feat: Use better lib for parsing and minor changes

2025-09-07 20:01:15 +02:00
parent fac36754fd
commit 6a6a32f47a
3 changed files with 1104 additions and 174 deletions
--- a/scrape/parse.js
+++ b/scrape/parse.js
@@ -1,152 +1,250 @@
-import XLSX from "xlsx";
-import fs from "fs";
-import parseAbsence from "./utils/parseAbsence.js";
-import parseTeachers from "./utils/parseTeachers.js";
+import ExcelJS from "exceljs"
+import fs from "fs"
+import parseAbsence from "./utils/parseAbsence.js"
+import parseTeachers from "./utils/parseTeachers.js"

 export default async function parseThisShit(downloadedFilePath) {
-  const workbook = XLSX.readFile(downloadedFilePath);
-  const sheetNames = workbook.SheetNames;
+  const workbook = new ExcelJS.Workbook()
+  await workbook.xlsx.readFile(downloadedFilePath)

-  const dateRegex = /^(pondělí|úterý|středa|čtvrtek|pátek)\s+(\d{1,2})\.\s*(\d{1,2})\.\s*(20\d{2})/i;
+  const sheetNames = workbook.worksheets.map((sheet) => sheet.name)
+
+  const dateRegex = /^(pondělí|úterý|středa|čtvrtek|pátek)\s+(\d{1,2})\.\s*(\d{1,2})\.\s*(20\d{2})/i

  // Get today's date for comparison
  function getCurrentDateObject() {
-    const now = new Date();
-    return new Date(now.getFullYear(), now.getMonth(), now.getDate());
+    const now = new Date()
+    return new Date(now.getFullYear(), now.getMonth(), now.getDate())
  }

-  const today = getCurrentDateObject();
+  const today = getCurrentDateObject()

-  const upcomingSheets = sheetNames.filter(name => {
-    const match = name.match(dateRegex);
-    if (!match) return false;
+  const upcomingSheets = sheetNames.filter((name) => {
+    const match = name.match(dateRegex)
+    if (!match) return false

-    const day = parseInt(match[2], 10);
-    const month = parseInt(match[3], 10) - 1; // JavaScript months are 0-indexed
-    const year = parseInt(match[4], 10);
+    const day = Number.parseInt(match[2], 10)
+    const month = Number.parseInt(match[3], 10) - 1 // JavaScript months are 0-indexed
+    const year = Number.parseInt(match[4], 10)

-    const sheetDate = new Date(year, month, day);
+    const sheetDate = new Date(year, month, day)

-    return sheetDate >= today;
-  });
+    return sheetDate >= today
+  })

-  const final = [];
+  const final = []

-  let finalIndex = 0;
+  let finalIndex = 0
  for (const key of upcomingSheets) {
-    const currentSheet = workbook.Sheets[key];
-    final.push({});
+    const currentSheet = workbook.getWorksheet(key)
+    final.push({})

-    const regex = /[AEC][0-4][a-c]?\s*\/.*/s;
-    const prefixRegex = /[AEC][0-4][a-c]?/;
-    const classes = [];
-    const matchingKeys = Object.keys(currentSheet).filter(key => {
-      const value = currentSheet[key];
+    const regex = /[AEC][0-4][a-c]?\s*\/.*/s
+    const prefixRegex = /[AEC][0-4][a-c]?/
+    const classes = []
+    const matchingKeys = []

-      const testResult = regex.test(value.v);
-      if (testResult) {
-        const prefixMatch = value.v.match(prefixRegex);
-        if (prefixMatch) {
-          const prefix = prefixMatch[0];
-          classes.push(prefix);
+    currentSheet.eachRow((row, rowNumber) => {
+      row.eachCell((cell, colNumber) => {
+        const cellAddress = cell.address
+        const value = cell.value
+
+        if (value && typeof value === "string") {
+          const testResult = regex.test(value)
+          if (testResult) {
+            const prefixMatch = value.match(prefixRegex)
+            if (prefixMatch) {
+              const prefix = prefixMatch[0]
+              classes.push(prefix)
+            }
+            matchingKeys.push(cellAddress)
+          }
        }
-      }
-
-      return testResult;
-    });
+      })
+    })

    function letterToNumber(letter) {
-      return letter.toLowerCase().charCodeAt(0) - 'a'.charCodeAt(0);
+      return letter.toLowerCase().charCodeAt(0) - "a".charCodeAt(0)
    }

    // For each class
-    let classI = 0;
+    let classI = 0
    for (const matchingKey of matchingKeys) {
-      const allKeys = Object.keys(currentSheet).filter(key => key !== matchingKey && key.replace(/[a-z]/gi, '') == matchingKey.replace(/[a-z]/gi, ''));
-      let final2 = [];
+      const matchingCell = currentSheet.getCell(matchingKey)
+      const rowNumber = matchingCell.row
+      const allKeys = []
+
+      // Get all cells in the same row
+      const row = currentSheet.getRow(rowNumber)
+      row.eachCell((cell, colNumber) => {
+        if (cell.address !== matchingKey) {
+          allKeys.push(cell.address)
+        }
+      })
+
+      let final2 = []

      for (const key of allKeys) {
-        const parsedKey = letterToNumber(key.replace(/[0-9]/gi, ''));
-        let d = true;
+        const cell = currentSheet.getCell(key)
+        const parsedKey = letterToNumber(key.replace(/[0-9]/gi, ""))
+        let d = true

        try {
-          const regex = /^úklid\s+(?:\d+\s+)?[A-Za-z]{2}$/;
-          if (regex.test(currentSheet[key]["w"].trim()) || currentSheet[key]["w"].trim().length == 0) {
-            d = false;
+          const regex = /^úklid\s+(?:\d+\s+)?[A-Za-z]{2}$/
+          const cellText = cell.text || ""
+          if (regex.test(cellText.trim()) || cellText.trim().length == 0 || cell.fill.fgColor === undefined) {
+            d = false
          }
        } catch {}

        if (d) {
-          final2[parsedKey] = currentSheet[key]["w"];
+          if (cell.fill.fgColor?.argb == "FFFFFF00") {
+            cell.text += "\n(bude upřesněno)";
+          }
+          final2[parsedKey] = cell.text || ""
        } else {
-          final2[parsedKey] = null;
+          final2[parsedKey] = null
        }
      }

-      final2 = Array.from(final2, item => item === undefined ? null : item);
+      final2 = Array.from(final2, (item) => (item === undefined ? null : item))
      while (final2.length < 10) {
-        final2.push(null);
+        final2.push(null)
      }

-      final[finalIndex][classes[classI]] = final2.slice(1, 11);;
+      final[finalIndex][classes[classI]] = final2.slice(1, 11)

-      classI++;
+      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, ''));
-    let i = 0;
-    for (const absenceKeyCur of allAbsenceKeys) {
-      if (i >= 10) {
-        break; // stop once 10 items are added
-      }
-      i++;
+    final[finalIndex]["ABSENCE"] = []
+    let absenceKey = null

-      const value = currentSheet[absenceKeyCur]["v"].trim();
-      if (value.length === 0) {
-        continue;
-      }
+    currentSheet.eachRow((row, rowNumber) => {
+      row.eachCell((cell, colNumber) => {
+        const value = (typeof cell.value === "string" ? cell.value : "").trim().toLowerCase()
+        if (value === "absence") {
+          absenceKey = cell.address
+        }
+      })
+    })

-      const data = parseAbsence(value, teacherMap);
-      final[finalIndex]["ABSENCE"].push(data);
+    if (absenceKey) {
+      const teacherMap = await parseTeachers()
+      const absenceCell = currentSheet.getCell(absenceKey)
+      const rowNumber = absenceCell.row
+      const allAbsenceKeys = []
+
+      // Get all cells in the same row as absence
+      const row = currentSheet.getRow(rowNumber)
+      row.eachCell((cell, colNumber) => {
+        if (cell.address !== absenceKey) {
+          allAbsenceKeys.push(cell.address)
+        }
+      })
+
+      let i = 0
+      for (const absenceKeyCur of allAbsenceKeys) {
+        if (i >= 10) {
+          break // stop once 10 items are added
+        }
+        i++
+
+        const cell = currentSheet.getCell(absenceKeyCur)
+        const value = (cell.value || "").toString().trim()
+        if (value.length === 0) {
+          continue
+        }
+
+        const data = parseAbsence(value, teacherMap)
+        final[finalIndex]["ABSENCE"].push(data)
+      }
    }

-    finalIndex++;
+    finalIndex++
  }

  const data = {
    schedule: final,
-    props: upcomingSheets.map(str => {
-      const dateMatch = str.match(/(\d{1,2})\.\s*(\d{1,2})\.\s*(\d{4})/);
+    props: upcomingSheets.map((str) => {
+      const dateMatch = str.match(/(\d{1,2})\.\s*(\d{1,2})\.\s*(\d{4})/)

-      let date = null;
+      let date = null

      if (dateMatch) {
-        const day = parseInt(dateMatch[1], 10);
-        const month = parseInt(dateMatch[2], 10);
-        const year = parseInt(dateMatch[3], 10);
+        const day = Number.parseInt(dateMatch[1], 10)
+        const month = Number.parseInt(dateMatch[2], 10)
+        const year = Number.parseInt(dateMatch[3], 10)

-        date = new Date(year, month - 1, day);
+        date = new Date(year, month - 1, day)
      }

-      const isPriprava = str.toLowerCase().normalize("NFD").replace(/[\u0300-\u036f]/g, "").includes("priprava");
+      const isPriprava = str
+        .toLowerCase()
+        .normalize("NFD")
+        .replace(/[\u0300-\u036f]/g, "")
+        .includes("priprava")

      return {
        date: date
-          ? `${date.getFullYear()}-${String(date.getMonth() + 1).padStart(2, '0')}-${String(date.getDate()).padStart(2, '0')}`
+          ? `${date.getFullYear()}-${String(date.getMonth() + 1).padStart(2, "0")}-${String(date.getDate()).padStart(2, "0")}`
          : null,
        priprava: isPriprava,
-      };
+      }
    }),
  }

-  fs.writeFileSync('db/current.json', JSON.stringify(data));
+  fs.writeFileSync("db/current.json", JSON.stringify(data))
 }

-// parseThisShit("downloads/table.xlsx");
+parseThisShit("downloads/table.xlsx")
+
+
+/**
+ * Get cell background color as hex (ARGB or theme + tint)
+ * @param {import('exceljs').Cell} cell
+ * @returns {string|null} Hex color like "#FF0000" or null if no fill
+ */
+function getCellHexColor(cell) {
+    if (!cell.fill || !cell.fill.fgColor) return null;
+
+    const color = cell.fill.fgColor;
+
+    if (color.argb) {
+        // Direct ARGB color
+        return `#${color.argb.substring(2)}`;
+    }
+
+    if (color.theme !== undefined) {
+        // Default Excel 2013 theme colors
+        const themeColors = [
+            'FFFFFF', '000000', 'EEECE1', '1F497D', '4F81BD', 'C0504D', '9BBB59', '8064A2', '4BACC6', 'F79646'
+        ];
+        let hex = themeColors[color.theme] || '000000';
+        const tint = color.tint || 0;
+
+        // Apply tint inline
+        let r = parseInt(hex.substring(0, 2), 16);
+        let g = parseInt(hex.substring(2, 4), 16);
+        let b = parseInt(hex.substring(4, 6), 16);
+
+        if (tint < 0) {
+            r = r * (1 + tint);
+            g = g * (1 + tint);
+            b = b * (1 + tint);
+        } else {
+            r = r + (255 - r) * tint;
+            g = g + (255 - g) * tint;
+            b = b + (255 - b) * tint;
+        }
+
+        r = Math.round(Math.min(255, Math.max(0, r)));
+        g = Math.round(Math.min(255, Math.max(0, g)));
+        b = Math.round(Math.min(255, Math.max(0, b)));
+
+        return `#${r.toString(16).padStart(2,'0')}${g.toString(16).padStart(2,'0')}${b.toString(16).padStart(2,'0')}`;
+    }
+
+    return null;
+}