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

update multi apps

Browse Source
This commit is contained in:
tsuky_chen
2024-03-05 22:46:56 +08:00
parent 0ad9c5b3a4
commit 69ef653a7c
9 changed files with 397 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
desktop_env/evaluators/metrics/__init__.py
View File

@@ -18,7 +18,8 @@ from .chrome import (
is_expected_search_query,
is_expected_active_tab,
is_expected_url_pattern_match,
is_added_to_steam_cart
is_added_to_steam_cart,
compare_pdf_images
)
from .docs import (
compare_font_names,
@@ -45,7 +46,8 @@ from .docs import (
is_first_line_centered,
check_file_exists,
check_tabstops,
compare_contains_image
compare_contains_image,
compare_docx_images
)
from .general import (
check_csv,
@@ -125,7 +127,8 @@ from .vscode import (
check_json_settings,
check_json_keybindings,
check_python_file_by_test_suite,
check_python_file_by_gold_file
check_python_file_by_gold_file,
compare_zip_files
)

33
desktop_env/evaluators/metrics/chrome.py
View File

@@ -129,6 +129,39 @@ def compare_pdfs(pdf1_path: Union[str, List[str]], pdf2_path: Union[str, List[st
logger.info(f"[ERROR]: unexpected error occurred when comparing PDF files: {e}")
return score / len(pdf2_path)
import fitz
from PIL import Image
from io import BytesIO
def compare_pdf_images(pdf1_path: str, pdf2_path: str, **kwargs) -> float:
def extract_images_from_pdf(pdf_path):
pdf_document = fitz.open(pdf_path)
images = []
for page_number in range(pdf_document.page_count):
page = pdf_document[page_number]
image_list = page.get_images(full=True)
for img_index, img_info in enumerate(image_list):
base_image = pdf_document.extract_image(img_index)
image_bytes = base_image["image"]
images.append(BytesIO(image_bytes))
return images
images1 = extract_images_from_pdf(pdf1_path)
images2 = extract_images_from_pdf(pdf2_path)
if len(images1) != len(images2):
return 0.
for i, (img1, img2) in enumerate(zip(images1, images2), 1):
if Image.open(img1).tobytes() != Image.open(img2).tobytes():
return 0.
return 1.
def compare_archive(pred_path: str, gold_path: str, **kwargs) -> float:
"""

23
desktop_env/evaluators/metrics/docs.py
View File

@@ -159,6 +159,29 @@ def compare_docx_tables(docx_file1, docx_file2):
return 1
from io import BytesIO
from PIL import Image
def compare_docx_images(docx_file1, docx_file2):
doc1 = Document(docx_file1)
doc2 = Document(docx_file2)
def extract_images(doc):
images = []
for rel in doc.part.rels.values():
if "image" in rel.reltype:
img_data = rel.target_part.blob
images.append(BytesIO(img_data))
return images
images1 = extract_images(doc1)
images2 = extract_images(doc2)
if len(images1) != len(images2):
return 0
for img1, img2 in zip(images1, images2):
if Image.open(img1).tobytes() != Image.open(img2).tobytes():
return 0
return 1
def compare_line_spacing(docx_file1, docx_file2):
if not compare_docx_files(docx_file1, docx_file2):

29
desktop_env/evaluators/metrics/vscode.py
View File

@@ -90,6 +90,35 @@ def compare_text_file(actual: str, expected: str, **options) -> float:
return 1.0
return 0.0
import zipfile
def compare_zip_files(actual: str, expected: str, **options) -> float:
"""
Args:
actual (str): path to result zip file
expected (str): path to gold zip file
Return:
float: the score
"""
if not actual:
return 0.
with zipfile.ZipFile(actual, 'r') as zip_file1, zipfile.ZipFile(expected, 'r') as zip_file2:
file_list1 = set(zip_file1.namelist())
file_list2 = set(zip_file2.namelist())
if file_list1 != file_list2:
return 0.0
for file_name in file_list1:
content1 = zip_file1.read(file_name)
content2 = zip_file2.read(file_name)
if content1 != content2:
return 0.0
return 1.0
def compare_config(actual: str, rules: Dict, **options) -> float:
if not actual:

58
evaluation_examples/examples/multi_apps/20236825-b5df-46e7-89bf-62e1d640a897.json Normal file
View File

@@ -0,0 +1,58 @@
{
"id": "20236825-b5df-46e7-89bf-62e1d640a897",
"snapshot": "vscode",
"instruction": "I am coding on my algorithm practice. The doc \"bubble_Sort_tutorial.docx\" is the document for it. Help me finish the function 'bubbleSort' in 'bubbleSort.py' on the Desktop save the output in 'res.txt' on Desktop.",
"source": "authors",
"config": [
{
"type": "download",
"parameters": {
"files": [
{
"url": "https://drive.usercontent.google.com/download?id=1AQtZ8Hrf7WtyUtaHMtDN-UErKRXnW64d&export=download&authuser=0&confirm=t&uuid=bf7f2b4b-ecf9-4260-b74e-db0cd40b58ca&at=APZUnTVimJzbmwJ8-4E1lq9ipJf_:1709624149129",
"path": "/home/user/Desktop/bubbleSort.zip"
},
{
"url": "https://drive.usercontent.google.com/download?id=168ZHCnK6v5PEZ8G5M25ZUW9fICk4OlfE&export=download&authuser=0&confirm=t&uuid=3642df08-dc40-4d37-93a0-8532e3012fb0&at=APZUnTUP1OTlq0kIgqcj7YSWw6MB:1709622592489",
"path": "/home/user/Desktop/Bubble_Sort_tutorial.docx"
}
]
}
},
{
"type": "execute",
"parameters": {
"command": [
"/bin/bash",
"-c",
"unzip /home/user/Desktop/bubbleSort.zip -d /home/user/Desktop/ && rm -rf /home/user/Desktop/bubbleSort.zip"
]
}
},
{
"type": "open",
"parameters": {
"path": "/home/user/Desktop/Bubble_Sort_tutorial.docx"
}
}
],
"trajectory": "trajectories/20236825-b5df-46e7-89bf-62e1d640a897",
"related_apps": [
"vscode",
"libreoffice_writer",
"os"
],
"evaluator": {
"func": "compare_text_file",
"expected": {
"type": "cloud_file",
"path": "https://drive.usercontent.google.com/download?id=1g2Trt9oxQyW_sx8aIztFA0zNsE4yNw2x&export=download&authuser=0&confirm=t&uuid=342751c4-54f1-4760-9326-e7388845ded0&at=APZUnTV5BcbaxIZrDglWbs84Oxln:1709623697315",
"dest": "res.txt"
},
"result": {
"type": "vm_file",
"path": "/home/user/Desktop/res.txt",
"dest": "res.txt"
}
}
}

69
evaluation_examples/examples/multi_apps/227d2f97-562b-4ccb-ae47-a5ec9e142fbb.json Normal file
View File

@@ -0,0 +1,69 @@
{
"id": "227d2f97-562b-4ccb-ae47-a5ec9e142fbb",
"snapshot": "gimp",
"instruction": "I have my .xcf file saved on Desktop. Could you help me copy the image and paste it into a Libreoffice Writer file? Save it as 'image.docx' on the Desktop.",
"source": "authors",
"config": [
{
"type": "download",
"parameters": {
"files": [
{
"path": "/home/user/Desktop/QTdHniCqfJbBLJe3L3nijU-1200-80.xcf",
"url": "https://drive.usercontent.google.com/download?id=1BGoDOu9bYIG7Twj5dVTxRIWDUgWzzDtP&export=download&authuser=0&confirm=t&uuid=235d1bb2-37a0-4d96-82bf-a87f31d03cb4&at=APZUnTX53EdR1stASFS3OH5luAtB:1709617456061"
}
]
}
}
],
"trajectory": "trajectories/227d2f97-562b-4ccb-ae47-a5ec9e142fbb",
"related_apps": [
"libreoffice_writer",
"gimp",
"os"
],
"evaluator": {
"postconfig": [
{
"type": "activate_window",
"parameters": {
"window_name": "image.docx - LibreOffice Writer",
"strict": true
}
},
{
"type": "sleep",
"parameters": {
"seconds": 0.5
}
},
{
"type": "execute",
"parameters": {
"command": [
"python",
"-c",
"import pyautogui; import time; pyautogui.hotkey(\"ctrl\", \"s\"); time.sleep(0.5);"
]
}
},
{
"type": "sleep",
"parameters": {
"seconds": 1.0
}
}
],
"func": "compare_docx_images",
"result": {
"type": "vm_file",
"path": "/home/user/Desktop/image.docx",
"dest": "image.docx"
},
"expected": {
"type": "cloud_file",
"path": "https://drive.usercontent.google.com/download?id=11kWQc1XFEqcIMuW0-NnZRSdv1199OmVI&export=download&authuser=0&confirm=t&uuid=694676fd-1ac9-4501-8acf-f48018494c7f&at=APZUnTV-koL51ka5dHum_HpGywv_:1709618406292",
"dest": "image.docx"
}
}
}

47
evaluation_examples/examples/multi_apps/8df7e444-8e06-4f93-8a1a-c5c974269d82.json Normal file
View File

@@ -0,0 +1,47 @@
{
"id": "8df7e444-8e06-4f93-8a1a-c5c974269d82",
"snapshot": "libreoffice_writer",
"instruction": "In the \"reminder.docx\" on Desktop is the submission instruction of our essay work. My essay is saved as docx file in /home/user. Please help me prepare the files for submission as required. ",
"source": "authors",
"config": [
{
"type": "download",
"parameters": {
"files": [
{
"url": "https://drive.usercontent.google.com/download?id=1wrCJd2f0xYnrHcj6KDzCe96X9JsN3VI3&export=download&authuser=0&confirm=t&uuid=277cb94d-1981-4f4d-b1ba-bceac8146001&at=APZUnTWKU5DBnr_6-_ZlEdsvhpCz:1709633482673",
"path": "/home/user/Recruitment_and_retention_of_health_professionals_across_Europe.docx"
},
{
"url": "https://drive.usercontent.google.com/download?id=1vzawJthEhQHcK4cUF0W9QT4zaFywO6aH&export=download&authuser=0&confirm=t&uuid=0fbb6a50-f9c1-44d2-b430-3af738d5fadc&at=APZUnTWyfv-N5f-EjnF8ob-VfCsD:1709633450986",
"path": "/home/user/Desktop/reminder.docx"
}
]
}
},
{
"type": "open",
"parameters": {
"path": "/home/user/Desktop/reminder.docx"
}
}
],
"trajectory": "trajectories/8df7e444-8e06-4f93-8a1a-c5c974269d82",
"related_apps": [
"libreoffice_writer",
"os"
],
"evaluator": {
"func": "compare_zip_files",
"expected": {
"type": "cloud_file",
"path": "https://drive.usercontent.google.com/download?id=1IKRu-dMFP4Aqzq5-4TOmOWVci0qvC27K&export=download&authuser=0&confirm=t&uuid=e2dabad2-5648-4bc3-a40f-f008089cd613&at=APZUnTVh5JD5nT3EvutwHIaSnJAT:1709633945616",
"dest": "Recruitment_and_retention_of_health_professionals_across_Europe.zip"
},
"result": {
"type": "vm_file",
"path": "/home/user/Recruitment_and_retention_of_health_professionals_across_Europe.zip",
"dest": "Recruitment_and_retention_of_health_professionals_across_Europe.zip"
}
}
}

47
evaluation_examples/examples/multi_apps/a503b07f-9119-456b-b75d-f5146737d24f.json Normal file
View File

@@ -0,0 +1,47 @@
{
"id": "a503b07f-9119-456b-b75d-f5146737d24f",
"snapshot": "os",
"instruction": "I am preparing for my paper reading report. Could you help me download this image at https://github.com/xlang-ai/OpenAgents/blob/main/pics/openagents_overview.png and convert it to PDF format. Save it as \"openagents_overview.pdf\" on desktop.",
"source": "authors",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"google-chrome",
"--remote-debugging-port=1337"
]
}
},
{
"type": "launch",
"parameters": {
"command": [
"socat",
"tcp-listen:9222,fork",
"tcp:localhost:1337"
]
}
}
],
"trajectory": "trajectories/a503b07f-9119-456b-b75d-f5146737d24f",
"related_apps": [
"os",
"chrome",
"gimp",
"pdf"
],
"evaluator": {
"func": "compare_pdfs",
"expected": {
"type": "cloud_file",
"path": "https://drive.usercontent.google.com/download?id=1gD1odFNuLT6TP6rpAv_hot86pKcI5wY1&export=download&authuser=0&confirm=t&uuid=9d17c689-95d3-45e4-b093-0165de4045b4&at=APZUnTV9SssKQCoYyYeVsi8e9zcX:1709649491796",
"dest": "openagents_overview.pdf"
},
"result": {
"type": "vm_file",
"path": "/home/user/Desktop/openagents_overview.pdf",
"dest": "openagents_overview.pdf"
}
}
}

85
evaluation_examples/examples/multi_apps/b337d106-053f-4d37-8da0-7f9c4043a66b.json Normal file
View File

@@ -0,0 +1,85 @@
{
"id": "b337d106-053f-4d37-8da0-7f9c4043a66b",
"snapshot": "os",
"instruction": "I've recently wanted to try using the Vim editor to edit code, but my Vim editor doesn't show relative line numbers by default. Please search the internet for a tutorial on how to display relative line numbers and make it the default setting for my local Vim.",
"source": "authors",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"google-chrome",
"--remote-debugging-port=1337"
]
}
},
{
"type": "launch",
"parameters": {
"command": [
"socat",
"tcp-listen:9222,fork",
"tcp:localhost:1337"
]
}
},
{
"type": "execute",
"parameters": {
"command": [
"python",
"-c",
"import pyautogui; import time; time.sleep(0.5); pyautogui.click(960, 540); time.sleep(0.5); pyautogui.hotkey('ctrl', 'alt', 't'); time.sleep(0.5)"
]
}
},
{
"type": "activate_window",
"parameters": {
"window_name": "Terminal"
}
}
],
"trajectory": "trajectories/b337d106-053f-4d37-8da0-7f9c4043a66b",
"related_apps": [
"os",
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "download",
"parameters": {
"files": [
{
"url": "https://drive.usercontent.google.com/download?id=1CyhWjUS2oov4Fzc0VRwTh6LiS2Qu-T_8&export=download&authuser=0&confirm=t&uuid=9d0e2c62-895c-4bb3-a057-30cae60329ed&at=APZUnTVngSwARjYsWSmhSyHAqwID:1709647023362",
"path": "eval.sh"
}
]
}
},
{
"type": "execute",
"parameters": {
"command": "chmod +x eval.sh",
"shell": true
}
}
],
"func": "check_include_exclude",
"result": {
"type": "vm_command_line",
"command": "bash eval.sh",
"shell": true
},
"expected": {
"type": "rule",
"rules": {
"include": [
"The File Has Set Number!"
],
"exclude": []
}
}
}
}