import pandas as pd

outputFile = r'/root/.openclaw/workspace-xiaoyan/output/30-34标注完成.xlsx'

# 读取结果文件
df = pd.read_excel(outputFile)

print("=== 结果文件样例 ===")
print(f"列名: {df.columns.tolist()}")
print(f"\n前20行数据:")
print(df.head(20))

# 统计一下
yes_count = (df['是否三级'].str.contains('【是】')).sum()
no_count = len(df) - yes_count

print(f"\n\n=== 统计结果 ===")
print(f"总单词数: {len(df)}")
print(f"是三级词汇: {yes_count}")
print(f"不是三级词汇: {no_count}")