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

ver Jan12thv2

Browse Source 
sqlite3 metric
This commit is contained in:
David Chang
2024-01-12 23:07:00 +08:00
parent 5160619783
commit e08df57129
4 changed files with 134 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
desktop_env/evaluators/metrics/__init__.py
View File

@@ -9,5 +9,5 @@ from .table import check_sheet_list, check_xlsx_freeze, check_xlsx_zoom
from .table import compare_table
from .vlc import is_vlc_playing, is_vlc_recordings_folder, is_vlc_fullscreen, compare_images, compare_audios, \
compare_videos
from .general import check_csv, check_accessibility_tree, check_list
from .general import check_csv, check_accessibility_tree, check_list, run_sqlite3
from .thunderbird import check_thunderbird_prefs, check_thunderbird_filter

7
desktop_env/evaluators/metrics/general.py
View File

@@ -13,6 +13,8 @@ from rapidfuzz import fuzz
from .utils import _match_record
import sqlite3
def exact_match(result, rules) -> float:
expect = rules["expected"]
print(result, expect)
@@ -146,3 +148,8 @@ def check_accessibility_tree(result: str, rules: Dict[str, Any]) -> float:
# def check_existence(result: str, *args) -> float:
# return 1. - (result is None)
def run_sqlite3(result: str, rules: Dict[str, Any]) -> float:
connection = sqlite3.connect(result)
connection.execute(rules["sql"])
return float(connection.fetchone()[0] or 0)