autobyteus 1.1.4__py3-none-any.whl → 1.1.5__py3-none-any.whl

autobyteus/tools/usage/registries/xml_tool_usage_parser_registry.py

@@ -1,30 +0,0 @@
-# file: autobyteus/autobyteus/tools/usage/registries/xml_tool_usage_parser_registry.py
-import logging
-from typing import Optional
-
-from autobyteus.llm.providers import LLMProvider
-from autobyteus.tools.usage.parsers.base_parser import BaseToolUsageParser
-from autobyteus.tools.usage.parsers.default_xml_tool_usage_parser import DefaultXmlToolUsageParser
-from autobyteus.utils.singleton import SingletonMeta
-
-logger = logging.getLogger(__name__)
-
-class XmlToolUsageParserRegistry(metaclass=SingletonMeta):
-    """A singleton registry for retrieving XML-based tool usage parsers."""
-
-    def __init__(self):
-        self._default_parser = DefaultXmlToolUsageParser()
-        logger.info("XmlToolUsageParserRegistry initialized.")
-
-    def get_parser(self, provider: Optional[LLMProvider] = None) -> BaseToolUsageParser:
-        """
-        Retrieves the appropriate XML parser.
-
-        Args:
-            provider: The LLMProvider enum member. Currently ignored but kept for API consistency.
-
-        Returns:
-            An instance of a class derived from BaseToolUsageParser.
-        """
-        # For now, there's only one XML format, so always return the default.
-        return self._default_parser

examples/workflow/__init__.py

@@ -1 +0,0 @@
-# file: autobyteus/examples/workflow/__init__.py

examples/workflow/run_basic_research_workflow.py

@@ -1,189 +0,0 @@
-# file: autobyteus/examples/workflow/run_basic_research_workflow.py
-import asyncio
-import logging
-import argparse
-from pathlib import Path
-import sys
-import os
-
-# --- Boilerplate to make the script runnable from the project root ---
-SCRIPT_DIR = Path(__file__).resolve().parent.parent
-PACKAGE_ROOT = SCRIPT_DIR.parent
-if str(PACKAGE_ROOT) not in sys.path:
-    sys.path.insert(0, str(PACKAGE_ROOT))
-
-# Load environment variables from .env file in the project root
-try:
-    from dotenv import load_dotenv
-    env_file_path = PACKAGE_ROOT / ".env"
-    if env_file_path.exists():
-        load_dotenv(env_file_path)
-        print(f"Loaded environment variables from: {env_file_path}")
-    else:
-        print(f"Info: No .env file found at: {env_file_path}. Relying on exported environment variables.")
-except ImportError:
-    print("Warning: python-dotenv not installed. Cannot load .env file.")
-
-# --- Imports for the Workflow Example ---
-try:
-    from autobyteus.agent.context import AgentConfig
-    from autobyteus.llm.models import LLMModel
-    from autobyteus.llm.llm_factory import default_llm_factory, LLMFactory
-    from autobyteus.workflow.workflow_builder import WorkflowBuilder
-    from autobyteus.cli import workflow_cli
-except ImportError as e:
-    print(f"Error importing autobyteus components: {e}", file=sys.stderr)
-    print("Please ensure that the autobyteus library is installed and accessible.", file=sys.stderr)
-    sys.exit(1)
-
-# --- Logging Setup ---
-logger = logging.getLogger("basic_workflow_example")
-
-def setup_logging(args: argparse.Namespace):
-    """Configures logging for the interactive session."""
-    loggers_to_clear = [
-        logging.getLogger(),
-        logging.getLogger("autobyteus"),
-        logging.getLogger("autobyteus.cli"),
-    ]
-    for l in loggers_to_clear:
-        if l.hasHandlers():
-            for handler in l.handlers[:]:
-                l.removeHandler(handler)
-                if hasattr(handler, 'close'): handler.close()
-
-    script_log_level = logging.DEBUG if args.debug else logging.INFO
-
-    console_formatter = logging.Formatter('%(asctime)s - %(levelname)s - %(name)s - %(message)s')
-    formatted_console_handler = logging.StreamHandler(sys.stdout)
-    formatted_console_handler.setFormatter(console_formatter)
-    
-    root_logger = logging.getLogger()
-    root_logger.addHandler(formatted_console_handler)
-    root_logger.setLevel(script_log_level) 
-    
-    # Configure the main log file
-    log_file_path = Path(args.log_file).resolve()
-    log_file_path.parent.mkdir(parents=True, exist_ok=True)
-    agent_file_handler = logging.FileHandler(log_file_path, mode='w')  
-    agent_file_formatter = logging.Formatter('%(asctime)s - %(levelname)s - %(name)s:%(lineno)d - %(message)s')
-    agent_file_handler.setFormatter(agent_file_formatter)
-    file_log_level = logging.DEBUG if args.debug else logging.INFO
-
-    autobyteus_logger = logging.getLogger("autobyteus")
-    autobyteus_logger.addHandler(agent_file_handler)
-    autobyteus_logger.setLevel(file_log_level)
-    autobyteus_logger.propagate = True
-    
-    logger.info(f"Core library logs redirected to: {log_file_path} (level: {logging.getLevelName(file_log_level)})")
-
-
-async def main(args: argparse.Namespace):
-    """Main function to configure and run the research workflow."""
-    logger.info("--- Starting Basic Research Workflow Example ---")
-
-    # 1. Create LLM instance for the agents
-    try:
-        _ = LLMModel[args.llm_model]
-    except KeyError:
-        logger.error(f"LLM Model '{args.llm_model}' is not valid. Use --help-models to see available models.")
-        sys.exit(1)
-
-    logger.info(f"Creating LLM instance for model: {args.llm_model}")
-    llm_instance = default_llm_factory.create_llm(model_identifier=args.llm_model)
-
-    # 2. Define the Agent Configurations
-    
-    # The Coordinator/Manager Agent
-    research_manager_config = AgentConfig(
-        name="ResearchManager",
-        role="Coordinator",
-        description="A manager agent that receives research goals and delegates them to specialists.",
-        llm_instance=llm_instance,
-        # The prompt is now simpler, as the workflow builder will handle context.
-        # The {{tools}} placeholder is essential for tool injection.
-        system_prompt=(
-            "You are the manager of a research team. Your job is to understand the user's research goal and delegate it to the correct specialist agent on your team. "
-            "Do not answer questions yourself; always delegate. "
-            "You will be provided a manifest of your team members and available tools.\n\n"
-            "{{tools}}"
-        ),
-    )
-
-    # The Worker/Specialist Agent
-    fact_checker_config = AgentConfig(
-        name="FactChecker",
-        role="Specialist",
-        description="An agent with a limited, internal knowledge base for answering direct factual questions.",
-        llm_instance=llm_instance,
-        system_prompt=(
-            "You are a fact-checking bot. You have the following knowledge:\n"
-            "- The capital of France is Paris.\n"
-            "- The tallest mountain on Earth is Mount Everest.\n"
-            "- The primary programming language for AutoByteUs is Python.\n"
-            "You MUST ONLY answer questions based on this knowledge. If you are asked something you do not know, you MUST respond with 'I do not have information on that topic.'"
-        )
-    )
-
-    # 3. Define and Build the Workflow using WorkflowBuilder
-    
-    research_workflow = (
-        WorkflowBuilder(
-            name="BasicResearchWorkflow",
-            description="A simple two-agent workflow for delegating and answering research questions."
-        )
-        .set_coordinator(research_manager_config)
-        .add_agent_node(fact_checker_config, dependencies=[])
-        .build()
-    )
-    
-    # 4. Run the Workflow
-    
-    logger.info(f"Workflow instance '{research_workflow.name}' created with ID: {research_workflow.workflow_id}")
-
-    try:
-        logger.info("Starting interactive workflow session...")
-        await workflow_cli.run_workflow(
-            workflow=research_workflow, 
-            initial_prompt=args.initial_prompt
-        )
-        logger.info("Interactive workflow session finished.")
-    except Exception as e:
-        logger.error(f"An error occurred during the workflow execution: {e}", exc_info=True)
-    
-    logger.info("--- Basic Research Workflow Example Finished ---")
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser(description="Run a basic two-agent research workflow.")
-    parser.add_argument("--llm-model", type=str, default="gpt-4o", help="The LLM model to use for the agents.")
-    parser.add_argument("--help-models", action="store_true", help="Display available LLM models and exit.")
-    parser.add_argument("--initial-prompt", type=str, help="An optional initial prompt to start the workflow automatically.")
-    parser.add_argument("--debug", action="store_true", help="Enable debug logging.")
-    parser.add_argument("--log-file", type=str, default="./workflow_logs.txt", 
-                       help="Path to the log file for autobyteus library logs.")
-
-    if "--help-models" in sys.argv:
-        try:
-            LLMFactory.ensure_initialized() 
-            print("Available LLM Models (you can use either name or value with --llm-model):")
-            all_models = sorted(list(LLMModel), key=lambda m: m.name)
-            if not all_models:
-                print("  No models found.")
-            for model in all_models:
-                print(f"  - Name: {model.name:<35} Value: {model.value}")
-        except Exception as e:
-            print(f"Error listing models: {e}")
-        sys.exit(0)
-
-    parsed_args = parser.parse_args()
-    
-    setup_logging(parsed_args)
-
-    try:
-        asyncio.run(main(parsed_args))
-    except (KeyboardInterrupt, SystemExit):
-        logger.info("Script interrupted by user. Exiting.")
-    except Exception as e:
-        logger.error(f"An unhandled error occurred at the top level: {e}", exc_info=True)
-    finally:
-        logger.info("Exiting script.")

examples/workflow/run_code_review_workflow.py

@@ -1,269 +0,0 @@
-# file: autobyteus/examples/workflow/run_code_review_workflow.py
-"""
-This example script demonstrates a simple software development workflow
-with a coordinator, an engineer, a code reviewer, a test writer, and a tester.
-"""
-import asyncio
-import logging
-import argparse
-from pathlib import Path
-import sys
-import os
-
-# --- Boilerplate to make the script runnable from the project root ---
-SCRIPT_DIR = Path(__file__).resolve().parent.parent
-PACKAGE_ROOT = SCRIPT_DIR.parent
-if str(PACKAGE_ROOT) not in sys.path:
-    sys.path.insert(0, str(PACKAGE_ROOT))
-
-# Load environment variables from .env file
-try:
-    from dotenv import load_dotenv
-    load_dotenv(PACKAGE_ROOT / ".env")
-except ImportError:
-    pass
-
-# --- Imports for the Workflow TUI Example ---
-try:
-    from autobyteus.agent.context import AgentConfig
-    from autobyteus.llm.models import LLMModel
-    from autobyteus.llm.llm_factory import default_llm_factory, LLMFactory
-    from autobyteus.workflow.workflow_builder import WorkflowBuilder
-    from autobyteus.cli.workflow_tui.app import WorkflowApp
-    from autobyteus.tools import file_writer, file_reader, bash_executor
-    from autobyteus.agent.workspace import BaseAgentWorkspace, WorkspaceConfig
-    from autobyteus.tools.parameter_schema import ParameterSchema, ParameterDefinition, ParameterType
-except ImportError as e:
-    print(f"Error importing autobyteus components: {e}", file=sys.stderr)
-    sys.exit(1)
-
-# --- A simple, self-contained workspace for this example ---
-class SimpleLocalWorkspace(BaseAgentWorkspace):
-    """A minimal workspace for local file system access."""
-
-    def __init__(self, config: WorkspaceConfig):
-        super().__init__(config)
-        self.root_path: str = config.get("root_path")
-        if not self.root_path:
-            raise ValueError("SimpleLocalWorkspace requires a 'root_path' in its config.")
-
-    def get_base_path(self) -> str:
-        return self.root_path
-
-    @classmethod
-    def get_workspace_type_name(cls) -> str:
-        return "simple_local_workspace_for_review"
-
-    @classmethod
-    def get_description(cls) -> str:
-        return "A basic workspace for local file access for the code review workflow."
-
-    @classmethod
-    def get_config_schema(cls) -> ParameterSchema:
-        schema = ParameterSchema()
-        schema.add_parameter(ParameterDefinition(
-            name="root_path",
-            param_type=ParameterType.STRING,
-            description="The absolute local file path for the workspace root.",
-            required=True
-        ))
-        return schema
-
-
-# --- Logging Setup ---
-def setup_file_logging() -> Path:
-    log_dir = PACKAGE_ROOT / "logs"
-    log_dir.mkdir(exist_ok=True)
-    log_file_path = log_dir / "code_review_workflow_tui_app.log"
-    logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s] %(name)s: %(message)s", filename=log_file_path, filemode="w")
-    logging.getLogger("asyncio").setLevel(logging.WARNING)
-    logging.getLogger("textual").setLevel(logging.WARNING)
-    return log_file_path
-
-def create_code_review_workflow(
-    coordinator_model: str, 
-    engineer_model: str, 
-    reviewer_model: str, 
-    test_writer_model: str,
-    tester_model: str,
-    workspace: BaseAgentWorkspace,
-    use_xml_tool_format: bool = True
-):
-    """Creates the code review workflow."""
-    
-    # --- AGENT CONFIGURATIONS ---
-
-    # Coordinator Agent
-    coordinator_config = AgentConfig(
-        name="ProjectManager", role="Coordinator", description="Manages the development process, assigning tasks to the team.",
-        llm_instance=default_llm_factory.create_llm(model_identifier=coordinator_model),
-        system_prompt=(
-            "You are the project manager for a software team. Your role is to manage a strict, sequential code development, review, and testing process. Your team consists of a SoftwareEngineer, a CodeReviewer, a TestWriter, and a Tester.\n\n"
-            "### Your Workflow\n"
-            "You must follow this workflow precisely:\n"
-            "1.  **Delegate to Engineer:** Receive a request from the user to write code to a specific filename. Instruct the `SoftwareEngineer` to write the code and save it.\n"
-            "2.  **Delegate to Reviewer:** After the engineer confirms completion, instruct the `CodeReviewer` to review the code. You must provide the filename to the reviewer.\n"
-            "3.  **Delegate to Test Writer:** After the review is complete, instruct the `TestWriter` to write pytest tests for the code. Provide the original source filename and tell them to save the tests in a new file, like `test_FILENAME.py`.\n"
-            "4.  **Delegate to Tester:** After the tests are written, instruct the `Tester` to run the tests. You must provide the filename of the test file.\n"
-            "5.  **Report to User:** Once you receive the test results, present the final status (code written, reviewed, and tests passed/failed) to the user.\n\n"
-            "**CRITICAL RULE:** This is a sequential process. You must wait for one agent to finish before contacting the next. You are the central point of communication.\n\n"
-            "{{tools}}"
-        ),
-        use_xml_tool_format=use_xml_tool_format
-    )
-
-    # Software Engineer Agent
-    engineer_config = AgentConfig(
-        name="SoftwareEngineer", role="Developer", description="Writes Python code based on instructions and saves it to a file.",
-        llm_instance=default_llm_factory.create_llm(model_identifier=engineer_model),
-        system_prompt=(
-            "You are a skilled Python software engineer. You receive tasks from your ProjectManager. "
-            "Your job is to write high-quality Python code to fulfill the request. "
-            "After writing the code, you MUST save it to the specified filename using the `FileWriter` tool. "
-            "Confirm completion once the file is saved.\n\n{{tools}}"
-        ),
-        tools=[file_writer],
-        workspace=workspace,
-        use_xml_tool_format=use_xml_tool_format
-    )
-    
-    # Code Reviewer Agent
-    reviewer_config = AgentConfig(
-        name="CodeReviewer", role="Senior Developer", description="Reads and reviews Python code from files for quality and correctness.",
-        llm_instance=default_llm_factory.create_llm(model_identifier=reviewer_model),
-        system_prompt=(
-            "You are a senior software engineer acting as a code reviewer. You will be given a file path to review. "
-            "You MUST use the `FileReader` tool to read the code from the file. "
-            "After reading the code, provide a constructive review, identifying any potential bugs, style issues, or areas for improvement.\n\n{{tools}}"
-        ),
-        tools=[file_reader],
-        workspace=workspace,
-        use_xml_tool_format=use_xml_tool_format
-    )
-
-    # Test Writer Agent
-    test_writer_config = AgentConfig(
-        name="TestWriter", role="QA Engineer", description="Writes pytest tests for Python code.",
-        llm_instance=default_llm_factory.create_llm(model_identifier=test_writer_model),
-        system_prompt=(
-            "You are a QA engineer specializing in testing. You will be given the path to a Python source file. "
-            "Your task is to read that file, write comprehensive tests for it using the `pytest` framework, and save the tests to a new file. "
-            "The test filename MUST start with `test_`. For example, if you are testing `code.py`, you should save the tests in `test_code.py`.\n\n{{tools}}"
-        ),
-        tools=[file_reader, file_writer],
-        workspace=workspace,
-        use_xml_tool_format=use_xml_tool_format
-    )
-
-    # Tester Agent
-    tester_config = AgentConfig(
-        name="Tester", role="QA Automation", description="Executes pytest tests and reports results.",
-        llm_instance=default_llm_factory.create_llm(model_identifier=tester_model),
-        system_prompt=(
-            "You are a QA automation specialist. Your job is to run tests. You will be given a test file to execute. "
-            "You MUST use the `BashExecutor` tool to run the command `pytest` on the given test file. "
-            "Report the full output from the command back to the Project Manager.\n\n{{tools}}"
-        ),
-        tools=[bash_executor],
-        workspace=workspace,
-        use_xml_tool_format=use_xml_tool_format
-    )
-
-
-    # --- BUILD THE WORKFLOW ---
-    
-    code_review_workflow = (
-        WorkflowBuilder(name="SoftwareDevWorkflow", description="A workflow for writing, reviewing, and testing code.")
-        .set_coordinator(coordinator_config)
-        .add_agent_node(engineer_config)
-        .add_agent_node(reviewer_config)
-        .add_agent_node(test_writer_config)
-        .add_agent_node(tester_config)
-        .build()
-    )
-
-    return code_review_workflow
-
-async def main(args: argparse.Namespace, log_file: Path):
-    """Main async function to create the workflow and run the TUI app."""
-    print("Setting up software development workflow...")
-    print(f"--> Logs will be written to: {log_file.resolve()}")
-
-    workspace_path = Path(args.output_dir).resolve()
-    workspace_path.mkdir(parents=True, exist_ok=True)
-    print(f"--> Agent workspace (output directory) is set to: {workspace_path}")
-    
-    workspace_config = WorkspaceConfig(params={"root_path": str(workspace_path)})
-    workspace = SimpleLocalWorkspace(config=workspace_config)
-
-    # Resolve models
-    coordinator_model = args.coordinator_model or args.llm_model
-    engineer_model = args.engineer_model or args.llm_model
-    reviewer_model = args.reviewer_model or args.llm_model
-    test_writer_model = args.test_writer_model or args.llm_model
-    tester_model = args.tester_model or args.llm_model
-    
-    print(f"--> Coordinator Model: {coordinator_model}")
-    print(f"--> Engineer Model: {engineer_model}")
-    print(f"--> Reviewer Model: {reviewer_model}")
-    print(f"--> Test Writer Model: {test_writer_model}")
-    print(f"--> Tester Model: {tester_model}")
-
-    use_xml_tool_format = not args.no_xml_tools
-    print(f"--> Using XML Tool Format: {use_xml_tool_format}")
-
-    try:
-        workflow = create_code_review_workflow(
-            coordinator_model=coordinator_model,
-            engineer_model=engineer_model,
-            reviewer_model=reviewer_model,
-            test_writer_model=test_writer_model,
-            tester_model=tester_model,
-            workspace=workspace,
-            use_xml_tool_format=use_xml_tool_format
-        )
-        app = WorkflowApp(workflow=workflow)
-        await app.run_async()
-    except Exception as e:
-        logging.critical(f"Failed to create or run workflow TUI: {e}", exc_info=True)
-        print(f"\nCRITICAL ERROR: {e}\nCheck log file for details: {log_file.resolve()}")
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser(
-        description="Run a software development workflow with a Textual TUI.",
-        formatter_class=argparse.RawTextHelpFormatter
-    )
-    parser.add_argument("--llm-model", type=str, default="kimi-latest", help="The default LLM model for all agents.")
-    parser.add_argument("--coordinator-model", type=str, help="Specific LLM model for the ProjectManager. Defaults to --llm-model.")
-    parser.add_argument("--engineer-model", type=str, help="Specific LLM model for the SoftwareEngineer. Defaults to --llm-model.")
-    parser.add_argument("--reviewer-model", type=str, help="Specific LLM model for the CodeReviewer. Defaults to --llm-model.")
-    parser.add_argument("--test-writer-model", type=str, help="Specific LLM model for the TestWriter. Defaults to --llm-model.")
-    parser.add_argument("--tester-model", type=str, help="Specific LLM model for the Tester. Defaults to --llm-model.")
-    parser.add_argument("--output-dir", type=str, default="./code_review_output", help="Directory for the shared workspace.")
-    parser.add_argument("--no-xml-tools", action="store_true", help="Disable XML-based tool formatting.")
-    parser.add_argument("--help-models", action="store_true", help="Display available LLM models and exit.")
-    
-    if "--help-models" in sys.argv:
-        try:
-            LLMFactory.ensure_initialized()
-            print("Available LLM Models (you can use either name or value with model arguments):")
-            all_models = sorted(list(LLMModel), key=lambda m: m.name)
-            if not all_models:
-                print("  No models found.")
-            for model in all_models:
-                print(f"  - Name: {model.name:<35} Value: {model.value}")
-        except Exception as e:
-            print(f"Error listing models: {e}")
-        sys.exit(0)
-
-    parsed_args = parser.parse_args()
-
-    log_file_path = setup_file_logging()
-    try:
-        asyncio.run(main(parsed_args, log_file_path))
-    except KeyboardInterrupt:
-        print("\nExiting application.")
-    except Exception as e:
-        logging.critical(f"Top-level application error: {e}", exc_info=True)
-        print(f"\nUNHANDLED ERROR: {e}\nCheck log file for details: {log_file_path.resolve()}")
-