| 1 | package control |
| 2 | |
| 3 | import ( |
| 4 | "bytes" |
| 5 | "context" |
| 6 | "crypto/sha256" |
| 7 | "encoding/base64" |
| 8 | "encoding/hex" |
| 9 | "fmt" |
| 10 | "html" |
| 11 | "io" |
| 12 | "net/http" |
| 13 | "os" |
| 14 | "os/exec" |
| 15 | "path/filepath" |
| 16 | "regexp" |
| 17 | "sort" |
| 18 | "strings" |
| 19 | "time" |
| 20 | |
| 21 | "reasonix/internal/fileref" |
| 22 | "reasonix/internal/instruction" |
| 23 | "reasonix/internal/proc" |
| 24 | "reasonix/internal/secrets" |
| 25 | ) |
| 26 | |
| 27 | // maxFileRefBytes caps how much of an @-referenced file is injected into a |
| 28 | // message, so "@somehuge.log" can't blow the context window. The head is kept |
| 29 | // and the rest noted as truncated. |
| 30 | const maxFileRefBytes = 64 * 1024 |
| 31 | |
| 32 | const pdfExtractTimeout = 8 * time.Second |
| 33 | const pdfExtractWaitDelay = 1 * time.Second |
| 34 | |
| 35 | var extractPDFText = extractPDFTextDefault |
| 36 | |
| 37 | type pdfExtractResult struct { |
| 38 | text string |
| 39 | tool string |
| 40 | truncated bool |
| 41 | } |
| 42 | |
| 43 | // refKind distinguishes the two things an @reference can resolve to. |
| 44 | type refKind int |
| 45 | |
| 46 | const ( |
| 47 | refResource refKind = iota // an MCP resource: @<server>:<uri> |
| 48 | refFile // a local file or directory: @<path> |
| 49 | refImage // a local image attachment: @.reasonix/attachments/<file> |
| 50 | ) |
| 51 | |
| 52 | // ref is a resolved @reference found in a submitted line. |
| 53 | type ref struct { |
| 54 | kind refKind |
| 55 | server string // refResource |
| 56 | uri string // refResource |
| 57 | path string // refFile, relative to baseDir when baseDir is set |
| 58 | baseDir string // refFile override for session-authorized external roots |
| 59 | displayPath string // refFile label/path exposed in the resolved context block |
| 60 | raw string // the original token after '@', for labelling |
| 61 | } |
| 62 | |
| 63 | // ExternalFolderRefEntry is a session-authorized entry under a dropped external |
| 64 | // folder. Path is the opaque @ token path to submit; display fields are safe for |
| 65 | // UI labels and transcripts. |
| 66 | type ExternalFolderRefEntry struct { |
| 67 | Name string |
| 68 | Path string |
| 69 | DisplayName string |
| 70 | DisplayPath string |
| 71 | IsDir bool |
| 72 | } |
| 73 | |
| 74 | var pathLocationSuffixRe = regexp.MustCompile(`:\d+(?::\d+)?:?$`) |
| 75 | |
| 76 | const externalFolderRefPrefix = "__reasonix_external_folder" |
| 77 | |
| 78 | // parseRefTokens extracts the deduped, punctuation-trimmed tokens following '@' |
| 79 | // in a line. A token is a run of non-whitespace bytes, except that a |
| 80 | // backslash-escaped space or tab is part of the token with the backslash |
| 81 | // dropped — that is how a path containing spaces survives the |
| 82 | // whitespace-delimited grammar (EscapeRefPath produces that form). Any other |
| 83 | // backslash stays literal so Windows separators keep their meaning. Pure: |
| 84 | // classification (server? file?) happens in classifyRef. |
| 85 | func parseRefTokens(line string) []string { |
| 86 | var toks []string |
| 87 | seen := map[string]bool{} |
| 88 | for i := 0; i < len(line); i++ { |
| 89 | if line[i] != '@' { |
| 90 | continue |
| 91 | } |
| 92 | var b strings.Builder |
| 93 | j := i + 1 |
| 94 | for j < len(line) { |
| 95 | ch := line[j] |
| 96 | if ch == '\\' && j+1 < len(line) && (line[j+1] == ' ' || line[j+1] == '\t') { |
| 97 | b.WriteByte(line[j+1]) |
| 98 | j += 2 |
| 99 | continue |
| 100 | } |
| 101 | if isRefTokenBoundary(ch) { |
| 102 | break |
| 103 | } |
| 104 | b.WriteByte(ch) |
| 105 | j++ |
| 106 | } |
| 107 | i = j - 1 |
| 108 | t := strings.TrimRight(b.String(), ".,;!?)]}") |
| 109 | if t == "" || seen[t] { |
| 110 | continue |
| 111 | } |
| 112 | seen[t] = true |
| 113 | toks = append(toks, t) |
| 114 | } |
| 115 | return toks |
| 116 | } |
| 117 | |
| 118 | // isRefTokenBoundary matches the whitespace class the old `@([^\s]+)` token |
| 119 | // regexp stopped at. |
| 120 | func isRefTokenBoundary(ch byte) bool { |
| 121 | switch ch { |
| 122 | case ' ', '\t', '\n', '\r', '\f': |
| 123 | return true |
| 124 | default: |
| 125 | return false |
| 126 | } |
| 127 | } |
| 128 | |
| 129 | // EscapeRefPath returns path with spaces and tabs backslash-escaped so the |
| 130 | // result survives whitespace-delimited @-token parsing (parseRefTokens |
| 131 | // reverses it). Every other byte, including backslashes, passes through |
| 132 | // unchanged so Windows separators keep their meaning. |
| 133 | func EscapeRefPath(path string) string { |
| 134 | if !strings.ContainsAny(path, " \t") { |
| 135 | return path |
| 136 | } |
| 137 | var b strings.Builder |
| 138 | b.Grow(len(path) + 8) |
| 139 | for i := 0; i < len(path); i++ { |
| 140 | if path[i] == ' ' || path[i] == '\t' { |
| 141 | b.WriteByte('\\') |
| 142 | } |
| 143 | b.WriteByte(path[i]) |
| 144 | } |
| 145 | return b.String() |
| 146 | } |
| 147 | |
| 148 | // UnescapeRefPath reverses EscapeRefPath: a backslash before a space or tab is |
| 149 | // dropped; any other backslash stays literal. |
| 150 | func UnescapeRefPath(path string) string { |
| 151 | if !strings.Contains(path, `\`) { |
| 152 | return path |
| 153 | } |
| 154 | var b strings.Builder |
| 155 | b.Grow(len(path)) |
| 156 | for i := 0; i < len(path); i++ { |
| 157 | if path[i] == '\\' && i+1 < len(path) && (path[i+1] == ' ' || path[i+1] == '\t') { |
| 158 | continue |
| 159 | } |
| 160 | b.WriteByte(path[i]) |
| 161 | } |
| 162 | return b.String() |
| 163 | } |
| 164 | |
| 165 | // classifyRef decides what a token refers to. A "server:uri" token whose server |
| 166 | // is connected is an MCP resource; otherwise a token that names an existing path |
| 167 | // is a file. Anything else (an @mention, an email) is not a reference. exists is |
| 168 | // injected so the rule is testable without touching the filesystem. |
| 169 | func classifyRef(token string, known map[string]bool, exists func(string) bool) (ref, bool) { |
| 170 | if i := strings.Index(token, ":"); i > 0 && i+1 < len(token) && known[token[:i]] { |
| 171 | return ref{kind: refResource, server: token[:i], uri: token[i+1:], raw: token}, true |
| 172 | } |
| 173 | if isAttachmentRef(token) && exists(token) { |
| 174 | if isImageAttachmentRef(token) { |
| 175 | return ref{kind: refImage, path: token, raw: token}, true |
| 176 | } |
| 177 | return ref{kind: refFile, path: token, raw: token}, true |
| 178 | } |
| 179 | if exists(token) { |
| 180 | return ref{kind: refFile, path: token, raw: token}, true |
| 181 | } |
| 182 | return ref{}, false |
| 183 | } |
| 184 | |
| 185 | func isAttachmentRef(token string) bool { |
| 186 | return strings.HasPrefix(filepath.ToSlash(token), ".reasonix/attachments/") |
| 187 | } |
| 188 | |
| 189 | func isImageAttachmentRef(token string) bool { |
| 190 | switch strings.ToLower(filepath.Ext(token)) { |
| 191 | case ".png", ".jpg", ".jpeg", ".gif", ".webp", ".bmp", ".svg", ".tif", ".tiff": |
| 192 | return true |
| 193 | } |
| 194 | return false |
| 195 | } |
| 196 | |
| 197 | // RegisterExternalFolderRef authorizes one dropped directory outside the |
| 198 | // workspace as a structured @reference for this controller session. The returned |
| 199 | // token is path-like and whitespace-free so it survives the existing @ token |
| 200 | // parser even when the real directory path contains spaces or Windows drive |
| 201 | // punctuation. |
| 202 | func (c *Controller) RegisterExternalFolderRef(path string) (token, displayPath string, err error) { |
| 203 | if c == nil { |
| 204 | return "", "", fmt.Errorf("controller is not ready") |
| 205 | } |
| 206 | abs, err := normalizeExternalFolderRoot(path) |
| 207 | if err != nil { |
| 208 | return "", "", err |
| 209 | } |
| 210 | token = externalFolderRefToken(abs) |
| 211 | c.externalFolderRefsMu.Lock() |
| 212 | if c.externalFolderRefs == nil { |
| 213 | c.externalFolderRefs = map[string]string{} |
| 214 | } |
| 215 | c.externalFolderRefs[token] = abs |
| 216 | c.externalFolderRefsMu.Unlock() |
| 217 | if c.externalFolderToolRefs != nil { |
| 218 | c.externalFolderToolRefs.RegisterReadRoot(token, abs) |
| 219 | } |
| 220 | return token, filepath.ToSlash(abs), nil |
| 221 | } |
| 222 | |
| 223 | func normalizeExternalFolderRoot(path string) (string, error) { |
| 224 | path = strings.TrimSpace(path) |
| 225 | if path == "" { |
| 226 | return "", os.ErrInvalid |
| 227 | } |
| 228 | abs, err := filepath.Abs(path) |
| 229 | if err != nil { |
| 230 | return "", err |
| 231 | } |
| 232 | abs = filepath.Clean(abs) |
| 233 | if resolved, err := filepath.EvalSymlinks(abs); err == nil { |
| 234 | abs = filepath.Clean(resolved) |
| 235 | } |
| 236 | info, err := os.Stat(abs) |
| 237 | if err != nil { |
| 238 | return "", err |
| 239 | } |
| 240 | if !info.IsDir() { |
| 241 | return "", fmt.Errorf("%s is not a directory", path) |
| 242 | } |
| 243 | return abs, nil |
| 244 | } |
| 245 | |
| 246 | func externalFolderRefToken(abs string) string { |
| 247 | sum := sha256.Sum256([]byte(filepath.Clean(abs))) |
| 248 | hash := hex.EncodeToString(sum[:])[:12] |
| 249 | name := safeExternalFolderRefComponent(filepath.Base(abs)) |
| 250 | return externalFolderRefPrefix + "/" + hash + "/" + name |
| 251 | } |
| 252 | |
| 253 | func safeExternalFolderRefComponent(name string) string { |
| 254 | name = strings.TrimSpace(name) |
| 255 | if name == "" || name == "." || name == string(filepath.Separator) { |
| 256 | return "folder" |
| 257 | } |
| 258 | var b strings.Builder |
| 259 | lastDash := false |
| 260 | for _, r := range name { |
| 261 | ok := r >= 'a' && r <= 'z' || r >= 'A' && r <= 'Z' || r >= '0' && r <= '9' || r == '.' || r == '_' || r == '-' |
| 262 | if ok { |
| 263 | b.WriteRune(r) |
| 264 | lastDash = false |
| 265 | continue |
| 266 | } |
| 267 | if !lastDash { |
| 268 | b.WriteByte('-') |
| 269 | lastDash = true |
| 270 | } |
| 271 | } |
| 272 | out := strings.Trim(b.String(), ".-") |
| 273 | if out == "" { |
| 274 | return "folder" |
| 275 | } |
| 276 | return out |
| 277 | } |
| 278 | |
| 279 | func normalizeExternalFolderRefToken(token string) string { |
| 280 | token = strings.TrimSpace(token) |
| 281 | token = strings.TrimPrefix(token, "@") |
| 282 | token = filepath.ToSlash(token) |
| 283 | token = strings.TrimRight(token, "/") |
| 284 | return token |
| 285 | } |
| 286 | |
| 287 | func (c *Controller) externalFolderRef(token string) (ref, bool) { |
| 288 | _, rel, abs, ok := c.externalFolderRefTarget(token) |
| 289 | if !ok { |
| 290 | return ref{}, false |
| 291 | } |
| 292 | displayPath := externalFolderDisplayPath(abs, rel) |
| 293 | return ref{kind: refFile, path: rel, baseDir: abs, displayPath: displayPath, raw: token}, true |
| 294 | } |
| 295 | |
| 296 | func (c *Controller) externalFolderRefTarget(token string) (rootToken, rel, abs string, ok bool) { |
| 297 | key := normalizeExternalFolderRefToken(token) |
| 298 | if !strings.HasPrefix(key, externalFolderRefPrefix+"/") { |
| 299 | return "", "", "", false |
| 300 | } |
| 301 | c.externalFolderRefsMu.RLock() |
| 302 | defer c.externalFolderRefsMu.RUnlock() |
| 303 | if abs, ok := c.externalFolderRefs[key]; ok { |
| 304 | return key, ".", abs, true |
| 305 | } |
| 306 | for registered, abs := range c.externalFolderRefs { |
| 307 | if !strings.HasPrefix(key, registered+"/") { |
| 308 | continue |
| 309 | } |
| 310 | sub, ok := cleanExternalFolderSubpath(strings.TrimPrefix(key, registered+"/")) |
| 311 | if !ok { |
| 312 | return "", "", "", false |
| 313 | } |
| 314 | return registered, sub, abs, true |
| 315 | } |
| 316 | return "", "", "", false |
| 317 | } |
| 318 | |
| 319 | func cleanExternalFolderSubpath(sub string) (string, bool) { |
| 320 | sub = strings.TrimPrefix(filepath.ToSlash(strings.TrimSpace(sub)), "/") |
| 321 | if sub == "" || sub == "." { |
| 322 | return ".", true |
| 323 | } |
| 324 | cleaned := filepath.Clean(filepath.FromSlash(sub)) |
| 325 | if cleaned == "." { |
| 326 | return ".", true |
| 327 | } |
| 328 | if !filepath.IsLocal(cleaned) { |
| 329 | return "", false |
| 330 | } |
| 331 | return filepath.ToSlash(cleaned), true |
| 332 | } |
| 333 | |
| 334 | func externalFolderDisplayPath(abs, rel string) string { |
| 335 | if rel == "" || rel == "." { |
| 336 | return filepath.ToSlash(abs) |
| 337 | } |
| 338 | return filepath.ToSlash(filepath.Join(abs, filepath.FromSlash(rel))) |
| 339 | } |
| 340 | |
| 341 | func externalFolderDisplayName(abs, rel string) string { |
| 342 | name := filepath.Base(abs) |
| 343 | if rel != "" && rel != "." { |
| 344 | name = filepath.ToSlash(filepath.Join(name, filepath.FromSlash(rel))) |
| 345 | } |
| 346 | return name |
| 347 | } |
| 348 | |
| 349 | // ListExternalFolderRefDir lists one directory level under a registered |
| 350 | // external folder token. handled is true only when tokenPath targets a |
| 351 | // registered external folder; callers can fall back to workspace listing when it |
| 352 | // is false. |
| 353 | func (c *Controller) ListExternalFolderRefDir(tokenPath string) (entries []ExternalFolderRefEntry, handled bool) { |
| 354 | rootToken, rel, abs, ok := c.externalFolderRefTarget(tokenPath) |
| 355 | if !ok { |
| 356 | return nil, false |
| 357 | } |
| 358 | root, err := os.OpenRoot(abs) |
| 359 | if err != nil { |
| 360 | return nil, true |
| 361 | } |
| 362 | defer root.Close() |
| 363 | info, err := root.Stat(rel) |
| 364 | if err != nil || !info.IsDir() { |
| 365 | return nil, true |
| 366 | } |
| 367 | f, err := root.Open(rel) |
| 368 | if err != nil { |
| 369 | return nil, true |
| 370 | } |
| 371 | dirEntries, err := f.ReadDir(-1) |
| 372 | f.Close() |
| 373 | if err != nil { |
| 374 | return nil, true |
| 375 | } |
| 376 | dirs, files := []ExternalFolderRefEntry{}, []ExternalFolderRefEntry{} |
| 377 | for _, e := range dirEntries { |
| 378 | name := e.Name() |
| 379 | if skipRefDirEntry(name, e.IsDir()) { |
| 380 | continue |
| 381 | } |
| 382 | childRel := name |
| 383 | if rel != "." { |
| 384 | childRel = filepath.ToSlash(filepath.Join(rel, name)) |
| 385 | } |
| 386 | item := ExternalFolderRefEntry{ |
| 387 | Name: name, |
| 388 | Path: rootToken + "/" + childRel, |
| 389 | DisplayName: name, |
| 390 | DisplayPath: externalFolderDisplayPath(abs, childRel), |
| 391 | IsDir: e.IsDir(), |
| 392 | } |
| 393 | if e.IsDir() { |
| 394 | dirs = append(dirs, item) |
| 395 | continue |
| 396 | } |
| 397 | info, err := e.Info() |
| 398 | if err != nil || !info.Mode().IsRegular() { |
| 399 | continue |
| 400 | } |
| 401 | files = append(files, item) |
| 402 | } |
| 403 | sortExternalFolderRefEntries(dirs) |
| 404 | sortExternalFolderRefEntries(files) |
| 405 | return append(dirs, files...), true |
| 406 | } |
| 407 | |
| 408 | // SearchExternalFolderRefs finds entries under all registered external folders. |
| 409 | // Returned Path values are opaque token paths, so selecting one stays within the |
| 410 | // current session's authorization boundary. |
| 411 | func (c *Controller) SearchExternalFolderRefs(query string, limit int) []ExternalFolderRefEntry { |
| 412 | query = strings.TrimSpace(query) |
| 413 | if limit <= 0 || len(query) < 2 || strings.ContainsAny(query, `/\`) { |
| 414 | return nil |
| 415 | } |
| 416 | c.externalFolderRefsMu.RLock() |
| 417 | roots := make([]struct { |
| 418 | token string |
| 419 | abs string |
| 420 | }, 0, len(c.externalFolderRefs)) |
| 421 | for token, abs := range c.externalFolderRefs { |
| 422 | roots = append(roots, struct { |
| 423 | token string |
| 424 | abs string |
| 425 | }{token: token, abs: abs}) |
| 426 | } |
| 427 | c.externalFolderRefsMu.RUnlock() |
| 428 | sort.Slice(roots, func(i, j int) bool { |
| 429 | return externalFolderDisplayPath(roots[i].abs, ".") < externalFolderDisplayPath(roots[j].abs, ".") |
| 430 | }) |
| 431 | out := make([]ExternalFolderRefEntry, 0, limit) |
| 432 | queryLower := strings.ToLower(query) |
| 433 | for _, root := range roots { |
| 434 | if len(out) >= limit { |
| 435 | break |
| 436 | } |
| 437 | if info, err := os.Stat(root.abs); err != nil || !info.IsDir() { |
| 438 | continue |
| 439 | } |
| 440 | if strings.Contains(strings.ToLower(filepath.Base(root.abs)), queryLower) { |
| 441 | out = append(out, ExternalFolderRefEntry{ |
| 442 | Name: filepath.Base(root.abs), |
| 443 | Path: root.token, |
| 444 | DisplayName: externalFolderDisplayName(root.abs, "."), |
| 445 | DisplayPath: externalFolderDisplayPath(root.abs, "."), |
| 446 | IsDir: true, |
| 447 | }) |
| 448 | if len(out) >= limit { |
| 449 | break |
| 450 | } |
| 451 | } |
| 452 | for _, result := range fileref.Search(root.abs, query, limit-len(out)) { |
| 453 | rel := filepath.ToSlash(result.Path) |
| 454 | out = append(out, ExternalFolderRefEntry{ |
| 455 | Name: rel, |
| 456 | Path: root.token + "/" + rel, |
| 457 | DisplayName: externalFolderDisplayName(root.abs, rel), |
| 458 | DisplayPath: externalFolderDisplayPath(root.abs, rel), |
| 459 | IsDir: result.IsDir, |
| 460 | }) |
| 461 | if len(out) >= limit { |
| 462 | break |
| 463 | } |
| 464 | } |
| 465 | } |
| 466 | return out |
| 467 | } |
| 468 | |
| 469 | // ExternalFolderRefLocalPath resolves a registered external-folder token path to |
| 470 | // the local filesystem path authorized for this controller session. |
| 471 | func (c *Controller) ExternalFolderRefLocalPath(tokenPath string) (path, displayPath string, ok bool) { |
| 472 | _, rel, abs, ok := c.externalFolderRefTarget(tokenPath) |
| 473 | if !ok { |
| 474 | return "", "", false |
| 475 | } |
| 476 | return filepath.Join(abs, filepath.FromSlash(rel)), externalFolderDisplayPath(abs, rel), true |
| 477 | } |
| 478 | |
| 479 | func sortExternalFolderRefEntries(entries []ExternalFolderRefEntry) { |
| 480 | sort.Slice(entries, func(i, j int) bool { |
| 481 | return strings.ToLower(entries[i].DisplayName) < strings.ToLower(entries[j].DisplayName) |
| 482 | }) |
| 483 | } |
| 484 | |
| 485 | func skipRefDirEntry(name string, isDir bool) bool { |
| 486 | switch name { |
| 487 | case ".DS_Store", "Thumbs.db": |
| 488 | return true |
| 489 | } |
| 490 | if !isDir { |
| 491 | return false |
| 492 | } |
| 493 | switch name { |
| 494 | case ".codex", ".git", ".idea", ".npm", ".pnpm-store", ".vscode", "__pycache__", "build", "dist", "node_modules": |
| 495 | return true |
| 496 | } |
| 497 | return false |
| 498 | } |
| 499 | |
| 500 | // detectRefs finds the @references in a line: MCP resources for connected |
| 501 | // servers, and local paths that exist on disk. |
| 502 | func (c *Controller) detectRefs(line string) []ref { |
| 503 | return c.detectRefsMode(line, false) |
| 504 | } |
| 505 | |
| 506 | func (c *Controller) detectRefsMode(line string, scopedOnly bool) []ref { |
| 507 | known := map[string]bool{} |
| 508 | for _, n := range c.mcp.serverNames() { |
| 509 | known[n] = true |
| 510 | } |
| 511 | |
| 512 | var refs []ref |
| 513 | for _, tok := range parseRefTokens(line) { |
| 514 | if i := strings.Index(tok, ":"); i > 0 && i+1 < len(tok) && known[tok[:i]] { |
| 515 | refs = append(refs, ref{kind: refResource, server: tok[:i], uri: tok[i+1:], raw: tok}) |
| 516 | continue |
| 517 | } |
| 518 | if r, ok := c.externalFolderRef(tok); ok { |
| 519 | refs = append(refs, r) |
| 520 | continue |
| 521 | } |
| 522 | if c.workspaceRoot != "" { |
| 523 | if rel, ok := workspaceRefPath(tok, c.workspaceRoot); ok { |
| 524 | kind := refFile |
| 525 | if isAttachmentRef(rel) && isImageAttachmentRef(rel) { |
| 526 | kind = refImage |
| 527 | } |
| 528 | refs = append(refs, ref{kind: kind, path: rel, raw: tok}) |
| 529 | } |
| 530 | continue |
| 531 | } |
| 532 | if scopedOnly { |
| 533 | continue |
| 534 | } |
| 535 | if r, ok := classifyRef(tok, known, func(p string) bool { |
| 536 | _, err := os.Stat(p) |
| 537 | return err == nil |
| 538 | }); ok { |
| 539 | refs = append(refs, r) |
| 540 | } |
| 541 | } |
| 542 | return refs |
| 543 | } |
| 544 | |
| 545 | // HasRefs reports whether a line contains any resolvable @references, so a |
| 546 | // frontend can decide to resolve off its event loop only when needed. |
| 547 | func (c *Controller) HasRefs(line string) bool { |
| 548 | return len(c.detectRefs(line)) > 0 |
| 549 | } |
| 550 | |
| 551 | // inputImages resolves image @-references in the turn input to data URLs so the |
| 552 | // turn can carry them to a vision-capable model. Best-effort: an unreadable image |
| 553 | // is skipped — the @ref still lands as text via ResolveRefs. |
| 554 | func (c *Controller) inputImages(line string) []string { |
| 555 | if !c.imageInputEnabled() { |
| 556 | return nil |
| 557 | } |
| 558 | var urls []string |
| 559 | for _, r := range c.detectRefs(line) { |
| 560 | baseDir := c.workspaceRoot |
| 561 | if r.baseDir != "" { |
| 562 | baseDir = r.baseDir |
| 563 | } |
| 564 | if url, err := visionRefImageDataURL(r, baseDir); err == nil { |
| 565 | urls = append(urls, url) |
| 566 | } |
| 567 | } |
| 568 | return urls |
| 569 | } |
| 570 | |
| 571 | func visionRefImageDataURL(r ref, baseDir string) (string, error) { |
| 572 | switch r.kind { |
| 573 | case refImage: |
| 574 | return visionImageDataURL(r.path) |
| 575 | case refFile: |
| 576 | return visionFileImageDataURL(r.path, baseDir) |
| 577 | default: |
| 578 | return "", fmt.Errorf("reference is not an image") |
| 579 | } |
| 580 | } |
| 581 | |
| 582 | func visionFileImageDataURL(path, baseDir string) (string, error) { |
| 583 | absPath, absBase, ok := resolveAbsRef(path, baseDir) |
| 584 | if !ok { |
| 585 | return "", os.ErrNotExist |
| 586 | } |
| 587 | if absBase == "" { |
| 588 | return "", fmt.Errorf("workspace root is required for file image references") |
| 589 | } |
| 590 | |
| 591 | root, err := os.OpenRoot(absBase) |
| 592 | if err != nil { |
| 593 | return "", err |
| 594 | } |
| 595 | defer root.Close() |
| 596 | |
| 597 | rel, err := filepath.Rel(absBase, absPath) |
| 598 | if err != nil { |
| 599 | return "", err |
| 600 | } |
| 601 | |
| 602 | info, err := root.Lstat(rel) |
| 603 | if err != nil { |
| 604 | return "", err |
| 605 | } |
| 606 | if info.Mode()&os.ModeSymlink != 0 { |
| 607 | return "", fmt.Errorf("image path must not be a symlink") |
| 608 | } |
| 609 | if info.IsDir() || info.Size() <= 0 || info.Size() > maxImageAttachmentBytes { |
| 610 | return "", fmt.Errorf("image must be between 1 byte and 10 MB") |
| 611 | } |
| 612 | f, err := root.Open(rel) |
| 613 | if err != nil { |
| 614 | return "", err |
| 615 | } |
| 616 | defer f.Close() |
| 617 | opened, err := f.Stat() |
| 618 | if err != nil { |
| 619 | return "", err |
| 620 | } |
| 621 | if !os.SameFile(info, opened) { |
| 622 | return "", fmt.Errorf("image changed while opening") |
| 623 | } |
| 624 | return dataURLFromImageReader(f, path) |
| 625 | } |
| 626 | |
| 627 | func dataURLFromImageReader(r io.Reader, path string) (string, error) { |
| 628 | raw, err := io.ReadAll(io.LimitReader(r, maxImageAttachmentBytes+1)) |
| 629 | if err != nil { |
| 630 | return "", err |
| 631 | } |
| 632 | if len(raw) == 0 || len(raw) > maxImageAttachmentBytes { |
| 633 | return "", fmt.Errorf("image must be between 1 byte and 10 MB") |
| 634 | } |
| 635 | mime := detectedImageMime(raw) |
| 636 | if mime == "" { |
| 637 | return "", fmt.Errorf("%s is not a supported image", path) |
| 638 | } |
| 639 | raw, mime = compressForVision(raw, mime) |
| 640 | return "data:" + mime + ";base64," + base64.StdEncoding.EncodeToString(raw), nil |
| 641 | } |
| 642 | |
| 643 | // resolveBareNames batch-resolves simple filenames (no path separator) that |
| 644 | // don't exist in cwd. It walks the working tree once and matches every |
| 645 | // unresolved name against the set, stopping when all are found. This runs in |
| 646 | // the async ResolveRefs path, never on the TUI event loop. |
| 647 | func resolveBareNames(refs []ref, workspaceRoot string) []ref { |
| 648 | need := map[string]*ref{} |
| 649 | var names []string |
| 650 | for i := range refs { |
| 651 | r := &refs[i] |
| 652 | if r.kind != refFile || r.path != "" || !isSafeBareRefName(r.raw) { |
| 653 | continue |
| 654 | } |
| 655 | if workspaceRoot != "" { |
| 656 | if rel, ok := workspaceRefPath(r.raw, workspaceRoot); ok { |
| 657 | r.path = rel |
| 658 | continue |
| 659 | } |
| 660 | } |
| 661 | need[r.raw] = r |
| 662 | names = append(names, r.raw) |
| 663 | } |
| 664 | if len(names) == 0 { |
| 665 | return refs |
| 666 | } |
| 667 | found := 0 |
| 668 | cwd := workspaceRoot |
| 669 | if cwd == "" { |
| 670 | cwd, _ = os.Getwd() |
| 671 | } |
| 672 | _ = filepath.WalkDir(cwd, func(p string, d os.DirEntry, wErr error) error { |
| 673 | if wErr != nil || found == len(names) { |
| 674 | return filepath.SkipAll |
| 675 | } |
| 676 | if d.IsDir() { |
| 677 | switch d.Name() { |
| 678 | case ".git", "node_modules", ".DS_Store", "__pycache__", ".idea", ".vscode": |
| 679 | return filepath.SkipDir |
| 680 | } |
| 681 | return nil |
| 682 | } |
| 683 | if r, ok := need[d.Name()]; ok { |
| 684 | rel, _ := filepath.Rel(cwd, p) |
| 685 | r.path = filepath.ToSlash(rel) |
| 686 | delete(need, d.Name()) |
| 687 | found++ |
| 688 | } |
| 689 | return nil |
| 690 | }) |
| 691 | return refs |
| 692 | } |
| 693 | |
| 694 | func isSafeBareRefName(name string) bool { |
| 695 | if name == "" || name == "." || name == ".." { |
| 696 | return false |
| 697 | } |
| 698 | if strings.ContainsAny(name, "/\\") || strings.Contains(name, "..") { |
| 699 | return false |
| 700 | } |
| 701 | return filepath.Base(name) == name && filepath.IsLocal(name) |
| 702 | } |
| 703 | |
| 704 | // FileRefLine reports whether a submitted line is nothing but a path to an |
| 705 | // existing file — a dragged or pasted file lands as its bare path, which on |
| 706 | // POSIX starts with '/' and would otherwise be misread as a slash command. The |
| 707 | // returned string is that path turned into an @reference so it attaches. |
| 708 | func FileRefLine(line string) (string, bool) { |
| 709 | p := strings.Trim(strings.TrimSpace(line), `"'`) |
| 710 | if p == "" { |
| 711 | return "", false |
| 712 | } |
| 713 | if info, err := os.Stat(p); err != nil || info.IsDir() { |
| 714 | return "", false |
| 715 | } |
| 716 | return "@" + EscapeRefPath(p), true |
| 717 | } |
| 718 | |
| 719 | // SlashCodeCommentLine reports whether a slash-prefixed line is ordinary source |
| 720 | // text rather than a Reasonix slash command. |
| 721 | func SlashCodeCommentLine(line string) bool { |
| 722 | trimmed := strings.TrimSpace(line) |
| 723 | return strings.HasPrefix(trimmed, "//") || strings.HasPrefix(trimmed, "/*") |
| 724 | } |
| 725 | |
| 726 | // SlashPathLineRef reports whether a slash-prefixed line starts with a local file |
| 727 | // path, including common compiler-location suffixes like ":12" or ":12:34". |
| 728 | // It returns an @reference for the file so diagnostics that begin with an |
| 729 | // absolute path can keep their original text while also attaching file context. |
| 730 | func SlashPathLineRef(line, baseDir string) (string, bool) { |
| 731 | token, ok := leadingSlashPathToken(line) |
| 732 | if !ok { |
| 733 | return "", false |
| 734 | } |
| 735 | for _, p := range pathTokenCandidates(token) { |
| 736 | if fileRefExists(p, baseDir) { |
| 737 | return "@" + p, true |
| 738 | } |
| 739 | } |
| 740 | return "", false |
| 741 | } |
| 742 | |
| 743 | // SlashPathLikeLine reports whether a slash-prefixed line looks like a POSIX |
| 744 | // absolute path rather than a slash command. It intentionally stays conservative: |
| 745 | // unknown "/foo" remains an unknown command, while "/foo/bar..." is sent as |
| 746 | // ordinary prompt text even if the path no longer exists. |
| 747 | func SlashPathLikeLine(line string) bool { |
| 748 | token, ok := leadingSlashPathToken(line) |
| 749 | if !ok { |
| 750 | return false |
| 751 | } |
| 752 | for _, p := range pathTokenCandidates(token) { |
| 753 | if strings.Contains(p[1:], "/") { |
| 754 | return true |
| 755 | } |
| 756 | } |
| 757 | return false |
| 758 | } |
| 759 | |
| 760 | func leadingSlashPathToken(line string) (string, bool) { |
| 761 | fields := strings.Fields(strings.TrimSpace(line)) |
| 762 | if len(fields) == 0 { |
| 763 | return "", false |
| 764 | } |
| 765 | token := strings.Trim(fields[0], `"'`) |
| 766 | if !strings.HasPrefix(token, "/") || strings.HasPrefix(token, "//") { |
| 767 | return "", false |
| 768 | } |
| 769 | return token, true |
| 770 | } |
| 771 | |
| 772 | func pathTokenCandidates(token string) []string { |
| 773 | token = strings.TrimRight(strings.Trim(token, `"'`), ".,;!?)]}") |
| 774 | if token == "" { |
| 775 | return nil |
| 776 | } |
| 777 | candidates := []string{token} |
| 778 | if stripped := pathLocationSuffixRe.ReplaceAllString(token, ""); stripped != token { |
| 779 | candidates = append(candidates, stripped) |
| 780 | } |
| 781 | return candidates |
| 782 | } |
| 783 | |
| 784 | func fileRefExists(path, baseDir string) bool { |
| 785 | if baseDir != "" { |
| 786 | rel, _, absBase, ok := workspaceRel(path, baseDir) |
| 787 | if !ok { |
| 788 | return false |
| 789 | } |
| 790 | root, err := os.OpenRoot(absBase) |
| 791 | if err != nil { |
| 792 | return false |
| 793 | } |
| 794 | defer root.Close() |
| 795 | info, err := root.Stat(rel) |
| 796 | return err == nil && !info.IsDir() |
| 797 | } |
| 798 | info, err := os.Stat(path) |
| 799 | return err == nil && !info.IsDir() |
| 800 | } |
| 801 | |
| 802 | func workspaceRefPath(path, baseDir string) (string, bool) { |
| 803 | rel, _, absBase, ok := workspaceRel(path, baseDir) |
| 804 | if !ok { |
| 805 | return "", false |
| 806 | } |
| 807 | root, err := os.OpenRoot(absBase) |
| 808 | if err != nil { |
| 809 | return "", false |
| 810 | } |
| 811 | defer root.Close() |
| 812 | if _, err := root.Stat(rel); err != nil { |
| 813 | return "", false |
| 814 | } |
| 815 | return filepath.ToSlash(rel), true |
| 816 | } |
| 817 | |
| 818 | func workspaceRel(path, baseDir string) (rel, absPath, absBase string, ok bool) { |
| 819 | absPath, absBase, ok = resolveAbsRef(path, baseDir) |
| 820 | if !ok || absBase == "" { |
| 821 | return "", "", "", false |
| 822 | } |
| 823 | rel, err := filepath.Rel(absBase, absPath) |
| 824 | if err != nil || !filepath.IsLocal(rel) { |
| 825 | return "", "", "", false |
| 826 | } |
| 827 | return rel, absPath, absBase, true |
| 828 | } |
| 829 | |
| 830 | // ResolveRefs resolves the @references in a line into a single tagged context |
| 831 | // block (file/dir contents, MCP resource bodies), plus per-reference error |
| 832 | // strings for any that failed. An empty block means no references resolved. |
| 833 | // Safe to call off a frontend's event loop; honours ctx for the resource reads. |
| 834 | func (c *Controller) ResolveRefs(ctx context.Context, line string) (block string, errs []string) { |
| 835 | return c.resolveRefs(ctx, line, false) |
| 836 | } |
| 837 | |
| 838 | // ResolveScopedRefs is the HTTP/frontend variant: file references are honored |
| 839 | // only when they can be resolved under the controller workspace root. |
| 840 | func (c *Controller) ResolveScopedRefs(ctx context.Context, line string) (block string, errs []string) { |
| 841 | return c.resolveRefs(ctx, line, true) |
| 842 | } |
| 843 | |
| 844 | func (c *Controller) resolveRefs(ctx context.Context, line string, scopedOnly bool) (block string, errs []string) { |
| 845 | refs := c.detectRefsMode(line, scopedOnly) |
| 846 | refs = resolveBareNames(refs, c.workspaceRoot) |
| 847 | var b strings.Builder |
| 848 | includedInstructionPaths := map[string]bool{} |
| 849 | includedInstructionBodies := map[string]bool{} |
| 850 | if current := c.memory.current(); current != nil { |
| 851 | for _, doc := range current.Docs { |
| 852 | includedInstructionPaths[cleanAbsPath(doc.Path)] = true |
| 853 | includedInstructionBodies[doc.Body] = true |
| 854 | } |
| 855 | } |
| 856 | for _, r := range refs { |
| 857 | switch r.kind { |
| 858 | case refResource: |
| 859 | text, err := c.mcp.readResource(ctx, r.server, r.uri) |
| 860 | if err != nil { |
| 861 | errs = append(errs, "@"+r.raw+" — "+err.Error()) |
| 862 | continue |
| 863 | } |
| 864 | appendRefBlock(&b, "resource", `ref="@`+r.raw+`"`, text) |
| 865 | case refFile: |
| 866 | baseDir := c.workspaceRoot |
| 867 | if r.baseDir != "" { |
| 868 | baseDir = r.baseDir |
| 869 | } |
| 870 | text, isDir, err := readFileRef(r.path, baseDir) |
| 871 | if err != nil { |
| 872 | errs = append(errs, "@"+r.raw+" — "+err.Error()) |
| 873 | continue |
| 874 | } |
| 875 | pathInstructions, diagnostics := c.resolveReferencedInstructions(r, baseDir, includedInstructionPaths, includedInstructionBodies) |
| 876 | if pathInstructions != "" { |
| 877 | appendRefBlock(&b, "path-instructions", `target="`+html.EscapeString(displayPathForRef(r))+`"`, pathInstructions) |
| 878 | } |
| 879 | for _, diagnostic := range diagnostics { |
| 880 | errs = append(errs, "@"+r.raw+" — "+diagnostic.Message) |
| 881 | } |
| 882 | tag := "file" |
| 883 | if isDir { |
| 884 | tag = "dir" |
| 885 | } |
| 886 | displayPath := r.path |
| 887 | if r.displayPath != "" { |
| 888 | displayPath = r.displayPath |
| 889 | } |
| 890 | appendRefBlock(&b, tag, `path="`+displayPath+`"`, text) |
| 891 | case refImage: |
| 892 | appendRefBlock(&b, "image", `path="`+r.path+`"`, "[image attachment available at @"+r.path+"; sent as direct model image input only when the selected model supports vision. Text-only models can still use an available OCR/image/vision tool with this local path; image bytes are not inlined into prompt text.]") |
| 893 | } |
| 894 | } |
| 895 | return b.String(), errs |
| 896 | } |
| 897 | |
| 898 | func (c *Controller) resolveReferencedInstructions(r ref, baseDir string, includedPaths, includedBodies map[string]bool) (string, []instruction.Diagnostic) { |
| 899 | mem := c.memory.current() |
| 900 | if mem == nil || strings.TrimSpace(c.workspaceRoot) == "" || r.baseDir != "" { |
| 901 | return "", nil |
| 902 | } |
| 903 | absPath, absBase, ok := resolveAbsRef(r.path, baseDir) |
| 904 | if !ok || cleanAbsPath(absBase) != cleanAbsPath(c.workspaceRoot) { |
| 905 | return "", nil |
| 906 | } |
| 907 | targetDir := absPath |
| 908 | if info, err := os.Stat(absPath); err != nil || !info.IsDir() { |
| 909 | targetDir = filepath.Dir(absPath) |
| 910 | } |
| 911 | resolved := instruction.Resolve(instruction.ResolveOptions{ |
| 912 | WorkspaceRoot: c.workspaceRoot, |
| 913 | TargetDir: targetDir, |
| 914 | UserDir: mem.UserDir, |
| 915 | }) |
| 916 | var delta []instruction.Document |
| 917 | for _, doc := range resolved.Documents { |
| 918 | pathKey := cleanAbsPath(doc.Path) |
| 919 | if includedPaths[pathKey] || includedBodies[doc.Body] { |
| 920 | continue |
| 921 | } |
| 922 | includedPaths[pathKey] = true |
| 923 | includedBodies[doc.Body] = true |
| 924 | delta = append(delta, doc) |
| 925 | } |
| 926 | return instruction.Block(delta), resolved.Diagnostics |
| 927 | } |
| 928 | |
| 929 | func displayPathForRef(r ref) string { |
| 930 | if r.displayPath != "" { |
| 931 | return r.displayPath |
| 932 | } |
| 933 | return r.path |
| 934 | } |
| 935 | |
| 936 | func cleanAbsPath(path string) string { |
| 937 | abs, err := filepath.Abs(path) |
| 938 | if err != nil { |
| 939 | return filepath.Clean(path) |
| 940 | } |
| 941 | return filepath.Clean(abs) |
| 942 | } |
| 943 | |
| 944 | func appendRefBlock(b *strings.Builder, tag, attr, body string) { |
| 945 | if b.Len() > 0 { |
| 946 | b.WriteString("\n\n") |
| 947 | } |
| 948 | fmt.Fprintf(b, "<%s %s>\n%s\n</%s>", tag, attr, body, tag) |
| 949 | } |
| 950 | |
| 951 | // maxDirEntries caps how many directory entries are injected so @some-huge-dir |
| 952 | // can't blow the context window. |
| 953 | const maxDirEntries = 100 |
| 954 | |
| 955 | const maxDirDepth = 16 |
| 956 | |
| 957 | func directoryRefNote() string { |
| 958 | return fmt.Sprintf("[directory listing only; file contents are not inlined. Mention a listed file path to read its content. Common generated/vendor folders are skipped. Listing is capped at %d entries and %d nested levels.]", maxDirEntries, maxDirDepth) |
| 959 | } |
| 960 | |
| 961 | // readFileRef reads an @-referenced path for injection. A directory yields a |
| 962 | // recursive listing capped at maxDirEntries; a binary file (NUL in the first |
| 963 | // 8 KiB) is noted rather than dumped; a large file is truncated to |
| 964 | // maxFileRefBytes with a marker. isDir lets the caller pick the wrapping tag. |
| 965 | // When baseDir is non-empty the read is sandboxed under it via os.Root so |
| 966 | // user-supplied paths cannot escape the workspace; otherwise the path is |
| 967 | // used as-is (CLI single-workspace compatibility). |
| 968 | func readFileRef(path, baseDir string) (content string, isDir bool, err error) { |
| 969 | absPath, absBase, ok := resolveAbsRef(path, baseDir) |
| 970 | if !ok { |
| 971 | return "", false, os.ErrNotExist |
| 972 | } |
| 973 | if absBase == "" { |
| 974 | return readFileRefUnscoped(absPath) |
| 975 | } |
| 976 | |
| 977 | root, rerr := os.OpenRoot(absBase) |
| 978 | if rerr != nil { |
| 979 | return "", false, rerr |
| 980 | } |
| 981 | defer root.Close() |
| 982 | |
| 983 | rel, rerr := filepath.Rel(absBase, absPath) |
| 984 | if rerr != nil { |
| 985 | return "", false, rerr |
| 986 | } |
| 987 | displayPath := filepath.ToSlash(rel) |
| 988 | |
| 989 | info, err := root.Stat(rel) |
| 990 | if err != nil { |
| 991 | return "", false, err |
| 992 | } |
| 993 | if info.IsDir() { |
| 994 | var b strings.Builder |
| 995 | b.WriteString(directoryRefNote()) |
| 996 | b.WriteString("\n\n") |
| 997 | n := 0 |
| 998 | err := walkRootDir(root, rel, rel, &b, &n, 0) |
| 999 | if n >= maxDirEntries { |
| 1000 | b.WriteString("\n…[truncated; directory has more entries]…") |
| 1001 | } |
| 1002 | if err != nil { |
| 1003 | return "", true, err |
| 1004 | } |
| 1005 | return b.String(), true, nil |
| 1006 | } |
| 1007 | |
| 1008 | if strings.EqualFold(filepath.Ext(rel), ".pdf") { |
| 1009 | return readPDFRef(absPath, info.Size()), false, nil |
| 1010 | } |
| 1011 | |
| 1012 | f, err := root.Open(rel) |
| 1013 | if err != nil { |
| 1014 | return "", false, err |
| 1015 | } |
| 1016 | defer f.Close() |
| 1017 | |
| 1018 | buf := make([]byte, maxFileRefBytes+1) |
| 1019 | n, rerr := io.ReadFull(f, buf) |
| 1020 | if rerr != nil && rerr != io.ErrUnexpectedEOF && rerr != io.EOF { |
| 1021 | return "", false, rerr |
| 1022 | } |
| 1023 | data := buf[:n] |
| 1024 | |
| 1025 | if mime := imageMime(data, rel); mime != "" { |
| 1026 | return imageFileRefNote(displayPath, mime, info.Size(), true), false, nil |
| 1027 | } |
| 1028 | if bytes.IndexByte(data[:min(n, 8192)], 0) >= 0 { |
| 1029 | return fmt.Sprintf("[binary file %s, %d bytes — not shown]", displayPath, info.Size()), false, nil |
| 1030 | } |
| 1031 | if n > maxFileRefBytes { |
| 1032 | return string(data[:maxFileRefBytes]) + fmt.Sprintf("\n…[truncated; file is %d bytes]…", info.Size()), false, nil |
| 1033 | } |
| 1034 | return string(data), false, nil |
| 1035 | } |
| 1036 | |
| 1037 | // readFileRefUnscoped is the legacy readFileRef body kept for CLI single-workspace |
| 1038 | // compatibility, where no controller-scoped sandbox is in effect. |
| 1039 | func readFileRefUnscoped(path string) (content string, isDir bool, err error) { |
| 1040 | info, err := os.Stat(path) |
| 1041 | if err != nil { |
| 1042 | return "", false, err |
| 1043 | } |
| 1044 | if info.IsDir() { |
| 1045 | var b strings.Builder |
| 1046 | b.WriteString(directoryRefNote()) |
| 1047 | b.WriteString("\n\n") |
| 1048 | n := 0 |
| 1049 | err := filepath.WalkDir(path, func(p string, d os.DirEntry, wErr error) error { |
| 1050 | if wErr != nil { |
| 1051 | return wErr |
| 1052 | } |
| 1053 | if n >= maxDirEntries { |
| 1054 | return filepath.SkipAll |
| 1055 | } |
| 1056 | if p == path { |
| 1057 | return nil |
| 1058 | } |
| 1059 | if skipRefDirEntry(d.Name(), d.IsDir()) { |
| 1060 | if d.IsDir() { |
| 1061 | return filepath.SkipDir |
| 1062 | } |
| 1063 | return nil |
| 1064 | } |
| 1065 | rel, rErr := filepath.Rel(path, p) |
| 1066 | if rErr != nil { |
| 1067 | rel = p |
| 1068 | } |
| 1069 | rel = strings.ReplaceAll(rel, string(os.PathSeparator), "/") |
| 1070 | if d.IsDir() { |
| 1071 | rel += "/" |
| 1072 | } |
| 1073 | b.WriteString(rel) |
| 1074 | b.WriteByte('\n') |
| 1075 | n++ |
| 1076 | return nil |
| 1077 | }) |
| 1078 | if n >= maxDirEntries { |
| 1079 | b.WriteString("\n…[truncated; directory has more entries]…") |
| 1080 | } |
| 1081 | if err != nil { |
| 1082 | return "", true, err |
| 1083 | } |
| 1084 | return b.String(), true, nil |
| 1085 | } |
| 1086 | |
| 1087 | if strings.EqualFold(filepath.Ext(path), ".pdf") { |
| 1088 | return readPDFRef(path, info.Size()), false, nil |
| 1089 | } |
| 1090 | |
| 1091 | f, err := os.Open(path) |
| 1092 | if err != nil { |
| 1093 | return "", false, err |
| 1094 | } |
| 1095 | defer f.Close() |
| 1096 | |
| 1097 | buf := make([]byte, maxFileRefBytes+1) |
| 1098 | n, rerr := io.ReadFull(f, buf) |
| 1099 | if rerr != nil && rerr != io.ErrUnexpectedEOF && rerr != io.EOF { |
| 1100 | return "", false, rerr |
| 1101 | } |
| 1102 | data := buf[:n] |
| 1103 | |
| 1104 | if mime := imageMime(data, path); mime != "" { |
| 1105 | return imageFileRefNote(path, mime, info.Size(), false), false, nil |
| 1106 | } |
| 1107 | if bytes.IndexByte(data[:min(n, 8192)], 0) >= 0 { |
| 1108 | return fmt.Sprintf("[binary file %s, %d bytes — not shown]", path, info.Size()), false, nil |
| 1109 | } |
| 1110 | if n > maxFileRefBytes { |
| 1111 | return string(data[:maxFileRefBytes]) + fmt.Sprintf("\n…[truncated; file is %d bytes]…", info.Size()), false, nil |
| 1112 | } |
| 1113 | return string(data), false, nil |
| 1114 | } |
| 1115 | |
| 1116 | func imageFileRefNote(displayPath, mime string, size int64, attached bool) string { |
| 1117 | if attached { |
| 1118 | return fmt.Sprintf("[image file %s, mime=%s, %d bytes — sent as direct model image input only when the selected model supports vision. Text-only models can still use an available OCR/image/vision tool with this local path; image bytes are not inlined into prompt text.]", displayPath, mime, size) |
| 1119 | } |
| 1120 | return fmt.Sprintf("[image file %s, mime=%s, %d bytes — not sent as direct model image input because no workspace root is available. Use a workspace-scoped file reference, image attachment, or an available OCR/image/vision tool with a readable local path.]", displayPath, mime, size) |
| 1121 | } |
| 1122 | |
| 1123 | // walkRootDir walks a directory under a sandboxed *os.Root and writes each |
| 1124 | // entry relative to base (skipping noisy ones like .git and node_modules) into b |
| 1125 | // until n hits maxDirEntries. |
| 1126 | func walkRootDir(root *os.Root, dir, base string, b *strings.Builder, n *int, depth int) error { |
| 1127 | if depth > maxDirDepth || *n >= maxDirEntries { |
| 1128 | return nil |
| 1129 | } |
| 1130 | f, err := root.Open(dir) |
| 1131 | if err != nil { |
| 1132 | return err |
| 1133 | } |
| 1134 | entries, err := f.ReadDir(-1) |
| 1135 | f.Close() |
| 1136 | if err != nil { |
| 1137 | return err |
| 1138 | } |
| 1139 | sort.Slice(entries, func(i, j int) bool { |
| 1140 | if entries[i].IsDir() != entries[j].IsDir() { |
| 1141 | return entries[i].IsDir() |
| 1142 | } |
| 1143 | return strings.ToLower(entries[i].Name()) < strings.ToLower(entries[j].Name()) |
| 1144 | }) |
| 1145 | for _, e := range entries { |
| 1146 | if *n >= maxDirEntries { |
| 1147 | return nil |
| 1148 | } |
| 1149 | name := e.Name() |
| 1150 | child := filepath.ToSlash(filepath.Join(dir, name)) |
| 1151 | entry := name |
| 1152 | if rel, err := filepath.Rel(base, child); err == nil && filepath.IsLocal(rel) { |
| 1153 | entry = filepath.ToSlash(rel) |
| 1154 | } |
| 1155 | if skipRefDirEntry(name, e.IsDir()) { |
| 1156 | continue |
| 1157 | } |
| 1158 | if e.IsDir() { |
| 1159 | entry += "/" |
| 1160 | } |
| 1161 | b.WriteString(entry) |
| 1162 | b.WriteByte('\n') |
| 1163 | *n++ |
| 1164 | if e.IsDir() { |
| 1165 | if err := walkRootDir(root, child, base, b, n, depth+1); err != nil { |
| 1166 | return err |
| 1167 | } |
| 1168 | } |
| 1169 | } |
| 1170 | return nil |
| 1171 | } |
| 1172 | |
| 1173 | // resolveAbsRef resolves the user-supplied @-reference path against baseDir |
| 1174 | // and returns the absolute path plus the absolute base root to sandbox I/O |
| 1175 | // under. With a baseDir, the path is confined under it (a relative path that |
| 1176 | // escapes via ".." is rejected). With an empty baseDir, the path is returned |
| 1177 | // as-is and the caller falls back to plain os.Stat/os.Open so CLI usage |
| 1178 | // (where there is no controller-scoped workspace) keeps working. |
| 1179 | func resolveAbsRef(path, baseDir string) (absPath, absBase string, ok bool) { |
| 1180 | if baseDir == "" { |
| 1181 | return path, "", true |
| 1182 | } |
| 1183 | absBase = baseDir |
| 1184 | if !filepath.IsAbs(absBase) { |
| 1185 | var err error |
| 1186 | absBase, err = filepath.Abs(absBase) |
| 1187 | if err != nil { |
| 1188 | return "", "", false |
| 1189 | } |
| 1190 | } |
| 1191 | cleaned := filepath.Clean(path) |
| 1192 | if !filepath.IsAbs(cleaned) { |
| 1193 | cleaned = filepath.Join(absBase, cleaned) |
| 1194 | } |
| 1195 | rel, err := filepath.Rel(absBase, cleaned) |
| 1196 | if err != nil || !filepath.IsLocal(rel) { |
| 1197 | return "", "", false |
| 1198 | } |
| 1199 | return cleaned, absBase, true |
| 1200 | } |
| 1201 | |
| 1202 | func readPDFRef(path string, size int64) string { |
| 1203 | result, err := extractPDFText(path) |
| 1204 | if err != nil { |
| 1205 | return fmt.Sprintf("[PDF file %s, %d bytes — text extraction unavailable: %v. If this is a scanned/image-only PDF, use OCR or an available multimodal/vision tool with this path.]", path, size, err) |
| 1206 | } |
| 1207 | text := strings.TrimSpace(result.text) |
| 1208 | if text == "" { |
| 1209 | return fmt.Sprintf("[PDF file %s, %d bytes — no extractable text found. It may be scanned/image-only; use OCR or an available multimodal/vision tool with this path.]", path, size) |
| 1210 | } |
| 1211 | var b strings.Builder |
| 1212 | fmt.Fprintf(&b, "[PDF text extracted from %s using %s", path, result.tool) |
| 1213 | if result.truncated { |
| 1214 | fmt.Fprintf(&b, "; truncated to the first %d bytes", maxFileRefBytes) |
| 1215 | } |
| 1216 | b.WriteString("]\n") |
| 1217 | b.WriteString(text) |
| 1218 | return b.String() |
| 1219 | } |
| 1220 | |
| 1221 | func extractPDFTextDefault(path string) (pdfExtractResult, error) { |
| 1222 | var firstErr error |
| 1223 | if pdftotext, err := exec.LookPath("pdftotext"); err == nil { |
| 1224 | if text, truncated, err := runPDFTextCommand(pdftotext, []string{"-enc", "UTF-8", "-layout", path, "-"}); err == nil { |
| 1225 | return pdfExtractResult{text: text, tool: "pdftotext", truncated: truncated}, nil |
| 1226 | } else { |
| 1227 | firstErr = err |
| 1228 | } |
| 1229 | } |
| 1230 | python, err := findPython() |
| 1231 | if err != nil { |
| 1232 | if firstErr != nil { |
| 1233 | return pdfExtractResult{}, fmt.Errorf("pdftotext failed (%v), and Python PDF libraries are not available", firstErr) |
| 1234 | } |
| 1235 | return pdfExtractResult{}, fmt.Errorf("pdftotext and Python PDF libraries are not available") |
| 1236 | } |
| 1237 | text, truncated, err := runPDFTextCommand(python, []string{"-c", pythonPDFExtractScript, path}) |
| 1238 | if err != nil { |
| 1239 | if firstErr != nil { |
| 1240 | return pdfExtractResult{}, fmt.Errorf("pdftotext failed (%v), Python PDF extraction failed (%w)", firstErr, err) |
| 1241 | } |
| 1242 | return pdfExtractResult{}, err |
| 1243 | } |
| 1244 | return pdfExtractResult{text: text, tool: "Python PDF library", truncated: truncated}, nil |
| 1245 | } |
| 1246 | |
| 1247 | func findPython() (string, error) { |
| 1248 | for _, name := range []string{"python3", "python", "py"} { |
| 1249 | if p, err := exec.LookPath(name); err == nil { |
| 1250 | return p, nil |
| 1251 | } |
| 1252 | } |
| 1253 | return "", fmt.Errorf("python not found") |
| 1254 | } |
| 1255 | |
| 1256 | func runPDFTextCommand(name string, args []string) (string, bool, error) { |
| 1257 | ctx, cancel := context.WithTimeout(context.Background(), pdfExtractTimeout) |
| 1258 | defer cancel() |
| 1259 | cmd := exec.CommandContext(ctx, name, args...) |
| 1260 | cmd.Env = secrets.ProcessEnv() |
| 1261 | setShellKillTree(cmd) |
| 1262 | cmd.WaitDelay = pdfExtractWaitDelay |
| 1263 | proc.HideWindow(cmd) |
| 1264 | var stdout limitedBuffer |
| 1265 | var stderr limitedBuffer |
| 1266 | cmd.Stdout = &stdout |
| 1267 | cmd.Stderr = &stderr |
| 1268 | waitErr := cmd.Run() |
| 1269 | if ctx.Err() == context.DeadlineExceeded { |
| 1270 | return "", false, fmt.Errorf("PDF text extraction timed out") |
| 1271 | } |
| 1272 | if waitErr != nil { |
| 1273 | msg := strings.TrimSpace(stderr.String()) |
| 1274 | if msg != "" { |
| 1275 | if stderr.Truncated() { |
| 1276 | msg += "\n…[truncated]…" |
| 1277 | } |
| 1278 | return "", false, fmt.Errorf("%w: %s", waitErr, msg) |
| 1279 | } |
| 1280 | return "", false, waitErr |
| 1281 | } |
| 1282 | return stdout.String(), stdout.Truncated(), nil |
| 1283 | } |
| 1284 | |
| 1285 | type limitedBuffer struct { |
| 1286 | buf bytes.Buffer |
| 1287 | truncated bool |
| 1288 | } |
| 1289 | |
| 1290 | func (b *limitedBuffer) Write(p []byte) (int, error) { |
| 1291 | remaining := maxFileRefBytes - b.buf.Len() |
| 1292 | if remaining > 0 { |
| 1293 | if len(p) > remaining { |
| 1294 | _, _ = b.buf.Write(p[:remaining]) |
| 1295 | b.truncated = true |
| 1296 | } else { |
| 1297 | _, _ = b.buf.Write(p) |
| 1298 | } |
| 1299 | } else if len(p) > 0 { |
| 1300 | b.truncated = true |
| 1301 | } |
| 1302 | return len(p), nil |
| 1303 | } |
| 1304 | |
| 1305 | func (b *limitedBuffer) String() string { return b.buf.String() } |
| 1306 | |
| 1307 | func (b *limitedBuffer) Truncated() bool { return b.truncated } |
| 1308 | |
| 1309 | const pythonPDFExtractScript = ` |
| 1310 | import sys |
| 1311 | |
| 1312 | path = sys.argv[1] |
| 1313 | |
| 1314 | try: |
| 1315 | from pypdf import PdfReader |
| 1316 | except Exception: |
| 1317 | try: |
| 1318 | from PyPDF2 import PdfReader |
| 1319 | except Exception: |
| 1320 | PdfReader = None |
| 1321 | |
| 1322 | if PdfReader is not None: |
| 1323 | reader = PdfReader(path) |
| 1324 | for page in reader.pages: |
| 1325 | text = page.extract_text() or "" |
| 1326 | if text: |
| 1327 | print(text) |
| 1328 | sys.exit(0) |
| 1329 | |
| 1330 | try: |
| 1331 | import pdfplumber |
| 1332 | except Exception as exc: |
| 1333 | raise SystemExit("no supported Python PDF library found") from exc |
| 1334 | |
| 1335 | with pdfplumber.open(path) as pdf: |
| 1336 | for page in pdf.pages: |
| 1337 | text = page.extract_text() or "" |
| 1338 | if text: |
| 1339 | print(text) |
| 1340 | ` |
| 1341 | |
| 1342 | func imageMime(data []byte, path string) string { |
| 1343 | mime := http.DetectContentType(data[:min(len(data), 512)]) |
| 1344 | if strings.HasPrefix(mime, "image/") { |
| 1345 | return mime |
| 1346 | } |
| 1347 | switch strings.ToLower(filepath.Ext(path)) { |
| 1348 | case ".png": |
| 1349 | return "image/png" |
| 1350 | case ".jpg", ".jpeg": |
| 1351 | return "image/jpeg" |
| 1352 | case ".gif": |
| 1353 | return "image/gif" |
| 1354 | case ".webp": |
| 1355 | return "image/webp" |
| 1356 | case ".tiff", ".tif": |
| 1357 | return "image/tiff" |
| 1358 | } |
| 1359 | return "" |
| 1360 | } |
| 1361 |