droidrun 0.3.10.dev5__py3-none-any.whl → 0.3.10.dev7__py3-none-any.whl

droidrun/cli/main.py

@@ -8,22 +8,22 @@ import os
 import warnings
 from contextlib import nullcontext
 from functools import wraps
-from pathlib import Path
 
 import click
 from adbutils import adb
 from rich.console import Console
 
 from droidrun.agent.droid import DroidAgent
-from droidrun.agent.utils.llm_picker import load_llm
+from droidrun.agent.utils.llm_picker import load_llm, load_llms_from_profiles
 from droidrun.cli.logs import LogHandler
+from droidrun.config_manager import ConfigManager
 from droidrun.config_manager.config_manager import (
     AgentConfig,
     CodeActConfig,
-    ManagerConfig,
-    ExecutorConfig,
     DeviceConfig,
+    ExecutorConfig,
     LoggingConfig,
+    ManagerConfig,
     ToolsConfig,
     TracingConfig,
 )
@@ -38,8 +38,6 @@ from droidrun.portal import (
 )
 from droidrun.telemetry import print_telemetry_message
 from droidrun.tools import AdbTools, IOSTools
-from droidrun.config_manager import ConfigManager
-
 
 # Suppress all warnings
 warnings.filterwarnings("ignore")
@@ -240,7 +238,7 @@ async def run_command(
                 if temperature is not None:
                     overrides = {name: {'temperature': temperature} for name in profile_names}
 
-                llms = config.load_all_llms(profile_names=profile_names, **overrides)
+                llms = load_llms_from_profiles(config.llm_profiles, profile_names=profile_names, **overrides)
                 logger.info(f"🧠 Loaded {len(llms)} agent-specific LLMs from profiles")
 
             # ================================================================
@@ -532,29 +530,43 @@ def run(
 ):
     """Run a command on your Android device using natural language."""
 
-    # Call our standalone function
-    return run_command(
-        command,
-        config,
-        device,
-        provider,
-        model,
-        steps,
-        base_url,
-        api_base,
-        vision,
-        manager_vision,
-        executor_vision,
-        codeact_vision,
-        reasoning,
-        tracing,
-        debug,
-        use_tcp,
-        temperature=temperature,
-        save_trajectory=save_trajectory,
-        allow_drag=allow_drag,
-        ios=ios if ios is not None else False,
-    )
+    try:
+        run_command(
+            command,
+            config,
+            device,
+            provider,
+            model,
+            steps,
+            base_url,
+            api_base,
+            vision,
+            manager_vision,
+            executor_vision,
+            codeact_vision,
+            reasoning,
+            tracing,
+            debug,
+            use_tcp,
+            temperature=temperature,
+            save_trajectory=save_trajectory,
+            allow_drag=allow_drag,
+            ios=ios if ios is not None else False,
+        )
+    finally:
+        # Disable DroidRun keyboard after execution
+        try:
+            if not (ios if ios is not None else False):
+                device_serial = adb.device().serial
+                if device_serial:
+                    tools = AdbTools(serial=device, use_tcp=use_tcp if use_tcp is not None else False)
+                    if hasattr(tools, 'device') and tools.device:
+                        tools.device.shell("ime disable com.droidrun.portal/.DroidrunKeyboardIME")
+                        click.echo("DroidRun keyboard disabled successfully")
+                    # Cleanup tools
+                    del tools
+        except Exception as disable_e:
+            click.echo(f"Warning: Failed to disable DroidRun keyboard: {disable_e}")
 
 
 @cli.command()
@@ -756,8 +768,206 @@ def ping(device: str | None, use_tcp: bool, debug: bool):
 cli.add_command(macro_cli, name="macro")
 
 
+async def test(command: str):
+    config = ConfigManager(path="config.yaml")
+    # Initialize logging first (use config default if debug not specified)
+    debug_mode = debug if debug is not None else config.logging.debug
+    log_handler = configure_logging(command, debug_mode, config.logging.rich_text)
+    logger = logging.getLogger("droidrun")
+
+    log_handler.update_step("Initializing...")
+
+    with log_handler.render():
+        try:
+            logger.info(f"🚀 Starting: {command}")
+            print_telemetry_message()
+
+            # ================================================================
+            # STEP 1: Build config objects with CLI overrides
+            # ================================================================
+
+            # Build agent-specific configs with vision overrides
+            if vision is not None:
+                # --vision flag overrides all agents
+                manager_vision_val = vision
+                executor_vision_val = vision
+                codeact_vision_val = vision
+                logger.debug(f"CLI override: vision={vision} (all agents)")
+            else:
+                # Use individual overrides or config defaults
+                manager_vision_val = config.agent.manager.vision
+                executor_vision_val = config.agent.executor.vision
+                codeact_vision_val = config.agent.codeact.vision
+
+            manager_cfg = ManagerConfig(
+                vision=manager_vision_val,
+                system_prompt="rev1.jinja2"
+            )
+
+            executor_cfg = ExecutorConfig(
+                vision=executor_vision_val,
+                system_prompt="rev1.jinja2"
+            )
+
+            codeact_cfg = CodeActConfig(
+                vision=codeact_vision_val,
+                system_prompt=config.agent.codeact.system_prompt,
+                user_prompt=config.agent.codeact.user_prompt
+            )
+
+            agent_cfg = AgentConfig(
+                max_steps=steps if steps is not None else config.agent.max_steps,
+                reasoning=reasoning if reasoning is not None else config.agent.reasoning,
+                after_sleep_action=config.agent.after_sleep_action,
+                wait_for_stable_ui=config.agent.wait_for_stable_ui,
+                prompts_dir=config.agent.prompts_dir,
+                manager=manager_cfg,
+                executor=executor_cfg,
+                codeact=codeact_cfg,
+                app_cards=config.agent.app_cards,
+            )
+
+            device_cfg = DeviceConfig(
+                serial=device if device is not None else config.device.serial,
+                use_tcp=use_tcp if use_tcp is not None else config.device.use_tcp,
+            )
+
+            tools_cfg = ToolsConfig(
+                allow_drag=allow_drag if allow_drag is not None else config.tools.allow_drag,
+            )
+
+            logging_cfg = LoggingConfig(
+                debug=debug if debug is not None else config.logging.debug,
+                save_trajectory=save_trajectory if save_trajectory is not None else config.logging.save_trajectory,
+                rich_text=config.logging.rich_text,
+            )
+
+            tracing_cfg = TracingConfig(
+                enabled=tracing if tracing is not None else config.tracing.enabled,
+            )
+
+            # ================================================================
+            # STEP 3: Load LLMs
+            # ================================================================
+
+            log_handler.update_step("Loading LLMs...")
+
+            # No custom provider/model - use profiles from config
+            logger.info("📋 Loading LLMs from config profiles...")
+
+            profile_names = ['manager', 'executor', 'codeact', 'text_manipulator', 'app_opener']
+
+            # Apply temperature override to all profiles if specified
+            overrides = {}
+            if temperature is not None:
+                overrides = {name: {'temperature': temperature} for name in profile_names}
+
+            llms = load_llms_from_profiles(config.llm_profiles, profile_names=profile_names, **overrides)
+            logger.info(f"🧠 Loaded {len(llms)} agent-specific LLMs from profiles")
+
+            # ================================================================
+            # STEP 4: Setup device and tools
+            # ================================================================
+
+            log_handler.update_step("Setting up tools...")
+
+            device_serial = device_cfg.serial
+            if device_serial is None and not ios:
+                logger.info("🔍 Finding connected device...")
+                devices = adb.list()
+                if not devices:
+                    raise ValueError("No connected devices found.")
+                device_serial = devices[0].serial
+                device_cfg = DeviceConfig(serial=device_serial, use_tcp=device_cfg.use_tcp)
+                logger.info(f"📱 Using device: {device_serial}")
+            elif device_serial is None and ios:
+                raise ValueError("iOS device not specified. Please specify device base url via --device")
+            else:
+                logger.info(f"📱 Using device: {device_serial}")
+
+            tools = (
+                AdbTools(
+                    serial=device_serial,
+                    use_tcp=device_cfg.use_tcp,
+                    app_opener_llm=llms.get('app_opener'),
+                    text_manipulator_llm=llms.get('text_manipulator')
+                )
+                if not ios
+                else IOSTools(url=device_serial)
+            )
+
+            excluded_tools = [] if tools_cfg.allow_drag else ["drag"]
+
+            # ================================================================
+            # STEP 5: Initialize DroidAgent with all settings
+            # ================================================================
+
+            log_handler.update_step("Initializing DroidAgent...")
+
+            mode = "planning with reasoning" if agent_cfg.reasoning else "direct execution"
+            logger.info(f"🤖 Agent mode: {mode}")
+            logger.info(f"👁️  Vision settings: Manager={agent_cfg.manager.vision}, "
+                       f"Executor={agent_cfg.executor.vision}, CodeAct={agent_cfg.codeact.vision}")
+
+            if tracing_cfg.enabled:
+                logger.info("🔍 Tracing enabled")
+
+            droid_agent = DroidAgent(
+                goal=command,
+                llms=llms,
+                tools=tools,
+                config=config,
+                agent_config=agent_cfg,
+                device_config=device_cfg,
+                tools_config=tools_cfg,
+                logging_config=logging_cfg,
+                tracing_config=tracing_cfg,
+                excluded_tools=excluded_tools,
+                timeout=1000,
+            )
+
+            # ================================================================
+            # STEP 6: Run agent
+            # ================================================================
+
+            logger.info("▶️  Starting agent execution...")
+            logger.info("Press Ctrl+C to stop")
+            log_handler.update_step("Running agent...")
+
+            try:
+                handler = droid_agent.run()
+
+                async for event in handler.stream_events():
+                    log_handler.handle_event(event)
+                result = await handler  # noqa: F841
+
+            except KeyboardInterrupt:
+                log_handler.is_completed = True
+                log_handler.is_success = False
+                log_handler.current_step = "Stopped by user"
+                logger.info("⏹️ Stopped by user")
+
+            except Exception as e:
+                log_handler.is_completed = True
+                log_handler.is_success = False
+                log_handler.current_step = f"Error: {e}"
+                logger.error(f"💥 Error: {e}")
+                if logging_cfg.debug:
+                    import traceback
+                    logger.debug(traceback.format_exc())
+
+        except Exception as e:
+            log_handler.current_step = f"Error: {e}"
+            logger.error(f"💥 Setup error: {e}")
+            debug_mode = debug if debug is not None else config.logging.debug
+            if debug_mode:
+                import traceback
+                logger.debug(traceback.format_exc())
+
+
+
 if __name__ == "__main__":
-    command = "Download clash royale app"
+    command = "set gboard to the default keyboard"
     device = None
     provider = "GoogleGenAI"
     model = "models/gemini-2.5-flash"
@@ -765,7 +975,7 @@ if __name__ == "__main__":
     api_key = os.getenv("GOOGLE_API_KEY")
     steps = 15
     vision = True
-    reasoning = True
+    reasoning = False
     tracing = True
     debug = True
     use_tcp = False
@@ -774,6 +984,6 @@ if __name__ == "__main__":
     ios = False
     save_trajectory = "none"
     allow_drag = False
-    run_command(
-        command
+    asyncio.run(
+        test(command)
     )

droidrun/config_manager/__init__.py

@@ -1,4 +1,3 @@
-from droidrun.config_manager.app_card_loader import AppCardLoader
 from droidrun.config_manager.config_manager import (
     AgentConfig,
     AppCardConfig,
@@ -23,5 +22,4 @@ __all__ = [
     "TracingConfig",
     "LoggingConfig",
     "ToolsConfig",
-    "AppCardLoader",
 ]

droidrun/config_manager/config_manager.py

@@ -3,7 +3,6 @@ from __future__ import annotations
 import os
 import threading
 from dataclasses import asdict, dataclass, field
-from pathlib import Path
 from typing import Any, Callable, Dict, List, Optional
 
 import yaml
@@ -35,30 +34,38 @@ agent:
     # Enable vision capabilities (screenshots)
     vision: false
     # System prompt filename (located in prompts_dir/codeact/)
-    system_prompt: system.md
+    system_prompt: system.jinja2
     # User prompt filename (located in prompts_dir/codeact/)
-    user_prompt: user.md
+    user_prompt: user.jinja2
 
   # Manager Agent Configuration
   manager:
     # Enable vision capabilities (screenshots)
     vision: false
     # System prompt filename (located in prompts_dir/manager/)
-    system_prompt: system.md
+    system_prompt: system.jinja2
 
   # Executor Agent Configuration
   executor:
     # Enable vision capabilities (screenshots)
     vision: false
     # System prompt filename (located in prompts_dir/executor/)
-    system_prompt: system.md
+    system_prompt: system.jinja2
 
   # App Cards Configuration
   app_cards:
     # Enable app-specific instruction cards
     enabled: true
-    # Directory containing app card files
+    # Mode: local (file-based), server (HTTP API), or composite (server with local fallback)
+    mode: local
+    # Directory containing app card files (for local/composite modes)
     app_cards_dir: config/app_cards
+    # Server URL for remote app cards (for server/composite modes)
+    server_url: null
+    # Server request timeout in seconds
+    server_timeout: 2.0
+    # Number of server retry attempts
+    server_max_retries: 2
 
 # === LLM Profiles ===
 # Define LLM configurations for each agent type
@@ -170,29 +177,33 @@ class LLMProfile:
 class CodeActConfig:
     """CodeAct agent configuration."""
     vision: bool = False
-    system_prompt: str = "system.md"
-    user_prompt: str = "user.md"
+    system_prompt: str = "system.jinja2"
+    user_prompt: str = "user.jinja2"
 
 
 @dataclass
 class ManagerConfig:
     """Manager agent configuration."""
     vision: bool = False
-    system_prompt: str = "system.md"
+    system_prompt: str = "system.jinja2"
 
 
 @dataclass
 class ExecutorConfig:
     """Executor agent configuration."""
     vision: bool = False
-    system_prompt: str = "system.md"
+    system_prompt: str = "system.jinja2"
 
 
 @dataclass
 class AppCardConfig:
     """App card configuration."""
     enabled: bool = True
+    mode: str = "local"  # local | server | composite
     app_cards_dir: str = "config/app_cards"
+    server_url: Optional[str] = None
+    server_timeout: float = 2.0
+    server_max_retries: int = 2
 
 
 @dataclass
@@ -210,20 +221,24 @@ class AgentConfig:
     app_cards: AppCardConfig = field(default_factory=AppCardConfig)
 
     def get_codeact_system_prompt_path(self) -> str:
-        """Get full path to CodeAct system prompt."""
-        return f"{self.prompts_dir}/codeact/{self.codeact.system_prompt}"
+        """Get resolved absolute path to CodeAct system prompt."""
+        path = f"{self.prompts_dir}/codeact/{self.codeact.system_prompt}"
+        return str(PathResolver.resolve(path, must_exist=True))
 
     def get_codeact_user_prompt_path(self) -> str:
-        """Get full path to CodeAct user prompt."""
-        return f"{self.prompts_dir}/codeact/{self.codeact.user_prompt}"
+        """Get resolved absolute path to CodeAct user prompt."""
+        path = f"{self.prompts_dir}/codeact/{self.codeact.user_prompt}"
+        return str(PathResolver.resolve(path, must_exist=True))
 
     def get_manager_system_prompt_path(self) -> str:
-        """Get full path to Manager system prompt."""
-        return f"{self.prompts_dir}/manager/{self.manager.system_prompt}"
+        """Get resolved absolute path to Manager system prompt."""
+        path = f"{self.prompts_dir}/manager/{self.manager.system_prompt}"
+        return str(PathResolver.resolve(path, must_exist=True))
 
     def get_executor_system_prompt_path(self) -> str:
-        """Get full path to Executor system prompt."""
-        return f"{self.prompts_dir}/executor/{self.executor.system_prompt}"
+        """Get resolved absolute path to Executor system prompt."""
+        path = f"{self.prompts_dir}/executor/{self.executor.system_prompt}"
+        return str(PathResolver.resolve(path, must_exist=True))
 
 
 @dataclass
@@ -372,6 +387,7 @@ class ConfigManager:
 
     Usage:
         from droidrun.config_manager.config_manager import ConfigManager
+        from droidrun.agent.utils.llm_picker import load_llms_from_profiles
 
         # Create config instance (singleton pattern)
         config = ConfigManager()
@@ -379,12 +395,19 @@ class ConfigManager:
         # Access typed config objects
         print(config.agent.max_steps)
 
-        # Load all LLMs
-        llms = config.load_all_llms()
+        # Load all LLMs from profiles
+        llms = load_llms_from_profiles(config.llm_profiles)
         manager_llm = llms['manager']
         executor_llm = llms['executor']
         codeact_llm = llms['codeact']
 
+        # Load specific profiles with overrides
+        llms = load_llms_from_profiles(
+            config.llm_profiles,
+            profile_names=['manager', 'executor'],
+            manager={'temperature': 0.1}
+        )
+
         # Modify and save
         config.save()
     """
@@ -494,86 +517,6 @@ class ConfigManager:
 
             return self._config.llm_profiles[profile_name]
 
-    def load_llm_from_profile(self, profile_name: str, **override_kwargs):
-        """
-        Load an LLM using a profile configuration.
-
-        Args:
-            profile_name: Name of the profile to use (fast, mid, smart, custom)
-            **override_kwargs: Additional kwargs to override profile settings
-
-        Returns:
-            Initialized LLM instance
-
-        Example:
-            # Use specific profile
-            llm = config.load_llm_from_profile("smart")
-
-            # Override specific settings
-            llm = config.load_llm_from_profile("fast", temperature=0.5)
-        """
-        from droidrun.agent.utils.llm_picker import load_llm
-
-        profile = self.get_llm_profile(profile_name)
-
-        # Get kwargs from profile
-        kwargs = profile.to_load_llm_kwargs()
-
-        # Override with any provided kwargs
-        kwargs.update(override_kwargs)
-
-        # Load the LLM
-        return load_llm(provider_name=profile.provider, **kwargs)
-
-    def load_all_llms(self, profile_names: Optional[list[str]] = None, **override_kwargs_per_profile):
-        """
-        Load multiple LLMs from profiles for different use cases.
-
-        Args:
-            profile_names: List of profile names to load. If None, loads agent-specific profiles
-            **override_kwargs_per_profile: Dict of profile-specific overrides
-                Example: manager={'temperature': 0.1}, executor={'max_tokens': 8000}
-
-        Returns:
-            Dict mapping profile names to initialized LLM instances
-
-        Example:
-            # Load all agent-specific profiles
-            llms = config.load_all_llms()
-            manager_llm = llms['manager']
-            executor_llm = llms['executor']
-            codeact_llm = llms['codeact']
-
-            # Load specific profiles
-            llms = config.load_all_llms(['manager', 'executor'])
-
-            # Load with overrides
-            llms = config.load_all_llms(
-                manager={'temperature': 0.1},
-                executor={'max_tokens': 8000}
-            )
-        """
-        from droidrun.agent.utils.llm_picker import load_llm
-
-        if profile_names is None:
-            profile_names = ["manager", "executor", "codeact", "text_manipulator", "app_opener"]
-
-        llms = {}
-        for profile_name in profile_names:
-            profile = self.get_llm_profile(profile_name)
-
-            # Get kwargs from profile
-            kwargs = profile.to_load_llm_kwargs()
-
-            # Apply profile-specific overrides if provided
-            if profile_name in override_kwargs_per_profile:
-                kwargs.update(override_kwargs_per_profile[profile_name])
-
-            # Load the LLM
-            llms[profile_name] = load_llm(provider_name=profile.provider, **kwargs)
-
-        return llms
-
     # ---------------- I/O ----------------
     def _ensure_file_exists(self) -> None:
         parent = self.path.parent
@@ -653,7 +596,7 @@ class ConfigManager:
 
         Example:
             >>> config.list_available_prompts("manager")
-            ['system.md', 'experimental.md', 'minimal.md']
+            ['system.jinja2', 'experimental.jinja2', 'minimal.jinja2']
         """
         agent_type = agent_type.lower()
         if agent_type not in ["codeact", "manager", "executor"]:
@@ -667,7 +610,7 @@ class ConfigManager:
             return []
 
         # List all .md files in the directory
-        return sorted([f.name for f in prompts_dir.glob("*.md")])
+        return sorted([f.name for f in prompts_dir.glob("*.jinja2")])
 
     # useful for tests to reset singleton state
     @classmethod

droidrun/config_manager/path_resolver.py

@@ -64,7 +64,7 @@ class PathResolver:
             output_dir = PathResolver.resolve("trajectories", create_if_missing=True)
 
             # Loading prompts (must exist, checks both locations)
-            prompt = PathResolver.resolve("config/prompts/system.md", must_exist=True)
+            prompt = PathResolver.resolve("config/prompts/system.jinja2", must_exist=True)
 
             # Absolute path (used as-is)
             abs_path = PathResolver.resolve("/tmp/output")