Merge pull request #26 from panregedit/feature/v2.0.0/framework_refactor

refactor the framework

.gitignore

@@ -150,4 +150,4 @@ tests
 data/
 tests/mathvista
 running_logs/
-.db
+*.db

README.md

@@ -74,7 +74,7 @@ For more details, check out our paper **[OmAgent: A Multi-modal Agent Framework
         logging.init_logger("omagent", "omagent", level="INFO")
         registry.import_module(project_root=Path(__file__).parent, custom=["./engine"])
         bot_builder = Builder.from_file("workflows/general") # General task processing workflow configuration directory
-        input = DnCInterface(bot_id="1", task=AgentTask(id=0, task=task))
+        input = BaseWorkflowContext(bot_id="1", task=AgentTask(id=0, task=task))
 
         bot_builder.run_bot(input)
         return input.last_output
@@ -145,7 +145,7 @@ For more details, check out our paper **[OmAgent: A Multi-modal Agent Framework
         logging.init_logger("omagent", "omagent", level="INFO")
         registry.import_module(project_root=Path(__file__).parent, custom=["./engine"])
         bot_builder = Builder.from_file("workflows/video_understanding") # Video understanding task workflow configuration directory
-        input = DnCInterface(bot_id="1", task=AgentTask(id=0, task=task))
+        input = BaseWorkflowContext(bot_id="1", task=AgentTask(id=0, task=task))
 
         bot_builder.run_bot(input)
         return input.last_output

README_FR.md

@@ -85,7 +85,7 @@ Pour plus d'informations, consultez notre article : **[OmAgent : Un cadre d'agen
        logging.init_logger("omagent", "omagent", level="INFO")
        registry.import_module(project_root=Path(__file__).parent, custom=["./engine"])
        bot_builder = Builder.from_file("workflows/general")
-       input = DnCInterface(bot_id="1", task=AgentTask(id=0, task=task))
+       input = BaseWorkflowContext(bot_id="1", task=AgentTask(id=0, task=task))
 
        bot_builder.run_bot(input)
        return input.last_output

README_JP.md

@@ -75,7 +75,7 @@ OmAgentは3つの主要なコンポーネントで構成されています：
         logging.init_logger("omagent", "omagent", level="INFO")
         registry.import_module(project_root=Path(__file__).parent, custom=["./engine"])
         bot_builder = Builder.from_file("workflows/general") # 一般タスク処理ワークフロー設定ディレクトリ
-        input = DnCInterface(bot_id="1", task=AgentTask(id=0, task=task))
+        input = BaseWorkflowContext(bot_id="1", task=AgentTask(id=0, task=task))
 
         bot_builder.run_bot(input)
         return input.last_output
@@ -146,7 +146,7 @@ OmAgentは3つの主要なコンポーネントで構成されています：
         logging.init_logger("omagent", "omagent", level="INFO")
         registry.import_module(project_root=Path(__file__).parent, custom=["./engine"])
         bot_builder = Builder.from_file("workflows/video_understanding") # Video understanding task workflow configuration directory
-        input = DnCInterface(bot_id="1", task=AgentTask(id=0, task=task))
+        input = BaseWorkflowContext(bot_id="1", task=AgentTask(id=0, task=task))
 
         bot_builder.run_bot(input)
         return input.last_output

README_ZH.md

@@ -74,7 +74,7 @@ OmAgent包括三个核心组成部分:
         logging.init_logger("omagent", "omagent", level="INFO")
         registry.import_module(project_root=Path(__file__).parent, custom=["./engine"])
         bot_builder = Builder.from_file("workflows/general") # 通用任务处理workflow配置目录
-        input = DnCInterface(bot_id="1", task=AgentTask(id=0, task=task))
+        input = BaseWorkflowContext(bot_id="1", task=AgentTask(id=0, task=task))
 
         bot_builder.run_bot(input)
         return input.last_output
@@ -146,7 +146,7 @@ OmAgent包括三个核心组成部分:
         logging.init_logger("omagent", "omagent", level="INFO")
         registry.import_module(project_root=Path(__file__).parent, custom=["./engine"])
         bot_builder = Builder.from_file("workflows/video_understanding") # 视频理解任务workflow配置目录
-        input = DnCInterface(bot_id="1", task=AgentTask(id=0, task=task))
+        input = BaseWorkflowContext(bot_id="1", task=AgentTask(id=0, task=task))
 
         bot_builder.run_bot(input)
         return input.last_output

engine/node/video_qa/conqueror/conqueror.py → examples/video_understanding/agent/conqueror/conqueror.py

@@ -4,13 +4,13 @@
 from typing import List, Tuple
 
 from colorama import Fore, Style
-from omagent_core.core.node.dnc import TaskConqueror
-from omagent_core.core.node.dnc.schemas import AgentTask, TaskStatus
-from omagent_core.core.prompt.prompt import PromptTemplate
-from omagent_core.core.tool_system.manager import ToolManager
-from omagent_core.handlers.data_handler.ltm import LTM
-from omagent_core.schemas.base import BaseInterface
+from omagent_core.advanced_components.node.conqueror.conqueror import TaskConqueror
+from omagent_core.engine.workflow.context import BaseWorkflowContext
+from omagent_core.models.llms.prompt import PromptTemplate
+from omagent_core.tool_system.manager import ToolManager
+from omagent_core.memories.ltms.ltm import LTM
 from omagent_core.utils.env import EnvVar
+from omagent_core.engine.task.agent_task import AgentTask, TaskStatus
 from omagent_core.utils.registry import registry
 from pydantic import Field
 from tenacity import (
@@ -45,7 +45,7 @@ class VideoConqueror(TaskConqueror):
         retry=retry_if_exception_message(message="LLM generation is not valid."),
         reraise=True,
     )
-    def _run(self, args: BaseInterface, ltm: LTM) -> Tuple[BaseInterface, str]:
+    def _run(self, args: BaseWorkflowContext, ltm: LTM) -> Tuple[BaseWorkflowContext, str]:
         task: AgentTask = args.task
         task.status = TaskStatus.RUNNING
 
@@ -197,7 +197,7 @@ def _run(self, args: BaseInterface, ltm: LTM) -> Tuple[BaseInterface, str]:
         retry=retry_if_exception_message(message="LLM generation is not valid."),
         reraise=True,
     )
-    async def _arun(self, args: BaseInterface, ltm: LTM) -> Tuple[BaseInterface, str]:
+    async def _arun(self, args: BaseWorkflowContext, ltm: LTM) -> Tuple[BaseWorkflowContext, str]:
         task: AgentTask = args.task
         task.status = TaskStatus.RUNNING
 

engine/node/video_qa/divider/divider.py → examples/video_understanding/agent/divider/divider.py

@@ -3,13 +3,13 @@
 from pathlib import Path
 from typing import List, Tuple
 
-from omagent_core.core.llm.base import BaseLLMBackend
-from omagent_core.core.node.base import BaseDecider
-from omagent_core.core.node.dnc.interface import DnCInterface
-from omagent_core.core.node.dnc.schemas import AgentTask
-from omagent_core.core.prompt.prompt import PromptTemplate
-from omagent_core.core.tool_system.manager import ToolManager
-from omagent_core.handlers.data_handler.ltm import LTM
+from omagent_core.models.llms.base import BaseLLMBackend
+from omagent_core.engine.node import BaseDecider
+from omagent_core.engine.workflow.context import BaseWorkflowContext
+from omagent_core.engine.task.agent_task import AgentTask
+from omagent_core.models.llms.prompt import PromptTemplate
+from omagent_core.tool_system.manager import ToolManager
+from omagent_core.memories.ltms.ltm import LTM
 from omagent_core.utils.env import EnvVar
 from omagent_core.utils.registry import registry
 from pydantic import Field
@@ -45,7 +45,7 @@ class VideoDivider(BaseLLMBackend, BaseDecider):
         retry=retry_if_exception_message(message="LLM generation is not valid."),
         reraise=True,
     )
-    def _run(self, args: DnCInterface, ltm: LTM) -> Tuple[DnCInterface, str]:
+    def _run(self, args: BaseWorkflowContext, ltm: LTM) -> Tuple[BaseWorkflowContext, str]:
         task: AgentTask = args.task
         if task.task_depth() >= EnvVar.MAX_TASK_DEPTH:
             args.last_output = "failed: Max subtask depth reached"
@@ -97,7 +97,7 @@ def _run(self, args: DnCInterface, ltm: LTM) -> Tuple[DnCInterface, str]:
         retry=retry_if_exception_message(message="LLM generation is not valid."),
         reraise=True,
     )
-    async def _arun(self, args: DnCInterface, ltm: LTM) -> Tuple[DnCInterface, str]:
+    async def _arun(self, args: BaseWorkflowContext, ltm: LTM) -> Tuple[BaseWorkflowContext, str]:
         task: AgentTask = args.task
         if task.task_depth() >= EnvVar.MAX_TASK_DEPTH:
             args.last_output = "failed: Max subtask depth reached"

engine/node/video_qa/qa.py → examples/video_understanding/agent/qa.py

@@ -3,11 +3,11 @@
 from pathlib import Path
 from typing import List
 
-from omagent_core.core.llm.base import BaseLLMBackend
-from omagent_core.core.node.base import BaseProcessor
-from omagent_core.core.node.dnc.interface import DnCInterface
-from omagent_core.core.prompt.prompt import PromptTemplate
-from omagent_core.handlers.data_handler.ltm import LTM
+from omagent_core.models.llms.base import BaseLLMBackend
+from omagent_core.engine.node import BaseProcessor
+from omagent_core.engine.workflow.context import BaseWorkflowContext
+from omagent_core.models.llms.prompt import PromptTemplate
+from omagent_core.memories.ltms.ltm import LTM
 from omagent_core.utils.registry import registry
 from pydantic import Field
 
@@ -27,7 +27,7 @@ class VideoQA(BaseProcessor, BaseLLMBackend):
         ]
     )
 
-    def _run(self, args: DnCInterface, ltm: LTM) -> DnCInterface:
+    def _run(self, args: BaseWorkflowContext, ltm: LTM) -> BaseWorkflowContext:
         video_md5 = args.kwargs.get("video_md5", None)
         self.stm.image_cache.clear()
         self.stm.former_results = {}

omagent-core/pyproject.toml

@@ -8,7 +8,7 @@ version = "0.0.1"
 authors = [
     { name = "panregedit", email = "[email protected]" },
 ]
-description = "Core package for OMbots"
+description = "Core package for OmAgent"
 readme = "README.md"
 requires-python = ">=3.10, <3.12"
 classifiers = [
@@ -40,4 +40,4 @@ dependencies = [
 ]
 
 [project.urls]
-"Homepage" = "https://git.linker.cc/research/misc/ombot_utils.git"
+"Homepage" = "https://github.com/om-ai-lab/OmAgent"

omagent-core/src/omagent_core/advanced_components/node/Infinite_loop/inf_loop.py

@@ -0,0 +1,12 @@
+from ....engine.node import BaseLoop, Node
+from ....memories.ltms.ltm import LTM
+from ....engine.workflow.context import BaseWorkflowContext
+from ....utils.registry import registry
+
+
+@registry.register_node()
+class InfLoop(BaseLoop):
+    loop_body: Node
+
+    def post_loop_exit(self, args: BaseWorkflowContext, ltm: LTM) -> bool:
+        return False

engine/node/conclude/conclude.py → omagent-core/src/omagent_core/advanced_components/node/conclude/conclude.py

@@ -1,12 +1,12 @@
 from pathlib import Path
 from typing import List
 
-from omagent_core.core.llm.base import BaseLLMBackend
-from omagent_core.core.node.base import BaseProcessor
-from omagent_core.core.node.dnc.interface import DnCInterface
-from omagent_core.core.prompt.prompt import PromptTemplate
-from omagent_core.handlers.data_handler.ltm import LTM
-from omagent_core.utils.registry import registry
+from ....models.llms.base import BaseLLMBackend
+from ....engine.node import BaseProcessor
+from ....engine.workflow.context import BaseWorkflowContext
+from ....models.llms.prompt import PromptTemplate
+from ....memories.ltms.ltm import LTM
+from ....utils.registry import registry
 from pydantic import Field
 
 CURRENT_PATH = root_path = Path(__file__).parents[0]
@@ -25,7 +25,7 @@ class Conclude(BaseLLMBackend, BaseProcessor):
         ]
     )
 
-    def _run(self, args: DnCInterface, ltm: LTM) -> DnCInterface:
+    def _run(self, args: BaseWorkflowContext, ltm: LTM) -> BaseWorkflowContext:
         chat_complete_res = self.simple_infer(
             task=args.task.find_root_task().task,
             result=args.last_output,
@@ -40,7 +40,7 @@ def _run(self, args: DnCInterface, ltm: LTM) -> DnCInterface:
             print(f"Usage of {key}: {value}")
         return args
 
-    async def _arun(self, args: DnCInterface, ltm: LTM) -> DnCInterface:
+    async def _arun(self, args: BaseWorkflowContext, ltm: LTM) -> BaseWorkflowContext:
         chat_complete_res = await self.simple_ainfer(
             task=args.task.find_root_task().task,
             result=args.last_output,

omagent-core/src/omagent_core/core/node/dnc/conqueror/conqueror.py → omagent-core/src/omagent_core/advanced_components/node/conqueror/conqueror.py

@@ -12,15 +12,15 @@
     stop_after_delay,
 )
 
-from .....handlers.data_handler.ltm import LTM
-from .....schemas.base import BaseInterface
-from .....utils.env import EnvVar
-from .....utils.registry import registry
-from ....llm.base import BaseLLMBackend
-from ....prompt.prompt import PromptTemplate
+from ....memories.ltms.ltm import LTM
+from ....engine.workflow.context import BaseWorkflowContext
+from ....utils.env import EnvVar
+from ....utils.registry import registry
+from ....models.llms.base import BaseLLMBackend
+from ....models.llms.prompt.prompt import PromptTemplate
 from ....tool_system.manager import ToolManager
-from ...base import BaseDecider
-from ..schemas import AgentTask, TaskStatus
+from ....engine.node.decider import BaseDecider
+from ....engine.task.agent_task import AgentTask, TaskStatus
 
 CURRENT_PATH = Path(__file__).parents[0]
 
@@ -47,7 +47,7 @@ class TaskConqueror(BaseLLMBackend, BaseDecider):
         retry=retry_if_exception_message(message="LLM generation is not valid."),
         reraise=True,
     )
-    def _run(self, args: BaseInterface, ltm: LTM) -> Tuple[BaseInterface, str]:
+    def _run(self, args: BaseWorkflowContext, ltm: LTM) -> Tuple[BaseWorkflowContext, str]:
         task: AgentTask = args.task
         task.status = TaskStatus.RUNNING
 
@@ -190,7 +190,7 @@ def _run(self, args: BaseInterface, ltm: LTM) -> Tuple[BaseInterface, str]:
         retry=retry_if_exception_message(message="LLM generation is not valid."),
         reraise=True,
     )
-    async def _arun(self, args: BaseInterface, ltm: LTM) -> Tuple[BaseInterface, str]:
+    async def _arun(self, args: BaseWorkflowContext, ltm: LTM) -> Tuple[BaseWorkflowContext, str]:
         task: AgentTask = args.task
         task.status = TaskStatus.RUNNING
 

omagent-core/src/omagent_core/core/node/dnc/divider/divider.py → omagent-core/src/omagent_core/advanced_components/node/divider/divider.py

@@ -11,15 +11,15 @@
     stop_after_delay,
 )
 
-from .....handlers.data_handler.ltm import LTM
-from .....utils.env import EnvVar
-from .....utils.registry import registry
-from ....llm.base import BaseLLMBackend
-from ....prompt.prompt import PromptTemplate
+from ....memories.ltms.ltm import LTM
+from ....utils.env import EnvVar
+from ....utils.registry import registry
+from ....models.llms.base import BaseLLMBackend
+from ....models.llms.prompt.prompt import PromptTemplate
 from ....tool_system.manager import ToolManager
-from ...base import BaseDecider
-from ..interface import DnCInterface
-from ..schemas import AgentTask
+from ....engine.node.decider import BaseDecider
+from ....engine.workflow.context import BaseWorkflowContext
+from ....engine.task.agent_task import AgentTask
 
 CURRENT_PATH = Path(__file__).parents[0]
 
@@ -46,7 +46,7 @@ class TaskDivider(BaseLLMBackend, BaseDecider):
         retry=retry_if_exception_message(message="LLM generation is not valid."),
         reraise=True,
     )
-    def _run(self, args: DnCInterface, ltm: LTM) -> Tuple[DnCInterface, str]:
+    def _run(self, args: BaseWorkflowContext, ltm: LTM) -> Tuple[BaseWorkflowContext, str]:
         task: AgentTask = args.task
         if task.task_depth() >= EnvVar.MAX_TASK_DEPTH:
             args.last_output = "failed: Max subtask depth reached"
@@ -97,7 +97,7 @@ def _run(self, args: DnCInterface, ltm: LTM) -> Tuple[DnCInterface, str]:
         retry=retry_if_exception_message(message="LLM generation is not valid."),
         reraise=True,
     )
-    async def _arun(self, args: DnCInterface, ltm: LTM) -> Tuple[DnCInterface, str]:
+    async def _arun(self, args: BaseWorkflowContext, ltm: LTM) -> Tuple[BaseWorkflowContext, str]:
         task: AgentTask = args.task
         if task.task_depth() >= EnvVar.MAX_TASK_DEPTH:
             args.last_output = "failed: Max subtask depth reached"

engine/loop/dnc_loop.py → omagent-core/src/omagent_core/advanced_components/node/dnc_loop/dnc_loop.py

@@ -1,14 +1,14 @@
-from omagent_core.core.node.base import BaseLoop, Node
-from omagent_core.core.node.dnc import DnCInterface
-from omagent_core.handlers.data_handler.ltm import LTM
-from omagent_core.utils.registry import registry
+from ....engine.node import BaseLoop, Node
+from ....memories.ltms.ltm import LTM
+from ....engine.workflow.context import BaseWorkflowContext
+from ....utils.registry import registry
 
 
 @registry.register_node()
 class DnCLoop(BaseLoop):
     loop_body: Node
 
-    def post_loop_exit(self, args: DnCInterface, ltm: LTM) -> bool:
+    def post_loop_exit(self, args: BaseWorkflowContext, ltm: LTM) -> bool:
         # self.stm.image_cache.clear()
         if args.task.status == "failed":
             return True