Scraping data from: http://www.weather.go.kr/weather/climate/past_cal.jsp

과거지상관측자료.py

# coding=utf-8
import re

import requests
from bs4 import BeautifulSoup, Tag

weather_pattern = re.compile(
    r'''(?P<name>[^<>]+?)             # name group
    \s*?:\s*?                         # split by ':'
    (?P<value>                        # value group
        (?:
            [-+]?                       # optional: '-' or '+'
            (?:\d*[.]\d+)|(?:\d+[.]?)   # (.d or d.d) or (d. or d) 
        ) | -                           # or just '-'
    )''',
    re.VERBOSE  # 정규표현식에 주석을 허용하고 공백은 무시
)
stn_table = {'서울': '108', '백령도': '102', '동두천': '98', '파주': '99', '인천': '112', '수원': '119',
             '강화': '201', '양평': '202', '이천': '203', '북춘천': '93', '철원': '95', '춘천': '101',
             '원주': '114', '영월': '121', '인제': '211', '홍천': '212', '북강릉': '104', '울릉도': '115',
             '강릉': '105', '속초': '90', '대관령': '100', '동해': '106', '태백': '216', '정선군': '217',
             '청주': '131', '충주': '127', '추풍령': '135', '제천': '221', '보은': '226', '홍성': '177',
             '대전': '133', '서산': '129', '천안': '232', '보령': '235', '부여': '236', '금산': '238',
             '전주': '146', '군산': '140', '부안': '243', '임실': '244', '정읍': '245', '남원': '247',
             '장수': '248', '순창': '254', '고창': '172', '고창(구)': '251', '광주': '156', '목포': '165',
             '흑산도': '169', '여수': '168', '완도': '170', '진도(첨찰산)': '175', '진도군': '268', '영광': '252',
             '순천': '174', '순천(구)': '256', '장흥': '260', '해남': '261', '고흥': '262', '강진군': '259',
             '보성군': '258', '광양': '266', '안동': '136', '포항': '138', '대구': '143', '대구(구)': '176',
             '울진': '130', '상주': '137', '봉화': '271', '영주': '272', '문경': '273', '영덕': '277',
             '의성': '278', '구미': '279', '영천': '281', '청송군': '276', '경주': '283', '부산': '159',
             '울산': '152', '창원': '155', '북창원': '255', '통영': '162', '진주': '192', '거창': '284',
             '합천': '285', '밀양': '288', '산청': '289', '거제': '294', '남해': '295', '김해시': '253',
             '양산': '257', '의령군': '263', '함양군': '264', '제주': '184', '고산': '185', '서귀포': '189',
             '성산': '188'}


def get_weather(local, year, month, raise_if_failure=False):
    stn_code = local if local.isdigit() else stn_table.get(local)
    if not stn_code:  # 없는 stn 코드
        return

    response = requests.get(
        'http://www.weather.go.kr/weather/climate/past_cal.jsp',
        params={'stn': stn_code, 'yy': str(year), 'mm': str(month), 'obs': '1'}
    )
    if raise_if_failure:
        response.raise_for_status()

    return response.text


def parse_weather(html):
    if not html:
        return
    soup = BeautifulSoup(html, 'html.parser')
    table = soup.find('table', {'class': 'table_develop'})  # type: Tag
    tr_object = iter(table.find_all('tr'))

    week_list = []
    for th_ in next(tr_object).find_all('th'):  # type: Tag
        week_list.append(th_.text.strip())

    def th_generator():
        while True:
            try:
                yield next(tr_object).find_all('td'), next(tr_object).find_all('td')
            except StopIteration:
                break

    for day_td_list, weather_td_list in th_generator():
        for day_tr, weather_tr, week in zip(day_td_list, weather_td_list, week_list):  # type: (Tag, Tag, str)
            day = day_tr.text.strip()
            if day:
                yield week, day, {rex.group('name'): rex.group('value') for rex in
                                  weather_pattern.finditer(str(weather_tr))}


if __name__ == '__main__':
    print('서울')
    html = get_weather('서울', 2019, 3)
    for week, day, data in parse_weather(html):
        print(week, day, data)

    print('대전')
    html = get_weather('대전', 2018, 2)
    for week, day, data in parse_weather(html):
        print(week, day, data)