Human Scraping (greasemonkey)

dirty_search.js

// ==UserScript==
// @name            Sample Script 2
// @namespace       https://daqua.jp
// @description  Human scraper
// @include      http://r-r.arclight.co.jp/backnumber
// @require https://code.jquery.com/jquery-3.5.1.min.js
// @require https://cdnjs.cloudflare.com/ajax/libs/underscore.js/1.11.0/underscore-min.js
// ==/UserScript==

const search_box = $('<input type="text" id="simple_search"></input>');
const result_box = $('<ul id="result_box"></ul>');

search_box.css({
    'z-index': 1000000,
    position: 'fixed',
    top: '10px',
    left: '10px',
    border: '1px solid red',
    'background-color': 'lightblue'
});
result_box.css({
    'z-index': 1000000,
    position: 'fixed',
    top: '36px',
    left: '10px',
    border: '1px solid black',
    'background-color': 'white',
    'list-style': 'none',
    'min-width': '100px',
    'max-width': '800px',
    padding: '5px',
    'max-height': '800px',
    'overflow-y': 'scroll',
    display: 'block'
});

$('#header').append(search_box).append(result_box);

$('#simple_search').on('change', (e) => {
    const search_word = e.currentTarget.value;
    result_box.empty();
    if (!search_word) {
    } else {
        _.each(search_source, (source) => {
            _.each(source.items, (item) => {
                if ((item.title.indexOf(search_word) > -1) || (item.text.indexOf(search_word) > -1)) {
                    result_box.append($(`<li style="margin-bottom: 2px;"><a href="http://r-r.arclight.co.jp/backnumber/${source.key}" target="_blank">${source.key}</a>
                    <span>title: ${item.title}</span>
                    <br>
                    <span>text: ${item.text}</span>
                    </li>`));
                }
            })
        });
    }
}).on('keydown', (e) => {
    if (e.key === 13) {
        debugger;
        e.preventDefault();
    }
});

const search_source = [];

_.each(Object.keys(localStorage), (key) => {
    search_source.push({
        key,
        items: JSON.parse(localStorage.getItem(key))
    });
});

safe_scrape.js

// ==UserScript==
// @name            Sample Script
// @namespace       https://daqua.jp
// @description  Human scraper
// @include      http://r-r.arclight.co.jp/backnumber/*
// @require http://ajax.googleapis.com/ajax/libs/jquery/1.5.2/jquery.min.js
// ==/UserScript==

const urlInfo = document.location.href.split('/');
const entry_id = urlInfo[urlInfo.length - 1];

if (entry_id === 'backnumber') {
    return;
}

const $elements = $('.backnumberText').children();
const dict = [];
let item = {title: '', text: []};
let i = 0;
while (i < $elements.length) {
    if ($elements[i].tagName === 'H1' || $elements[i].tagName === 'H3') {
        if (!!item.title) {
            dict.push({title: item.title, text: item.text.join('\n')});
            item = {title: '', text: []};
        }

        item.title = $elements[i].innerText;
    } else {
        item.text.push($elements[i].innerText);
    }
    i = i + 1;
}
if (!!item.title) {
    dict.push({title: item.title, text: item.text.join('\n')});
}

// console.log(JSON.stringify(dict));

localStorage.setItem(entry_id, JSON.stringify(dict));

// alert('complete');
$('#wrapper').css('background-color', 'pink')