Enhance TikTok heuristic parser for unstructured single-line captions

server.js

@@ -1360,8 +1360,39 @@ function isTikTokUrl(url) {
   }
 }
 
+// Helper: Clean raw TikTok caption by stripping trailing hashtags and emojis
+function cleanTikTokCaption(raw) {
+  if (!raw) return '';
+  let text = raw.replace(/(?:\s*#[^\s#]+)+\s*$/g, '').trim();
+  text = text.replace(/[\uFE00-\uFE0F\uD83C-\uDBFF\uDC00-\uDFFF\u2600-\u26FF\u2700-\u27BF]+$/, '').trim();
+  return text;
+}
+
+// Helper: Split single-line or unstructured social captions into readable candidate phrases
+function splitCaptionIntoPhrases(text) {
+  const breakPhrases = [
+    'Mix it', 'Whisk', 'Add in', 'Add the', 'Add one', 'In a separate bowl:?', 'In a separate',
+    'In a measuring cup', 'In a greased and floured', 'In a greased', 'In a large bowl',
+    'For the apple:?', 'For the glaze:?', 'For the sauce:?', 'For the dough:?',
+    'Then add', 'Then other', 'Then 1/2', 'Swirl it', 'Sprinkle',
+    'Bake at', 'Glaze:?'
+  ];
+
+  let working = text;
+  for (const phrase of breakPhrases) {
+    const re = new RegExp('\\b(' + phrase + ')', 'gi');
+    working = working.replace(re, '\n$1');
+  }
+
+  const qtyBreakRegex = /(?<=[a-zA-Z\)\,\.\!])\s+(?=(?:(?:\d+\s*(?:&|\+)?\s*)?(?:\d+[\/\.]\d+|[¼½¾⅓⅔⅛⅜⅝⅞]|\d+)\s*(?:cups?|c|tbsp|tablespoons?|tsp|teaspoons?|oz|ounces?|lbs?|sticks?|large|small|medium|cans?)\b))/gi;
+  working = working.replace(qtyBreakRegex, '\n');
+
+  return working.split('\n').map(s => s.trim()).filter(Boolean);
+}
+
 // Helper: Heuristic parser to extract recipe structure from a social video caption
-function parseRecipeCaptionHeuristic(caption, author, cover, sourceUrl) {
+function parseRecipeCaptionHeuristic(rawCaption, author, cover, sourceUrl) {
+  const caption = cleanTikTokCaption(rawCaption);
   if (!caption) return null;
 
   let title = '';
@@ -1369,84 +1400,62 @@ function parseRecipeCaptionHeuristic(caption, author, cover, sourceUrl) {
   const ingredients = [];
   const instructions = [];
 
-  const ingMatch = caption.match(/\b(?:ingredients?|what you(?:'ll)? need)\b/i);
+  const ingHeaderMatch = caption.match(/\b(?:ingredients?|what you(?:'ll)? need)\s*[:\-]?/i);
-  const instMatch = caption.match(/\b(?:instructions?|directions?|method|steps?)\b/i);
+  let remainingText = caption;
 
-  if (ingMatch && instMatch && instMatch.index > ingMatch.index) {
+  if (ingHeaderMatch) {
-    const introRaw = caption.slice(0, ingMatch.index).trim();
+    const rawTitle = caption.slice(0, ingHeaderMatch.index).trim();
-    const ingRaw = caption.slice(ingMatch.index + ingMatch[0].length, instMatch.index).trim();
+    title = rawTitle.replace(/[:\-.]+$/, '').trim();
-    const instRaw = caption.slice(instMatch.index + instMatch[0].length).trim();
+    remainingText = caption.slice(ingHeaderMatch.index + ingHeaderMatch[0].length).trim();
-
-    description = introRaw.replace(/#[\w-]+/g, '').trim();
-    const firstSentence = description.match(/^([^.!?\n]+[.!?])/);
-    title = firstSentence ? firstSentence[1].replace(/[:\-.]+$/, '').trim() : (description.slice(0, 60) || 'TikTok Recipe');
-
-    // Parse Ingredients (handles newlines OR inline quantity splits)
-    if (ingRaw.includes('\n')) {
-      ingRaw.split(/\r?\n/).forEach(line => {
-        const cleaned = line.replace(/^[•◦⁃∙*\-+]\s*/, '').replace(/#[\w-]+/g, '').trim();
-        if (cleaned) ingredients.push(cleaned);
-      });
-    } else {
-      const splitRegex = /(?<=[a-zA-Z\)\,\.\!])\s+(?=(?:\d+|[¼½¾⅓⅔⅛⅜⅝⅞]))/;
-      const parts = ingRaw.split(new RegExp(splitRegex, 'g')).map(p => p.trim()).filter(Boolean);
-      parts.forEach(part => {
-        const cleaned = part.replace(/^[•◦⁃∙*\-+]\s*/, '').replace(/#[\w-]+/g, '').trim();
-        if (cleaned.length > 2) ingredients.push(cleaned);
-      });
-    }
-
-    // Parse Instructions
-    if (instRaw.includes('\n')) {
-      instRaw.split(/\r?\n/).forEach(line => {
-        const cleaned = line.replace(/^\d+[\.\)]\s*/, '').replace(/^[•◦⁃∙*\-+]\s*/, '').replace(/#[\w-]+/g, '').trim();
-        if (cleaned) instructions.push(cleaned);
-      });
-    } else {
-      const stepRegex = /(?:\b\d+[\.\)]\s+)/g;
-      const steps = instRaw.split(stepRegex).map(s => s.replace(/#[\w-]+/g, '').trim()).filter(Boolean);
-      steps.forEach(step => {
-        if (step.length > 3) instructions.push(step);
-      });
-    }
   } else {
-    const lines = caption.split(/\r?\n/).map(l => l.trim()).filter(Boolean);
+    const firstSentence = caption.match(/^([^.!?\n]+[.!?])/);
-    let currentSection = 'intro';
+    title = firstSentence ? firstSentence[1].replace(/[:\-.]+$/, '').trim() : caption.slice(0, 50).trim();
-    for (const line of lines) {
+    remainingText = caption.slice(title.length).trim();
-      if (/^(?:ingredients?|what you need)/i.test(line)) {
+  }
-        currentSection = 'ingredients';
+
-        continue;
+  const segments = splitCaptionIntoPhrases(remainingText);
-      }
+
-      if (/^(?:instructions?|directions?|method|steps?)/i.test(line)) {
+  const instructionStarters = /^(?:mix|whisk|add|bake|stir|fold|combine|in a|then|swirl|sprinkle|pour|heat|cook|chill|serve|place|cut|slice|dice)\b/i;
-        currentSection = 'instructions';
+  const isIngredientPattern = /^(?:(?:\d+\s*(?:&|\+)?\s*)?(?:\d+[\/\.]\d+|[¼½¾⅓⅔⅛⅜⅝⅞]|\d+)\s*(?:cups?|c|tbsp|tablespoons?|tsp|teaspoons?|oz|ounces?|lbs?|sticks?|large|small|medium|cans?|pinch|pkg)?\b|[•◦⁃∙*\-+])/i;
-        continue;
+
-      }
+  for (let seg of segments) {
-      if (currentSection === 'ingredients') {
+    seg = seg.trim().replace(/^[:\-•◦⁃∙*\-+]\s*/, '').replace(/#[\w-]+/g, '').trim();
-        const cleaned = line.replace(/^[•◦⁃∙*\-+]\s*/, '').replace(/#[\w-]+/g, '').trim();
+    if (!seg || seg.length < 2) continue;
-        if (cleaned && !cleaned.startsWith('#')) ingredients.push(cleaned);
+
-      } else if (currentSection === 'instructions') {
+    if (instructionStarters.test(seg) || /^\d+[\.\)]\s+/.test(seg)) {
-        const cleaned = line.replace(/^\d+[\.\)]\s*/, '').replace(/^[•◦⁃∙*\-+]\s*/, '').replace(/#[\w-]+/g, '').trim();
+      instructions.push(seg.replace(/^\d+[\.\)]\s*/, ''));
-        if (cleaned && !cleaned.startsWith('#')) instructions.push(cleaned);
+    } else if (isIngredientPattern.test(seg)) {
+      if (/batter|pan/i.test(seg)) {
+        instructions.push(seg);
       } else {
-        if (!title) title = line.replace(/#[\w-]+/g, '').trim();
+        ingredients.push(seg);
-        else description += ' ' + line.replace(/#[\w-]+/g, '').trim();
+      }
+    } else if (/^(?:glaze|for the )/i.test(seg)) {
+      instructions.push(seg);
+    } else {
+      if (/(?:bowl|pan|oven|bake|mins?|minutes?|degrees?|350|mix|batter)/i.test(seg)) {
+        instructions.push(seg);
+      } else {
+        ingredients.push(seg);
       }
     }
   }
 
-  if (!title) {
+  let cookTime = '';
-    title = author ? `Recipe by ${author}` : 'TikTok Recipe';
+  const bakeMatch = caption.match(/\b(?:bake(?:\s+at\s+\d+)?\s+(?:for\s+)?(\d+[\s\-to]+\d+\s*(?:mins?|minutes?|hours?|hrs?)))\b/i);
+  if (bakeMatch) {
+    cookTime = bakeMatch[1].trim();
   }
 
   return {
-    title,
+    title: title || (author ? `Recipe by ${author}` : 'TikTok Recipe'),
-    description: description.trim(),
+    description: caption.slice(0, 160).trim(),
     image: cover || '',
     images: cover ? [cover] : [],
     servings: '',
     prepTime: '',
-    cookTime: '',
+    cookTime,
-    totalTime: '',
+    totalTime: cookTime,
     ingredients,
     instructions,
     sourceUrl: sourceUrl || ''