fix: Absence parser

scrape/utils/parseAbsence.js

@@ -1,84 +1,106 @@
 export default function parseAbsence(str, teacherMap) {
-  str = str.trim().replace(/\s+/g, " ");
+  const LAST_HOUR = 10;
 
+  const s = (str ?? "").trim().replace(/\s+/g, " ");
   const result = {
-    teacher: null,
-    teacherCode: null,
-    type: null, // "wholeDay" | "range" | "single" | "invalid"
-    hours: null
+    teacher: null,        // String or null
+    teacherCode: null,    // String or null
+    type: null,           // "wholeDay" | "range" | "single" | "invalid"
+    hours: null           // { from, to } or number
   };
 
-  // Patterns (teacher can be before or after the hours)
-  const wholeDayPattern = /^([A-Za-z]+)$/;
-  const rangePattern = /^(\d+)\s*-\s*(\d+)\s+([A-Za-z]+)$|^([A-Za-z]+)\s+(\d+)\s*-\s*(\d+)$/;
-  const singleHourPattern = /^(\d+)\s+([A-Za-z]+)$|^([A-Za-z]+)\s+(\d+)$/;
-  const plusPattern = /^(\d+)\+\s+([A-Za-z]+)$|^([A-Za-z]+)\s+(\d+)\+$/;
-
-  if (rangePattern.test(str)) {
-    const match = str.match(rangePattern);
-    // Case 1: hours first, then teacher
-    if (match[1] && match[2] && match[3]) {
-      const [, from, to, teacherCode] = match;
-      result.teacherCode = teacherCode;
-      result.teacher = teacherMap[teacherCode.toLowerCase()];
-      result.type = "range";
-      result.hours = { from: parseInt(from), to: parseInt(to) };
-    }
-    // Case 2: teacher first, then hours
-    else {
-      const [, , , teacherCode, from, to] = match;
-      result.teacherCode = teacherCode;
-      result.teacher = teacherMap[teacherCode.toLowerCase()];
-      result.type = "range";
-      result.hours = { from: parseInt(from), to: parseInt(to) };
-    }
-  }
-  else if (plusPattern.test(str)) {
-    const match = str.match(plusPattern);
-    if (match[1] && match[2]) {
-      // hours first
-      const [, from, teacherCode] = match;
-      result.teacherCode = teacherCode;
-      result.teacher = teacherMap[teacherCode.toLowerCase()];
-      result.type = "range";
-      result.hours = { from: parseInt(from), to: 10 };
-    } else {
-      // teacher first
-      const [, , teacherCode, from] = match;
-      result.teacherCode = teacherCode;
-      result.teacher = teacherMap[teacherCode.toLowerCase()];
-      result.type = "range";
-      result.hours = { from: parseInt(from), to: 10 };
-    }
-  }
-  else if (singleHourPattern.test(str)) {
-    const match = str.match(singleHourPattern);
-    if (match[1] && match[2]) {
-      // hours first
-      const [, hour, teacherCode] = match;
-      result.teacherCode = teacherCode;
-      result.teacher = teacherMap[teacherCode.toLowerCase()];
-      result.type = "single";
-      result.hours = parseInt(hour);
-    } else {
-      // teacher first
-      const [, , teacherCode, hour] = match;
-      result.teacherCode = teacherCode;
-      result.teacher = teacherMap[teacherCode.toLowerCase()];
-      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 {
+  if (!s) {
     result.type = "invalid";
-    result.original = str;
+    result.original = s;
+    return result;
   }
 
+  const isTeacher = (t) => /^[A-Za-z]+$/.test(t);
+
+  const parseSpec = (spec) => {
+    let m;
+    // "5+"  -> range 5..LAST_HOUR
+    if ((m = spec.match(/^(\d+)\+$/))) {
+      const from = parseInt(m[1], 10);
+      if (from >= 1 && from <= LAST_HOUR) {
+        return { kind: "range", value: { from, to: LAST_HOUR } };
+      }
+      return null;
+    }
+    // "5-7" -> range 5..7
+    if ((m = spec.match(/^(\d+)-(\d+)$/))) {
+      const from = parseInt(m[1], 10);
+      const to = parseInt(m[2], 10);
+      if (from >= 1 && to >= 1 && from <= to && to <= LAST_HOUR) {
+        return { kind: "range", value: { from, to } };
+      }
+      return null;
+    }
+    // "5"   -> single 5
+    if ((m = spec.match(/^(\d+)$/))) {
+      const hour = parseInt(m[1], 10);
+      if (hour >= 1 && hour <= LAST_HOUR) {
+        return { kind: "single", value: hour };
+      }
+      return null;
+    }
+    return null;
+  };
+
+  const parts = s.split(" ");
+  if (parts.length === 1) {
+    const t = parts[0];
+    if (isTeacher(t)) {
+      result.teacherCode = t;
+      result.teacher = teacherMap?.[t.toLowerCase()] ?? null;
+      result.type = "wholeDay";
+      return result;
+    }
+    result.type = "invalid";
+    result.original = s;
+    return result;
+  }
+
+  if (parts.length === 2) {
+    const [a, b] = parts;
+
+    // Teacher first: "Nm 5", "Nm 5-7", "Nm 5+"
+    if (isTeacher(a)) {
+      const spec = parseSpec(b);
+      if (!spec) {
+        result.type = "invalid";
+        result.original = s;
+        return result;
+      }
+      result.teacherCode = a;
+      result.teacher = teacherMap?.[a.toLowerCase()] ?? null;
+      result.type = spec.kind === "range" ? "range" : "single";
+      result.hours = spec.value;
+      return result;
+    }
+
+    // Teacher last: "5 Nm", "5-7 Nm", "5+ Nm"
+    if (isTeacher(b)) {
+      const spec = parseSpec(a);
+      if (!spec) {
+        result.type = "invalid";
+        result.original = s;
+        return result;
+      }
+      result.teacherCode = b;
+      result.teacher = teacherMap?.[b.toLowerCase()] ?? null;
+      result.type = spec.kind === "range" ? "range" : "single";
+      result.hours = spec.value;
+      return result;
+    }
+
+    result.type = "invalid";
+    result.original = s;
+    return result;
+  }
+
+  // Anything else is invalid
+  result.type = "invalid";
+  result.original = s;
   return result;
 }