From 5d16021557738c71d79501a18e82e43410455de2 Mon Sep 17 00:00:00 2001
From: "Torsten Schulz (local)" <tsschulz@gmx.net>
Date: Wed, 15 Apr 2026 11:12:18 +0200
Subject: [PATCH] feat(bisaya-course): implement core pattern extraction and
 merging for exercises

- Added functions to derive core patterns from various exercise types, including gap-fill, transformation, and multiple-choice.
- Implemented logic to merge derived core patterns with existing lesson patterns, ensuring a minimum count for effective lesson content.
- Enhanced the `createBisayaCourseContent` function to update lessons with merged core patterns and provide detailed logging based on the VOCAB_STRICT_AUDIT environment variable.
---
 .../scripts/create-bisaya-course-content.js   | 133 ++++++++++++++++++
 1 file changed, 133 insertions(+)

diff --git a/backend/scripts/create-bisaya-course-content.js b/backend/scripts/create-bisaya-course-content.js
index 4dd4bc5..0b11724 100644
--- a/backend/scripts/create-bisaya-course-content.js
+++ b/backend/scripts/create-bisaya-course-content.js
@@ -85,6 +85,127 @@ function buildCorePatternGlossLookup(didactics) {
   return map;
 }
 
+function phraseWordCount(value) {
+  return countWords(value);
+}
+
+function isPhraseLike(value) {
+  return phraseWordCount(value) >= 2;
+}
+
+function normalizeQuotedPrompt(value) {
+  return normalizeText(String(value || '').replace(/[„“"']/g, ''));
+}
+
+function firstQuotedContent(value) {
+  const text = String(value || '');
+  const doubleQuote = text.match(/"([^"]+)"/);
+  if (doubleQuote?.[1]) return normalizeText(doubleQuote[1]);
+  const deQuote = text.match(/„([^“]+)“/);
+  if (deQuote?.[1]) return normalizeText(deQuote[1]);
+  const singleQuote = text.match(/'([^']+)'/);
+  if (singleQuote?.[1]) return normalizeText(singleQuote[1]);
+  return '';
+}
+
+function addUniquePattern(out, seen, target, gloss) {
+  const t = normalizeText(target);
+  const g = normalizeText(gloss);
+  if (!t || !g) return;
+  if (!isPhraseLike(t)) return;
+  const key = `${t.toLowerCase()}|${g.toLowerCase()}`;
+  if (seen.has(key)) return;
+  seen.add(key);
+  out.push({ target: t, gloss: g });
+}
+
+function extractPairsFromGapFill(questionData, answerData, out, seen) {
+  const text = String(questionData?.text || '');
+  const answers = Array.isArray(answerData?.answers) ? answerData.answers : [];
+  if (!text || answers.length === 0) return;
+
+  const hints = [];
+  const hintRegex = /\(([^)]+)\)/g;
+  let match = hintRegex.exec(text);
+  while (match) {
+    hints.push(normalizeText(match[1]));
+    match = hintRegex.exec(text);
+  }
+
+  for (let i = 0; i < Math.min(answers.length, hints.length); i += 1) {
+    addUniquePattern(out, seen, answers[i], hints[i]);
+  }
+}
+
+function extractPairsFromTransformation(questionData, answerData, out, seen) {
+  const src = normalizeText(questionData?.text || '');
+  const trg = normalizeText(answerData?.correct || '');
+  if (!src || !trg) return;
+  const sourceLanguage = normalizeText(questionData?.sourceLanguage || '').toLowerCase();
+  const targetLanguage = normalizeText(questionData?.targetLanguage || '').toLowerCase();
+  if (sourceLanguage === 'deutsch' || targetLanguage === 'bisaya') {
+    addUniquePattern(out, seen, trg, src);
+    return;
+  }
+  addUniquePattern(out, seen, src, trg);
+}
+
+function extractPairsFromMultipleChoice(questionData, answerData, out, seen) {
+  const question = normalizeText(questionData?.question || '');
+  const options = Array.isArray(questionData?.options) ? questionData.options : [];
+  const index = Number(answerData?.correctAnswer);
+  const correct = normalizeText(options[index] || '');
+  if (!question || !correct) return;
+
+  const quoted = firstQuotedContent(question);
+  const lower = question.toLowerCase();
+  if (!quoted) return;
+
+  if (lower.startsWith('was bedeutet')) {
+    addUniquePattern(out, seen, quoted, correct);
+    return;
+  }
+  if (lower.startsWith('wie sagt man')) {
+    addUniquePattern(out, seen, correct, normalizeQuotedPrompt(quoted));
+  }
+}
+
+function deriveLessonCorePatternsFromExercises(exercises) {
+  const out = [];
+  const seen = new Set();
+  (Array.isArray(exercises) ? exercises : []).forEach((exercise) => {
+    const questionData = exercise?.questionData || {};
+    const answerData = exercise?.answerData || {};
+    const type = String(questionData?.type || '');
+    if (type === 'gap_fill') {
+      extractPairsFromGapFill(questionData, answerData, out, seen);
+    } else if (type === 'transformation') {
+      extractPairsFromTransformation(questionData, answerData, out, seen);
+    } else if (type === 'multiple_choice') {
+      extractPairsFromMultipleChoice(questionData, answerData, out, seen);
+    }
+  });
+  return out;
+}
+
+function mergeCorePatternsForLesson(didactics, exerciseDerived, minCount = 8) {
+  const out = [];
+  const seen = new Set();
+  const basePatterns = Array.isArray(didactics?.corePatterns) ? didactics.corePatterns : [];
+
+  basePatterns.forEach((entry) => {
+    const normalized = normalizeCorePatternEntry(entry);
+    if (!normalized?.target || !normalized?.gloss) return;
+    addUniquePattern(out, seen, normalized.target, normalized.gloss);
+  });
+  exerciseDerived.forEach((entry) => {
+    addUniquePattern(out, seen, entry?.target, entry?.gloss);
+  });
+
+  if (out.length >= minCount) return out;
+  return out;
+}
+
 function sanitizeGapFillHintText(lessonTitle, text, answers, glossLookup) {
   const source = String(text || '');
   const normalizedAnswers = Array.isArray(answers)
@@ -4454,6 +4575,18 @@ async function createBisayaCourseContent() {
       
       // Erstelle Übungen
       const lessonDidactics = getLessonDidactics(lesson);
+      const derivedCorePatterns = deriveLessonCorePatternsFromExercises(exercises);
+      const mergedCorePatterns = mergeCorePatternsForLesson(lessonDidactics, derivedCorePatterns, 8);
+      if (mergedCorePatterns.length >= 8) {
+        await lesson.update({ corePatterns: mergedCorePatterns });
+        if (process.env.VOCAB_STRICT_AUDIT === '1') {
+          console.log(`   ✅ [${lesson.title}] corePatterns auf ${mergedCorePatterns.length} Satzphrasen aktualisiert`);
+        }
+      } else if (process.env.VOCAB_STRICT_AUDIT === '1') {
+        console.warn(
+          `   ⚠️  [${lesson.title}] Nur ${mergedCorePatterns.length} Satzphrasen mit sicherer Gloss gefunden (Ziel: 8)`
+        );
+      }
       let exerciseNumber = 1;
       for (const exerciseData of exercises) {
         const { exercise, fixes, warnings } = sanitizeExerciseForConsistency(