report.py

#coding: utf-8
import requests
import pandas as pd
import numpy as np
from bs4 import BeautifulSoup as bs


def financial_statement(year, season, type='綜合損益彙總表'):

    if year >= 1000:
        year -= 1911
        
    if type == '綜合損益彙總表':
        url = 'http://mops.twse.com.tw/mops/web/ajax_t163sb04'
    elif type == '資產負債彙總表':
        url = 'http://mops.twse.com.tw/mops/web/ajax_t163sb05'
    elif type == '營益分析彙總表':
        url = 'http://mops.twse.com.tw/mops/web/ajax_t163sb06'
    else:
        print('type does not match')

    r = requests.post(url, {
        'encodeURIComponent':1,
        'step':1,
        'firstin':1,
        'off':1,
        'TYPEK':'sii',
        'year':str(year),
        'season':str(season),
    })
    
    r.encoding = 'utf8'
    soup = bs(r.text, 'lxml')
    
    all_df = []
    for tb in soup.select('table')[1:]:
        dfs = pd.read_html(tb.prettify('utf-8'), encoding= 'utf-8')
        all_df.append(pd.concat(dfs))
    return pd.concat(all_df)


df = financial_statement(2019, 1)
df.to_excel("test.xls")