@zodic/shared 0.0.231 → 0.0.232

package/app/services/ConceptService.ts

@@ -551,10 +551,21 @@ export class ConceptService {
     structuredContentEN: StructuredConceptContent;
     structuredContentPT: StructuredConceptContent;
   } {
-    console.log(`📌 [START] Parsing structured content for ${conceptSlug} from ChatGPT response.`);
+    console.log(
+      `📌 [START] Parsing structured content for ${conceptSlug} from ChatGPT response.`
+    );
   
-    // ✅ Step 1: Log the entire raw response before processing
-    console.log('🔹 [RAW RESPONSE]:', response);
+    // ✅ Step 1: Clean artifacts before processing
+    let cleanedResponse = response
+      .replace(/[*#•]/g, '') // Remove bullet points, bold markers, headings
+      .replace(/---+/g, '') // Remove dividers (like "---")
+      .replace(/\n\s*\n/g, '\n\n') // Normalize multiple line breaks
+      .trim();
+  
+    console.log(
+      '🔹 [CLEANED RESPONSE]:',
+      cleanedResponse.slice(0, 500) + '...'
+    );
   
     // ✅ Step 2: Define Section Titles Per Concept
     const conceptSections: Record<string, { en: string[]; pt: string[] }> = {
@@ -664,31 +675,37 @@ export class ConceptService {
     }
   
     // ✅ Step 3: Extract English and Portuguese content separately
-    const enMatches = response.match(/EN:\s*([\s\S]+?)\s*PT:/);
-    const ptMatches = response.match(/PT:\s*([\s\S]+)/);
+    const enMatches = cleanedResponse.match(/EN:\s*([\s\S]+?)\s*PT:/);
+    const ptMatches = cleanedResponse.match(/PT:\s*([\s\S]+)/);
   
     if (!enMatches || !ptMatches) {
-      console.error('❌ [ERROR] Missing English or Portuguese content in response.');
+      console.error(
+        '❌ [ERROR] Missing English or Portuguese content in response.'
+      );
       throw new Error('❌ Missing English or Portuguese content in response.');
     }
   
-    let enContent = enMatches[1].trim();
-    let ptContent = ptMatches[1].trim();
-  
-    console.log('✅ [MATCH SUCCESS] Extracted English Content:', enContent.slice(0, 500) + '...');
-    console.log('✅ [MATCH SUCCESS] Extracted Portuguese Content:', ptContent.slice(0, 500) + '...');
+    const enContent = enMatches[1].trim();
+    const ptContent = ptMatches[1].trim();
   
-    // ✅ Step 4: Debug **why** PT content has "EN:"
-    if (ptContent.includes('EN:')) {
-      console.warn('⚠️ PT content contains "EN:". Check if parsing was incorrect.');
-    }
+    console.log(
+      '✅ [MATCH SUCCESS] Extracted English Content:',
+      enContent.slice(0, 500) + '...'
+    );
+    console.log(
+      '✅ [MATCH SUCCESS] Extracted Portuguese Content:',
+      ptContent.slice(0, 500) + '...'
+    );
   
-    // ✅ Step 5: Extract structured sections
+    // ✅ Step 4: Extract structured sections
     function extractSections(text: string, sectionTitles: string[]) {
       return sectionTitles.map((title) => {
         console.log(`🔍 [PROCESSING] Extracting section: "${title}"`);
   
-        const regex = new RegExp(`\\d+\\.\\s*${title}:\\s*([\\s\\S]+?)(?=\\n\\d+\\.\\s*[A-Z]|$)`);
+        // ✅ Improved regex: Detects sections even if AI formatting changes
+        const regex = new RegExp(
+          `\\d+\\.\\s*${title}:\\s*([\\s\\S]+?)(?=\\n\\d+\\.\\s*[A-Z]|$)`
+        );
         const match = text.match(regex);
   
         if (!match) {
@@ -696,11 +713,12 @@ export class ConceptService {
           return { type: 'section', title, content: ['❌ Section Missing'] };
         }
   
+        // ✅ Split content into paragraphs and clean them
         const paragraphs = match[1]
           .trim()
-          .split(/\n{2,}/)
+          .split(/\n{2,}/) // Handles cases where paragraphs are separated by multiple new lines
           .map((p) => p.trim())
-          .filter((p) => p.length > 0);
+          .filter((p) => p.length > 0); // Removes empty paragraphs
   
         console.log(`✅ [EXTRACTED] "${title}" - Parsed Content:`, paragraphs);
   
@@ -708,13 +726,22 @@ export class ConceptService {
       });
     }
   
+    // ✅ Return parsed and cleaned structured content
     const structuredContentEN = extractSections(enContent, sectionsEN);
     const structuredContentPT = extractSections(ptContent, sectionsPT);
   
-    console.log('🎯 [FINAL RESULT] Parsed English Content:', JSON.stringify(structuredContentEN, null, 2));
-    console.log('🎯 [FINAL RESULT] Parsed Portuguese Content:', JSON.stringify(structuredContentPT, null, 2));
+    console.log(
+      '🎯 [FINAL RESULT] Parsed English Content:',
+      JSON.stringify(structuredContentEN, null, 2)
+    );
+    console.log(
+      '🎯 [FINAL RESULT] Parsed Portuguese Content:',
+      JSON.stringify(structuredContentPT, null, 2)
+    );
   
-    console.log('✅ [COMPLETE] Structured content parsing finished successfully.');
+    console.log(
+      '✅ [COMPLETE] Structured content parsing finished successfully.'
+    );
   
     return {
       structuredContentEN,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@zodic/shared",
-  "version": "0.0.231",
+  "version": "0.0.232",
   "module": "index.ts",
   "type": "module",
   "publishConfig": {