| 1 | package tinygit |
| 2 | |
| 3 | import ( |
| 4 | "bufio" |
| 5 | "bytes" |
| 6 | "context" |
| 7 | "errors" |
| 8 | "fmt" |
| 9 | "io" |
| 10 | "os" |
| 11 | "os/exec" |
| 12 | "path" |
| 13 | "sort" |
| 14 | "strconv" |
| 15 | "strings" |
| 16 | "unicode/utf8" |
| 17 | ) |
| 18 | |
| 19 | const previewBytes = 256 * 1024 |
| 20 | |
| 21 | type BrowseRequest struct { |
| 22 | Owner string `json:"owner"` |
| 23 | Repo string `json:"repo"` |
| 24 | Ref string `json:"ref"` |
| 25 | Path string `json:"path"` |
| 26 | View string `json:"view"` |
| 27 | Offset int `json:"offset"` |
| 28 | Limit int `json:"limit"` |
| 29 | } |
| 30 | |
| 31 | type BrowseEntry struct { |
| 32 | Name string `json:"name"` |
| 33 | Path string `json:"path"` |
| 34 | Mode string `json:"mode"` |
| 35 | Type string `json:"type"` |
| 36 | OID string `json:"oid"` |
| 37 | Size int64 `json:"size"` |
| 38 | } |
| 39 | |
| 40 | type BrowseCommit struct { |
| 41 | OID string `json:"oid"` |
| 42 | Parents []string `json:"parents"` |
| 43 | Author string `json:"author"` |
| 44 | Time int64 `json:"time"` |
| 45 | Subject string `json:"subject"` |
| 46 | } |
| 47 | |
| 48 | type BrowsePage struct { |
| 49 | Owner string `json:"owner"` |
| 50 | Identifier string `json:"identifier"` |
| 51 | Head string `json:"head"` |
| 52 | Private bool `json:"private"` |
| 53 | Clone []string `json:"clone"` |
| 54 | Relays []string `json:"relays"` |
| 55 | // Maintainers is filled by the host, which knows about agent grants. |
| 56 | Maintainers []Maintainer `json:"maintainers,omitempty"` |
| 57 | Refs map[string]string `json:"refs"` |
| 58 | Commit string `json:"commit"` |
| 59 | Path string `json:"path"` |
| 60 | View string `json:"view"` |
| 61 | Entries []BrowseEntry `json:"entries"` |
| 62 | Content string `json:"content"` |
| 63 | Binary bool `json:"binary"` |
| 64 | Truncated bool `json:"truncated"` |
| 65 | Commits []BrowseCommit `json:"commits"` |
| 66 | Diff string `json:"diff"` |
| 67 | NextOffset int `json:"next_offset"` |
| 68 | } |
| 69 | |
| 70 | // Repositories returns detached snapshots. Callers must enforce tenant and |
| 71 | // repository read permissions before revealing these records. |
| 72 | func (g *GitRelay) Repositories() []Repository { |
| 73 | g.mu.RLock() |
| 74 | defer g.mu.RUnlock() |
| 75 | rows := make([]Repository, 0, len(g.repos)) |
| 76 | for _, r := range g.repos { |
| 77 | rows = append(rows, cloneBrowseRepo(r)) |
| 78 | } |
| 79 | sort.Slice(rows, func(i, j int) bool { |
| 80 | return key(rows[i].Owner, rows[i].Identifier) < key(rows[j].Owner, rows[j].Identifier) |
| 81 | }) |
| 82 | return rows |
| 83 | } |
| 84 | |
| 85 | func cloneBrowseRepo(r Repository) Repository { |
| 86 | refs := make(map[string]string, len(r.Refs)) |
| 87 | for name, oid := range r.Refs { |
| 88 | refs[name] = oid |
| 89 | } |
| 90 | r.Refs = refs |
| 91 | r.Clone = append([]string(nil), r.Clone...) |
| 92 | r.Relays = append([]string(nil), r.Relays...) |
| 93 | r.Maintainers = append([]string(nil), r.Maintainers...) |
| 94 | return r |
| 95 | } |
| 96 | |
| 97 | func (g *GitRelay) BrowseRepository(owner, identifier string) (Repository, error) { |
| 98 | g.mu.RLock() |
| 99 | defer g.mu.RUnlock() |
| 100 | r, ok := g.repos[key(owner, identifier)] |
| 101 | if !ok { |
| 102 | return Repository{}, os.ErrNotExist |
| 103 | } |
| 104 | return cloneBrowseRepo(r), nil |
| 105 | } |
| 106 | |
| 107 | func validBrowsePath(value string) bool { |
| 108 | return value == "" || (!strings.ContainsRune(value, 0) && !strings.HasPrefix(value, "/" ) && path.Clean(value) == value && value != ".." && !strings.HasPrefix(value, "../" )) |
| 109 | } |
| 110 | |
| 111 | func newBrowsePage(r Repository, q BrowseRequest) BrowsePage { |
| 112 | return BrowsePage{Owner: r.Owner, Identifier: r.Identifier, Head: r.Head, Private: r.Private, Clone: r.Clone, Relays: r.Relays, Refs: r.Refs, Path: q.Path, View: q.View, NextOffset: -1} |
| 113 | } |
| 114 | |
| 115 | // Browse only reads objects reachable from materialized, signed public refs. |
| 116 | // It never checks out a working tree, invokes a shell, or follows a symlink. |
| 117 | func (g *GitRelay) Browse(ctx context.Context, q BrowseRequest) (BrowsePage, error) { |
| 118 | if !validBrowsePath(q.Path) || q.Offset < 0 { |
| 119 | return BrowsePage{}, errors.New("invalid: repository path or offset" ) |
| 120 | } |
| 121 | if q.View == "" { |
| 122 | q.View = "tree" |
| 123 | } |
| 124 | if q.Limit <= 0 || q.Limit > 200 { |
| 125 | q.Limit = 50 |
| 126 | } |
| 127 | r, err := g.BrowseRepository(q.Owner, q.Repo) |
| 128 | if err != nil { |
| 129 | return BrowsePage{}, err |
| 130 | } |
| 131 | page := newBrowsePage(r, q) |
| 132 | r.Refs, err = g.browseRefs(ctx, r) |
| 133 | if err != nil { |
| 134 | return page, err |
| 135 | } |
| 136 | page.Refs = r.Refs |
| 137 | if len(r.Refs) == 0 { |
| 138 | return page, nil |
| 139 | } |
| 140 | page.Commit, err = g.browseCommit(ctx, r, q.Ref) |
| 141 | if err != nil { |
| 142 | return page, err |
| 143 | } |
| 144 | switch q.View { |
| 145 | case "tree" : |
| 146 | err = g.browseTree(ctx, r, q, &page) |
| 147 | case "file" : |
| 148 | err = g.browseFile(ctx, r, q, &page) |
| 149 | case "history" : |
| 150 | err = g.browseHistory(ctx, r, q, &page) |
| 151 | case "commit" : |
| 152 | err = g.browseDiff(ctx, r, &page) |
| 153 | case "activity" : |
| 154 | default: |
| 155 | err = errors.New("invalid: unknown repository view" ) |
| 156 | } |
| 157 | return page, err |
| 158 | } |
| 159 | |
| 160 | func (g *GitRelay) browseCommand(ctx context.Context, r Repository, args ...string) *exec.Cmd { |
| 161 | base := []string{"--no-pager" , "--literal-pathspecs" , "-c" , "core.quotePath=false" , "-c" , "log.showSignature=false" , "--git-dir" , g.repoPath(r)} |
| 162 | return exec.CommandContext(ctx, "git" , append(base, args...)...) |
| 163 | } |
| 164 | |
| 165 | func (g *GitRelay) browseRefs(ctx context.Context, r Repository) (map[string]string, error) { |
| 166 | b, err := g.browseCommand(ctx, r, "for-each-ref" , "--format=%(refname) %(objectname)" , "refs/heads/" , "refs/tags/" , "refs/nostr/" ).Output() |
| 167 | if err != nil { |
| 168 | return nil, fmt.Errorf("read repository refs: %w" , err) |
| 169 | } |
| 170 | refs := make(map[string]string) |
| 171 | for _, line := range strings.Split(string(b), "\n" ) { |
| 172 | fields := strings.Fields(line) |
| 173 | if len(fields) == 2 && r.Refs[fields[0]] == fields[1] { |
| 174 | refs[fields[0]] = fields[1] |
| 175 | } |
| 176 | } |
| 177 | return refs, nil |
| 178 | } |
| 179 | |
| 180 | func (g *GitRelay) browseCommit(ctx context.Context, r Repository, ref string) (string, error) { |
| 181 | if ref == "" || ref == "HEAD" { |
| 182 | ref = strings.TrimPrefix(r.Head, "ref: " ) |
| 183 | } |
| 184 | for _, candidate := range []string{ref, "refs/heads/" + ref, "refs/tags/" + ref} { |
| 185 | if oid, ok := r.Refs[candidate]; ok { |
| 186 | return g.peelBrowseCommit(ctx, r, oid) |
| 187 | } |
| 188 | } |
| 189 | if !hexObject(ref) { |
| 190 | return "" , errors.New("not found: published ref" ) |
| 191 | } |
| 192 | oid, err := g.peelBrowseCommit(ctx, r, ref) |
| 193 | if err != nil { |
| 194 | return "" , err |
| 195 | } |
| 196 | cmd := g.browseCommand(ctx, r, "rev-list" , "--max-count=1" , "--stdin" ) |
| 197 | var input strings.Builder |
| 198 | input.WriteString(oid + "\n" ) |
| 199 | for _, tip := range r.Refs { |
| 200 | input.WriteString("^" + tip + "\n" ) |
| 201 | } |
| 202 | cmd.Stdin = strings.NewReader(input.String()) |
| 203 | b, err := cmd.Output() |
| 204 | if err != nil || len(bytes.TrimSpace(b)) != 0 { |
| 205 | return "" , errors.New("not found: published commit" ) |
| 206 | } |
| 207 | return oid, nil |
| 208 | } |
| 209 | |
| 210 | func hexObject(value string) bool { |
| 211 | if len(value) != 40 && len(value) != 64 { |
| 212 | return false |
| 213 | } |
| 214 | for _, ch := range value { |
| 215 | if !(ch >= '0' && ch <= '9' ) && !(ch >= 'a' && ch <= 'f' ) { |
| 216 | return false |
| 217 | } |
| 218 | } |
| 219 | return true |
| 220 | } |
| 221 | |
| 222 | func (g *GitRelay) peelBrowseCommit(ctx context.Context, r Repository, oid string) (string, error) { |
| 223 | if !hexObject(oid) { |
| 224 | return "" , errors.New("invalid: object id" ) |
| 225 | } |
| 226 | b, err := g.browseCommand(ctx, r, "rev-parse" , "--verify" , "--end-of-options" , oid+"^{commit}" ).Output() |
| 227 | if err != nil { |
| 228 | return "" , errors.New("not found: commit" ) |
| 229 | } |
| 230 | return strings.TrimSpace(string(b)), nil |
| 231 | } |
| 232 | |
| 233 | func readBrowseOutput(cmd *exec.Cmd, limit int64) ([]byte, bool, error) { |
| 234 | out, err := cmd.StdoutPipe() |
| 235 | if err != nil { |
| 236 | return nil, false, err |
| 237 | } |
| 238 | if err = cmd.Start(); err != nil { |
| 239 | return nil, false, err |
| 240 | } |
| 241 | b, readErr := io.ReadAll(io.LimitReader(out, limit+1)) |
| 242 | truncated := int64(len(b)) > limit |
| 243 | if truncated || readErr != nil { |
| 244 | _ = out.Close() |
| 245 | _ = cmd.Process.Kill() |
| 246 | } |
| 247 | waitErr := cmd.Wait() |
| 248 | if readErr != nil { |
| 249 | return nil, false, readErr |
| 250 | } |
| 251 | if truncated { |
| 252 | return b[:limit], true, nil |
| 253 | } |
| 254 | if waitErr != nil { |
| 255 | return nil, false, fmt.Errorf("read Git object: %w" , waitErr) |
| 256 | } |
| 257 | return b, false, nil |
| 258 | } |
| 259 | |
| 260 | func (g *GitRelay) browseTree(ctx context.Context, r Repository, q BrowseRequest, page *BrowsePage) error { |
| 261 | tree := page.Commit + ":" + q.Path |
| 262 | cmd := g.browseCommand(ctx, r, "ls-tree" , "-z" , "-l" , tree) |
| 263 | out, err := cmd.StdoutPipe() |
| 264 | if err != nil { |
| 265 | return err |
| 266 | } |
| 267 | if err := cmd.Start(); err != nil { |
| 268 | return err |
| 269 | } |
| 270 | reader := bufio.NewReader(out) |
| 271 | readErr := readBrowseTree(reader, q, page) |
| 272 | if readErr != nil || page.NextOffset >= 0 { |
| 273 | // Closing this page cancels only its read-only Git child. |
| 274 | _ = out.Close() |
| 275 | _ = cmd.Process.Kill() |
| 276 | } |
| 277 | waitErr := cmd.Wait() |
| 278 | if readErr != nil { |
| 279 | return readErr |
| 280 | } |
| 281 | if page.NextOffset >= 0 { |
| 282 | return nil |
| 283 | } |
| 284 | return waitErr |
| 285 | } |
| 286 | |
| 287 | func readBrowseTree(reader *bufio.Reader, q BrowseRequest, page *BrowsePage) error { |
| 288 | for index := 0; ; index++ { |
| 289 | row, err := reader.ReadString(0) |
| 290 | if errors.Is(err, io.EOF) && row == "" { |
| 291 | return nil |
| 292 | } |
| 293 | if err != nil { |
| 294 | return err |
| 295 | } |
| 296 | if index < q.Offset { |
| 297 | continue |
| 298 | } |
| 299 | if len(page.Entries) >= q.Limit { |
| 300 | page.NextOffset = index |
| 301 | return nil |
| 302 | } |
| 303 | meta, name, ok := strings.Cut(strings.TrimSuffix(row, "\x00" ), "\t" ) |
| 304 | fields := strings.Fields(meta) |
| 305 | if !ok || len(fields) != 4 { |
| 306 | return errors.New("invalid Git tree entry" ) |
| 307 | } |
| 308 | var size int64 |
| 309 | if fields[3] != "-" { |
| 310 | size, err = strconv.ParseInt(fields[3], 10, 64) |
| 311 | if err != nil { |
| 312 | return err |
| 313 | } |
| 314 | } |
| 315 | page.Entries = append(page.Entries, BrowseEntry{Name: name, Path: path.Join(q.Path, name), Mode: fields[0], Type: fields[1], OID: fields[2], Size: size}) |
| 316 | } |
| 317 | } |
| 318 | |
| 319 | func (g *GitRelay) fileObject(ctx context.Context, r Repository, commit, file string) (string, error) { |
| 320 | if file == "" || !validBrowsePath(file) { |
| 321 | return "" , errors.New("invalid: file path" ) |
| 322 | } |
| 323 | b, err := g.browseCommand(ctx, r, "rev-parse" , "--verify" , "--end-of-options" , commit+":" +file).Output() |
| 324 | if err != nil { |
| 325 | return "" , os.ErrNotExist |
| 326 | } |
| 327 | oid := strings.TrimSpace(string(b)) |
| 328 | if !hexObject(oid) { |
| 329 | return "" , os.ErrNotExist |
| 330 | } |
| 331 | kind, err := g.browseCommand(ctx, r, "cat-file" , "-t" , oid).Output() |
| 332 | if err != nil || strings.TrimSpace(string(kind)) != "blob" { |
| 333 | return "" , errors.New("not found: source file" ) |
| 334 | } |
| 335 | return oid, nil |
| 336 | } |
| 337 | |
| 338 | func (g *GitRelay) browseFile(ctx context.Context, r Repository, q BrowseRequest, page *BrowsePage) error { |
| 339 | oid, err := g.fileObject(ctx, r, page.Commit, q.Path) |
| 340 | if err != nil { |
| 341 | return err |
| 342 | } |
| 343 | b, truncated, err := readBrowseOutput(g.browseCommand(ctx, r, "cat-file" , "blob" , oid), previewBytes) |
| 344 | if err != nil { |
| 345 | return err |
| 346 | } |
| 347 | page.Binary = bytes.IndexByte(b, 0) >= 0 || (!utf8.Valid(b) && !truncated) |
| 348 | page.Truncated = truncated |
| 349 | if !page.Binary { |
| 350 | page.Content = strings.ToValidUTF8(string(b), "�" ) |
| 351 | } |
| 352 | return nil |
| 353 | } |
| 354 | |
| 355 | func (g *GitRelay) browseHistory(ctx context.Context, r Repository, q BrowseRequest, page *BrowsePage) error { |
| 356 | args := []string{"log" , "--format=%H%x00%P%x00%an%x00%at%x00%s%x00" , "--max-count=" + strconv.Itoa(q.Limit+1), "--skip=" + strconv.Itoa(q.Offset), page.Commit, "--" } |
| 357 | if q.Path != "" { |
| 358 | args = append(args, q.Path) |
| 359 | } |
| 360 | b, truncated, err := readBrowseOutput(g.browseCommand(ctx, r, args...), previewBytes) |
| 361 | if err != nil { |
| 362 | return err |
| 363 | } |
| 364 | if truncated { |
| 365 | return errors.New("commit metadata exceeds preview capacity" ) |
| 366 | } |
| 367 | fields := strings.Split(string(b), "\x00" ) |
| 368 | for i := 0; i+4 < len(fields); i += 5 { |
| 369 | if len(page.Commits) >= q.Limit { |
| 370 | page.NextOffset = q.Offset + q.Limit |
| 371 | break |
| 372 | } |
| 373 | stamp, err := strconv.ParseInt(fields[i+3], 10, 64) |
| 374 | if err != nil { |
| 375 | return err |
| 376 | } |
| 377 | page.Commits = append(page.Commits, BrowseCommit{OID: strings.TrimSpace(fields[i]), Parents: strings.Fields(fields[i+1]), Author: fields[i+2], Time: stamp, Subject: fields[i+4]}) |
| 378 | } |
| 379 | return nil |
| 380 | } |
| 381 | |
| 382 | func (g *GitRelay) browseDiff(ctx context.Context, r Repository, page *BrowsePage) error { |
| 383 | if err := g.browseHistory(ctx, r, BrowseRequest{Limit: 1}, page); err != nil { |
| 384 | return err |
| 385 | } |
| 386 | b, truncated, err := readBrowseOutput(g.browseCommand(ctx, r, "show" , "--format=" , "--no-ext-diff" , "--no-textconv" , "--no-renames" , page.Commit, "--" ), previewBytes) |
| 387 | page.Diff, page.Truncated = strings.ToValidUTF8(string(b), "�" ), truncated |
| 388 | return err |
| 389 | } |
| 390 | |
| 391 | // WriteSource streams a published source blob directly to the destination. |
| 392 | func (g *GitRelay) WriteSource(ctx context.Context, q BrowseRequest, dst io.Writer) error { |
| 393 | r, err := g.BrowseRepository(q.Owner, q.Repo) |
| 394 | if err != nil { |
| 395 | return err |
| 396 | } |
| 397 | r.Refs, err = g.browseRefs(ctx, r) |
| 398 | if err != nil { |
| 399 | return err |
| 400 | } |
| 401 | commit, err := g.browseCommit(ctx, r, q.Ref) |
| 402 | if err != nil { |
| 403 | return err |
| 404 | } |
| 405 | oid, err := g.fileObject(ctx, r, commit, q.Path) |
| 406 | if err != nil { |
| 407 | return err |
| 408 | } |
| 409 | cmd := g.browseCommand(ctx, r, "cat-file" , "blob" , oid) |
| 410 | cmd.Stdout = dst |
| 411 | if err := cmd.Run(); err != nil { |
| 412 | return fmt.Errorf("stream source: %w" , err) |
| 413 | } |
| 414 | return nil |
| 415 | } |