jecnarozvrh/scrape/parse/v3.js

/*
 * Copyright (C) 2025 Jakub Žitník
 *
 * This program is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 */

import fs from "fs";
import parseAbsence from "../utils/parseAbsence.js"
import parseTeachers from "../utils/parseTeachers.js"
import ExcelJS from "exceljs"

export default async function parseV3(downloadedFilePath) {
  const workbook = new ExcelJS.Workbook();
  await workbook.xlsx.readFile(downloadedFilePath);

  const teacherMap = await parseTeachers();

  const upcoming = getUpcomingSheets(workbook);
  const resolvedDays = groupSheetsByDate(upcoming);

  const schedule = {};

  for (const { dateKey, sheet } of resolvedDays) {
    const { changes, absence, inWork, takesPlace, reservedRooms } = extractDaySchedule(sheet, teacherMap);

    schedule[dateKey] = {
      info: { inWork },
      changes,
      absence,
      takesPlace,
      reservedRooms,
    };
  }

  const data = {
    status: { lastUpdated: formatNowTime() },
    schedule,
  };

  fs.writeFileSync("db/v3.json", JSON.stringify(data, null, 2));
}

//
// ────────────────────────────────────────────────────────────
//   SHEET FILTERING
// ────────────────────────────────────────────────────────────
//

function getUpcomingSheets(workbook) {
  const dateRegex = /^(pondělí|úterý|středa|čtvrtek|pátek|po|út|ut|st|čt|ct|pa|pá)\s+(\d{1,2})\.\s*(\d{1,2})\.\s*(20\d{2})/i;

  const today = new Date();
  const todayMidnight = new Date(today.getFullYear(), today.getMonth(), today.getDate());

  const result = [];

  for (const sheet of workbook.worksheets) {
    const match = sheet.name.match(dateRegex);
    if (!match) continue;

    const day = Number(match[2]);
    const month = Number(match[3]) - 1;
    const year = Number(match[4]);

    const sheetDate = new Date(year, month, day);
    if (sheetDate < todayMidnight) continue;

    const dateKey = `${year}-${String(month + 1).padStart(2, "0")}-${String(day).padStart(2, "0")}`;
    result.push({ dateKey, sheet });
  }

  return result;
}

function groupSheetsByDate(items) {
  const map = {};

  for (const item of items) {
    map[item.dateKey] ??= [];
    map[item.dateKey].push(item.sheet);
  }

  return Object.entries(map).map(([dateKey, sheets]) => {
    const chosen =
      sheets.length === 1
        ? sheets[0]
        : sheets.find((s) => s.state !== "hidden") ?? sheets[0];

    return { dateKey, sheet: chosen };
  });
}

//
// ────────────────────────────────────────────────────────────
//   DAY PARSING
// ────────────────────────────────────────────────────────────
//

function extractDaySchedule(sheet, teacherMap) {
  return {
    changes: extractClassChanges(sheet),
    absence: extractAbsence(sheet, teacherMap),
    inWork: isPripravaSheet(sheet.name),
    takesPlace: extractTakesPlace(sheet),
    reservedRooms: extractReservedRooms(sheet)
  };
}

function isPripravaSheet(name) {
  return name
    .toLowerCase()
    .normalize("NFD")
    .replace(/[\u0300-\u036f]/g, "")
    .includes("priprava");
}

//
// ────────────────────────────────────────────────────────────
//   CLASS CHANGES
// ────────────────────────────────────────────────────────────
//

function extractClassChanges(sheet) {
  const classRegex = /[AEC][0-4][a-c]?\s*\/.*/s;
  const prefixRegex = /[AEC][0-4][a-c]?/;

  const classes = [];
  const classCells = [];

  sheet.eachRow((row) => {
    row.eachCell((cell) => {
      const value = cell.value;
      if (typeof value === "string" && classRegex.test(value) && cell.address.startsWith("A")) {
        const prefixMatch = value.match(prefixRegex);
        if (prefixMatch) classes.push(prefixMatch[0]);
        classCells.push(cell.address);
      }
    });
  });

  const changes = {};

  classCells.forEach((address, index) => {
    const row = sheet.getRow(sheet.getCell(address).row);
    changes[classes[index]] = buildLessonArray(row, address);
  });

  return changes;
}

function buildLessonArray(row, ignoreAddress) {
  const lessons = [];

  row.eachCell((cell) => {
    if (cell.address === ignoreAddress) return;

    const colIndex = letterToNumber(cell.address.replace(/[0-9]/g, ""));
    lessons[colIndex] = parseLessonCell(cell);
  });

  const normalized = Array.from(lessons, (x) => (x === undefined ? null : x));
  while (normalized.length < 11) normalized.push(null);

  return normalized.slice(1, 11);
}

function parseLessonCell(cell) {
  try {
    const text = (cell.text || "").trim();
    const cleanupRegex = /^úklid\s+(?:\d+\s+)?[A-Za-z]{2}$/;

    if (!text || cleanupRegex.test(text) || !cell.fill?.fgColor) return null;

    return {
      text,
      backgroundColor: cell.fill.fgColor.argb === undefined ? undefined : `#${cell.fill.fgColor.argb}`,
      willBeSpecified: cell.fill.fgColor.argb === "FFFFFF00" ? true : undefined,
    };
  } catch {
    return null;
  }
}


function extractTakesPlace(sheet) {
  const cell = sheet.getCell("B4");

  if (!cell.isMerged) {
    return "";
  }

  let str = cell.master.value.trim();

  let i = 5;
  while (true) {
    const nextCell = sheet.getCell(`B${i}`);

    if (!nextCell.isMerged) {
      break;
    }

    const cell = nextCell.master

    str += `\n${cell.value.trim()}`
    i++;
  }

  return str;
}

function extractReservedRooms(sheet) {
  const result = [];

  const cells = [];

  sheet.eachRow((row) => {
    row.eachCell((cell) => {
      const value = cell.value;
      if (typeof value === "string" && value.trim() === "rezervace" && cell.address.startsWith("A")) {
        cells.push(cell.address);
      }
    });
  });

  cells.forEach((address) => {
    const row = sheet.getRow(sheet.getCell(address).row);

    row.eachCell((cell) => {
      if (cell.address === address) return;

      result.push(cell.value.trim().length == 0 ? null : cell.value)
    });
  });

  return result;
}

//
// ────────────────────────────────────────────────────────────
//   ABSENCE
// ────────────────────────────────────────────────────────────
//

function extractAbsence(sheet, teacherMap) {
  let absenceAddress = null;

  sheet.eachRow((row) => {
    row.eachCell((cell) => {
      if ((cell.value || "").toString().trim().toLowerCase() === "absence") {
        absenceAddress = cell.address;
      }
    });
  });

  if (!absenceAddress) return [];

  const row = sheet.getRow(sheet.getCell(absenceAddress).row);
  const results = [];
  let i = 0;

  row.eachCell((cell) => {
    if (cell.address === absenceAddress || i >= 10) return;
    i++;

    const value = (cell.value || "").toString().trim();
    if (!value) return;

    results.push(...parseAbsence(value, teacherMap));
  });

  return results;
}

//
// ────────────────────────────────────────────────────────────
//   UTILS
// ────────────────────────────────────────────────────────────
//

function letterToNumber(letter) {
  return letter.toLowerCase().charCodeAt(0) - 97;
}

function formatNowTime() {
  const now = new Date();
  return (
    now.getHours().toString().padStart(2, "0") +
    ":" +
    now.getMinutes().toString().padStart(2, "0")
  );
}

// parseV3("db/current.xlsx")