# %% 提取數(shù)據(jù) 從sqlite
import pandas as pd
from sqlalchemy import create_engine
'''創(chuàng)建sqlite3連接'''
eg = create_engine('sqlite:///I:/jianguoyun/Udoc/WorkDb/dataset2022.db')  #創(chuàng)建數(shù)據(jù)庫連接
# 讀取數(shù)據(jù)庫文件
q_store = """
select *  from pj_store20220525
"""
# 讀取表
cols= ['貨號','品類','節(jié)慶','庫位','合格','換包裝','錯(cuò)碼','維修','總數(shù)量' ]
df = pd.read_sql(q_store,eg)[cols]

# df.duplicated 計(jì)算重復(fù)項(xiàng)
cond = df.duplicated('貨號') # 查看貨號相同的項(xiàng) df.duplicated返回的是布爾值
def_line = df[cond]
def_line

#有多少個(gè)重復(fù)行
dep_data = df.duplicated('貨號').sum() 
dep_data

# 用df.drop_duplicates提取夏季品類
df_category = df.query('節(jié)慶=="夏季"').drop_duplicates('品類',keep='last',inplace=False) 
df_category['品類'].sort_values()