@retrivora-ai/rag-engine 1.7.9 → 1.8.1

package/src/utils/SchemaMapper.ts

@@ -0,0 +1,129 @@
+import { ILLMProvider } from '../llm/ILLMProvider';
+
+export interface SchemaMap {
+  name?: string;
+  price?: string;
+  brand?: string;
+  image?: string;
+  stock?: string;
+  description?: string;
+  category?: string;
+}
+
+/**
+ * SchemaMapper — Uses AI to "train" the plugin on unknown database schemas.
+ * It maps arbitrary metadata keys to standard UI properties.
+ */
+export class SchemaMapper {
+  private static cache = new Map<string, SchemaMap>();
+
+  /**
+   * Descriptions of standard UI properties to help the AI map fields accurately.
+   */
+  private static readonly TARGET_PROPERTIES: Record<string, string> = {
+    name: 'The primary title, name, or label of the item',
+    price: 'The numeric cost, price, MSRP, or amount',
+    brand: 'The manufacturer, vendor, brand, or maker',
+    image: 'The URL to an image, thumbnail, photo, or picture',
+    stock: 'The availability, inventory count, quantity, or stock level',
+    description: 'The detailed text content, summary, or body info',
+    category: 'The group, department, type, or category name',
+  };
+
+  /**
+   * Trains the plugin on a set of keys.
+   * This is done once per schema and cached.
+   */
+  static async train(
+    llm: ILLMProvider,
+    projectId: string,
+    keys: string[]
+  ): Promise<SchemaMap> {
+    const cacheKey = `${projectId}:${keys.sort().join(',')}`;
+    if (this.cache.has(cacheKey)) {
+      return this.cache.get(cacheKey)!;
+    }
+
+    console.log(`[SchemaMapper] 🧠 Training AI on new schema keys: ${keys.join(', ')}`);
+
+    // Dynamically build the property list for the prompt
+    const propertyList = Object.entries(this.TARGET_PROPERTIES)
+      .map(([prop, desc]) => `- ${prop} (${desc})`)
+      .join('\n');
+
+    const prompt = `
+Given these metadata keys from a database: [${keys.join(', ')}]
+
+Identify which keys best correspond to these standard UI properties:
+${propertyList}
+
+Return ONLY a JSON object where the keys are the UI properties and the values are the matching database keys. 
+If no good match is found for a property, omit it.
+
+Example:
+{
+  "name": "Product_Title",
+  "price": "MSRP_USD",
+  "brand": "VendorName"
+}
+`;
+
+    try {
+      const response = await llm.chat(
+        [
+          { role: 'system', content: 'You are a database schema expert. You ONLY respond with valid JSON.' },
+          { role: 'user', content: prompt }
+        ],
+        ''
+      );
+
+      // Extract the first balanced JSON object from the response
+      const startIdx = response.indexOf('{');
+      if (startIdx !== -1) {
+        let braceCount = 0;
+        let endIdx = -1;
+
+        for (let i = startIdx; i < response.length; i++) {
+          if (response[i] === '{') braceCount++;
+          else if (response[i] === '}') {
+            braceCount--;
+            if (braceCount === 0) {
+              endIdx = i;
+              break;
+            }
+          }
+        }
+
+        if (endIdx !== -1) {
+          const jsonContent = response.substring(startIdx, endIdx + 1);
+          const cleanJson = this.sanitizeJson(jsonContent);
+          const mapping = JSON.parse(cleanJson) as SchemaMap;
+          this.cache.set(cacheKey, mapping);
+          return mapping;
+        }
+      }
+    } catch (error) {
+      console.warn('[SchemaMapper] AI training failed, falling back to heuristics:', error);
+    }
+
+    return {};
+  }
+
+  /**
+   * Forgiving JSON parser that fixes common AI formatting mistakes.
+   */
+  private static sanitizeJson(s: string): string {
+    return s
+      .replace(/\/\*[\s\S]*?\*\/|([^:]|^)\/\/.*$/gm, '') // Remove comments
+      .replace(/[\u201C\u201D\u2018\u2019]/g, '"') // Replace smart quotes
+      .replace(/'/g, '"') // Replace single quotes
+      .replace(/([{,]\s*)([a-zA-Z0-9_]+?)\s*:/g, '$1"$2":') // Add quotes to keys
+      .replace(/,\s*([}\]])/g, '$1') // Remove trailing commas
+      .trim();
+  }
+
+  static getCached(projectId: string, keys: string[]): SchemaMap | undefined {
+    const cacheKey = `${projectId}:${keys.sort().join(',')}`;
+    return this.cache.get(cacheKey);
+  }
+}