feat: v2

Support of teacher absence format: "za Vn zastupuje Jk"

scrape/parse.js

@@ -12,8 +12,8 @@
  * GNU General Public License for more details.
  */
 
-import parseV1 from "./parse/v1.js";
+import parseV1V2 from "./parse/v1_v2.js";
 
 export default async function parseThisShit(downloadedFilePath) {
-  await parseV1(downloadedFilePath)
+  await parseV1V2(downloadedFilePath)
 }

scrape/parse/v1_v2.js

@@ -17,7 +17,7 @@ import fs from "fs"
 import parseAbsence from "../utils/parseAbsence.js"
 import parseTeachers from "../utils/parseTeachers.js"
 
-export default async function parseV1(downloadedFilePath) {
+export default async function parseV1V2(downloadedFilePath) {
   const workbook = new ExcelJS.Workbook();
   await workbook.xlsx.readFile(downloadedFilePath);
   const teacherMap = await parseTeachers();
@@ -216,7 +216,29 @@ export default async function parseV1(downloadedFilePath) {
     }
   }
 
-  fs.writeFileSync("db/v1.json", JSON.stringify(data, null, 2));
+  fs.writeFileSync("db/v2.json", JSON.stringify(data, null, 2));
+
+  // Modify the data for v1
+  const copy = JSON.parse(JSON.stringify(data));
+
+  copy.schedule.forEach(day => {
+    if (!Array.isArray(day.ABSENCE)) return;
+
+    day.ABSENCE = day.ABSENCE.map(old => {
+      if (old.type === "zastoupen") {
+        console.log("pepa")
+        return {
+          teacher: old.teacher,
+          teacherCode: old.teacherCode,
+          type: "wholeDay",
+          hours: null
+        };
+      }
+      return old;
+    });
+  });
+
+  fs.writeFileSync("db/v1.json", JSON.stringify(copy, null, 2))
 }
 
-//parseThisShit("downloads/table.xlsx")
+parseV1V2("db/current.xlsx")

scrape/scraper.js

@@ -67,7 +67,7 @@ async function handleError(page, err) {
   let browser, page;
   try {
     browser = await puppeteer.launch({
-      headless: 'new',
+      headless: false,
       userDataDir: VOLUME_PATH,
       args: ['--no-sandbox', '--disable-setuid-sandbox']
     });

scrape/utils/parseAbsence.js

@@ -98,6 +98,7 @@ export default function parseAbsence(input, teacherMap = {}) {
   const markConsumed = (start, end) => consumed.push([start, end]);
   const isConsumed = (i) => consumed.some(([a, b]) => i >= a && i < b);
 
+  // 1. Teachers with specific hours (e.g. "Ab 1-4")
   const teacherListThenSpecRe =
     /([A-Za-z]+(?:[,;]\s?[A-Za-z]+)*)(?:\s*)(\d+(?:\+|-\d+|,\d+)?)(?![A-Za-z])/g;
 
@@ -115,6 +116,7 @@ export default function parseAbsence(input, teacherMap = {}) {
     markConsumed(matchStart, matchEnd);
   }
 
+  // 2. Teachers with "-exk" suffix
   const teacherExkRe = /([A-Za-z]+)-exk/gi;
   while ((m = teacherExkRe.exec(s)) !== null) {
     const matchStart = m.index;
@@ -132,7 +134,34 @@ export default function parseAbsence(input, teacherMap = {}) {
     markConsumed(matchStart, matchEnd);
   }
 
-  // Standalone teachers → whole day
+  // ---------------------------------------------------------
+  // 3. Substitution Pattern: "za Vn zastupuje Jk"
+  // ---------------------------------------------------------
+  const substitutionRe = /za\s+([A-Za-z]+)\s+zastupuje\s+([A-Za-z]+)/gi;
+  while ((m = substitutionRe.exec(s)) !== null) {
+    const matchStart = m.index;
+    const matchEnd = substitutionRe.lastIndex;
+    if (isConsumed(matchStart)) continue;
+
+    const missingCode = m[1];
+    const substituteCode = m[2];
+    
+    const missingResolved = resolveTeacher(missingCode, teacherMap);
+    const subResolved = resolveTeacher(substituteCode, teacherMap);
+
+    results.push({
+      teacher: missingResolved.name,
+      teacherCode: missingResolved.code.toLowerCase(),
+      type: "zastoupen",
+      zastupuje: {
+        teacher: subResolved.name,
+        teacherCode: subResolved.code.toLowerCase()
+      }
+    });
+
+    markConsumed(matchStart, matchEnd);
+  }
+
   const teacherOnlyRe = /([A-Za-z]+(?:[,;]\s?[A-Za-z]+)*)/g;
   while ((m = teacherOnlyRe.exec(s)) !== null) {
     const matchStart = m.index;
@@ -141,6 +170,9 @@ export default function parseAbsence(input, teacherMap = {}) {
 
     const tList = m[1].split(/[,;]\s*/).filter(Boolean);
     tList.forEach((t) => {
+      const lowerT = t.toLowerCase();
+      if (lowerT === 'za' || lowerT === 'zastupuje') return;
+
       if (isTeacherToken(t)) results.push(makeResult(t, null, teacherMap));
       else
         results.push({
@@ -154,12 +186,12 @@ export default function parseAbsence(input, teacherMap = {}) {
     markConsumed(matchStart, matchEnd);
   }
 
-  // Bare specs without teacher → invalid
+  // 5. Bare specs without teacher → invalid
   const specOnlyRe = /\b(\d+(?:\+|-\d+|,\d+)?)\b/g;
   while ((m = specOnlyRe.exec(s)) !== null) {
     const matchStart = m.index;
     if (isConsumed(matchStart)) continue;
-    if (m[1].trim() == "exk") continue; // Exkurze, will be implemented later
+    if (m[1].trim() == "exk") continue; 
 
     results.push({
       type: "invalid",