From a94e5ca592fd1e83d320bc2d8ae0bf2c34996282 Mon Sep 17 00:00:00 2001
From: Jun-te Kim <junte.kim@mealcraft.com>
Date: Tue, 10 Feb 2026 08:04:57 +0000
Subject: [PATCH] s3 url processing

---
 backend/postcode_splitter/main.py | 43 ++++++++++++-------------------
 1 file changed, 17 insertions(+), 26 deletions(-)

diff --git a/backend/postcode_splitter/main.py b/backend/postcode_splitter/main.py
index 1d0e56a0..adb8e5c9 100644
--- a/backend/postcode_splitter/main.py
+++ b/backend/postcode_splitter/main.py
@@ -23,41 +23,32 @@ def parse_s3_console_url(s3_uri: str) -> tuple[str, str]:
 
     Format: https://account-id-hash.region.console.aws.amazon.com/s3/object/bucket?region=...&prefix=path
     """
-    logger.info(f"Parsing S3 URI: {s3_uri}")
-
-    if "console.aws.amazon.com" not in s3_uri:
-        logger.error("URI does not contain 'console.aws.amazon.com'")
-        raise ValueError(f"Could not parse S3 URI: {s3_uri}")
-
-    if "?prefix=" not in s3_uri:
-        logger.error("URI does not contain '?prefix='")
-        raise ValueError(f"Could not parse S3 URI: {s3_uri}")
+    logger.info("Parsing S3 console URL")
 
     try:
+        # Split base URL and query string
+        if "?" not in s3_uri:
+            raise ValueError("No query string found")
+
         base, query = s3_uri.split("?", 1)
-        logger.debug(f"Base: {base}")
-        logger.debug(f"Query: {query}")
+
+        # Extract bucket from base URL
+        if "/s3/object/" not in base:
+            raise ValueError("No '/s3/object/' found in URL path")
 
         path_parts = base.split("/s3/object/")
-        logger.debug(f"Path parts: {path_parts}")
+        bucket = path_parts[1]
+        logger.info(f"Extracted bucket: {bucket}")
 
-        if len(path_parts) > 1:
-            bucket = path_parts[1]
-            logger.info(f"Extracted bucket: {bucket}")
+        # Extract prefix from query parameters
+        params = dict(item.split("=") for item in query.split("&") if "=" in item)
+        key = unquote(params.get("prefix", ""))
+        logger.info(f"Extracted key: {key}")
 
-            params = dict(item.split("=") for item in query.split("&") if "=" in item)
-            logger.debug(f"Query params: {params}")
-
-            key = unquote(params.get("prefix", ""))
-            logger.info(f"Extracted key: {key}")
-
-            return bucket, key
-        else:
-            logger.error(f"Could not find '/s3/object/' in URI")
-            raise ValueError(f"Could not parse S3 URI: {s3_uri}")
+        return bucket, key
     except Exception as e:
         logger.error(f"Error parsing S3 URI: {type(e).__name__}: {e}")
-        raise ValueError(f"Could not parse S3 URI: {s3_uri}") from e
+        raise ValueError(f"Could not parse S3 URI") from e
 
 
 def sanitise_postcode(postcode: str) -> str | None: