lzy/sci-gui-agent-benchmark
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat&fix: enhance task status handling and update logging configuration

Browse Source
This commit is contained in:
adlsdztony
2025-06-05 09:33:36 +00:00
parent 2bfb4af8b5
commit 3b1540ed23
6 changed files with 17 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
mm_agents/openai_cua_agent.py
View File

@@ -1,28 +1,15 @@
import base64
import json
import logging
import os
import re
import tempfile
import time
import xml.etree.ElementTree as ET
from http import HTTPStatus
from io import BytesIO
from typing import Dict, List
import backoff
import dashscope
import google.generativeai as genai
import openai
import requests
import tiktoken
from PIL import Image
from google.api_core.exceptions import InvalidArgument, ResourceExhausted, InternalServerError, BadRequest
from groq import Groq
from requests.exceptions import SSLError
from typing import Any, Optional, Union, Tuple
from mm_agents.accessibility_tree_wrap.heuristic_retrieve import filter_nodes, draw_bounding_boxes
from mm_agents.prompts import SYS_PROMPT_IN_SCREENSHOT_OUT_CODE, SYS_PROMPT_IN_SCREENSHOT_OUT_ACTION, \
SYS_PROMPT_IN_A11Y_OUT_CODE, SYS_PROMPT_IN_A11Y_OUT_ACTION, \
SYS_PROMPT_IN_BOTH_OUT_CODE, SYS_PROMPT_IN_BOTH_OUT_ACTION, \

4
monitor/main.py
View File

@@ -122,6 +122,8 @@ def get_task_status(task_type, task_id):
status = "Error"
elif log_data.get("exit_condition") and "message_exit: True" in log_data.get("exit_condition", ""):
status = "Done (Message Exit)"
elif log_data.get("exit_condition") and "thought_exit: True" in log_data.get("exit_condition", ""):
status = "Done (Thought Exit)"
elif len(steps) >= MAX_STEPS:
status = "Done (Max Steps)"
else:
@@ -231,6 +233,8 @@ def get_task_status_brief(task_type, task_id):
log_tail = result.stdout
if "message_exit: True" in log_tail:
status = "Done (Message Exit)"
elif "thought_exit: True" in log_tail:
status = "Done (Thought Exit)"
except:
pass

8
monitor/static/index.js
View File

@@ -85,6 +85,7 @@ function updateTaskStatus(data) {
case 'Done':
case 'Done (Message Exit)':
case 'Done (Max Steps)':
case 'Done (Thought Exit)':
statusClass = 'status-completed';
statusIcon = 'fa-check-circle';
break;
@@ -183,7 +184,7 @@ function updateStatistics(data) {
tasks.forEach(task => {
if (task.status.status === 'Running' || task.status.status === 'Preparing' || task.status.status === 'Initializing') {
activeTasks++;
} else if (task.status.status === 'Done' || task.status.status === 'Done (Message Exit)' || task.status.status === 'Done (Max Steps)') {
} else if (task.status.status === 'Done' || task.status.status === 'Done (Message Exit)' || task.status.status === 'Done (Max Steps)' || task.status.status === 'Done (Thought Exit)') {
completedTasks++;
// Calculate score if task is completed
if (task.status.result) {
@@ -241,7 +242,7 @@ function renderTasks(data) {
if (currentFilter === 'active') {
filteredTasks = tasks.filter(task => ['Running', 'Preparing', 'Initializing'].includes(task.status.status));
} else if (currentFilter === 'completed') {
filteredTasks = tasks.filter(task => task.status.status === 'Done' || task.status.status === 'Done (Message Exit)' || task.status.status === 'Done (Max Steps)');
filteredTasks = tasks.filter(task => task.status.status === 'Done' || task.status.status === 'Done (Message Exit)' || task.status.status === 'Done (Max Steps)'|| task.status.status === 'Done (Thought Exit)');
} else if (currentFilter === 'error') {
filteredTasks = tasks.filter(task => task.status.status === 'Error');
}
@@ -264,7 +265,7 @@ function renderTasks(data) {
tasks.forEach(task => {
if (task.status.status === 'Running' || task.status.status === 'Preparing' || task.status.status === 'Initializing') {
runningCount++;
} else if (task.status.status === 'Done' || task.status.status === 'Done (Message Exit)' || task.status.status === 'Done (Max Steps)') {
} else if (task.status.status === 'Done' || task.status.status === 'Done (Message Exit)' || task.status.status === 'Done (Max Steps)' || task.status.status === 'Done (Thought Exit)') {
completedCount++;
} else if (task.status.status === 'Error') {
errorCount++;
@@ -346,6 +347,7 @@ function renderTasks(data) {
case 'Done':
case 'Done (Message Exit)':
case 'Done (Max Steps)':
case 'Done (Thought Exit)':
statusClass = 'status-completed';
statusIcon = 'fa-check-circle';
break;

2
monitor/static/task_detail.css
View File

@@ -173,7 +173,7 @@ pre {
.status-not-started { background: linear-gradient(135deg, #f0f0f0, #e6e6e6); color: #555; }
.status-preparing, .status-initializing { background: linear-gradient(135deg, #fff7e0, #ffe8a3); color: #8a6d00; }
.status-running { background: linear-gradient(135deg, #e3f2fd, #bbdefb); color: #0d47a1; }
.status-done, .status-done-message-exit, .status-done-max-steps { background: linear-gradient(135deg, #e8f5e9, #c8e6c9); color: #1b5e20; }
.status-done, .status-done-message-exit, .status-done-max-steps, .status-done-thought-exit { background: linear-gradient(135deg, #e8f5e9, #c8e6c9); color: #1b5e20; }
.status-error { background: linear-gradient(135deg, #ffebee, #ffcdd2); color: #b71c1c; }
.step-intent {

5
monitor/templates/task_detail.html
View File

@@ -41,6 +41,11 @@
<i class="fas fa-question-circle"></i>
<span class="tooltip-text">Maximum steps reached, task completed</span>
</span>
{% elif task_status.status == 'Done (Thought Exit)' %}
<span class="tooltip">
<i class="fas fa-question-circle"></i>
<span class="tooltip-text">Task completed with a thought exit condition</span>
</span>
{% endif %}
</dd>
<dt>Current Step</dt>

16
run_multienv_openaicua.py
View File

@@ -33,41 +33,29 @@ if os.path.exists(".env"):
# Logger Configs {{{ #
logger = logging.getLogger()
logger.setLevel(logging.DEBUG)
logger.setLevel(logging.INFO)
datetime_str: str = datetime.datetime.now().strftime("%Y%m%d@%H%M%S")
file_handler = logging.FileHandler(
os.path.join("logs", "normal-{:}.log".format(datetime_str)), encoding="utf-8"
)
debug_handler = logging.FileHandler(
os.path.join("logs", "debug-{:}.log".format(datetime_str)), encoding="utf-8"
)
stdout_handler = logging.StreamHandler(sys.stdout)
sdebug_handler = logging.FileHandler(
os.path.join("logs", "sdebug-{:}.log".format(datetime_str)), encoding="utf-8"
)
file_handler.setLevel(logging.INFO)
debug_handler.setLevel(logging.DEBUG)
stdout_handler.setLevel(logging.INFO)
sdebug_handler.setLevel(logging.DEBUG)
formatter = logging.Formatter(
fmt="\x1b[1;33m[%(asctime)s \x1b[31m%(levelname)s \x1b[32m%(module)s/%(lineno)d-%(processName)s\x1b[1;33m] \x1b[0m%(message)s"
)
file_handler.setFormatter(formatter)
debug_handler.setFormatter(formatter)
stdout_handler.setFormatter(formatter)
sdebug_handler.setFormatter(formatter)
stdout_handler.addFilter(logging.Filter("desktopenv"))
sdebug_handler.addFilter(logging.Filter("desktopenv"))
logger.addHandler(file_handler)
logger.addHandler(debug_handler)
logger.addHandler(stdout_handler)
logger.addHandler(sdebug_handler)
# }}} Logger Configs #
logger = logging.getLogger("desktopenv.experiment")