From 46d1ca337551d410706a435cc0c8a4fb305b5b9a Mon Sep 17 00:00:00 2001
From: dlawler489 <104159223@student.swin.edu.au>
Date: Sat, 13 Jun 2026 13:05:17 +1000
Subject: [PATCH] Fix size-aware product suggestions in resolver

Previously a different-size variant was boosted to 0.95 confidence
('same product, different size'), so a Small order would be confidently
suggested the Large product. Now suggestions require the same size; a
known different size is suppressed entirely (you create the right one
instead). Also handle 'Size: X for ...' phrasing, medium sizes, and
American 'Color:' spelling.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---
 client/src/utils/productMatcher.ts | 112 +++++++++++++++--------------
 1 file changed, 57 insertions(+), 55 deletions(-)

diff --git a/client/src/utils/productMatcher.ts b/client/src/utils/productMatcher.ts
index 533d3fb..05f0281 100644
--- a/client/src/utils/productMatcher.ts
+++ b/client/src/utils/productMatcher.ts
@@ -260,6 +260,31 @@ const calculateTitleSimilarity = (title1: string, title2: string): number => {
   return matchingWords / totalUniqueWords;
 };
 
+/**
+ * Extract a size token (large/med/small) from a title or variant string.
+ * Prefers an explicit "Size: X" so trailing words like "Large for med plants"
+ * don't get misread.
+ */
+export const extractSize = (text: string): string => {
+  const lower = text.toLowerCase();
+  const explicit = lower.match(/size:\s*(large|medium|med|small)/);
+  let token = explicit ? explicit[1] : '';
+  if (!token) {
+    if (lower.includes('large') && !lower.includes('small')) token = 'large';
+    else if (lower.includes('small') && !lower.includes('large')) token = 'small';
+  }
+  return token === 'medium' ? 'med' : token;
+};
+
+const cleanForCompare = (title: string): string =>
+  title
+    .replace(/Colou?r:\s*[^,\s-]+(?:\s+[^,\s-]+)*/i, '') // strip Color/Colour
+    .replace(/Size:\s*[^,\s-]+(?:\s+(?:for|plants?|succulents?)\b[^,]*)?/i, '') // strip "Size: X for ..."
+    .replace(/3D-Printed/i, '')
+    .replace(/\s+/g, ' ')
+    .trim()
+    .toLowerCase();
+
 /**
  * Find potential matches for a missing product
  * This helps suggest existing products that might be similar
@@ -270,77 +295,54 @@ export const findPotentialMatches = (
   maxSuggestions: number = 5
 ): Array<{product: any; confidence: number; reason: string}> => {
   const suggestions: Array<{product: any; confidence: number; reason: string}> = [];
-  
-  const itemTitle = missingProductTitle.toLowerCase();
-  let itemSize = '';
-  
-  // Extract size from missing product
-  if (itemTitle.includes('large') && !itemTitle.includes('small')) {
-    itemSize = 'large';
-  } else if (itemTitle.includes('small') && !itemTitle.includes('large')) {
-    itemSize = 'small';
-  }
-  
-  const cleanItemTitle = missingProductTitle
-    .replace(/Colour:\s*[^,\s-]+(?:\s+[^,\s-]+)*/i, '')
-    .replace(/Size:\s*[^,\s-]+/i, '')
-    .replace(/3D-Printed/i, '')
-    .replace(/\s+/g, ' ')
-    .trim()
-    .toLowerCase();
-  
+
+  const itemSize = extractSize(missingProductTitle);
+  const cleanItemTitle = cleanForCompare(missingProductTitle);
+
   for (const product of existingProducts) {
-    const productTitle = product.title.toLowerCase();
-    let productSize = '';
-    
-    if (productTitle.includes('large') && !productTitle.includes('small')) {
-      productSize = 'large';
-    } else if (productTitle.includes('small') && !productTitle.includes('large')) {
-      productSize = 'small';
-    }
-    
-    const cleanProductTitle = product.title
-      .replace(/Colour:\s*[^,\s-]+(?:\s+[^,\s-]+)*/i, '')
-      .replace(/Size:\s*[^,\s-]+/i, '')
-      .replace(/3D-Printed/i, '')
-      .replace(/\s+/g, ' ')
-      .trim()
-      .toLowerCase();
-    
+    const productSize = extractSize(`${product.title} ${(product.aliases || []).join(' ')}`);
+    const cleanProductTitle = cleanForCompare(product.title);
+
     const titleSimilarity = calculateTitleSimilarity(cleanItemTitle, cleanProductTitle);
-    
+
     let reason = '';
-    let confidence = titleSimilarity;
-    
-    // Check if it's the same product family but different size
+    let confidence = 0;
+
     if (titleSimilarity > 0.8) {
-      if (itemSize && productSize && itemSize !== productSize) {
-        reason = `Same product, different size (${productSize} vs ${itemSize})`;
-        confidence = 0.95; // Very high confidence - likely meant to be this size variant
+      if (itemSize && productSize) {
+        if (itemSize === productSize) {
+          reason = `Same product, same size (${itemSize})`;
+          confidence = Math.max(titleSimilarity, 0.95);
+        } else {
+          // Both sizes known and they differ — this is NOT the right product.
+          // Suppress it so a wrong-size variant is never suggested.
+          confidence = 0;
+        }
       } else if (itemSize && !productSize) {
-        reason = `Same product, missing size specification`;
-        confidence = 0.85;
+        reason = `Same product (existing has no size set)`;
+        confidence = 0.75;
       } else if (!itemSize && productSize) {
         reason = `Same product, has size: ${productSize}`;
-        confidence = 0.8;
+        confidence = 0.7;
       } else {
         reason = `Very similar product`;
         confidence = titleSimilarity;
       }
     } else if (titleSimilarity > 0.6) {
-      reason = `Similar product`;
-      confidence = titleSimilarity;
+      // Don't suggest a different-size variant even on a looser title match
+      if (itemSize && productSize && itemSize !== productSize) {
+        confidence = 0;
+      } else {
+        reason = `Similar product`;
+        confidence = titleSimilarity;
+      }
     }
-    
+
     if (confidence > 0.6) {
-      suggestions.push({
-        product,
-        confidence,
-        reason
-      });
+      suggestions.push({ product, confidence, reason });
     }
   }
-  
+
   // Sort by confidence and return top suggestions
   return suggestions
     .sort((a, b) => b.confidence - a.confidence)