From 5a86b3c96b3c0d9732ba7a60aea5a3db840cf458 Mon Sep 17 00:00:00 2001
From: Adam Moussa <166072409+amoussa1229@users.noreply.github.com>
Date: Thu, 30 Apr 2026 14:42:09 -0400
Subject: [PATCH] Add site_code and structured address fields to extraction
 prompt

The Claude extraction prompt now explicitly asks for site_code (the
Amazon facility code) and structured ship_to address fields (street,
city, state, zip). The site-extractor Lambda prefers these direct
fields when available, falling back to regex for older PO records.
---
 lambdas/email_processor/handler.py | 12 ++++++++++++
 lambdas/site_extractor/handler.py  | 16 ++++++++++++++++
 2 files changed, 28 insertions(+)

diff --git a/lambdas/email_processor/handler.py b/lambdas/email_processor/handler.py
index 4f28086..eea5a8e 100644
--- a/lambdas/email_processor/handler.py
+++ b/lambdas/email_processor/handler.py
@@ -52,9 +52,14 @@ Analyze the following email and extract structured data. Return ONLY valid JSON
   "supplier": {
     "name": "string or null"
   },
+  "site_code": "string or null",
   "ship_to": {
     "name": "string or null",
     "address": "string or null",
+    "street": "string or null",
+    "city": "string or null",
+    "state": "string or null",
+    "zip": "string or null",
     "location_code": "string or null",
     "attn": "string or null"
   },
@@ -80,6 +85,13 @@ Rules:
 - Extract the PO number from the email (e.g., "2D-18206023")
 - Extract all line items with their descriptions, amounts, and metadata
 - Ship-to address should include the full address, location code, and attention line
+- Ship-to street, city, state, and zip should be parsed from the address into separate fields
+- "site_code" is the Amazon facility code (e.g., "SNY5", "DFW6", "WND1") — a 3-5 character alphanumeric code identifying the delivery site. Look for it in:
+  - The ship-to name, e.g., "Amazon.com Services LLC - SNY5" or "Amazon.com Services LLC (WFB1)"
+  - The ATTN line, e.g., "ATTN: Wagon Wheel DS - WTN1"
+  - Line item descriptions, e.g., "WND1 - 2024 - Plumbing PM"
+  - Anywhere else in the email where a facility code appears
+  - If the ship-to name IS the site code (e.g., just "DBU2"), use that
 - total_amount should be the numeric total in USD
 - If a field is not present in the email, set it to null
 - Do NOT invent or infer data that is not explicitly in the email
diff --git a/lambdas/site_extractor/handler.py b/lambdas/site_extractor/handler.py
index 7d8027f..2a187d1 100644
--- a/lambdas/site_extractor/handler.py
+++ b/lambdas/site_extractor/handler.py
@@ -27,6 +27,11 @@ def deserialize_image(image: dict) -> dict:
 
 
 def extract_site_code(record: dict) -> str | None:
+    # Prefer the explicit site_code field (set by updated extraction prompt)
+    direct = (record.get("site_code") or "").strip().upper()
+    if direct and SITE_CODE_PATTERN.match(direct):
+        return direct
+
     ship_to = record.get("ship_to") or {}
     ship_to_name = ship_to.get("name", "")
 
@@ -68,6 +73,17 @@ def parse_address(record: dict) -> dict:
     }
 
     ship_to = record.get("ship_to") or {}
+
+    # Prefer structured fields if the extraction prompt provided them
+    if ship_to.get("street") and ship_to.get("state"):
+        result["address"] = ship_to["street"]
+        result["city"] = ship_to.get("city")
+        result["state"] = ship_to["state"]
+        result["zip"] = ship_to.get("zip")
+        parts = [p for p in [result["address"], result["city"], result["state"], result["zip"]] if p]
+        result["fullAddress"] = ", ".join(parts)
+        return result
+
     address_text = ship_to.get("address", "")
     if not address_text:
         return result