Ling0925
/
suna
mirror of https://github.com/kortix-ai/suna.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Handle URL in see_image tool

This commit is contained in:
Krishav Raj Singh 2025-06-29 22:44:07 +05:30
parent 4b99c23c18
commit 577872974c
2 changed files with 102 additions and 42 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

134
backend/agent/tools/sb_vision_tool.py
View File

@ -4,11 +4,12 @@ import mimetypes
from typing import Optional, Tuple
from io import BytesIO
from PIL import Image
from urllib.parse import urlparse
from agentpress.tool import ToolResult, openapi_schema, xml_schema
from sandbox.tool_base import SandboxToolsBase
from agentpress.thread_manager import ThreadManager
import json
import requests
# Add common image MIME types if mimetypes module is limited
mimetypes.add_type("image/webp", ".webp")
@ -100,17 +101,55 @@ class SandboxVisionTool(SandboxToolsBase):
print(f"[SeeImage] Failed to compress image: {str(e)}. Using original.")
return image_bytes, mime_type
def is_url(self, file_path: str) -> bool:
"""check if the file path is url"""
parsed_url = urlparse(file_path)
return parsed_url.scheme in ('http', 'https')
def download_image_from_url(self, url: str) -> Tuple[bytes, str]:
"""Download image from a URL"""
try:
headers = {
"User-Agent": "Mozilla/5.0" # Some servers block default Python
}
# HEAD request to get the image size
head_response = requests.head(url, timeout=10, headers=headers, stream=True)
head_response.raise_for_status()
# Check content length
content_length = int(head_response.headers.get('Content-Length'))
if content_length and content_length > MAX_IMAGE_SIZE:
raise Exception(f"Image is too large ({(content_length)/(1024*1024):.2f}MB) for the maximum allowed size of {MAX_IMAGE_SIZE/(1024*1024):.2f}MB")
# Download the image
response = requests.get(url, timeout=10, headers=headers, stream=True)
response.raise_for_status()
image_bytes = response.content
if len(image_bytes) > MAX_IMAGE_SIZE:
raise Exception(f"Downloaded image is too large ({(len(image_bytes))/(1024*1024):.2f}MB). Maximum allowed size of {MAX_IMAGE_SIZE/(1024*1024):.2f}MB")
# Get MIME type
mime_type = response.headers.get('Content-Type')
if not mime_type or not mime_type.startswith('image/'):
raise Exception(f"URL does not point to an image (Content-Type: {mime_type}): {url}")
return image_bytes, mime_type
except Exception as e:
return self.fail_response(f"Failed to download image from URL: {str(e)}")
@openapi_schema({
"type": "function",
"function": {
"name": "see_image",
"description": "Allows the agent to 'see' an image file located in the /workspace directory. Provide the relative path to the image. The image will be compressed before sending to reduce token usage. The image content will be made available in the next turn's context.",
"description": "Allows the agent to 'see' an image file located in the /workspace directory or from a URL. Provide either a relative path to a local image or the URL to an image. The image will be compressed before sending to reduce token usage. The image content will be made available in the next turn's context.",
"parameters": {
"type": "object",
"properties": {
"file_path": {
"type": "string",
"description": "The relative path to the image file within the /workspace directory (e.g., 'screenshots/image.png'). Supported formats: JPG, PNG, GIF, WEBP. Max size: 10MB."
"description": "Either a relative path to the image file within the /workspace directory (e.g., 'screenshots/image.png') or a URL to an image (e.g., 'https://example.com/image.jpg'). Supported formats: JPG, PNG, GIF, WEBP. Max size: 10MB."
}
},
"required": ["file_path"]
@ -123,53 +162,72 @@ class SandboxVisionTool(SandboxToolsBase):
{"param_name": "file_path", "node_type": "attribute", "path": "."}
],
example='''
<!-- Example: Request to see an image named 'diagram.png' inside the 'docs' folder -->
<!-- Example: Request to see a local image named 'diagram.png' inside the 'docs' folder -->
<function_calls>
<invoke name="see_image">
<parameter name="file_path">docs/diagram.png</parameter>
</invoke>
</function_calls>
<!-- Example: Request to see an image from a URL -->
<function_calls>
<invoke name="see_image">
<parameter name="file_path">https://example.com/image.jpg</parameter>
</invoke>
</function_calls>
'''
)
async def see_image(self, file_path: str) -> ToolResult:
"""Reads an image file, compresses it, converts it to base64, and adds it as a temporary message."""
"""Reads an image file from local file system or from a URL, compresses it, converts it to base64, and adds it as a temporary message."""
try:
# Ensure sandbox is initialized
await self._ensure_sandbox()
is_url = self.is_url(file_path)
if is_url:
try:
image_bytes, mime_type = self.download_image_from_url(file_path)
original_size = len(image_bytes)
cleaned_path = file_path
except Exception as e:
return self.fail_response(f"Failed to download image from URL: {str(e)}")
else:
# Ensure sandbox is initialized
await self._ensure_sandbox()
# Clean and construct full path
cleaned_path = self.clean_path(file_path)
full_path = f"{self.workspace_path}/{cleaned_path}"
# Clean and construct full path
cleaned_path = self.clean_path(file_path)
full_path = f"{self.workspace_path}/{cleaned_path}"
# Check if file exists and get info
try:
file_info = self.sandbox.fs.get_file_info(full_path)
if file_info.is_dir:
return self.fail_response(f"Path '{cleaned_path}' is a directory, not an image file.")
except Exception as e:
return self.fail_response(f"Image file not found at path: '{cleaned_path}'")
# Check if file exists and get info
try:
file_info = self.sandbox.fs.get_file_info(full_path)
if file_info.is_dir:
return self.fail_response(f"Path '{cleaned_path}' is a directory, not an image file.")
except Exception as e:
return self.fail_response(f"Image file not found at path: '{cleaned_path}'")
# Check file size
if file_info.size > MAX_IMAGE_SIZE:
return self.fail_response(f"Image file '{cleaned_path}' is too large ({file_info.size / (1024*1024):.2f}MB). Maximum size is {MAX_IMAGE_SIZE / (1024*1024)}MB.")
# Check file size
if file_info.size > MAX_IMAGE_SIZE:
return self.fail_response(f"Image file '{cleaned_path}' is too large ({file_info.size / (1024*1024):.2f}MB). Maximum size is {MAX_IMAGE_SIZE / (1024*1024)}MB.")
# Read image file content
try:
image_bytes = self.sandbox.fs.download_file(full_path)
except Exception as e:
return self.fail_response(f"Could not read image file: {cleaned_path}")
# Read image file content
try:
image_bytes = self.sandbox.fs.download_file(full_path)
except Exception as e:
return self.fail_response(f"Could not read image file: {cleaned_path}")
# Determine MIME type
mime_type, _ = mimetypes.guess_type(full_path)
if not mime_type or not mime_type.startswith('image/'):
# Basic fallback based on extension if mimetypes fails
ext = os.path.splitext(cleaned_path)[1].lower()
if ext == '.jpg' or ext == '.jpeg': mime_type = 'image/jpeg'
elif ext == '.png': mime_type = 'image/png'
elif ext == '.gif': mime_type = 'image/gif'
elif ext == '.webp': mime_type = 'image/webp'
else:
return self.fail_response(f"Unsupported or unknown image format for file: '{cleaned_path}'. Supported: JPG, PNG, GIF, WEBP.")
# Determine MIME type
mime_type, _ = mimetypes.guess_type(full_path)
if not mime_type or not mime_type.startswith('image/'):
# Basic fallback based on extension if mimetypes fails
ext = os.path.splitext(cleaned_path)[1].lower()
if ext == '.jpg' or ext == '.jpeg': mime_type = 'image/jpeg'
elif ext == '.png': mime_type = 'image/png'
elif ext == '.gif': mime_type = 'image/gif'
elif ext == '.webp': mime_type = 'image/webp'
else:
return self.fail_response(f"Unsupported or unknown image format for file: '{cleaned_path}'. Supported: JPG, PNG, GIF, WEBP.")
original_size = file_info.size
# Compress the image
compressed_bytes, compressed_mime_type = self.compress_image(image_bytes, mime_type, cleaned_path)
@ -186,7 +244,7 @@ class SandboxVisionTool(SandboxToolsBase):
"mime_type": compressed_mime_type,
"base64": base64_image,
"file_path": cleaned_path, # Include path for context
"original_size": file_info.size,
"original_size": original_size,
"compressed_size": len(compressed_bytes)
}
@ -200,7 +258,7 @@ class SandboxVisionTool(SandboxToolsBase):
)
# Inform the agent the image will be available next turn
return self.success_response(f"Successfully loaded and compressed the image '{cleaned_path}' (reduced from {file_info.size / 1024:.1f}KB to {len(compressed_bytes) / 1024:.1f}KB).")
return self.success_response(f"Successfully loaded and compressed the image '{cleaned_path}' (reduced from {original_size / 1024:.1f}KB to {len(compressed_bytes) / 1024:.1f}KB).")
except Exception as e:
return self.fail_response(f"An unexpected error occurred while trying to see the image: {str(e)}")

10
frontend/src/components/thread/tool-views/see-image-tool/_utils.ts
View File

@ -364,6 +364,12 @@ export function constructImageUrl(filePath: string, project?: { sandbox?: { sand
}
const cleanPath = filePath.replace(/^['"](.*)['"]$/, '$1');
// Check if it's a URL first, before trying to construct sandbox paths
if (cleanPath.startsWith('http')) {
return cleanPath;
}
const sandboxId = typeof project?.sandbox === 'string'
? project.sandbox
: project?.sandbox?.id;
@ -390,10 +396,6 @@ export function constructImageUrl(filePath: string, project?: { sandbox?: { sand
return fullUrl;
}
if (cleanPath.startsWith('http')) {
return cleanPath;
}
console.warn('No sandbox URL or ID available, using path as-is:', cleanPath);
return cleanPath;
}