gistfile1.txt

input_data = {
    'format': 
    [
        {
            'item_name': 'name', 
            'xpath': '//*[@id="topicPath"]/li[2]/text()', 
            'regex_match': 'Null', 
            'regex_item': 'Null', 
            'remove_tag': ''},
        {
            'item_name': 'zip', 
            'xpath': '//*[@id="header"]/p', 
            'regex_match': '〒([\d\-]+)\s*', 
            'regex_item': '$1 ', 
            'remove_tag': ''
            }, 
        {
            'item_name': 'address', 
            'xpath': '//*[@id="header"]/p', 
            'regex_match': '〒[\d\-]+\s*(.+)\s*TEL', 
            'regex_item': '$1 ', 
            'remove_tag': ''}, 
        {
            'item_name': 'tel', 
            'xpath': '//*[@id="header"]/p', 
            'regex_match': '〒[\d\-]+\s*.+\s*TEL\s*\:*\s*([\d\-\(\)\s]+)', 
            'regex_item': '$1', 
            'remove_tag': ''}
    ],

    'target': [

            {
                'id': '1', 
                'url': 'https://www.takashimaya.co.jp/rakusai/index.html'
                }, 
            {
                'id': '2', 
                'url': 'https://www.takashimaya.co.jp/okayama/index.html'
                },
            {
                'id': '3', 
                'url': 'https://www.takashimaya.co.jp/tachikawa/index.html'}, 
            {   
                'id': '4', 
                'url': 'https://www.takashimaya.co.jp/yokohama/index.html'}
        ],

    'coordination':[
            {
                 "item_name": "coordination",
                 "mapurl_xpath": '//*[@id="storeInfo"]/div/p/a',
                 "xpath": '//*[@id="mapDiv"]/div/div/div[10]/div/div/div/div[7]/div/a',
                 "regex_match": "https:\/\/maps\.google\.com\/maps\?.*ll=([\d\.]+),([\d\.]+)",
                 "lat": "$1",
                 "lon": "$2",
                 "mapurl_regex": "\/[^\/]+\/access\/index.html"
            }
        ] 
    

}

input_data = {
    'format': [

        {
            'item_name': 'name', 
            'xpath': '//*[@id="storeMap"]/header[3]/div/h2/text()', 
            'regex_match': 'Null', 
            'regex_item': 'Null', 
            'remove_tag': ''}, 

        {
            'item_name': 'zip', 
            'xpath': '//*[@id="storeMap"]/section/div/div[1]/dl/dd[1]', 
            'regex_match': '〒([\\d\\-]+)', 
            'regex_item': '$1 ', 
            'remove_tag': ''}, 

        {
            'item_name': 'address', 
            'xpath': '//*[@id="storeMap"]/section/div/div[1]/dl/dd[1]', 
            'regex_match': '〒[\\d\\-]+<br>([^<]+)', 
            'regex_item': '$1 ', 
            'remove_tag': ''}, 

        {
            'item_name': 'tel', 
            'xpath': '//*[@id="storeMap"]/section/div/div[1]/dl/dd[2]/text()', 
            'regex_match': 'Null', 
            'regex_item': 'Null', 
            'remove_tag': ''}, 
        {
            'item_name': 'hour', 
            'xpath': '//*[@id="storeMap"]/section/div/div[1]/dl/dd[3]', 
            'regex_match': 'Null', 
            'regex_item': 'Null', 
            'remove_tag': '1'},
        {
            'item_name': 'name', 
            'xpath': '//*[@id="storeMap"]/header[3]/div/h2/text()', 
            'regex_match': 'Null', 
            'regex_item': 'Null', 
            'remove_tag': ''}
        ], 

    'target': [
        {
            'id': '1', 
            'url': 'https://www.burgerkingjapan.co.jp/stores/detail.html?sn=89'}, 
        {
            'id': '2', 
            'url': 'https://www.burgerkingjapan.co.jp/stores/detail.html?sn=124'}, 
        {
            'id': '3', 
            'url': 'https://www.burgerkingjapan.co.jp/stores/detail.html?sn=38'}, 
        {
            'id': '4', 
            'url': 'https://www.burgerkingjapan.co.jp/stores/detail.html?sn=12'
        }, 
        {
            'id': '5', 
            'url': 'https://www.burgerkingjapan.co.jp/stores/detail.html?sn=91'}
        ],

    'coordination': [
        {
            "item_name":"coordinates",
            "xpath": '//*[@id="gmap"]/div/div/div[2]/a',
            "regex_match": "https:\/\/maps\.google\.com\/maps\?.*ll=([\d¥.]+),([\d\.]+)",
            "lat": "$1",
            "lon": "$2",
            "mapurl_regex": ""

        }
    ]
}