Psubhashish/tools/Prepare words for Lingua Libre

The HTML code below is for a converter to prepare words for Lingua Libre. It is created considering a language written in a writing system other than Latin. It has five major functions:

Remove punctuation and convert words into new lines
Remove words typed in Latin script
Remove duplicates
Sort alphabetically
Replace lines with # signs

Your source text can be copied into the box below the "Input Text". You can then press 1. Remove punctuation & words to new line followed by the other buttons. After the fifth step click on the button "Copy text" below the "Output Text" box. Go to LinguaLibre's RecordWizard and paste the copied text to generate a word list.

<!DOCTYPE html>
<html>
<head>
    <title>Prepare words for Lingua Libre</title>
    <meta charset="UTF-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
</head>
<body>
    <h1>Prepare words for Lingua Libre</h1>
    <label for="input-text">Input Text:</label><br>
    <textarea id="input-text" rows="10" cols="50"></textarea><br>
    <button onclick="removePunctuationAndWords()">1. Remove punctuation & words to new line</button><br/><br/>
    <button onclick="removeLatinWords()">2. Remove Latin character words</button><br/><br/>
    <button onclick="removeDuplicates()">3. Remove duplicates</button><br/><br/>
    <button onclick="sortAlphabetically()">4. Sort alphabetically</button><br/><br/>
    <button onclick="replaceLinesWithHash()">5. Replace lines with #</button><br/><br/>
    <button onclick="copyText()">Copy text</button><br><br/>
    <label for="output-text">Output Text:</label><br>
    <textarea id="output-text" rows="10" cols="50"></textarea><br>

    <script>
        function removePunctuationAndWords() {
            var inputText = document.getElementById("input-text").value;

            // Replace punctuation with a space, except for hyphen and apostrophe
            var textWithSpaces = inputText.replace(/[।“”\(\)\[\]\{\}<>.,\/#!$%\^&\*;:{}=\-_`~]+/g, " ").replace(/(^[-])|([-]$)/g, "");

            // Convert spaces to new line breaks
            var textWithLineBreaks = textWithSpaces.replace(/[\s]+/g, "\n");

            document.getElementById("output-text").value = textWithLineBreaks;
        }

        function removeLatinWords() {
            var inputText = document.getElementById("output-text").value;
            var outputText = inputText.replace(/[^\u0B00-\u0B7F\s]+/g, "");
            document.getElementById("output-text").value = outputText.trim();
        }

        function removeDuplicates() {
            var inputText = document.getElementById("output-text").value;
            var words = inputText.split(/\s+/);
            var uniqueWords = [];
            for (var i = 0; i < words.length; i++) {
                if (uniqueWords.indexOf(words[i]) === -1) {
                    uniqueWords.push(words[i]);
                }
            }
            var outputText = uniqueWords.join("\n");
            document.getElementById("output-text").value = outputText.trim();
        }

        function sortAlphabetically() {
            var inputText = document.getElementById("output-text").value;
            var words = inputText.split(/\s+/);
            words.sort();
            var outputText = words.join("\n");
            document.getElementById("output-text").value = outputText.trim();
        }

        function replaceLinesWithHash() {
            var inputText = document.getElementById("output-text").value;
            var outputText = inputText.replace(/\n+/g, "#");
            document.getElementById("output-text").value = outputText.trim();
        }

        function copyText() {
            var outputText = document.getElementById("output-text");
            outputText.select();
            document.execCommand("copy");
            alert("Text copied to clipboard!");
        }
    </script>
</body>
</html>