| 1 | // Package builtin provides Reasonix's compile-time built-in tools. Each tool |
| 2 | // self-registers via init(); main blank-imports this package to wire them in. |
| 3 | package builtin |
| 4 | |
| 5 | import ( |
| 6 | "bufio" |
| 7 | "bytes" |
| 8 | "context" |
| 9 | "crypto/sha256" |
| 10 | "encoding/hex" |
| 11 | "encoding/json" |
| 12 | "fmt" |
| 13 | "io" |
| 14 | "path/filepath" |
| 15 | "strconv" |
| 16 | "strings" |
| 17 | |
| 18 | "golang.org/x/text/transform" |
| 19 | |
| 20 | fileenc "reasonix/internal/fileutil/encoding" |
| 21 | "reasonix/internal/tool" |
| 22 | ) |
| 23 | |
| 24 | const ( |
| 25 | readFileBinaryPeek = 8 * 1024 // bytes scanned for NUL before reading further |
| 26 | readFileDetectSample = 256 * 1024 // bytes sampled for encoding detection before streaming |
| 27 | readFileMaxLineBytes = 1024 * 1024 |
| 28 | readFileMaxFormattedBytes = 8 << 20 |
| 29 | ) |
| 30 | |
| 31 | func init() { tool.RegisterBuiltin(readFile{}) } |
| 32 | |
| 33 | // readFile reads a text file. workDir, when non-empty, is the directory a |
| 34 | // relative path is resolved against (see resolveIn). paths maps session-scoped |
| 35 | // external read aliases to local roots without changing the model-visible tool |
| 36 | // schema. forbidRoots lists directories the tool may not read from (resolved, |
| 37 | // absolute paths). |
| 38 | type readFile struct { |
| 39 | workDir string |
| 40 | paths *PathResolver |
| 41 | forbidRoots []string |
| 42 | // overlay, when non-nil, serves content from the host transport (unsaved |
| 43 | // editor buffers) before falling back to disk. Consulted only after path |
| 44 | // resolution and read confinement, and never for external alias paths. |
| 45 | overlay FileOverlay |
| 46 | captured *tool.ReadResultSource |
| 47 | } |
| 48 | |
| 49 | const ( |
| 50 | readFileDefaultLimit = 2000 // lines returned when limit is unset |
| 51 | ) |
| 52 | |
| 53 | // readFileParams is one validated read_file call with defaults applied. |
| 54 | type readFileParams struct { |
| 55 | Path string |
| 56 | Intent tool.ReadIntent |
| 57 | WindowGiven bool |
| 58 | Offset int |
| 59 | Limit int |
| 60 | } |
| 61 | |
| 62 | const ( |
| 63 | readFileEmptyOutput = "(empty file)" |
| 64 | readFilePastEOFTail = " is past EOF — file has " |
| 65 | ) |
| 66 | |
| 67 | // readWindowGiven reports whether the call named an explicit line window. |
| 68 | func readWindowGiven(args json.RawMessage) bool { |
| 69 | var fields map[string]json.RawMessage |
| 70 | if err := json.Unmarshal(args, &fields); err != nil { |
| 71 | return false |
| 72 | } |
| 73 | _, offset := fields["offset"] |
| 74 | _, limit := fields["limit"] |
| 75 | return offset || limit |
| 76 | } |
| 77 | |
| 78 | // parseReadFileParams validates one read_file call and applies the documented |
| 79 | // defaults, so Execute and ReadEnvelope agree on what was requested. |
| 80 | func parseReadFileParams(args json.RawMessage) (readFileParams, error) { |
| 81 | var p struct { |
| 82 | Path string `json:"path"` |
| 83 | Intent string `json:"intent,omitempty"` |
| 84 | Offset int `json:"offset,omitempty"` |
| 85 | Limit int `json:"limit,omitempty"` |
| 86 | } |
| 87 | if err := json.Unmarshal(args, &p); err != nil { |
| 88 | return readFileParams{}, fmt.Errorf("invalid args: %w", err) |
| 89 | } |
| 90 | if p.Path == "" { |
| 91 | return readFileParams{}, fmt.Errorf("path is required") |
| 92 | } |
| 93 | windowGiven := readWindowGiven(args) |
| 94 | intent, err := readIntentFor(p.Intent, windowGiven) |
| 95 | if err != nil { |
| 96 | return readFileParams{}, err |
| 97 | } |
| 98 | if p.Offset < 0 { |
| 99 | p.Offset = 0 |
| 100 | } |
| 101 | if p.Limit <= 0 { |
| 102 | p.Limit = readFileDefaultLimit |
| 103 | } |
| 104 | return readFileParams{Path: p.Path, Intent: intent, WindowGiven: windowGiven, Offset: p.Offset, Limit: p.Limit}, nil |
| 105 | } |
| 106 | |
| 107 | // readIntentFor resolves the effective read intent and rejects combinations |
| 108 | // that would leave the caller unsure which promise the call made. |
| 109 | func readIntentFor(explicit string, windowGiven bool) (tool.ReadIntent, error) { |
| 110 | switch tool.ReadIntent(strings.TrimSpace(explicit)) { |
| 111 | case "": |
| 112 | if windowGiven { |
| 113 | return tool.ReadIntentRange, nil |
| 114 | } |
| 115 | return tool.ReadIntentInspect, nil |
| 116 | case tool.ReadIntentInspect: |
| 117 | return tool.ReadIntentInspect, nil |
| 118 | case tool.ReadIntentRange: |
| 119 | if !windowGiven { |
| 120 | return "", fmt.Errorf("intent=range requires an explicit offset or limit; pass the window to read, or use intent=inspect for a bounded preview") |
| 121 | } |
| 122 | return tool.ReadIntentRange, nil |
| 123 | case tool.ReadIntentFull: |
| 124 | // Kept as a compatibility hint. Every call remains one bounded window; |
| 125 | // the host no longer creates a whole-file debt or completion gate. |
| 126 | return tool.ReadIntentFull, nil |
| 127 | default: |
| 128 | return "", fmt.Errorf("intent must be inspect, range, or full (got %q)", explicit) |
| 129 | } |
| 130 | } |
| 131 | |
| 132 | func (readFile) Name() string { return "read_file" } |
| 133 | |
| 134 | func (readFile) Description() string { |
| 135 | return "Read one bounded text window with optional line offset/limit. Output prefixes each line with its 1-based number. Any successful window observes the current file version for later structured edits. Use the next-window hint to page only when more content is useful. Legacy intent and cursor fields are accepted as navigation hints and never create a whole-file completion requirement." |
| 136 | } |
| 137 | |
| 138 | func (readFile) Schema() json.RawMessage { |
| 139 | return json.RawMessage(`{ |
| 140 | "type":"object", |
| 141 | "properties":{ |
| 142 | "path":{"type":"string","description":"File path"}, |
| 143 | "intent":{"type":"string","enum":["inspect","range","full"],"description":"Compatibility hint. Every value reads only this bounded window and creates no whole-file obligation."}, |
| 144 | "cursor":{"type":"string","description":"Optional continuation cursor from a prior result. Invalid legacy cursors should be replaced with an explicit offset and limit."}, |
| 145 | "offset":{"type":"integer","description":"0-based line offset to start reading from (default 0)","minimum":0}, |
| 146 | "limit":{"type":"integer","description":"Maximum lines to return (default 2000)","minimum":1} |
| 147 | }, |
| 148 | "required":["path"] |
| 149 | }`) |
| 150 | } |
| 151 | |
| 152 | func (readFile) ReadOnly() bool { return true } |
| 153 | |
| 154 | // ReadEnvelope reports what one read_file call delivered. The source identity |
| 155 | // comes from the store that actually served the content, the snapshot stays |
| 156 | // constant across the pages of one logical read, and the window digest covers |
| 157 | // only this page's delivered lines. read_id, result_ref and workspace_id are |
| 158 | // host identity the agent fills in. |
| 159 | func (r readFile) ReadEnvelope(ctx context.Context, args json.RawMessage, output string) (tool.ReadResultEnvelope, bool) { |
| 160 | p, err := parseReadFileParams(args) |
| 161 | if err != nil { |
| 162 | return tool.ReadResultEnvelope{}, false |
| 163 | } |
| 164 | rp := resolveReadablePath(r.workDir, p.Path, r.paths) |
| 165 | env := tool.ReadResultEnvelope{ |
| 166 | ProtocolVersion: tool.ReadResultProtocolVersion, |
| 167 | Source: tool.ReadResultSource{CanonicalPath: rp.Path}, |
| 168 | Intent: p.Intent, |
| 169 | } |
| 170 | if p.WindowGiven { |
| 171 | requested := tool.ReadRange{Start: p.Offset, End: p.Offset + p.Limit} |
| 172 | env.RequestedRange = &requested |
| 173 | } |
| 174 | |
| 175 | // The store that served the content owns the identity: an unsaved editor |
| 176 | // buffer must never be proven by the disk file's identity. |
| 177 | if r.captured != nil { |
| 178 | env.Source = *r.captured |
| 179 | } |
| 180 | env.Source.Snapshot = tool.SourceSnapshot(env.Source.Kind, rp.Path, env.Source.Identity) |
| 181 | |
| 182 | window, hasWindow := tool.ParseReadWindow(output) |
| 183 | if hasWindow { |
| 184 | env.DeliveredRanges = []tool.ReadRange{window.Range()} |
| 185 | env.WindowDigest = tool.WindowDigest(rp.Path, window) |
| 186 | } |
| 187 | trailer := tool.ParseReadTrailer(output) |
| 188 | env.HasMore = trailer.HasMore |
| 189 | env.EOF = !trailer.HasMore |
| 190 | switch { |
| 191 | case trailer.LocalSafety: |
| 192 | env.SourceCut = tool.ReadCutSafetyPage |
| 193 | case trailer.HasMore: |
| 194 | env.SourceCut = tool.ReadCutPageLimit |
| 195 | } |
| 196 | if env.EOF { |
| 197 | if end, ok := readSourceEnd(output, window, hasWindow); ok { |
| 198 | env.SourceEnd = &end |
| 199 | } |
| 200 | } |
| 201 | if trailer.HasMore { |
| 202 | env.NextCursor = tool.EncodeReadCursor(tool.ReadCursor{ |
| 203 | Path: rp.Path, |
| 204 | Snapshot: env.Source.Snapshot, |
| 205 | NextStart: trailer.NextOffset, |
| 206 | }) |
| 207 | } |
| 208 | return env, true |
| 209 | } |
| 210 | |
| 211 | // overlayText mirrors Execute's overlay routing so the envelope names the same |
| 212 | // store that produced the delivered bytes. |
| 213 | func (r readFile) overlayText(ctx context.Context, rp ResolvedPath) (string, bool) { |
| 214 | if r.overlay == nil || rp.External || !filepath.IsAbs(rp.Path) { |
| 215 | return "", false |
| 216 | } |
| 217 | return r.overlay.ReadTextFile(ctx, rp.Path) |
| 218 | } |
| 219 | |
| 220 | func digestText(content string) string { |
| 221 | sum := sha256.Sum256([]byte(content)) |
| 222 | return hex.EncodeToString(sum[:]) |
| 223 | } |
| 224 | |
| 225 | // readSourceEnd recovers the source's zero-based end line index from the |
| 226 | // reader's own result text: a complete window ends at its last line, and the |
| 227 | // empty-file / past-EOF markers state the count directly. |
| 228 | func readSourceEnd(output string, window tool.ReadWindow, hasWindow bool) (int, bool) { |
| 229 | if hasWindow { |
| 230 | return window.Range().End, true |
| 231 | } |
| 232 | trimmed := strings.TrimSpace(output) |
| 233 | if trimmed == readFileEmptyOutput { |
| 234 | return 0, true |
| 235 | } |
| 236 | if rest, ok := strings.CutPrefix(trimmed, "(offset "); ok { |
| 237 | if _, tail, found := strings.Cut(rest, readFilePastEOFTail); found { |
| 238 | if n, err := strconv.Atoi(strings.TrimSuffix(strings.TrimSpace(tail), " lines)")); err == nil && n >= 0 { |
| 239 | return n, true |
| 240 | } |
| 241 | } |
| 242 | } |
| 243 | return 0, false |
| 244 | } |
| 245 | |
| 246 | // SnipHint front-loads file content: the most relevant lines are near the top, |
| 247 | // so keep a generous head and a short tail when an old read is shortened. |
| 248 | func (readFile) SnipHint() tool.SnipHint { |
| 249 | return tool.SnipHint{Head: 120, Tail: 12, HeadChars: 12000, TailChars: 2000} |
| 250 | } |
| 251 | |
| 252 | func (r readFile) Execute(ctx context.Context, args json.RawMessage) (string, error) { |
| 253 | output, _, err := r.ExecuteRead(ctx, args) |
| 254 | return output, err |
| 255 | } |
| 256 | |
| 257 | func (r readFile) scanEncoded(f io.Reader, offset, limit int) (string, error) { |
| 258 | |
| 259 | // Peek the first 8 KiB to reject binary files cheaply (a NUL byte) before |
| 260 | // reading further — keeps a multi-GB archive from being slurped just to be |
| 261 | // discarded. |
| 262 | peek := make([]byte, readFileBinaryPeek) |
| 263 | pn, perr := io.ReadFull(f, peek) |
| 264 | peek = peek[:pn] |
| 265 | peekEOF := perr != nil // whole file fit in the peek (EOF / ErrUnexpectedEOF) |
| 266 | |
| 267 | // BOM check first: UTF-16 files contain 0x00 for every ASCII character, so a |
| 268 | // naive NUL check would misidentify them as binary. |
| 269 | switch fileenc.DetectQuick(peek) { |
| 270 | case fileenc.UTF16LE, fileenc.UTF16BE: |
| 271 | enc := fileenc.DetectQuick(peek) |
| 272 | return r.scan(transform.NewReader(io.MultiReader(bytes.NewReader(peek), f), fileenc.Decoder(enc)), offset, limit) |
| 273 | case fileenc.UTF8BOM: |
| 274 | // Strip the 3-byte BOM; the content is valid UTF-8 and streams directly. |
| 275 | body := peek |
| 276 | if len(body) >= 3 { |
| 277 | body = body[3:] |
| 278 | } |
| 279 | return r.scan(io.MultiReader(bytes.NewReader(body), f), offset, limit) |
| 280 | } |
| 281 | |
| 282 | // BOM-less UTF-16 (Windows source files) has a NUL for every ASCII char but |
| 283 | // no BOM, so it reaches here; recognise it by its NUL pattern and decode it |
| 284 | // rather than rejecting it as binary. |
| 285 | if k, ok := fileenc.DetectUTF16NoBOM(peek); ok { |
| 286 | return r.scan(transform.NewReader(io.MultiReader(bytes.NewReader(peek), f), fileenc.Decoder(k)), offset, limit) |
| 287 | } |
| 288 | |
| 289 | if bytes.IndexByte(peek, 0) >= 0 { |
| 290 | return "", fmt.Errorf("binary file (NUL byte detected); use a binary inspection tool") |
| 291 | } |
| 292 | |
| 293 | // Read up to a bounded sample for encoding detection, then stream the rest — |
| 294 | // so a large text file isn't slurped whole just to return a few lines. |
| 295 | head := peek |
| 296 | if !peekEOF { |
| 297 | more := make([]byte, readFileDetectSample-len(peek)) |
| 298 | mn, merr := io.ReadFull(f, more) |
| 299 | head = append(peek, more[:mn]...) |
| 300 | peekEOF = merr != nil |
| 301 | } |
| 302 | |
| 303 | // Detect from a char-safe slice: when more file follows, trim to the last |
| 304 | // newline so the sample never ends mid multi-byte sequence (UTF-8 and GB18030 |
| 305 | // are ASCII-transparent, so '\n' is always a clean boundary). |
| 306 | sample := head |
| 307 | if !peekEOF { |
| 308 | if i := bytes.LastIndexByte(head, '\n'); i >= 0 { |
| 309 | sample = head[:i+1] |
| 310 | } |
| 311 | } |
| 312 | enc, _ := fileenc.Detect(sample) |
| 313 | |
| 314 | src := io.MultiReader(bytes.NewReader(head), f) |
| 315 | if dec := fileenc.Decoder(enc); dec != nil { |
| 316 | return r.scan(transform.NewReader(src, dec), offset, limit) |
| 317 | } |
| 318 | return r.scan(src, offset, limit) |
| 319 | } |
| 320 | |
| 321 | // scan reads lines from src and returns the formatted output with line numbers. |
| 322 | func (r readFile) scan(src io.Reader, offset, limit int) (string, error) { |
| 323 | scanner := bufio.NewScanner(src) |
| 324 | scanner.Buffer(make([]byte, 0, 64*1024), readFileMaxLineBytes) |
| 325 | |
| 326 | var collected []string |
| 327 | textBytes := 0 |
| 328 | lineNo := 0 |
| 329 | hasMore := false |
| 330 | safetyPaged := false |
| 331 | requestedEnd := offset + limit |
| 332 | for scanner.Scan() { |
| 333 | lineNo++ |
| 334 | if lineNo <= offset { |
| 335 | continue |
| 336 | } |
| 337 | if len(collected) < limit { |
| 338 | line := scanner.Text() |
| 339 | count := len(collected) + 1 |
| 340 | width := len(strconv.Itoa(offset + count)) |
| 341 | nextOffset := offset + count |
| 342 | bodyBytes := textBytes + len(line) + count*(width+len("→")+1) |
| 343 | trailer := readFileSafetyTrailer(nextOffset, requestedEnd) |
| 344 | if bodyBytes+len(trailer) > readFileMaxFormattedBytes { |
| 345 | hasMore = true |
| 346 | safetyPaged = true |
| 347 | break |
| 348 | } |
| 349 | collected = append(collected, line) |
| 350 | textBytes += len(line) |
| 351 | continue |
| 352 | } |
| 353 | // A line past the requested window exists — stop here rather than reading |
| 354 | // the rest of the file just to count the remainder. |
| 355 | hasMore = true |
| 356 | break |
| 357 | } |
| 358 | if err := scanner.Err(); err != nil { |
| 359 | if strings.Contains(err.Error(), "token too long") { |
| 360 | return "", fmt.Errorf("scan: source line exceeds the 1 MiB local safety limit: %w", err) |
| 361 | } |
| 362 | return "", fmt.Errorf("scan: %w", err) |
| 363 | } |
| 364 | |
| 365 | if lineNo == 0 { |
| 366 | return readFileEmptyOutput, nil |
| 367 | } |
| 368 | if len(collected) == 0 { |
| 369 | return fmt.Sprintf("(offset %d%s%d lines)", offset, readFilePastEOFTail, lineNo), nil |
| 370 | } |
| 371 | |
| 372 | maxShown := offset + len(collected) |
| 373 | w := len(fmt.Sprint(maxShown)) |
| 374 | |
| 375 | var b strings.Builder |
| 376 | for i, line := range collected { |
| 377 | fmt.Fprintf(&b, "%*d→%s\n", w, offset+i+1, line) |
| 378 | } |
| 379 | if safetyPaged { |
| 380 | b.WriteString(readFileSafetyTrailer(offset+len(collected), requestedEnd)) |
| 381 | } else if hasMore { |
| 382 | fmt.Fprintf(&b, "\n[PARTIAL view: showing lines %d-%d of at least %d; pass offset=%d to continue. A partial window may be sufficient for local work.]\n", offset+1, maxShown, lineNo, maxShown) |
| 383 | } |
| 384 | return b.String(), nil |
| 385 | } |
| 386 | |
| 387 | func readFileSafetyTrailer(nextOffset, requestedEnd int) string { |
| 388 | return fmt.Sprintf("\n[read_file local safety page; next_offset=%d requested_end=%d]\n", nextOffset, requestedEnd) |
| 389 | } |
| 390 |