YouTube_captions_search_engine/website/index.php

<?php

    function echoUrl($url)
    {
        echo "<a href=\"$url\">$url</a>";
    }

    ?>

See <?php echoUrl('https://gitea.lemnoslife.com/Benjamin_Loison/YouTube_captions_search_engine'); ?> for more information.<br/>

Access raw data with: <?php echoUrl('channels/'); ?>.<br/>
Access found channels with: <?php echoUrl('channels.txt'); ?>.

<form id="form">
    <input type="text" autofocus id="search" size="23" placeholder="Your search regex"></input><br/>
    <input type="text" autofocus id="path-search" size="23" placeholder="Your path regex (default: *empty*)"></input><br/>
    <input type="submit" id="search" value="Search">
    <input type="submit" id="search-only-captions" value="Search only captions">
</form>

Progress: <span id="progress"></span> channels

<ul id="channels">
</ul>

<script>
    var firstRun = true;
    var conn;
    // Could parse DOM instead of using following variable.
    var channels = [];

    function createA(text, href) {
        var a = document.createElement('a');
        var text = document.createTextNode(text);
        a.appendChild(text);
        a.href = href;
        return a;
    }

    function treatLine(line) {
        console.log(line);
        if (line.startsWith('progress:')) {
            document.getElementById('progress').innerHTML = line.replace('progress:', '');
        } else {
            var channelsDom = document.getElementById('channels');
            var timestamp = [];
            const lineParts = line.split('|');
            if (lineParts.length > 0) {
                timestamps = lineParts.slice(1).map(linePart => parseInt(linePart));
                line = lineParts[0];
            }
            const channelFileParts = line.split('/');
            const channel = channelFileParts[0];
            const channelFile = channelFileParts.slice(1).join('/');
            const channelHref = `channels/${channel}`;
            if (!channels.includes(channel)) {
                channels.push(channel);
                channelDom = document.createElement('li');
                var a = createA(channel, channelHref);
                channelDom.appendChild(a);
                var channelFilesDom = document.createElement('ul');
                channelDom.appendChild(channelFilesDom);
                channelsDom.appendChild(channelDom);
            }
            var channelDom = channelsDom.lastChild;
            var channelFilesDom = channelDom.lastChild;
            var channelFileDom = document.createElement('li');
            var a = createA(channelFile, `${channelHref}/${channelFile}`);
            channelFileDom.appendChild(a);
            const id = channelFileParts[2];
            for(var timestampsIndex = 0; timestampsIndex < timestamps.length; timestampsIndex++) {
                const space = document.createTextNode('\u00A0');
                channelFileDom.appendChild(space);
                const timestamp = timestamps[timestampsIndex];
                var a = createA(`${timestamp} s`, `https://www.youtube.com/watch?v=${id}&t=${timestamp}`);
                channelFileDom.appendChild(a);
            }
            channelFilesDom.appendChild(channelFileDom);
        }
    }

    function search(event) {
        // We don't want to refresh the webpage which is the default behavior.
        event.preventDefault();
        const query = event.submitter.id + ' ' + document.getElementById('path-search').value + ' ' + document.getElementById('search').value;
        if (firstRun) {
            firstRun = false;
            conn = new WebSocket('wss://crawler.yt.lemnoslife.com/websocket');
            conn.onmessage = function(e) {
                e.data.split('\n').forEach(treatLine);
            };
            // We can't directly proceed with `conn.send`, as the connection may not be already established.
            conn.onopen = function(e) { conn.send(query); };
        } else {
            // We assume at this point that the connection is established.
            channels = [];
            document.getElementById('channels').innerHTML = '';
            conn.send(query);
        }
    }

    var form = document.getElementById('form');
    form.addEventListener('submit', search);
</script>