pupget/pupget.json

A simple webscrapping package to extract data from a DSL

{
    "base_page": "https://acm.timus.ru/problemset.aspx?space=1&page=all",
    "output": {
        "folder": "exercises",
        "filename": "exercise_",
        "autoid": "true"
    },
    "steps": [{
        "page": "",
        "auth": "false",
        "query": {
            "query": "td.name a",
            "inject_text": "",
            "clickable": "false",
            "output_field": "exercise",
            "results": [{
                    "query": "h2",
                    "output_field": "title"
                },
                {
                    "query": ".problem_limits",
                    "output_field": "limits"
                },
                {
                    "query": "div.problem_par:nth-of-type(1) div, h3:nth-of-type(1), div.problem_par:nth-of-type(2) div, h3:nth-of-type(2), div:nth-of-type(3) div",
                    "output_field": "description"
                },
                {
                    "query": ".sample td:nth-of-type(1) pre",
                    "output_field": "input"
                },
                {
                    "query": ".sample td:nth-of-type(2) pre",
                    "output_field": "output"
                }
            ]
        }
    }]
}