fix(llmExtract): remove unsupported JSON schema properties (#1335)

apps/api/src/__tests__/snips/extract.test.ts

@@ -29,6 +29,28 @@ describe("Extract tests", () => {
             expect(typeof res.data.is_open_source).toBe("boolean");
             expect(res.data.is_open_source).toBe(true);
         }, 60000);
+
+        it.concurrent("works with unsupported JSON schema parameters", async () => {
+            const res = await extract({
+                urls: ["https://firecrawl.dev"],
+                schema: {
+                    "type": "object",
+                    "properties": {
+                        "company_name": {
+                            "type": "string",
+                            "pattern": "^[a-zA-Z0-9]+$"
+                        },
+                    },
+                    "required": [
+                        "company_name"
+                    ]
+                },
+                origin: "api-sdk",
+            });
+
+            expect(res.data).toHaveProperty("company_name");
+            expect(typeof res.data.company_name).toBe("string")
+        }, 60000);
     } else {
         it.concurrent("dummy test", () => {
             expect(true).toBe(true);

apps/api/src/scraper/scrapeURL/transformers/llmExtract.ts

@@ -365,7 +365,37 @@ export async function performLLMExtract(
 export function removeDefaultProperty(schema: any): any {
   if (typeof schema !== "object" || schema === null) return schema;
 
-  const { default: _, ...rest } = schema;
+  const rest = { ...schema };
+  
+  // unsupported global keys
+  delete rest.default;
+
+  // unsupported object keys
+  delete rest.patternProperties;
+  delete rest.unevaluatedProperties;
+  delete rest.propertyNames;
+  delete rest.minProperties;
+  delete rest.maxProperties;
+
+  // unsupported string keys
+  delete rest.minLength;
+  delete rest.maxLength;
+  delete rest.pattern;
+  delete rest.format;
+
+  // unsupported number keys
+  delete rest.minimum;
+  delete rest.maximum;
+  delete rest.multipleOf;
+
+  // unsupported array keys
+  delete rest.unevaluatedItems;
+  delete rest.contains;
+  delete rest.minContains;
+  delete rest.maxContains;
+  delete rest.minItems;
+  delete rest.maxItems;
+  delete rest.uniqueItems;
 
   for (const key in rest) {
     if (Array.isArray(rest[key])) {