WestLaw Scraper

westlaw_scraper.js

// ==UserScript==
// @name        WestLaw Scraper
// @namespace   http://userscripts.org/users/
// @include     http://weblinks.westlaw.com/result/*
// @require     http://code.jquery.com/jquery-1.8.1.min.js
// @require     https://raw.github.com/tmcw/happen/master/src/happen.js
// @version     1
// @grant       GM_log
// ==/UserScript==

var html_content = '';

var laws = localStorage.getItem('laws');
if (laws) {
    laws = JSON.parse(laws);
    console.log(laws.length, ' scraped');
} else {
    laws = [];
}

$('.DocumentTextGroupBox *').contents().filter(function() {
    return this.nodeType === 3;
}).each(function(i, t) {
    html_content += t.nodeValue + '\n';
});

var doc = {
    content: html_content,
    title: $('#headerTitleTruncate1').text()
};

laws.push(doc);

localStorage.setItem('laws', JSON.stringify(laws));

function download_results() {
    var laws = localStorage.getItem('laws');
    document.location = 'data:Application/octet-stream,' +
        encodeURIComponent(laws);
}

var $next = $('a[title="Next Document"]');
if ($next) {
    console.log('next available');
    happen.click($next[0]);
}