Feature: Add BrowserAutomationConnector with smolagents integration

logos/connectors.py

@@ -299,6 +299,130 @@ class NeMoAgentConnector:
             return f"[Vision Error] {e}"
 
 
+# ==========================================
+# BROWSER AUTOMATION CONNECTOR (smolagents)
+# ==========================================
+
+class BrowserAutomationConnector:
+    """
+    Adapter for HuggingFace smolagents + helium for browser automation.
+    Enables autonomous web navigation, form filling, and data extraction.
+    """
+    
+    def __init__(self, headless: bool = True):
+        self.headless = headless
+        self._driver = None
+        self._agent = None
+    
+    def _ensure_browser(self):
+        """Initialize browser with helium."""
+        if self._driver is None:
+            try:
+                from helium import start_chrome, go_to, click, write, get_driver
+                from selenium.webdriver.chrome.options import Options
+                
+                options = Options()
+                if self.headless:
+                    options.add_argument('--headless')
+                    options.add_argument('--no-sandbox')
+                    options.add_argument('--disable-dev-shm-usage')
+                
+                start_chrome(headless=self.headless)
+                self._driver = get_driver()
+            except ImportError:
+                raise ImportError("helium/selenium not installed. Run: pip install helium selenium")
+        return self._driver
+    
+    def navigate(self, url: str) -> str:
+        """Navigate to a URL and return page title."""
+        try:
+            from helium import go_to, get_driver
+            self._ensure_browser()
+            go_to(url)
+            return f"Navigated to: {get_driver().title}"
+        except Exception as e:
+            return f"[Navigation Error] {e}"
+    
+    def search_page(self, query: str) -> str:
+        """Search for text on current page."""
+        try:
+            driver = self._ensure_browser()
+            page_source = driver.page_source.lower()
+            if query.lower() in page_source:
+                return f"Found '{query}' on page"
+            else:
+                return f"'{query}' not found on page"
+        except Exception as e:
+            return f"[Search Error] {e}"
+    
+    def get_page_text(self) -> str:
+        """Extract visible text from current page."""
+        try:
+            driver = self._ensure_browser()
+            from bs4 import BeautifulSoup
+            soup = BeautifulSoup(driver.page_source, 'html.parser')
+            text = soup.get_text(separator=' ', strip=True)
+            return text[:5000]  # Limit to 5k chars
+        except Exception as e:
+            return f"[Extraction Error] {e}"
+    
+    def click_element(self, text: str) -> str:
+        """Click an element by its text."""
+        try:
+            from helium import click
+            self._ensure_browser()
+            click(text)
+            return f"Clicked: {text}"
+        except Exception as e:
+            return f"[Click Error] {e}"
+    
+    def type_text(self, field: str, text: str) -> str:
+        """Type text into a field."""
+        try:
+            from helium import write
+            self._ensure_browser()
+            write(text, into=field)
+            return f"Typed into {field}"
+        except Exception as e:
+            return f"[Type Error] {e}"
+    
+    def close(self):
+        """Close the browser."""
+        try:
+            from helium import kill_browser
+            kill_browser()
+            self._driver = None
+        except:
+            pass
+    
+    def run_agent_task(self, task: str, model: str = "Qwen/Qwen2.5-Coder-32B-Instruct") -> str:
+        """
+        Run a browser automation task using smolagents Code Agent.
+        
+        Args:
+            task: Natural language task description
+            model: HF model for the code agent
+            
+        Returns:
+            Task result
+        """
+        try:
+            from smolagents import CodeAgent, HfApiModel
+            from smolagents.tools import Tool
+            
+            # Create agent with browser tools
+            model_instance = HfApiModel(model)
+            agent = CodeAgent(tools=[], model=model_instance)
+            
+            # Run the task
+            result = agent.run(task)
+            return str(result)
+        except ImportError:
+            return "[Error] smolagents not installed. Run: pip install smolagents"
+        except Exception as e:
+            return f"[Agent Error] {e}"
+
+
 # ==========================================
 # FACTORY
 # ==========================================
@@ -308,7 +432,7 @@ def get_connector(connector_type: str, **kwargs) -> Any:
     Factory function for connectors.
     
     Args:
-        connector_type: One of 'hf', 'ocr', 'vision', 'nemo'
+        connector_type: One of 'hf', 'ocr', 'vision', 'nemo', 'browser'
         **kwargs: Connector-specific arguments
         
     Returns:
@@ -318,7 +442,8 @@ def get_connector(connector_type: str, **kwargs) -> Any:
         'hf': HuggingFaceConnector,
         'ocr': OCRConnector,
         'vision': VisionConnector,
-        'nemo': NeMoAgentConnector
+        'nemo': NeMoAgentConnector,
+        'browser': BrowserAutomationConnector
     }
     
     if connector_type not in connectors:
@@ -355,5 +480,11 @@ AVAILABLE_CONNECTORS = {
         'capabilities': ['chat', 'analyze_diagram', 'react_agent', 'tool_calling'],
         'requires': ['nvidia-nat', 'openai'],
         'env_vars': ['NVIDIA_API_KEY']
+    },
+    'browser': {
+        'name': 'Browser Automation (smolagents)',
+        'capabilities': ['navigate', 'click_element', 'type_text', 'get_page_text', 'run_agent_task'],
+        'requires': ['smolagents', 'helium', 'selenium', 'beautifulsoup4'],
+        'env_vars': []
     }
 }

requirements.txt

@@ -8,3 +8,6 @@ plotly
 sympy
 easyocr
 nvidia-nat
+smolagents
+helium
+beautifulsoup4