Koikatsu

/* -----------------------------------------------
-- MOON RATING
--------------------------------------------------
-- I'm using this because it was the easier way
-- to simulate rating stars.
-------------------------------------------------- */

function moonRating(number) {

  // Is it a number?
  if (!isNaN(number)) {
    var votes;

    // Is it an interger?
    if (number % 1 === 0) {
      votes="🌕".repeat(number);
      votes+="🌑".repeat(5-number);

    // Or is it a float?
    } else {
      let n = Math.floor(number);
      votes="🌕".repeat(n);
      votes+="🌗";
      votes+="🌑".repeat(4-n);
    }
    return votes;
  } else {
    // If no votes
    return "------"
  }
}

/* --------------------------------------------------------------------
-- Scraping data using Cheerio library
-- (https://github.com/tani/cheeriogs)
-- Script ID: 1ReeQ6WO8kKNxoaA_O0XEQ589cIrRvEBA9qcWpNqdOP17i47u6N9M5Xh0
----------------------------------------------------------------------- */

// Cheerio method to scrape
function getContent(url) {
    return UrlFetchApp.fetch(url).getContentText()
}

// Actual scrape function
function scrapeData() {

  // Selects the active sheet
  var sheet = SpreadsheetApp.getActiveSpreadsheet().getActiveSheet();

  // Get selected cells
  var selectedRange = sheet.getActiveRange();

  // Get the values from the selected cells
  var selectedValues = selectedRange.getValues();

  for (var i = 0; i < selectedValues.length; i++) {

    // It assumes the URLs are in the first column of the selected column
    var url = selectedValues[i][0];

    // If the value is not empty...
    if (url) {
      // Grabs the HTML data
      var content = getContent(url);

      // Parses the HTML data
      var $ = Cheerio.load(content);

      /* -----------------------------------------------
      -- [TITLE]
      -------------------------------------------------- */
      var threadTitle = $('h1.p-title-value');

      var title = threadTitle.contents().filter(function() {
        return this.type === 'text';
      }).text();

      /* -----------------------------------------------
      -- [STATUS] (ACTIVE - ABANDONED - ONHOLD)
      -------------------------------------------------- */
      // Default status
      var status = 'Active';

      // Search for "Abandoned" or "Onhold"
      threadTitle.find('span').each(function() {
        var spanText = $(this).text();
        if (spanText === 'abandoned' || spanText === 'Onhold') {
          status = spanText;
          return false;
        }
      });

      /* -----------------------------------------------
      -- [ENGINE]
      -------------------------------------------------- */
      // If there's no engine info, it's unknown
      var engine = 'Unknown';

      // Checks if there's one of these strings
      threadTitle.find('span').each(function() {
        var spanText = $(this).text();
        if (['ADRIFT','Flash','Java','Others','QSP','RAGS','RPGM',"Ren'Py",'Tads','Unity','Unreal Engine','WebGL','Wolf RPG'].includes(spanText)) {
          engine = spanText;
          return false; // Stop searching once a match is found
        }
      });

      /* -----------------------------------------------
      -- [TAGS]
      ----------------------------------------------- */
      // Selects the element containing the tags
      var tagList = $('span.js-tagList');

      // Grabs the tags and separate them by comma
      var tags = tagList.find('a.tagItem').map(function() {
        return $(this).text();
      }).get().join(', ');

      /* -----------------------------------------------
      -- GAMEPLAY
      -------------------------------------------------- */
      var gameplay;

      // Assumes the [Gameplay] type by using the tags
      switch (true) {
        case tags.includes('sandbox'):
          gameplay = 'Sandbox';
          break;
        case tags.includes('turn based combat'):
          gameplay = 'TBC';
          break;
        case tags.includes('management'):
          gameplay = 'Management';
          break;
        case tags.includes('simulator'):
          gameplay = 'Simulator';
          break;
        case tags.includes('rpg'):
          gameplay = 'RPG';
          break;
        default:
          gameplay = 'Visual Novel';
      }

      /* -----------------------------------------------
      -- EXTRA INFORMATION
      --------------------------------------------------
      -- - Thread [Updated] Date
      -- - [Release] Date
      -- - [Dev] name
      -- - [Censored]
      -------------------------------------------------- */
      // Selects the first message body (which is the one containing the game info)
      var articleText = $('article.message-body').first().text();

      // Uses regular expressions to match and capture the information
      // When it was updated
      var updatedREGEX = /Thread Updated: (\d{4}-\d{2}-\d{2})/.exec(articleText);
      // When it was released
      var releaseREGEX = /Release Date: (\d{4}-\d{2}-\d{2})/.exec(articleText);
      // Who is the dev
      var devREGEX = /Developer: ([^\s]+)/.exec(articleText);
      // It is censored?
      var censoredREGEX = /Censored: (.+?)(?=\n|\r|\r\n)/.exec(articleText);

      // Extract the captured values or provide default values if not found
      var updated = updatedREGEX ? updatedREGEX[1] : "Not found";
      var release = releaseREGEX ? releaseREGEX[1] : "Not found";
      var dev = devREGEX ? devREGEX[1] : "Not found";
      var censored = censoredREGEX ? censoredREGEX[1] : "Not found";

      /* -----------------------------------------------
      -- [RATING] & [VOTES]
      -------------------------------------------------- */
      // Grabs the JSON generated by F95 forum
      var jsonInfo = $('script[type="application/ld+json"]').last();

      // Uses regular expression to match and caputre the information
      var ratingREGEX = /"ratingValue": "(\d+(?:\.\d+)?)"/.exec(jsonInfo);
      var votesREGEX = /"ratingCount": "(\d+)"/.exec(jsonInfo);

      // Extract the captured values or provide default values if not found
      var rating = ratingREGEX ? ratingREGEX[1] : "No Rating";
      var votes = votesREGEX ? votesREGEX[1] : "0";

      /* -----------------------------------------------
      -- OUTPUTS THE VALUE INTO DIFFERENT CELLS
      -------------------------------------------------- */

      // Write the scraped data to the same row
      var outputRow = selectedRange.getRow() + i;

      // Write the scraped data to adjacent columns
      var outputColumn = selectedRange.getColumn() + 1;

      // All values that will be written
      var values = [[
                      title,
                      status,
                      engine,
                      gameplay,
                      tags,
                      moonRating(rating),
                      votes,
                      updated,
                      release,
                      dev,
                      censored
                    ]];

      // Where it will be written
      sheet.getRange(outputRow, outputColumn, 1, values[0].length).setValues(values);
    }
  }

}

/* -----------------------------------------------
-- CREATES AN EXTRA MENU OPTION
-------------------------------------------------- */

function onOpen() {
  var ui = SpreadsheetApp.getUi();
  ui.createMenu('Scrape Data')
      .addItem('Scrape Data', 'scrapeData')
      .addToUi();
}