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