Update app.py

app.py

@@ -29,8 +29,11 @@ def upload_pdf(path):
 # ------------------ Prompt ---------------------
 def build_prompt():
     return (
-        "Extract structured JSON from this shipping document. "
-        "Return ONLY valid JSON:\n"
+        "You are an advanced extraction system for logistics, rail, lumber, shipping, trucking, "
+        "inventory, and packing documents. Your task is to read the attached PDF or image and "
+        "extract ONLY the information that is explicitly present.\n\n"
+
+        "You must return STRICT JSON in the EXACT structure below:\n"
         "{\n"
         "  \"po_number\": string|null,\n"
         "  \"ship_from\": string|null,\n"
@@ -38,11 +41,41 @@ def build_prompt():
         "  \"rail_car_number\": string|null,\n"
         "  \"total_quantity\": number|null,\n"
         "  \"inventories\": [\n"
-        "     {\"productName\":string,\"productCode\":string|null,\"pcs\":number|null,\"dimensions\":string|null}\n"
+        "     {\n"
+        "       \"productName\": string,\n"
+        "       \"productCode\": string|null,\n"
+        "       \"pcs\": number|null,\n"
+        "       \"dimensions\": string|null\n"
+        "     }\n"
         "  ],\n"
         "  \"custom_fields\": {}\n"
-        "}\n"
-        "Use ONLY the text visible in the document. Do NOT hallucinate values."
+        "}\n\n"
+
+        "YOUR RULES (MUST FOLLOW EXACTLY):\n"
+        "1. Do NOT guess or hallucinate. Only extract values explicitly shown in the document.\n"
+        "2. If a field is not present or cannot be confirmed → output null.\n"
+        "3. PO Number may appear under wording like 'PO', 'Purchase Order', 'P.O.', 'Customer PO', etc.\n"
+        "4. Ship From may appear as 'Origin', 'From', 'Exporter', 'Ship From', 'Supplier', etc.\n"
+        "5. Carrier Type may appear as 'Carrier', 'Carrier Type', 'Routing', 'Mode', 'Transport Type', "
+        "'RAIL', 'TRUCK', 'CN', 'BNSF', 'CP', 'Truckload', etc.\n"
+        "6. Rail Car Number may appear as 'Railcar', 'Rail Car #', 'Car Number', 'Car #', etc.\n"
+        "7. Total Quantity must be ONLY the explicit total PCS/pieces count if it appears. "
+        "If the only total shown is FBM/weight/volume → DO NOT treat that as quantity.\n"
+        "8. Inventories must capture every unique product line that appears. Extract product name, "
+        "item description, dimensions like '2x4', '23/32', and PCS when available.\n"
+        "9. Dimensions may appear as '2 X 4', '2x6', '48x96', '23/32', etc. Normalize to a single "
+        "string representation.\n"
+        "10. custom_fields must contain ANY additional fields not part of the main schema (dates, mills, "
+        "FBM, weights, routing codes, package counts, etc.). Key names must be lower_snake_case.\n"
+        "11. JSON MUST be valid, must not include comments, and must not include text outside the JSON object.\n\n"
+
+        "ADDITIONAL RULES FOR COMPLEX TABLES:\n"
+        "- If multiple product variants exist, create multiple inventory objects.\n"
+        "- If tables list PCS per package × number of packages, you MAY compute total PCS.\n"
+        "- Never compute derived values unless the math is explicitly possible.\n"
+        "- If a value is ambiguous, set it to null.\n\n"
+
+        "Final requirement: Return ONLY the JSON object. No explanation, no markdown.\n"
     )