| 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 | "encoding/json" |
| 10 | "fmt" |
| 11 | "io" |
| 12 | "os" |
| 13 | "path/filepath" |
| 14 | "strings" |
| 15 | |
| 16 | "golang.org/x/text/transform" |
| 17 | |
| 18 | fileenc "reasonix/internal/fileutil/encoding" |
| 19 | "reasonix/internal/tool" |
| 20 | ) |
| 21 | |
| 22 | const ( |
| 23 | readFileBinaryPeek = 8 * 1024 // bytes scanned for NUL before reading further |
| 24 | readFileDetectSample = 256 * 1024 // bytes sampled for encoding detection before streaming |
| 25 | ) |
| 26 | |
| 27 | func init() { tool.RegisterBuiltin(readFile{}) } |
| 28 | |
| 29 | // readFile reads a text file. workDir, when non-empty, is the directory a |
| 30 | // relative path is resolved against (see resolveIn). paths maps session-scoped |
| 31 | // external read aliases to local roots without changing the model-visible tool |
| 32 | // schema. forbidRoots lists directories the tool may not read from (resolved, |
| 33 | // absolute paths). |
| 34 | type readFile struct { |
| 35 | workDir string |
| 36 | paths *PathResolver |
| 37 | forbidRoots []string |
| 38 | // overlay, when non-nil, serves content from the host transport (unsaved |
| 39 | // editor buffers) before falling back to disk. Consulted only after path |
| 40 | // resolution and read confinement, and never for external alias paths. |
| 41 | overlay FileOverlay |
| 42 | } |
| 43 | |
| 44 | const ( |
| 45 | readFileDefaultLimit = 2000 // lines returned when limit is unset |
| 46 | ) |
| 47 | |
| 48 | func (readFile) Name() string { return "read_file" } |
| 49 | |
| 50 | func (readFile) Description() string { |
| 51 | return "Read a text file with optional line offset/limit. Output prefixes each line with its 1-based number (e.g. ` 42→...`) so subsequent edit_file calls can target exact lines. Use `offset` and `limit` to page through large files; the tool reports total length and pagination hints in a trailer." |
| 52 | } |
| 53 | |
| 54 | func (readFile) Schema() json.RawMessage { |
| 55 | return json.RawMessage(`{ |
| 56 | "type":"object", |
| 57 | "properties":{ |
| 58 | "path":{"type":"string","description":"File path"}, |
| 59 | "offset":{"type":"integer","description":"0-based line offset to start reading from (default 0)","minimum":0}, |
| 60 | "limit":{"type":"integer","description":"Maximum lines to return (default 2000)","minimum":1} |
| 61 | }, |
| 62 | "required":["path"] |
| 63 | }`) |
| 64 | } |
| 65 | |
| 66 | func (readFile) ReadOnly() bool { return true } |
| 67 | |
| 68 | // SnipHint front-loads file content: the most relevant lines are near the top, |
| 69 | // so keep a generous head and a short tail when an old read is shortened. |
| 70 | func (readFile) SnipHint() tool.SnipHint { |
| 71 | return tool.SnipHint{Head: 120, Tail: 12, HeadChars: 12000, TailChars: 2000} |
| 72 | } |
| 73 | |
| 74 | func (r readFile) Execute(ctx context.Context, args json.RawMessage) (string, error) { |
| 75 | var p struct { |
| 76 | Path string `json:"path"` |
| 77 | Offset int `json:"offset,omitempty"` |
| 78 | Limit int `json:"limit,omitempty"` |
| 79 | } |
| 80 | if err := json.Unmarshal(args, &p); err != nil { |
| 81 | return "", fmt.Errorf("invalid args: %w", err) |
| 82 | } |
| 83 | if p.Path == "" { |
| 84 | return "", fmt.Errorf("path is required") |
| 85 | } |
| 86 | rp := resolveReadablePath(r.workDir, p.Path, r.paths) |
| 87 | p.Path = rp.Path |
| 88 | displayPath := rp.DisplayPath |
| 89 | if confineRead(r.forbidRoots, p.Path) { |
| 90 | err := &os.PathError{Op: "open", Path: p.Path, Err: os.ErrNotExist} |
| 91 | if rp.External { |
| 92 | return "", fmt.Errorf("read %s: %s", displayPath, rp.ErrorText(err)) |
| 93 | } |
| 94 | return "", err |
| 95 | } |
| 96 | if p.Offset < 0 { |
| 97 | p.Offset = 0 |
| 98 | } |
| 99 | if p.Limit <= 0 { |
| 100 | p.Limit = readFileDefaultLimit |
| 101 | } |
| 102 | |
| 103 | // The host overlay (unsaved editor buffers) wins over the disk when it can |
| 104 | // serve the path. Content arrives already decoded as text, so the encoding |
| 105 | // and binary-detection pipeline below applies to the disk fallback only. |
| 106 | if r.overlay != nil && !rp.External && filepath.IsAbs(p.Path) { |
| 107 | if content, ok := r.overlay.ReadTextFile(ctx, p.Path); ok { |
| 108 | return r.scan(strings.NewReader(content), p.Offset, p.Limit) |
| 109 | } |
| 110 | } |
| 111 | |
| 112 | // A directory can be os.Open'd but not read as text — catch it up front with |
| 113 | // an actionable message (and avoid the doubled "read X: read X:" the scanner's |
| 114 | // error would otherwise produce) so the model switches to the ls tool. |
| 115 | if info, err := os.Stat(p.Path); err == nil && info.IsDir() { |
| 116 | return "", fmt.Errorf("%s is a directory, not a file — use the ls tool to list it, or read a specific file inside it", displayPath) |
| 117 | } |
| 118 | |
| 119 | f, err := os.Open(p.Path) |
| 120 | if err != nil { |
| 121 | if rp.External { |
| 122 | return "", fmt.Errorf("read %s: %s", displayPath, rp.ErrorText(err)) |
| 123 | } |
| 124 | return "", fmt.Errorf("read %s: %w", displayPath, err) |
| 125 | } |
| 126 | defer f.Close() |
| 127 | |
| 128 | // Peek the first 8 KiB to reject binary files cheaply (a NUL byte) before |
| 129 | // reading further — keeps a multi-GB archive from being slurped just to be |
| 130 | // discarded. |
| 131 | peek := make([]byte, readFileBinaryPeek) |
| 132 | pn, perr := io.ReadFull(f, peek) |
| 133 | peek = peek[:pn] |
| 134 | peekEOF := perr != nil // whole file fit in the peek (EOF / ErrUnexpectedEOF) |
| 135 | |
| 136 | // BOM check first: UTF-16 files contain 0x00 for every ASCII character, so a |
| 137 | // naive NUL check would misidentify them as binary. |
| 138 | switch fileenc.DetectQuick(peek) { |
| 139 | case fileenc.UTF16LE, fileenc.UTF16BE: |
| 140 | // UTF-16 is not self-synchronising and can't be streamed line-by-line, so |
| 141 | // buffer it fully (these files are rare and usually small). |
| 142 | rest, rerr := io.ReadAll(f) |
| 143 | if rerr != nil { |
| 144 | if rp.External { |
| 145 | return "", fmt.Errorf("read %s: %s", displayPath, rp.ErrorText(rerr)) |
| 146 | } |
| 147 | return "", fmt.Errorf("read %s: %w", displayPath, rerr) |
| 148 | } |
| 149 | all := append(peek, rest...) |
| 150 | bom := fileenc.DetectQuick(all) |
| 151 | return r.scan(bytes.NewReader(fileenc.Decode(all, bom)), p.Offset, p.Limit) |
| 152 | case fileenc.UTF8BOM: |
| 153 | // Strip the 3-byte BOM; the content is valid UTF-8 and streams directly. |
| 154 | body := peek |
| 155 | if len(body) >= 3 { |
| 156 | body = body[3:] |
| 157 | } |
| 158 | return r.scan(io.MultiReader(bytes.NewReader(body), f), p.Offset, p.Limit) |
| 159 | } |
| 160 | |
| 161 | // BOM-less UTF-16 (Windows source files) has a NUL for every ASCII char but |
| 162 | // no BOM, so it reaches here; recognise it by its NUL pattern and decode it |
| 163 | // rather than rejecting it as binary. |
| 164 | if k, ok := fileenc.DetectUTF16NoBOM(peek); ok { |
| 165 | rest, rerr := io.ReadAll(f) |
| 166 | if rerr != nil { |
| 167 | if rp.External { |
| 168 | return "", fmt.Errorf("read %s: %s", displayPath, rp.ErrorText(rerr)) |
| 169 | } |
| 170 | return "", fmt.Errorf("read %s: %w", displayPath, rerr) |
| 171 | } |
| 172 | all := append(peek, rest...) |
| 173 | return r.scan(bytes.NewReader(fileenc.Decode(all, k)), p.Offset, p.Limit) |
| 174 | } |
| 175 | |
| 176 | if bytes.IndexByte(peek, 0) >= 0 { |
| 177 | if rp.External { |
| 178 | return "", fmt.Errorf("binary file %s (NUL byte detected); not shown by read_file", displayPath) |
| 179 | } |
| 180 | return "", fmt.Errorf("binary file %s (NUL byte detected); use `bash hexdump` or another tool", displayPath) |
| 181 | } |
| 182 | |
| 183 | // Read up to a bounded sample for encoding detection, then stream the rest — |
| 184 | // so a large text file isn't slurped whole just to return a few lines. |
| 185 | head := peek |
| 186 | if !peekEOF { |
| 187 | more := make([]byte, readFileDetectSample-len(peek)) |
| 188 | mn, merr := io.ReadFull(f, more) |
| 189 | head = append(peek, more[:mn]...) |
| 190 | peekEOF = merr != nil |
| 191 | } |
| 192 | |
| 193 | // Detect from a char-safe slice: when more file follows, trim to the last |
| 194 | // newline so the sample never ends mid multi-byte sequence (UTF-8 and GB18030 |
| 195 | // are ASCII-transparent, so '\n' is always a clean boundary). |
| 196 | sample := head |
| 197 | if !peekEOF { |
| 198 | if i := bytes.LastIndexByte(head, '\n'); i >= 0 { |
| 199 | sample = head[:i+1] |
| 200 | } |
| 201 | } |
| 202 | enc, _ := fileenc.Detect(sample) |
| 203 | |
| 204 | src := io.MultiReader(bytes.NewReader(head), f) |
| 205 | if dec := fileenc.Decoder(enc); dec != nil { |
| 206 | return r.scan(transform.NewReader(src, dec), p.Offset, p.Limit) |
| 207 | } |
| 208 | return r.scan(src, p.Offset, p.Limit) |
| 209 | } |
| 210 | |
| 211 | // scan reads lines from src and returns the formatted output with line numbers. |
| 212 | func (r readFile) scan(src io.Reader, offset, limit int) (string, error) { |
| 213 | scanner := bufio.NewScanner(src) |
| 214 | scanner.Buffer(make([]byte, 0, 64*1024), 1024*1024) |
| 215 | |
| 216 | var collected []string |
| 217 | lineNo := 0 |
| 218 | hasMore := false |
| 219 | for scanner.Scan() { |
| 220 | lineNo++ |
| 221 | if lineNo <= offset { |
| 222 | continue |
| 223 | } |
| 224 | if len(collected) < limit { |
| 225 | collected = append(collected, scanner.Text()) |
| 226 | continue |
| 227 | } |
| 228 | // A line past the requested window exists — stop here rather than reading |
| 229 | // the rest of the file just to count the remainder. |
| 230 | hasMore = true |
| 231 | break |
| 232 | } |
| 233 | if err := scanner.Err(); err != nil { |
| 234 | return "", fmt.Errorf("scan: %w", err) |
| 235 | } |
| 236 | |
| 237 | if lineNo == 0 { |
| 238 | return "(empty file)", nil |
| 239 | } |
| 240 | if len(collected) == 0 { |
| 241 | return fmt.Sprintf("(offset %d is past EOF — file has %d lines)", offset, lineNo), nil |
| 242 | } |
| 243 | |
| 244 | maxShown := offset + len(collected) |
| 245 | w := len(fmt.Sprint(maxShown)) |
| 246 | |
| 247 | var b strings.Builder |
| 248 | for i, line := range collected { |
| 249 | fmt.Fprintf(&b, "%*d→%s\n", w, offset+i+1, line) |
| 250 | } |
| 251 | if hasMore { |
| 252 | fmt.Fprintf(&b, "\n[more lines below; pass offset=%d to continue]\n", offset+len(collected)) |
| 253 | } |
| 254 | return b.String(), nil |
| 255 | } |
| 256 |