patchpal 0.23.1__tar.gz → 0.24.0__tar.gz

{patchpal-0.23.1/patchpal.egg-info → patchpal-0.24.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: patchpal
-Version: 0.23.1
+Version: 0.24.0
 Summary: An agentic coding and automation assistant, supporting both local and cloud LLMs
 Author: PatchPal Contributors
 License-Expression: Apache-2.0
@@ -38,6 +38,9 @@ Requires-Dist: ruff==0.14.13; extra == "dev"
 Requires-Dist: pre-commit>=3.0.0; extra == "dev"
 Provides-Extra: mcp
 Requires-Dist: mcp>=0.9.0; extra == "mcp"
+Provides-Extra: browser
+Requires-Dist: playwright>=1.40.0; extra == "browser"
+Requires-Dist: nest-asyncio>=1.5.0; extra == "browser"
 Provides-Extra: docs
 Requires-Dist: mkdocs<2.0,>=1.5.0; extra == "docs"
 Requires-Dist: mkdocs-material>=9.5.0; extra == "docs"
@@ -62,6 +65,7 @@ Most agent frameworks are [built in TypeScript](https://news.ycombinator.com/ite
 - [Sandbox Mode](https://amaiya.github.io/patchpal/usage/sandbox/) for secure container isolation
 - [Python SDK](https://amaiya.github.io/patchpal/usage/python-api/) for flexibility and extensibility
 - [Built-In](https://amaiya.github.io/patchpal/features/tools/) and [Custom Tools](https://amaiya.github.io/patchpal/features/custom-tools/)
+- [Browser Automation](https://amaiya.github.io/patchpal/features/tools/#browser-automation-14-tools-optional) for interactive web tasks, form filling, and JavaScript-heavy sites
 - [Skills System](https://amaiya.github.io/patchpal/features/skills/) and [MCP Integration](https://amaiya.github.io/patchpal/features/mcp/)
 - [Autopilot Mode](https://amaiya.github.io/patchpal/usage/autopilot/) using [Ralph Wiggum loops](https://github.com/amaiya/patchpal/tree/main/examples/ralph/)
 - [Project Memory](https://amaiya.github.io/patchpal/features/memory/) automatically loads project context from `MEMORY.md` (repository root or `~/.patchpal/repos/<repo-name>/MEMORY.md`) at startup.
@@ -108,6 +112,7 @@ docker run -it --rm \
 
 ## Setup
 0. **Install**: `pip install patchpal`
+   - **Optional**: For browser automation (13 additional tools): `pip install patchpal[browser] && python -m playwright install chromium`
 1. **Get an API key or a Local LLM Engine**:
    - **[Cloud]** For Anthropic models (default): Sign up at https://console.anthropic.com/
    - **[Cloud]** For OpenAI models: Get a key from https://platform.openai.com/

{patchpal-0.23.1 → patchpal-0.24.0}/README.md

@@ -14,6 +14,7 @@ Most agent frameworks are [built in TypeScript](https://news.ycombinator.com/ite
 - [Sandbox Mode](https://amaiya.github.io/patchpal/usage/sandbox/) for secure container isolation
 - [Python SDK](https://amaiya.github.io/patchpal/usage/python-api/) for flexibility and extensibility
 - [Built-In](https://amaiya.github.io/patchpal/features/tools/) and [Custom Tools](https://amaiya.github.io/patchpal/features/custom-tools/)
+- [Browser Automation](https://amaiya.github.io/patchpal/features/tools/#browser-automation-14-tools-optional) for interactive web tasks, form filling, and JavaScript-heavy sites
 - [Skills System](https://amaiya.github.io/patchpal/features/skills/) and [MCP Integration](https://amaiya.github.io/patchpal/features/mcp/)
 - [Autopilot Mode](https://amaiya.github.io/patchpal/usage/autopilot/) using [Ralph Wiggum loops](https://github.com/amaiya/patchpal/tree/main/examples/ralph/)
 - [Project Memory](https://amaiya.github.io/patchpal/features/memory/) automatically loads project context from `MEMORY.md` (repository root or `~/.patchpal/repos/<repo-name>/MEMORY.md`) at startup.
@@ -60,6 +61,7 @@ docker run -it --rm \
 
 ## Setup
 0. **Install**: `pip install patchpal`
+   - **Optional**: For browser automation (13 additional tools): `pip install patchpal[browser] && python -m playwright install chromium`
 1. **Get an API key or a Local LLM Engine**:
    - **[Cloud]** For Anthropic models (default): Sign up at https://console.anthropic.com/
    - **[Cloud]** For OpenAI models: Get a key from https://platform.openai.com/

{patchpal-0.23.1 → patchpal-0.24.0}/patchpal/__init__.py

@@ -1,6 +1,6 @@
 """PatchPal - An open-source Claude Code clone implemented purely in Python."""
 
-__version__ = "0.23.1"
+__version__ = "0.24.0"
 
 from patchpal.agent import create_agent, create_react_agent
 from patchpal.cli.autopilot import autopilot_loop

{patchpal-0.23.1 → patchpal-0.24.0}/patchpal/agent/function_calling.py

@@ -1496,6 +1496,92 @@ Repository root takes priority if both exist."""
                                     "\033[2m❓ Asking user a question...\033[0m",
                                     flush=True,
                                 )
+                            elif tool_name == "browser_navigate":
+                                print(
+                                    f"\033[2m🌐 Browser navigating: {tool_args.get('url', '')}\033[0m",
+                                    flush=True,
+                                )
+                            elif tool_name == "browser_click":
+                                print(
+                                    f"\033[2m🖱️  Browser clicking: {tool_args.get('selector', '')}\033[0m",
+                                    flush=True,
+                                )
+                            elif tool_name == "browser_fill":
+                                print(
+                                    f"\033[2m⌨️  Browser filling: {tool_args.get('selector', '')}\033[0m",
+                                    flush=True,
+                                )
+                            elif tool_name == "browser_screenshot":
+                                print(
+                                    f"\033[2m📸 Browser screenshot: {tool_args.get('path', '') or 'default path'}\033[0m",
+                                    flush=True,
+                                )
+                            elif tool_name == "browser_get_text":
+                                print(
+                                    "\033[2m📄 Browser extracting page text...\033[0m",
+                                    flush=True,
+                                )
+                            elif tool_name == "browser_get_html":
+                                print(
+                                    "\033[2m📋 Browser extracting HTML source...\033[0m",
+                                    flush=True,
+                                )
+                            elif tool_name == "browser_scroll":
+                                direction = tool_args.get("direction", "down")
+                                selector = tool_args.get("selector", "")
+                                if selector:
+                                    scroll_desc = f"to '{selector}'"
+                                else:
+                                    scroll_desc = direction
+                                print(
+                                    f"\033[2m📜 Browser scrolling {scroll_desc}...\033[0m",
+                                    flush=True,
+                                )
+                            elif tool_name == "browser_execute_script":
+                                script_preview = tool_args.get("script", "")[:50]
+                                if len(tool_args.get("script", "")) > 50:
+                                    script_preview += "..."
+                                print(
+                                    f"\033[2m⚡ Browser executing script: {script_preview}\033[0m",
+                                    flush=True,
+                                )
+                            elif tool_name == "browser_wait":
+                                selector = tool_args.get("selector", "")
+                                if selector:
+                                    wait_desc = f"for '{selector}'"
+                                else:
+                                    wait_desc = f"{tool_args.get('milliseconds', 1000)}ms"
+                                print(
+                                    f"\033[2m⏳ Browser waiting {wait_desc}...\033[0m",
+                                    flush=True,
+                                )
+                            elif tool_name == "browser_close":
+                                print(
+                                    "\033[2m🚪 Closing browser...\033[0m",
+                                    flush=True,
+                                )
+                            elif tool_name == "browser_list_frames":
+                                print(
+                                    "\033[2m🖼️  Browser listing frames...\033[0m",
+                                    flush=True,
+                                )
+                            elif tool_name == "browser_switch_frame":
+                                frame_info = ""
+                                if "index" in tool_args and tool_args["index"] is not None:
+                                    frame_info = f"index {tool_args['index']}"
+                                elif "name" in tool_args and tool_args["name"] is not None:
+                                    frame_info = f"'{tool_args['name']}'"
+                                else:
+                                    frame_info = "main page"
+                                print(
+                                    f"\033[2m🖼️  Browser switching to frame: {frame_info}\033[0m",
+                                    flush=True,
+                                )
+                            elif tool_name == "browser_dismiss_modals":
+                                print(
+                                    "\033[2m🚫 Browser dismissing modals...\033[0m",
+                                    flush=True,
+                                )
                             else:
                                 # Check if this is an MCP tool by looking for __mcp_server__ attribute
                                 if hasattr(tool_func, "__mcp_server__"):

{patchpal-0.23.1 → patchpal-0.24.0}/patchpal/cli/interactive.py

@@ -575,6 +575,32 @@ Supported models: Any LiteLLM-supported model
                     pass  # Don't fail if audit logging fails
 
                 print("\nGoodbye!")
+
+                # Clean up the browser if it was opened during the session.
+                # An open Playwright browser keeps its own asyncio machinery
+                # alive; closing it here releases the browser process/loop so
+                # they don't linger into interpreter shutdown.
+                try:
+                    from patchpal.tools.browser_tools import _BrowserState
+
+                    if _BrowserState.is_open():
+                        _BrowserState.close(silent=True)
+                except Exception:
+                    pass  # Browser tools optional / already closed
+
+                # Suppress the benign "Task exception was never retrieved"
+                # warning that asyncio emits at shutdown: prompt_toolkit's
+                # pending Application.run_async() task gets cancelled with a
+                # KeyboardInterrupt during teardown, and nobody retrieves it.
+                # This is cosmetic and happens AFTER all work is done.
+                try:
+                    import asyncio
+
+                    _loop = asyncio.get_event_loop()
+                    _loop.set_exception_handler(lambda loop, context: None)
+                except Exception:
+                    pass
+
                 break
 
             # Handle /help command - show available commands
@@ -1633,7 +1659,25 @@ Supported models: Any LiteLLM-supported model
             )
             continue
         except Exception as e:
-            print(f"\n\033[1;31mError:\033[0m {e}")
+            # Debug: Print full exception info to help diagnose asyncio errors
+            import traceback
+
+            error_str = str(e)
+            print(f"\n\033[1;31mError:\033[0m {error_str}")
+
+            # If it's the asyncio error, print more debug info
+            if "asyncio.run() cannot be called from a running event loop" in error_str:
+                print("\n\033[1;33m[DEBUG] AsyncIO Error Details:\033[0m")
+                print(
+                    "This error occurs when asyncio.run() is called while an event loop is already running."
+                )
+                print("Stack trace:")
+                traceback.print_exc()
+                print(
+                    "\n\033[1;33m[DEBUG] If browser tools were just used, this may be a Playwright cleanup issue.\033[0m"
+                )
+                print("Try: Close any open browser windows and type 'exit' to restart PatchPal.\n")
+
             print("Please try again or type 'exit' to quit.")
 
 

{patchpal-0.23.1 → patchpal-0.24.0}/patchpal/config.py

@@ -79,6 +79,16 @@ class Config:
         """Enable web search and fetch tools (default: true)."""
         return _get_env_bool("PATCHPAL_ENABLE_WEB", "true")
 
+    @property
+    def ENABLE_BROWSER(self) -> bool:
+        """Enable browser automation tools (default: true).
+
+        Browser tools require web access, so they are only available when
+        ENABLE_WEB is also true (and Playwright is installed). Set this to
+        false to disable the browser tools while keeping web_search/web_fetch.
+        """
+        return _get_env_bool("PATCHPAL_ENABLE_BROWSER", "true")
+
     @property
     def ENABLE_MCP(self) -> bool:
         """Enable Model Context Protocol tools (default: true)."""

{patchpal-0.23.1 → patchpal-0.24.0}/patchpal/permissions.py

@@ -358,6 +358,23 @@ class PermissionManager:
         if self._check_grant_list(self.persistent_grants, tool_name, pattern, full_command):
             return True
 
+        # Special case: "browser" pattern applies across all browser tools
+        # If browser pattern was granted to ANY browser tool, apply to all
+        if pattern == "browser":
+            browser_tools = [
+                "browser_navigate",
+                "browser_click",
+                "browser_fill",
+                "browser_execute_script",
+            ]
+            for browser_tool in browser_tools:
+                if self._check_grant_list(self.session_grants, browser_tool, pattern, full_command):
+                    return True
+                if self._check_grant_list(
+                    self.persistent_grants, browser_tool, pattern, full_command
+                ):
+                    return True
+
         return False
 
     def _grant_permission(

{patchpal-0.23.1 → patchpal-0.24.0}/patchpal/tools/__init__.py

@@ -67,6 +67,42 @@ from patchpal.tools.web_tools import (
     web_search,
 )
 
+# Browser tools (optional - only available if Playwright installed)
+try:
+    from patchpal.tools.browser_tools import (
+        PLAYWRIGHT_AVAILABLE,
+        browser_click,
+        browser_close,
+        browser_dismiss_modals,
+        browser_execute_script,
+        browser_fill,
+        browser_get_html,
+        browser_get_text,
+        browser_list_frames,
+        browser_navigate,
+        browser_press_key,
+        browser_screenshot,
+        browser_scroll,
+        browser_switch_frame,
+        browser_wait,
+    )
+except ImportError:
+    PLAYWRIGHT_AVAILABLE = False
+    browser_navigate = None
+    browser_click = None
+    browser_fill = None
+    browser_screenshot = None
+    browser_get_text = None
+    browser_get_html = None
+    browser_scroll = None
+    browser_execute_script = None
+    browser_press_key = None
+    browser_wait = None
+    browser_close = None
+    browser_dismiss_modals = None
+    browser_list_frames = None
+    browser_switch_frame = None
+
 __all__ = [
     # File operations
     "read_file",
@@ -96,6 +132,22 @@ __all__ = [
     "web_fetch",
     "web_search",
     "get_url_tracker",
+    # Browser tools (optional)
+    "PLAYWRIGHT_AVAILABLE",
+    "browser_navigate",
+    "browser_click",
+    "browser_fill",
+    "browser_screenshot",
+    "browser_get_text",
+    "browser_get_html",
+    "browser_scroll",
+    "browser_execute_script",
+    "browser_press_key",
+    "browser_wait",
+    "browser_close",
+    "browser_dismiss_modals",
+    "browser_list_frames",
+    "browser_switch_frame",
     # Shell tools
     "run_shell",
     # User interaction