simple draft

gui/main.py

@@ -12,102 +12,77 @@
 # limitations under the License.
 # =========== Copyright 2024 @ CAMEL-AI.org. All Rights Reserved. ===========
 import warnings
+from pathlib import Path
+from uuid import uuid4
 
-from crab import (
-    BenchmarkConfig,
-    create_benchmark,
-)
-from crab.actions.crab_actions import complete
-from crab.actions.visual_prompt_actions import (
-    get_elements_prompt,
-    groundingdino_easyocr,
-)
-from crab.environments.macos import mac_env
-from gui.envs import ANDROID_ENV, UBUNTU_ENV, WINDOWS_ENV
-from gui.host_os import HostOS
+import customtkinter as ctk
+
+from crab import Experiment
+from crab.agents.backend_models import OpenAIModel
+from crab.agents.policies import SingleAgentPolicy
+from gui.utils import get_benchmark
 
 warnings.filterwarnings("ignore")
 
 
-def check_host_os() -> HostOS:
-    return HostOS.WINDOWS
-
-
-def get_benchmark(env: str, ubuntu_url: str):
-    ubuntu_tool = {
-        "screenshot": groundingdino_easyocr(font_size=16) >> get_elements_prompt
-    }
-    android_tool = {
-        "screenshot": groundingdino_easyocr(font_size=40) >> get_elements_prompt
-    }
-    mac_tool = {
-        "screenshot": groundingdino_easyocr(font_size=24) >> get_elements_prompt
-    }
-
-    if env == "ubuntu":
-        prompting_tools = {"ubuntu": ubuntu_tool}
-        benchmark_config = BenchmarkConfig(
-            name="ubuntu_benchmark",
-            tasks=[],
-            environments=[UBUNTU_ENV],
-            prompting_tools=prompting_tools,
-            root_action_space=[complete],
-            multienv=True,
-        )
-    elif env == "android":
-        prompting_tools = {"android": android_tool}
-        benchmark_config = BenchmarkConfig(
-            name="android_benchmark",
-            tasks=[],
-            environments=[ANDROID_ENV],
-            prompting_tools=prompting_tools,
-            root_action_space=[complete],
-            multienv=True,
-        )
-    elif env == "cross":
-        prompting_tools = {
-            "android": android_tool,
-            "ubuntu": ubuntu_tool,
-        }
-        benchmark_config = BenchmarkConfig(
-            name="ubuntu_android_benchmark",
-            tasks=[],
-            environments=[UBUNTU_ENV, ANDROID_ENV],
-            prompting_tools=prompting_tools,
-            root_action_space=[complete],
-            multienv=True,
-        )
-    elif env == "mac":
-        prompting_tools = {"macos": mac_tool}
-        benchmark_config = BenchmarkConfig(
-            name="mac_benchmark",
-            tasks=[],
-            environments=[mac_env, ANDROID_ENV],
-            prompting_tools=prompting_tools,
-            root_action_space=[complete],
-            multienv=True,
-        )
-    elif env == "windows":
-        prompting_tools = {"windows": ubuntu_tool}
-        benchmark_config = BenchmarkConfig(
-            name="windows_benchmark",
-            tasks=[],
-            environments=[WINDOWS_ENV],
-            prompting_tools=prompting_tools,
-            root_action_space=[complete],
-            multienv=True,
-        )
-    else:
-        raise ValueError("Env not support")
+def assign_task():
+    task_description = input_entry.get()
+    input_entry.delete(0, "end")
+    display_message(task_description)
 
-    benchmark_config.step_limit = 15
-    return create_benchmark(benchmark_config)
+    task_id = str(uuid4())
+    benchmark = get_benchmark(task_id, task_description)
+    expeirment = Experiment(
+        benchmark=benchmark,
+        task_id=task_id,
+        agent_policy=agent_policy,
+        log_dir=log_dir,
+    )
+    expeirment.start_benchmark()
 
 
-def main():
-    host_os = check_host_os()
-    print(f"Host OS: {host_os}")
+def display_message(message, sender="user"):
+    chat_display.configure(state="normal")
+    if sender == "user":
+        chat_display.insert("end", f"User: {message}\n", "user")
+    else:
+        chat_display.insert("end", f"AI: {message}\n", "ai")
+    chat_display.tag_config("user", justify="left", foreground="blue")
+    chat_display.tag_config("ai", justify="right", foreground="green")
+    chat_display.configure(state="disabled")
+    chat_display.see("end")
+    app.update_idletasks()
 
 
 if __name__ == "__main__":
-    main()
+    model = OpenAIModel(model="gpt-4o", history_messages_len=2)
+    agent_policy = SingleAgentPolicy(model_backend=model)
+    log_dir = (Path(__file__).parent / "logs").resolve()
+
+    ctk.set_appearance_mode("System")
+    ctk.set_default_color_theme("blue")
+
+    app = ctk.CTk()
+    app.title("CRAB")
+    app.geometry("400x500")
+
+    chat_display_frame = ctk.CTkFrame(app, width=380, height=380)
+    chat_display_frame.pack(pady=10)
+    chat_display = ctk.CTkTextbox(
+        chat_display_frame, width=380, height=380, state="disabled"
+    )
+    chat_display.pack()
+
+    # Frame for input and send button side-by-side
+    input_frame = ctk.CTkFrame(app)
+    input_frame.pack(pady=10, padx=10, fill="x")
+
+    # Entry widget for user input
+    input_entry = ctk.CTkEntry(
+        input_frame, placeholder_text="Type your message here..."
+    )
+    input_entry.pack(side="left", fill="x", expand=True, padx=(0, 10))
+
+    send_button = ctk.CTkButton(input_frame, text="Send", command=assign_task)
+    send_button.pack(side="right")
+    app.mainloop()

gui/utils.py

@@ -0,0 +1,110 @@
+# =========== Copyright 2024 @ CAMEL-AI.org. All Rights Reserved. ===========
+# Licensed under the Apache License, Version 2.0 (the “License”);
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an “AS IS” BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# =========== Copyright 2024 @ CAMEL-AI.org. All Rights Reserved. ===========
+
+from crab import (
+    Benchmark,
+    BenchmarkConfig,
+    Task,
+    create_benchmark,
+    evaluator,
+)
+from crab.actions.crab_actions import complete
+from crab.actions.visual_prompt_actions import (
+    get_elements_prompt,
+    groundingdino_easyocr,
+)
+from gui.envs import MAC_ENV, UBUNTU_ENV, WINDOWS_ENV
+from gui.host_os import HostOS
+
+
+def check_host_os() -> HostOS:
+    # TODO: Check the host OS and return the corresponding HostOS enum
+    return HostOS.WINDOWS
+
+
+@evaluator(env_name="ubuntu")
+def empty_evaluator_linux() -> bool:
+    return False
+
+
+@evaluator(env_name="mac")
+def empty_evaluator_mac() -> bool:
+    return False
+
+
+@evaluator(env_name="windows")
+def empty_evaluator_windows() -> bool:
+    return False
+
+
+def get_benchmark(task_id: str, task_description: str) -> Benchmark:
+    host_os = check_host_os()
+
+    ubuntu_tool = {
+        "screenshot": groundingdino_easyocr(font_size=16) >> get_elements_prompt
+    }
+    mac_tool = {
+        "screenshot": groundingdino_easyocr(font_size=24) >> get_elements_prompt
+    }
+
+    if host_os == HostOS.LINUX:
+        prompting_tools = {"ubuntu": ubuntu_tool}
+        benchmark_config = BenchmarkConfig(
+            name="ubuntu_benchmark",
+            tasks=[
+                Task(
+                    id=task_id,
+                    description=task_description,
+                    evaluator=empty_evaluator_linux,
+                )
+            ],
+            environments=[UBUNTU_ENV],
+            prompting_tools=prompting_tools,
+            root_action_space=[complete],
+        )
+    elif host_os == HostOS.MAC:
+        prompting_tools = {"macos": mac_tool}
+        benchmark_config = BenchmarkConfig(
+            name="mac_benchmark",
+            tasks=[
+                Task(
+                    id=task_id,
+                    description=task_description,
+                    evaluator=empty_evaluator_mac,
+                )
+            ],
+            environments=[MAC_ENV],
+            prompting_tools=prompting_tools,
+            root_action_space=[complete],
+        )
+    elif host_os == HostOS.WINDOWS:
+        prompting_tools = {"windows": ubuntu_tool}
+        benchmark_config = BenchmarkConfig(
+            name="windows_benchmark",
+            tasks=[
+                Task(
+                    id=task_id,
+                    description=task_description,
+                    evaluator=empty_evaluator_windows,
+                )
+            ],
+            environments=[WINDOWS_ENV],
+            prompting_tools=prompting_tools,
+            root_action_space=[complete],
+        )
+    else:
+        raise ValueError(f"Host OS {host_os} is not supported")
+
+    benchmark_config.step_limit = 15
+    return create_benchmark(benchmark_config)