GooglePoiCrawlerV1/utility/parseutils.py

# -*- coding: utf-8 -*-
DB_NAME = 'google_poi'
SHOP_LIST_TABLE = 'shop_list3'
SHOP_LIST_TABLE_COL = ['name', 'fid', 'city', 'area', 
                      'rating', 'user_ratings_total', 'category', 'price_level', 
                      'addr', 'tel', 'services', 'products', 'choices', 
                      'facilities', 'groups', 'plans', 'payments', 'safeties', 
                      'specials', 'barrierlevels', 'items' ,
                      'open_now', 'periods', 'weekday_text','reviews',
                      'shop_photo','menu_photo',
                      'google_url', 'item_url', 'keyword', 'crawler_date']
                      
# SHOP_LIST_TABLE_COL = ['unique_id', 'name', 'lon', 'lat', 'city', 'area', 
#                       'rating', 'user_ratings_total', 'category', 'price_level', 
#                       'addr', 'tel', 'services', 'products', 'choices', 
#                       'facilities', 'groups', 'plans', 'payments', 'safeties', 
#                       'specials', 'barrierlevels', 'items' ,
#                       'open_now', 'periods', 'weekday_text','reviews',
#                       'shop_photo','menu_photo',
#                       'google_url', 'item_url', 'keyword', 'crawler_date']

element_list = {
    'category': ['button', {'jsaction':'pane.rating.category'}],
    'rating': ['ol', {}, 'aria-label'],
    'user_ratings_total': ['span', {'jsaction':'pane.rating.moreReviews'}],
    'price_level':['span', {'jsan':'0.aria-label'}]
}


intro_list = {
    '服務選項': ['services','service'],
    '產品/服務': ['products','product'],
    '用餐選擇': ['choices','choice'],
    '設施': ['facilities','facility'],
    '客層族群':['groups','group'],
    '規劃':['plans','plan'],
    '付款方式':['payments','payment'],
    '健康與安全':['safeties','safety'],
    '特色':['specials','special'],
    '無障礙程度':['barrierlevels','barrierlevel'],
    '詳細資料':['items','item']
}

week_list = {
    '星期日': 0, 
    '星期一': 1, 
    '星期二': 2, 
    '星期三': 3,
    '星期四': 4, 
    '星期五': 5, 
    '星期六': 6,  
}


def blank_check(value):
    while value.startswith(' '):
        value = value[1:]

    while value.endswith(' '):
        value = value[:-1]
    return value


def value_check(key, value):
    value = blank_check(value)
    if key == 'rating':
        value = float(value.replace(' 星級',''))
    elif key == 'user_ratings_total':
        value = int(value.replace(' 則評論','').replace(',',''))
    elif key == 'price_level':
        value = len(['$' for i in value if i == '$'])
        if value == 0:
            value = ''
        else:
            value = int(value)

    return value