feat: Absences

scrape/parse.js

@@ -1,7 +1,9 @@
 const XLSX = require('xlsx');
 const fs = require("fs");
+const parseAbsence = require('./utils/parseAbsence');
+const parseTeachers = require('./utils/parseTeachers');
 
-function parseThisShit(downloadedFilePath) {
+async function parseThisShit(downloadedFilePath) {
   const workbook = XLSX.readFile(downloadedFilePath);
   const sheetNames = workbook.SheetNames;
 
@@ -60,7 +62,7 @@ function parseThisShit(downloadedFilePath) {
     // For each class
     let classI = 0;
     for (const matchingKey of matchingKeys) {
-      const allKeys = Object.keys(currentSheet).filter(key => key !== matchingKey && key.endsWith(matchingKey.replace(/[a-z]/gi, '')));
+      const allKeys = Object.keys(currentSheet).filter(key => key !== matchingKey && key.replace(/[a-z]/gi, '') == matchingKey.replace(/[a-z]/gi, ''));
 
       const final2 = [];
 
@@ -86,6 +88,24 @@ function parseThisShit(downloadedFilePath) {
       classI++;
     }
 
+    // ABSENCE
+    final[finalIndex]["ABSENCE"] = [];
+    const absenceKey = Object.keys(currentSheet).find(key => {
+      const value = (typeof currentSheet[key].v == "string" ? currentSheet[key].v : "").trim().toLowerCase();
+      return value == "absence";
+    });
+    const teacherMap = await parseTeachers();
+    const allAbsenceKeys = Object.keys(currentSheet).filter(key => key !== absenceKey && key.replace(/[a-z]/gi, '') == absenceKey.replace(/[a-z]/gi, ''));
+    for (const absenceKeyCur of allAbsenceKeys) {
+      const value = currentSheet[absenceKeyCur]["v"].trim();
+      if (value.length === 0) {
+        continue;
+      }
+
+      const data = parseAbsence(value, teacherMap);
+      final[finalIndex]["ABSENCE"].push(data);
+    }
+
     finalIndex++;
   }
 
@@ -119,3 +139,5 @@ function parseThisShit(downloadedFilePath) {
 }
 
 module.exports = parseThisShit;
+
+// parseThisShit("downloads/table.xlsx");

scrape/scraper.js

@@ -144,7 +144,7 @@ async function clearDownloadsFolder() {
   console.log('Waiting for file:', downloadedFilePath);
   await waitForFile(downloadedFilePath);
 
-  parseThisShit(downloadedFilePath);
+  await parseThisShit(downloadedFilePath);
 
   await clearDownloadsFolder();
 

scrape/utils/parseAbsence.js

@@ -0,0 +1,44 @@
+function parseAbsence(str, teacherMap) {
+  str = str.trim().replace(/\s+/g, " ");
+
+  const result = {
+    teacher: null, // String or null
+    teacherCode: null, // String or null
+    type: null,   // "wholeDay" | "range" | "single" | "invalid"
+    hours: null // { from: number, to: number } or number
+  };
+
+  // Regex patterns (with flexible spaces)
+  const wholeDayPattern = /^([A-Za-z]+)$/;                   
+  const rangePattern = /^(\d+)\s*-\s*(\d+)\s+([A-Za-z]+)$/;  
+  const singleHourPattern = /^(\d+)\s+([A-Za-z]+)$/;         
+
+  if (rangePattern.test(str)) {
+    const [, from, to, teacherCode] = str.match(rangePattern);
+    result.teacher = teacherMap[teacherCode.toLowerCase()];
+    result.teacherCode = teacherCode;
+    result.type = "range";
+    result.hours = { from: parseInt(from), to: parseInt(to) };
+  } 
+  else if (singleHourPattern.test(str)) {
+    const [, hour, teacherCode] = str.match(singleHourPattern);
+    result.teacher = teacherMap[teacherCode.toLowerCase()];
+    result.teacherCode = teacherCode;
+    result.type = "single";
+    result.hours = parseInt(hour);
+  } 
+  else if (wholeDayPattern.test(str)) {
+    const [, teacherCode] = str.match(wholeDayPattern);
+    result.teacher = teacherMap[teacherCode.toLowerCase()];
+    result.teacherCode = teacherCode;
+    result.type = "wholeDay";
+  } 
+  else {
+    result.type = "invalid";
+    result.original = str;
+  }
+
+  return result;
+}
+
+module.exports = parseAbsence;

scrape/utils/parseTeachers.js

@@ -0,0 +1,25 @@
+const { default: axios } = require("axios");
+const cheerio = require("cheerio");
+
+async function parseTeachers() {
+  const url = "https://spsejecna.cz/ucitel";
+  const { data } = await axios.get(url);
+  const $ = cheerio.load(data);
+
+  const map = {};
+
+  $("main .contentLeftColumn li, main .contentRightColumn li").each((_, el) => {
+    const link = $(el).find("a");
+    const href = link.attr("href"); // e.g. "/ucitel/PA"
+    const text = link.text().trim(); // e.g. "Ing. Bc. Šárka Páltiková"
+
+    if (href) {
+      const key = href.split("/").pop().toLowerCase(); // get "pa"
+      map[key] = text;
+    }
+  });
+
+  return map;
+}
+
+module.exports = parseTeachers;