merge-html

#!/usr/bin/env node

const usage = 'file...'
const onetab_dateformats = ['M-D-Y, h:m:s A', 'Y-M-D HH:m:s', 'D/M/Y, HH:m:s']

if (process.argv.length == 2) {
    console.log('Usage:', usage)
    process.exit()
}

var cheerio = require('cheerio');
var fs = require('fs-extra-promise');
var moment = require('moment');

merge_html_files(process.argv.slice(2))

function merge_html_files(files) {
    // return console.log(files);
    try {
        var onetabs = files.map(file => load_html(file))
            // onetabs.forEach(onetab => console.log(onetab.summary))

        var wins = []
        for (let onetab of onetabs) {
            for (let win of onetab.windows) {
                // win.file = onetab.file
                wins.push(win)
            }
        }
        wins.sort((a, b) => a.created - b.created)

        // for (let win of wins) console.log(win.created, win.file);
        // // for (let win of wins) console.log(win);
        // console.log(wins.length);

        console.log(JSON.stringify(wins));
    }
    catch (e) {
        console.error(e);
    }
}

function load_html(file) {
    // console.log('load', file);
    var html = fs.readFileSync(file)
    var $ = cheerio.load(html)
    var onetab = {
        file: file,
        windows: [],
    }
    $('#contentAreaDiv > div').each((idx, div) => {
        if (div.children.length != 3) return;
        var createdText = $(':first-child :first-child :nth-child(5) :first-child', div).text().replace('Created ', '')
        var created = null;
        var dtformat = onetab_dateformats.find(format => {
            var m = moment(createdText, format)
            if (!m.isValid()) return false
            created = m.toDate()
            let y = created.getFullYear()
            if (!(y > 2010 && y < 2020)) return false
            return true
        })
        if (!dtformat) created = null
        if (!created) {
            console.error('unfound date format', createdText);
            return;
        }
        var window = {
            created: created,
            tabs: [],
        }
        onetab.windows.push(window)
        $(':nth-child(2) a', div).each((idx, a) => {
            a = $(a)
            window.tabs.push({
                url: a.attr('href'),
                title: a.text().trim(),
            })
        })
    })
    onetab.summary = {
        file: file,
        windows: onetab.windows.length,
        tabs: onetab.windows.reduce((count, w) => count + w.tabs.length, 0),
    }
    return onetab;
}

// function merge(onetabs) {
//     var map = {}
//     for (let onetab of onetabs) {
//         for (let window of onetab.windows) {
//             let id = window.created.toISOString() + ' ' + onetab.file
//             let tabs = map[id]
//             if (!tabs) {
//                 tabs = []
//                 map[id] = tabs
//             }
//             for (let tab of window.tabs) {
//                 tabs.push(tab)
//             }
//         }
//     }
//     for (let id of Object.keys(map).sort().reverse()) {
//         console.log();
//         // console.log('\n' + id);
//         for (let tab of map[id]) {
//             console.log(tab.url + ' | ' + tab.title);
//         }
//     }
// }