vibesurf 0.1.35__py3-none-any.whl → 0.1.37__py3-none-any.whl

vibe_surf/tools/browser_use_tools.py

@@ -35,6 +35,7 @@ from browser_use.tools.views import (
     StructuredOutputAction,
     SwitchTabAction,
     UploadFileAction,
+    NavigateAction
 )
 from browser_use.llm.base import BaseChatModel
 from browser_use.llm.messages import UserMessage, ContentPartTextParam, ContentPartImageParam, ImageURL
@@ -72,7 +73,7 @@ class BrowserUseTools(Tools, VibeSurfTools):
             self.display_files_in_done_text = display_files_in_done_text
 
             @self.registry.action(
-                'Complete task - with return text and if the task is finished (success=True) or not yet completely finished (success=False), because last step is reached',
+                'Complete task with structured output.',
                 param_model=StructuredOutputAction[output_model],
             )
             async def done(params: StructuredOutputAction):
@@ -94,7 +95,7 @@ class BrowserUseTools(Tools, VibeSurfTools):
         else:
 
             @self.registry.action(
-                'Complete task - provide a summary of results for the user. Set success=True if task completed successfully, false otherwise. Text should be your response to the user summarizing results. Include files in files_to_display if you would like to display to the user or there files are important for the task result.',
+                'Complete task.',
                 param_model=DoneAction,
             )
             async def done(params: DoneAction, file_system: CustomFileSystem):
@@ -143,8 +144,8 @@ class BrowserUseTools(Tools, VibeSurfTools):
     def _register_browser_actions(self):
         """Register custom browser actions"""
 
-        @self.registry.action('Upload file to interactive element with file path', param_model=UploadFileAction)
-        async def upload_file_to_element(
+        @self.registry.action('', param_model=UploadFileAction)
+        async def upload_file(
                 params: UploadFileAction, browser_session: BrowserSession, file_system: FileSystem
         ):
 
@@ -250,7 +251,8 @@ class BrowserUseTools(Tools, VibeSurfTools):
 
             # Dispatch upload file event with the file input node
             try:
-                event = browser_session.event_bus.dispatch(UploadFileEvent(node=file_input_node, file_path=full_file_path))
+                event = browser_session.event_bus.dispatch(
+                    UploadFileEvent(node=file_input_node, file_path=full_file_path))
                 await event
                 await event.event_result(raise_if_any=True, raise_if_none=False)
                 msg = f'Successfully uploaded file to index {params.index}'
@@ -264,10 +266,10 @@ class BrowserUseTools(Tools, VibeSurfTools):
                 raise BrowserError(f'Failed to upload file: {e}')
 
         @self.registry.action(
-            'Hover over an element',
+            '',
             param_model=HoverAction,
         )
-        async def hover_element(params: HoverAction, browser_session: AgentBrowserSession):
+        async def hover(params: HoverAction, browser_session: AgentBrowserSession):
             """Hovers over the element specified by its index from the cached selector map or by XPath."""
             try:
                 if params.xpath:
@@ -370,7 +372,7 @@ class BrowserUseTools(Tools, VibeSurfTools):
         # =======================
 
         @self.registry.action(
-            'Search the query using the specified search engine. Defaults to DuckDuckGo (recommended) to avoid reCAPTCHA. Options: duckduckgo, google, bing. Query should be concrete and not vague or super long.',
+            '',
             param_model=SearchAction,
         )
         async def search(params: SearchAction, browser_session: AgentBrowserSession):
@@ -386,11 +388,11 @@ class BrowserUseTools(Tools, VibeSurfTools):
                 'bing': f'https://www.bing.com/search?q={encoded_query}',
             }
 
-            if params.search_engine.lower() not in search_engines:
+            if params.engine.lower() not in search_engines:
                 return ActionResult(
-                    error=f'Unsupported search engine: {params.search_engine}. Options: duckduckgo, google, bing')
+                    error=f'Unsupported search engine: {params.engine}. Options: duckduckgo, google, bing')
 
-            search_url = search_engines[params.search_engine.lower()]
+            search_url = search_engines[params.engine.lower()]
 
             try:
                 # Use AgentBrowserSession's direct navigation method
@@ -404,10 +406,10 @@ class BrowserUseTools(Tools, VibeSurfTools):
                 return ActionResult(error=f'Failed to search Google for "{params.query}": {str(e)}')
 
         @self.registry.action(
-            'Navigate to URL, set new_tab=True to open in new tab, False to navigate in current tab',
-            param_model=GoToUrlAction
+            '',
+            param_model=NavigateAction
         )
-        async def go_to_url(params: GoToUrlAction, browser_session: AgentBrowserSession):
+        async def navigate(params: NavigateAction, browser_session: AgentBrowserSession):
             try:
                 # Use AgentBrowserSession's direct navigation method
                 await browser_session.navigate_to_url(params.url, new_tab=params.new_tab)
@@ -426,9 +428,10 @@ class BrowserUseTools(Tools, VibeSurfTools):
                 return ActionResult(error=f'Navigation failed: {str(e)}')
 
         @self.registry.action(
-            'Go back',
+            '',
+            param_model=NoParamsAction
         )
-        async def go_back(browser_session: AgentBrowserSession):
+        async def go_back(_: NoParamsAction, browser_session: AgentBrowserSession):
             try:
                 cdp_session = await browser_session.get_or_create_cdp_session()
                 history = await cdp_session.cdp_client.send.Page.getNavigationHistory(session_id=cdp_session.session_id)
@@ -458,18 +461,12 @@ class BrowserUseTools(Tools, VibeSurfTools):
                 return ActionResult(error=f'Failed to go back: {str(e)}')
 
         @self.registry.action(
-            'Switch tab',
+            '',
             param_model=SwitchTabAction
         )
-        async def switch_tab(params: SwitchTabAction, browser_session: AgentBrowserSession):
+        async def switch(params: SwitchTabAction, browser_session: AgentBrowserSession):
             try:
-
-                if params.tab_id:
-                    target_id = await browser_session.get_target_id_from_tab_id(params.tab_id)
-                elif params.url:
-                    target_id = await browser_session.get_target_id_from_url(params.url)
-                else:
-                    target_id = await browser_session.get_most_recently_opened_target_id()
+                target_id = await browser_session.get_target_id_from_tab_id(params.tab_id)
 
                 # Switch to target using CDP
                 await browser_session.get_or_create_cdp_session(target_id, focus=True)
@@ -488,7 +485,7 @@ class BrowserUseTools(Tools, VibeSurfTools):
         async def take_screenshot(_: NoParamsAction, browser_session: AgentBrowserSession, file_system: FileSystem):
             try:
                 # Take screenshot using browser session
-                screenshot = await browser_session.take_screenshot()
+                screenshot_bytes = await browser_session.take_screenshot()
 
                 # Generate timestamp for filename
                 timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
@@ -507,7 +504,7 @@ class BrowserUseTools(Tools, VibeSurfTools):
                 filepath = screenshots_dir / filename
 
                 with open(filepath, "wb") as f:
-                    f.write(base64.b64decode(screenshot))
+                    f.write(screenshot_bytes)
 
                 msg = f'📸 Screenshot saved to path: {str(filepath.relative_to(fs_dir))}'
                 logger.info(msg)
@@ -530,23 +527,23 @@ class BrowserUseTools(Tools, VibeSurfTools):
             try:
                 # Get file system directory path (Path type)
                 fs_dir = file_system.get_dir()
-                
+
                 # Create downloads directory if it doesn't exist
                 downloads_dir = fs_dir / "downloads"
                 downloads_dir.mkdir(exist_ok=True)
-                
+
                 # Download the file and detect format
                 async with aiohttp.ClientSession() as session:
                     async with session.get(params.url) as response:
                         if response.status != 200:
                             raise Exception(f"HTTP {response.status}: Failed to download from {params.url}")
-                        
+
                         # Get content
                         content = await response.read()
-                        
+                        headers_dict = dict(response.headers)
                         # Detect file format and extension
-                        file_extension = await self._detect_file_format(params.url, response.headers, content)
-                        
+                        file_extension = await self._detect_file_format(params.url, headers_dict, content)
+
                         # Generate filename
                         if params.filename:
                             # Use provided filename, add extension if missing
@@ -557,7 +554,7 @@ class BrowserUseTools(Tools, VibeSurfTools):
                             # Generate filename from URL or timestamp
                             url_path = urllib.parse.urlparse(params.url).path
                             url_filename = os.path.basename(url_path)
-                            
+
                             if url_filename and not url_filename.startswith('.'):
                                 # Use URL filename, ensure correct extension
                                 filename = url_filename
@@ -568,19 +565,19 @@ class BrowserUseTools(Tools, VibeSurfTools):
                                 # Generate timestamp-based filename
                                 timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
                                 filename = f"media_{timestamp}{file_extension}"
-                        
+
                         # Sanitize filename
                         filename = sanitize_filename(filename)
                         filepath = downloads_dir / filename
-                        
+
                         # Save file
                         with open(filepath, "wb") as f:
                             f.write(content)
-                        
+
                         # Calculate file size for display
                         file_size = len(content)
                         size_str = self._format_file_size(file_size)
-                        
+
                         msg = f'📥 Downloaded media to: {str(filepath.relative_to(fs_dir))} ({size_str})'
                         logger.info(msg)
                         return ActionResult(
@@ -588,7 +585,7 @@ class BrowserUseTools(Tools, VibeSurfTools):
                             include_in_memory=True,
                             long_term_memory=f'Downloaded media from {params.url} to {str(filepath.relative_to(fs_dir))}',
                         )
-                        
+
             except Exception as e:
                 error_msg = f'❌ Failed to download media: {str(e)}'
                 logger.error(error_msg)
@@ -666,9 +663,9 @@ class BrowserUseTools(Tools, VibeSurfTools):
         if url_path:
             ext = os.path.splitext(url_path)[1].lower()
             if ext in ['.jpg', '.jpeg', '.png', '.gif', '.webp', '.svg', '.bmp', '.tiff',
-                      '.mp4', '.webm', '.avi', '.mov', '.wmv', '.flv',
-                      '.mp3', '.wav', '.ogg', '.aac', '.flac',
-                      '.pdf', '.doc', '.docx', '.txt']:
+                       '.mp4', '.webm', '.avi', '.mov', '.wmv', '.flv',
+                       '.mp3', '.wav', '.ogg', '.aac', '.flac',
+                       '.pdf', '.doc', '.docx', '.txt']:
                 return ext
 
         # Default fallback

vibe_surf/tools/file_system.py

@@ -6,7 +6,9 @@ from pathlib import Path
 from browser_use.filesystem.file_system import FileSystem, FileSystemError, INVALID_FILENAME_ERROR_MESSAGE, \
     FileSystemState
 from browser_use.filesystem.file_system import BaseFile, MarkdownFile, TxtFile, JsonFile, CsvFile, PdfFile
+from vibe_surf.logger import get_logger
 
+logger = get_logger(__name__)
 
 class PythonFile(BaseFile):
     """Plain text file implementation"""
@@ -315,9 +317,10 @@ class CustomFileSystem(FileSystem):
         """Save extracted content to a numbered file"""
         initial_filename = f'extracted_content_{self.extracted_content_count}'
         extracted_filename = f'{initial_filename}.md'
-        await self.write_file(initial_filename, content)
+        write_result = await self.write_file(extracted_filename, content)
+        logger.info(write_result)
         self.extracted_content_count += 1
-        return f'Extracted content saved to file {extracted_filename} successfully.'
+        return extracted_filename
 
     async def list_directory(self, directory_path: str = "") -> str:
         """List contents of a directory within the file system (data_dir only)"""

vibe_surf/tools/utils.py

@@ -0,0 +1,118 @@
+from bs4 import BeautifulSoup
+from browser_use.dom.service import EnhancedDOMTreeNode
+
+def clean_html_basic(page_html_content, max_text_length=100):
+    soup = BeautifulSoup(page_html_content, 'html.parser')
+
+    for script in soup(["script", "style"]):
+        script.decompose()
+
+    from bs4 import Comment
+    comments = soup.findAll(text=lambda text: isinstance(text, Comment))
+    for comment in comments:
+        comment.extract()
+
+    for text_node in soup.find_all(string=True):
+        if text_node.parent.name not in ['script', 'style']:
+            clean_text = ' '.join(text_node.split())
+
+            if len(clean_text) > max_text_length:
+                clean_text = clean_text[:max_text_length].rstrip() + "..."
+
+            if clean_text != text_node:
+                text_node.replace_with(clean_text)
+
+    important_attrs = ['id', 'class', 'name', 'role', 'type',
+                       'colspan', 'rowspan', 'headers', 'scope',
+                       'href', 'src', 'alt', 'title']
+
+    for tag in soup.find_all():
+        attrs_to_keep = {}
+        for attr in list(tag.attrs.keys()):
+            if (attr in important_attrs or
+                    attr.startswith('data-') or
+                    attr.startswith('aria-')):
+                attrs_to_keep[attr] = tag.attrs[attr]
+        tag.attrs = attrs_to_keep
+
+    return str(soup)
+
+
+def get_sibling_position(node: EnhancedDOMTreeNode) -> int:
+    """Get the position of node among its siblings with the same tag"""
+    if not node.parent_node:
+        return 1
+
+    tag_name = node.tag_name
+    position = 1
+
+    # Find siblings with same tag name before this node
+    for sibling in node.parent_node.children:
+        if sibling == node:
+            break
+        if sibling.tag_name == tag_name:
+            position += 1
+
+    return position
+
+
+def extract_css_hints(node: EnhancedDOMTreeNode) -> dict:
+    """Extract CSS selector construction hints"""
+    hints = {}
+
+    if "id" in node.attributes:
+        hints["id"] = f"#{node.attributes['id']}"
+
+    if "class" in node.attributes:
+        classes = node.attributes["class"].split()
+        hints["class"] = f".{'.'.join(classes[:3])}"  # Limit class count
+
+    # Attribute selector hints
+    for attr in ["name", "data-testid", "type"]:
+        if attr in node.attributes:
+            hints[f"attr_{attr}"] = f"[{attr}='{node.attributes[attr]}']"
+
+    return hints
+
+
+def convert_selector_map_for_llm(selector_map) -> dict:
+    """
+    Convert complex selector_map to simplified format suitable for LLM understanding and JS code writing
+    """
+    simplified_elements = []
+
+    for element_index, node in selector_map.items():
+        if node.is_visible and node.element_index is not None:  # Only include visible interactive elements
+            element_info = {
+                "tag": node.tag_name,
+                "text": node.get_meaningful_text_for_llm()[:200],  # Limit text length
+
+                # Selector information - most needed for JS code
+                "selectors": {
+                    "xpath": node.xpath,
+                    "css_hints": extract_css_hints(node),  # Extract id, class etc
+                },
+
+                # Element semantics
+                "role": node.ax_node.role if node.ax_node else None,
+                "type": node.attributes.get("type"),
+                "aria_label": node.attributes.get("aria-label"),
+
+                # Key attributes
+                "attributes": {k: v for k, v in node.attributes.items()
+                               if k in ["id", "class", "name", "href", "src", "value", "placeholder", "data-testid"]},
+
+                # Interactivity
+                "is_clickable": node.snapshot_node.is_clickable if node.snapshot_node else False,
+                "is_input": node.tag_name.lower() in ["input", "textarea", "select"],
+
+                # Structure information
+                "parent_tag": node.parent_node.tag_name if node.parent_node else None,
+                "position_info": f"{node.tag_name}[{get_sibling_position(node)}]"
+            }
+            simplified_elements.append(element_info)
+
+    return {
+        "page_elements": simplified_elements,
+        "total_elements": len(simplified_elements)
+    }