
		branch = "main"
	}
	if !ValidRef(branch) {
		return fmt.Errorf("%q: %w", branch, ErrInvalidRef)
	}
	m := g.lock(name)
	m.Lock()
	defer m.Unlock()
	_, err = g.run(ctx, runOpts{}, "init", "--bare", "--initial-branch="+branch, p)
	return err
}

// EnsureBare sets core.bare on a repo discovered on disk.
func (g *Git) EnsureBare(ctx context.Context, name string) error {
	p, err := g.repoDir(name)
	if err != nil {
		return err
	}
	cfgFile := filepath.Join(p, "config")
	m := g.lock(name)
	m.Lock()
	defer m.Unlock()
	if cur, err := g.line(ctx, "config", "--file", cfgFile, "--get", "core.bare"); err == nil && cur == "true" {
		return nil
	}
	_, err = g.run(ctx, runOpts{}, "config", "--file", cfgFile, "core.bare", "true")
	return err
}

// asBadRef maps git's "this ref does not resolve" stderr onto ErrBadRef.
// It covers an unknown revision, a bad default HEAD and a repo with no
// commits. Any other failure is returned unchanged.
func asBadRef(ref string, err error) error {
	msg := err.Error()
	switch {
	case strings.Contains(msg, "unknown revision"),
		strings.Contains(msg, "not a valid object name"),
		strings.Contains(msg, "bad revision"),
		strings.Contains(msg, "bad object"),
		strings.Contains(msg, "bad default revision"),
		strings.Contains(msg, "does not have any commits yet"),
		strings.Contains(msg, "ambiguous argument"):
		return fmt.Errorf("%q: %w", ref, ErrBadRef)
	}
	return err
}

// Log returns up to limit commits starting at ref, skipping skip.
func (g *Git) Log(ctx context.Context, name, ref string, limit, skip int) ([]Commit, error) {
	p, err := g.repoDir(name)
	if err != nil {
		return nil, err
	}
	if ref == "" {
		ref = "HEAD"
	}
	if !ValidRef(ref) {
		return nil, fmt.Errorf("%q: %w", ref, ErrInvalidRef)
	}
	args := append(g.verifyArgs(), "-C", p, "log",
		"--format=%H%x1f%s%x1f%an%x1f%ai%x1f%G?",
		"--max-count="+strconv.Itoa(limit),
		"--skip="+strconv.Itoa(skip),
		// Everything after --end-of-options is data, never an option.
		"--end-of-options", ref, "--")
	out, err := g.text(ctx, args...)
	if err != nil {
		return nil, fmt.Errorf("log %s: %w", ref, asBadRef(ref, err))
	}
	var commits []Commit
	for _, line := range splitLines(out) {
		parts := strings.Split(line, "\x1f")
		commits = append(commits, Commit{
			Hash:      field(parts, 0),
			Subject:   field(parts, 1),
			Author:    field(parts, 2),
			Date:      field(parts, 3),
			SigStatus: parseSigStatus(field(parts, 4)),
		})
	}
	return commits, nil
}

// LsTree lists one directory level. subpath "" means the repo root.
func (g *Git) LsTree(ctx context.Context, name, ref, subpath string) ([]TreeEntry, error) {
	p, err := g.repoDir(name)
	if err != nil {
		return nil, err
	}
	if !ValidRef(ref) {
		return nil, fmt.Errorf("%q: %w", ref, ErrInvalidRef)
	}
	// A trailing `--` with no pathspec means "match nothing" to ls-tree,
	// so only add the separator when there is a path.
	args := []string{"-C", p, "ls-tree", "--long", "--end-of-options", ref}
	if subpath != "" {
		if !ValidPath(subpath) {
			return nil, fmt.Errorf("%q: %w", subpath, ErrInvalidRef)
		}
		args = append(args, "--", subpath+"/")
	}
	out, err := g.text(ctx, args...)
	if err != nil {
		return nil, fmt.Errorf("ls-tree %s: %w", ref, asBadRef(ref, err))
	}
	prefix := subpath + "/"
	var entries []TreeEntry
	for _, line := range splitLines(out) {
		// format: <mode> SP <type> SP <object> SP <size> TAB <file>
		tab := strings.IndexByte(line, '\t')
		if tab < 0 {
			continue
		}
		meta := strings.Fields(line[:tab])
		e := TreeEntry{
			Mode: field(meta, 0),
			Type: field(meta, 1),
			Hash: field(meta, 2),
			Size: field(meta, 3),
			Name: line[tab+1:],
		}
		if subpath != "" {
			e.Name = strings.TrimPrefix(e.Name, prefix)
		}
		entries = append(entries, e)
	}
	return entries, nil
}

// Show returns the blob contents at ref:filePath.
func (g *Git) Show(ctx context.Context, name, ref, filePath string) ([]byte, error) {
	p, err := g.repoDir(name)
	if err != nil {
		return nil, err
	}
	if !ValidRef(ref) {
		return nil, fmt.Errorf("%q: %w", ref, ErrInvalidRef)
	}
	if !ValidPath(filePath) {
		return nil, fmt.Errorf("%q: %w", filePath, ErrNotFound)
	}
	out, err := g.run(ctx, runOpts{}, "-C", p, "show", "--end-of-options", ref+":"+filePath)
	if err != nil {
		// A missing path is a normal answer, for example probing for a CI config.
		return nil, fmt.Errorf("show %s:%s: %w", ref, filePath, ErrNotFound)
	}
	return out, nil
}

// Diff returns the patch text for one commit.
func (g *Git) Diff(ctx context.Context, name, sha string) (string, error) {
	p, err := g.repoDir(name)
	if err != nil {
		return "", err
	}
	if !ValidRef(sha) {
		return "", fmt.Errorf("%q: %w", sha, ErrInvalidRef)
	}
	return g.text(ctx, "-C", p, "diff-tree", "--no-commit-id", "-r", "-p", "-M", "--root", "--end-of-options", sha, "--")
}

// BlobSize returns the size of a blob object, 0 for the all-zero hash.
func (g *Git) BlobSize(ctx context.Context, name, hash string) (int64, error) {
	p, err := g.repoDir(name)
	if err != nil {
		return 0, err
	}
	if strings.Trim(hash, "0") == "" {
		return 0, nil
	}
	if !ValidRef(hash) {
		return 0, fmt.Errorf("%q: %w", hash, ErrInvalidRef)
	}
	out, err := g.line(ctx, "-C", p, "cat-file", "-s", "--end-of-options", hash)
	if err != nil {
		return 0, fmt.Errorf("cat-file: %w", err)
	}
	return strconv.ParseInt(out, 10, 64)
}

// FileSize returns the size of the blob at ref:filePath. A path that is not a
// blob, a directory for example, is reported as not found.
func (g *Git) FileSize(ctx context.Context, name, ref, filePath string) (int64, error) {
	p, err := g.repoDir(name)
	if err != nil {
		return 0, err
	}
	if !ValidRef(ref) || !ValidPath(filePath) {
		return 0, ErrInvalidRef
	}
	// --batch-check reports the type as well as the size in one process.
	// Without the type a directory would answer with the tree's size, and
	// the streaming readers would then send an empty body.
	out, err := g.run(ctx, runOpts{stdin: []byte(ref + ":" + filePath + "\n")},
		"-C", p, "cat-file", "--batch-check")
	if err != nil {
		return 0, fmt.Errorf("%s:%s: %w", ref, filePath, ErrNotFound)
	}
	fields := strings.Fields(string(out))
	if len(fields) != 3 || fields[1] != "blob" {
		return 0, fmt.Errorf("%s:%s: %w", ref, filePath, ErrNotFound)
	}
	return strconv.ParseInt(fields[2], 10, 64)
}

// cachedRefs serves a ref list from cache, or fills it via load.
func (g *Git) cachedRefs(cache *util.Cache[string, []string], name string, load func() ([]string, error)) ([]string, error) {
	if v, ok := cache.Get(name); ok {
		return v, nil
	}
	value, err := load()
	if err != nil {
		return nil, err
	}
	cache.Set(name, value)
	return value, nil
}

// InvalidateRefCache drops the cached branch and tag lists for a repo.
func (g *Git) InvalidateRefCache(name string) {
	g.branches.Delete(name)
	g.tags.Delete(name)
}

func (g *Git) Branches(ctx context.Context, name string) ([]string, error) {
	p, err := g.repoDir(name)
	if err != nil {
		return nil, err
	}
	return g.cachedRefs(g.branches, name, func() ([]string, error) {
		out, err := g.text(ctx, "-C", p, "for-each-ref", "--count="+strconv.Itoa(MaxRefList),
			"--format=%(refname:short)", "refs/heads/")
		if err != nil {
			return nil, fmt.Errorf("branches: %w", err)
		}
		return splitLines(out), nil
	})
}

func (g *Git) Tags(ctx context.Context, name string) ([]string, error) {
	p, err := g.repoDir(name)
	if err != nil {
		return nil, err
	}
	return g.cachedRefs(g.tags, name, func() ([]string, error) {
		out, err := g.text(ctx, "-C", p, "for-each-ref", "--count="+strconv.Itoa(MaxRefList),
			"--format=%(refname:short)", "refs/tags/")
		if err != nil {
			return nil, fmt.Errorf("tags: %w", err)
		}
		return splitLines(out), nil
	})
}

func (g *Git) BranchesWithInfo(ctx context.Context, name string, maxCount int) ([]BranchInfo, error) {
	p, err := g.repoDir(name)
	if err != nil {
		return nil, err
	}
	if maxCount <= 0 {
		maxCount = MaxRefList
	}
	// for-each-ref has no %x1f escape, so embed the separator byte directly.
	const f = "%(refname:short)\x1f%(objectname:short)\x1f%(contents:subject)\x1f%(authorname)\x1f%(authordate:iso8601)"
	out, err := g.text(ctx, "-C", p, "for-each-ref", "--sort=-creatordate",
		"--count="+strconv.Itoa(maxCount), "--format="+f, "refs/heads/")
	if err != nil {
		return nil, fmt.Errorf("branchesWithInfo: %w", err)
	}
	var list []BranchInfo
	for _, line := range splitLines(out) {
		parts := strings.Split(line, "\x1f")
		list = append(list, BranchInfo{
			Name: field(parts, 0), ShortHash: field(parts, 1), Subject: field(parts, 2),
			AuthorName: field(parts, 3), Date: field(parts, 4),
		})
	}
	return list, nil
}

func (g *Git) TagsWithInfo(ctx context.Context, name string, maxCount int) ([]TagInfo, error) {
	p, err := g.repoDir(name)
	if err != nil {
		return nil, err
	}
	if maxCount <= 0 {
		maxCount = MaxRefList
	}
	// %(*objectname:short) resolves annotated tags to their commit. It is
	// empty for lightweight tags, which is how we tell the two apart.
	const f = "%(refname:short)\x1f%(*objectname:short)\x1f%(objectname:short)\x1f%(contents:subject)\x1f%(taggername)\x1f%(creatordate:iso8601)"
	out, err := g.text(ctx, "-C", p, "for-each-ref", "--sort=-creatordate",
		"--count="+strconv.Itoa(maxCount), "--format="+f, "refs/tags/")
	if err != nil {
		return nil, fmt.Errorf("tagsWithInfo: %w", err)
	}
	var list []TagInfo
	for _, line := range splitLines(out) {
		parts := strings.Split(line, "\x1f")
		deref := strings.TrimSpace(field(parts, 1))
		own := strings.TrimSpace(field(parts, 2))
		hash := own
		if deref != "" {
			hash = deref
		}
		list = append(list, TagInfo{
			Name: field(parts, 0), ShortHash: hash, Subject: field(parts, 3),
			TaggerName: field(parts, 4), Date: field(parts, 5), IsAnnotated: deref != "",
		})
	}
	return list, nil
}

// DefaultBranch trusts HEAD only when it names a branch that exists.
func (g *Git) DefaultBranch(ctx context.Context, name string) string {
	p, err := g.repoDir(name)
	if err != nil {
		return "main"
	}
	branches, err := g.Branches(ctx, name)
	if err != nil {
		return "main"
	}
	head, _ := g.line(ctx, "-C", p, "symbolic-ref", "--short", "HEAD")
	for _, b := range branches {
		if b == head {
			return head
		}
	}
	for _, want := range []string{"main", "master"} {
		for _, b := range branches {
			if b == want {
				return want
			}
		}
	}
	if len(branches) > 0 {
		return branches[0]
	}
	return "main"
}

// ResolveRef returns the object id a ref points at.
func (g *Git) ResolveRef(ctx context.Context, name, ref string) (string, error) {
	p, err := g.repoDir(name)
	if err != nil {
		return "", err
	}
	if !ValidRef(ref) {
		return "", fmt.Errorf("%q: %w", ref, ErrInvalidRef)
	}
	out, err := g.line(ctx, "-C", p, "rev-parse", "--verify", "--end-of-options", ref)
	if err != nil || out == "" {
		return "", fmt.Errorf("%q: %w", ref, ErrBadRef)
	}
	return out, nil
}

// HasCommits reports whether the repo has at least one commit.
func (g *Git) HasCommits(ctx context.Context, name string) bool {
	p, err := g.repoDir(name)
	if err != nil {
		return false
	}
	out, err := g.line(ctx, "-C", p, "log", "--oneline", "-1", "--")
	return err == nil && out != ""
}

// CommitMeta returns the full detail for one commit, including its
// signature badge.
func (g *Git) CommitMeta(ctx context.Context, name, sha string) (*CommitMeta, error) {
	p, err := g.repoDir(name)
	if err != nil {
		return nil, err
	}
	if !ValidRef(sha) {
		return nil, fmt.Errorf("%q: %w", sha, ErrInvalidRef)
	}
	args := append(g.verifyArgs(), "-C", p, "show", "--no-patch",
		"--format=%H%x1f%an%x1f%ae%x1f%ai%x1f%cn%x1f%ce%x1f%ci%x1f%P%x1f%G?",
		"--end-of-options", sha, "--")
	metaOut, err := g.line(ctx, args...)
	if err != nil {
		return nil, fmt.Errorf("commitMeta %s: %w", sha, ErrNotFound)
	}
	msgOut, err := g.text(ctx, "-C", p, "log", "--format=%B", "-1", "--end-of-options", sha, "--")
	if err != nil {
		return nil, fmt.Errorf("commitMeta message %s: %w", sha, err)
	}
	parts := strings.Split(metaOut, "\x1f")
	full := strings.TrimRight(msgOut, "\n")
	subject, body, _ := strings.Cut(full, "\n")
	hash := field(parts, 0)
	if hash == "" {
		hash = sha
	}
	return &CommitMeta{
		Hash:           hash,
		Subject:        subject,
		Body:           strings.TrimSpace(body),
		Author:         field(parts, 1),
		Email:          field(parts, 2),
		Date:           field(parts, 3),
		Committer:      field(parts, 4),
		CommitterEmail: field(parts, 5),
		CommitterDate:  field(parts, 6),
		Parents:        strings.Fields(field(parts, 7)),
		SigStatus:      parseSigStatus(field(parts, 8)),
	}, nil
}

// SetHead points HEAD at a branch.
func (g *Git) SetHead(ctx context.Context, name, branch string) error {
	p, err := g.repoDir(name)
	if err != nil {
		return err
	}
	if !ValidRef(branch) {
		return fmt.Errorf("%q: %w", branch, ErrInvalidRef)
	}
	_, err = g.run(ctx, runOpts{}, "-C", p, "symbolic-ref", "HEAD", "refs/heads/"+branch)
	return err
}