a6298d6 · crush · git.iko.soy

git.iko.soy/crush

fix: fix fetch and view tool

- don't allow binary files - limit the fetch content

commita6298d6d1d2a07afd8737a420466ea6d02b6eabe authorKujtim Hoxha <kujtimii.h@gmail.com> dateThu 17 Jul 2025, 14:54 (1 year ago) parenta700b64

2 files changed+39−10

compare with main

unifiedsplit

internal/llm/tools/fetch.go+33−9 internal/llm/tools/view.go+6−1

internal/llm/tools/fetch.go+33−9

@@ -8,6 +8,7 @@ import (

88 "net/http"

99 "strings"

1010 "time"

11+ "unicode/utf8"

1112

1213 md "github.com/JohannesKaufmann/html-to-markdown"

1314 "github.com/PuerkitoBio/goquery"

@@ -182,6 +183,11 @@ func (t *fetchTool) Run(ctx context.Context, call ToolCall) (ToolResponse, error

182183 }

183184

184185 content := string(body)

186+

187+ isValidUt8 := utf8.ValidString(content)

188+ if !isValidUt8 {

189+ return NewTextErrorResponse("Response content is not valid UTF-8"), nil

190+ }

185191 contentType := resp.Header.Get("Content-Type")

186192

187193 switch format {

@@ -191,9 +197,8 @@ func (t *fetchTool) Run(ctx context.Context, call ToolCall) (ToolResponse, error

191197 if err != nil {

192198 return NewTextErrorResponse("Failed to extract text from HTML: " + err.Error()), nil

193199 }

194− return NewTextResponse(text), nil

200+ content = text

195201 }

196− return NewTextResponse(content), nil

197202

198203 case "markdown":

199204 if strings.Contains(contentType, "text/html") {

@@ -201,17 +206,36 @@ func (t *fetchTool) Run(ctx context.Context, call ToolCall) (ToolResponse, error

201206 if err != nil {

202207 return NewTextErrorResponse("Failed to convert HTML to Markdown: " + err.Error()), nil

203208 }

204− return NewTextResponse(markdown), nil

209+ content = markdown

205210 }

206211

207− return NewTextResponse("```\n" + content + "\n```"), nil

212+ content = "```\n" + content + "\n```"

208213

209214 case "html":

210− return NewTextResponse(content), nil

211−

212− default:

213− return NewTextResponse(content), nil

215+ // return only the body of the HTML document

216+ if strings.Contains(contentType, "text/html") {

217+ doc, err := goquery.NewDocumentFromReader(strings.NewReader(content))

218+ if err != nil {

219+ return NewTextErrorResponse("Failed to parse HTML: " + err.Error()), nil

220+ }

221+ body, err := doc.Find("body").Html()

222+ if err != nil {

223+ return NewTextErrorResponse("Failed to extract body from HTML: " + err.Error()), nil

224+ }

225+ if body == "" {

226+ return NewTextErrorResponse("No body content found in HTML"), nil

227+ }

228+ content = "<html>\n<body>\n" + body + "\n</body>\n</html>"

229+ }

230+ }

231+ // calculate byte size of content

232+ contentSize := int64(len(content))

233+ if contentSize > MaxReadSize {

234+ content = content[:MaxReadSize]

235+ content += fmt.Sprintf("\n\n[Content truncated to %d bytes]", MaxReadSize)

214236 }

237+

238+ return NewTextResponse(content), nil

215239}

216240

217241func extractTextFromHTML(html string) (string, error) {

@@ -220,7 +244,7 @@ func extractTextFromHTML(html string) (string, error) {

220244 return "", err

221245 }

222246

223− text := doc.Text()

247+ text := doc.Find("body").Text()

224248 text = strings.Join(strings.Fields(text), " ")

225249

226250 return text, nil

internal/llm/tools/view.go+6−1

@@ -9,6 +9,7 @@ import (

99 "os"

1010 "path/filepath"

1111 "strings"

12+ "unicode/utf8"

1213

1314 "github.com/charmbracelet/crush/internal/lsp"

1415)

@@ -173,11 +174,15 @@ func (v *viewTool) Run(ctx context.Context, call ToolCall) (ToolResponse, error)

173174 isImage, imageType := isImageFile(filePath)

174175 // TODO: handle images

175176 if isImage {

176− return NewTextErrorResponse(fmt.Sprintf("This is an image file of type: %s\nUse a different tool to process images", imageType)), nil

177+ return NewTextErrorResponse(fmt.Sprintf("This is an image file of type: %s\n", imageType)), nil

177178 }

178179

179180 // Read the file content

180181 content, lineCount, err := readTextFile(filePath, params.Offset, params.Limit)

182+ isValidUt8 := utf8.ValidString(content)

183+ if !isValidUt8 {

184+ return NewTextErrorResponse("File content is not valid UTF-8"), nil

185+ }

181186 if err != nil {

182187 return ToolResponse{}, fmt.Errorf("error reading file: %w", err)

183188 }