| 1 | package builtin |
| 2 | |
| 3 | import ( |
| 4 | "context" |
| 5 | "encoding/base64" |
| 6 | "encoding/json" |
| 7 | "fmt" |
| 8 | "io" |
| 9 | "os" |
| 10 | "strings" |
| 11 | |
| 12 | "reasonix/internal/attachment" |
| 13 | "reasonix/internal/tool" |
| 14 | ) |
| 15 | |
| 16 | // Bound encoded payload size consistently with MCP image results. |
| 17 | const viewImageMaxBytes = 3 << 20 |
| 18 | |
| 19 | type viewImage struct { |
| 20 | workDir string |
| 21 | paths *PathResolver |
| 22 | forbidRoots []string |
| 23 | } |
| 24 | |
| 25 | func init() { tool.RegisterBuiltin(viewImage{}) } |
| 26 | func (viewImage) Name() string { return "view_image" } |
| 27 | func (viewImage) Description() string { |
| 28 | return "Read a local PNG, JPEG, GIF, or WebP image by path and return visual content through native vision or the configured image-understanding model. Use this for image paths instead of read_file. Maximum file size: 3 MiB; maximum dimensions: 40 million pixels." |
| 29 | } |
| 30 | func (viewImage) Schema() json.RawMessage { |
| 31 | return json.RawMessage(`{"type":"object","properties":{"path":{"type":"string","description":"Image file path"}},"required":["path"]}`) |
| 32 | } |
| 33 | func (viewImage) ReadOnly() bool { return true } |
| 34 | func (v viewImage) Execute(ctx context.Context, args json.RawMessage) (string, error) { |
| 35 | text, _, err := v.ExecuteWithImages(ctx, args) |
| 36 | if err == nil { |
| 37 | text += "\nVisual content requires a structured image channel and an image-capable model." |
| 38 | } |
| 39 | return text, err |
| 40 | } |
| 41 | func (v viewImage) ExecuteWithImages(ctx context.Context, args json.RawMessage) (string, []string, error) { |
| 42 | var p struct { |
| 43 | Path string `json:"path"` |
| 44 | } |
| 45 | if err := json.Unmarshal(args, &p); err != nil { |
| 46 | return "", nil, fmt.Errorf("invalid args: %w", err) |
| 47 | } |
| 48 | if strings.TrimSpace(p.Path) == "" { |
| 49 | return "", nil, fmt.Errorf("path is required") |
| 50 | } |
| 51 | if err := ctx.Err(); err != nil { |
| 52 | return "", nil, err |
| 53 | } |
| 54 | rp := resolveReadablePath(v.workDir, p.Path, v.paths) |
| 55 | if confineRead(v.forbidRoots, rp.Path) { |
| 56 | return "", nil, fmt.Errorf("read %s: file not found", rp.DisplayPath) |
| 57 | } |
| 58 | info, err := os.Stat(rp.Path) |
| 59 | if err != nil { |
| 60 | return "", nil, fmt.Errorf("read %s: %s", rp.DisplayPath, rp.ErrorText(err)) |
| 61 | } |
| 62 | if !info.Mode().IsRegular() { |
| 63 | return "", nil, fmt.Errorf("%s is not a regular image file", rp.DisplayPath) |
| 64 | } |
| 65 | if info.Size() > viewImageMaxBytes { |
| 66 | return "", nil, fmt.Errorf("image exceeds 3 MiB limit") |
| 67 | } |
| 68 | f, err := os.Open(rp.Path) |
| 69 | if err != nil { |
| 70 | return "", nil, fmt.Errorf("read %s: %s", rp.DisplayPath, rp.ErrorText(err)) |
| 71 | } |
| 72 | defer f.Close() |
| 73 | data, err := io.ReadAll(io.LimitReader(f, viewImageMaxBytes+1)) |
| 74 | if err != nil { |
| 75 | return "", nil, fmt.Errorf("read %s: %s", rp.DisplayPath, rp.ErrorText(err)) |
| 76 | } |
| 77 | if len(data) > viewImageMaxBytes { |
| 78 | return "", nil, fmt.Errorf("image exceeds 3 MiB limit") |
| 79 | } |
| 80 | if err := ctx.Err(); err != nil { |
| 81 | return "", nil, err |
| 82 | } |
| 83 | mime, width, height, err := attachment.ValidateImage(data, "", attachment.ViewImagePolicy()) |
| 84 | if err != nil { |
| 85 | return "", nil, fmt.Errorf("invalid or unsupported image: %w", err) |
| 86 | } |
| 87 | return fmt.Sprintf("[image: %s, %dx%d] %s", mime, width, height, rp.DisplayPath), []string{"data:" + mime + ";base64," + base64.StdEncoding.EncodeToString(data)}, nil |
| 88 | } |
| 89 |