scraping and data

finder/main.py

@@ -57,6 +57,16 @@ def parse_args() -> argparse.Namespace:
         default=DATA_DIR,
         help=f"Directory for parquet output. Defaults to {DATA_DIR}.",
     )
+    parser.add_argument(
+        "--outcodes",
+        type=str,
+        default=None,
+        help=(
+            "Comma-separated outcodes to scrape (e.g. 'SW9' or 'SW9,E14,BR1') "
+            "instead of the full Greater London set. Must fall within the "
+            "London-ish areas; takes precedence over --test/--limit-outcodes."
+        ),
+    )
     parser.add_argument(
         "--limit-outcodes",
         type=int,
@@ -116,17 +126,32 @@ def main() -> int:
     from scraper import (
         build_postcode_coords,
         build_postcode_index,
+        filter_londonish_outcodes,
         load_outcodes,
         run_scrape,
     )
 
-    outcodes = load_outcodes()
-    if args.test and args.limit_outcodes is None:
-        preferred = [outcode for outcode in TEST_OUTCODES if outcode in set(outcodes)]
-        if preferred:
-            outcodes = preferred
-    if args.limit_outcodes is not None:
-        outcodes = outcodes[: args.limit_outcodes]
+    if args.outcodes is not None:
+        requested = [code.strip().upper() for code in args.outcodes.split(",") if code.strip()]
+        if not requested:
+            raise SystemExit("--outcodes was empty")
+        outcodes = filter_londonish_outcodes(requested)
+        dropped = sorted(set(requested) - set(outcodes))
+        if dropped:
+            log.warning("Ignoring outcodes outside the Greater London-ish areas: %s", ", ".join(dropped))
+        if not outcodes:
+            raise SystemExit(
+                "None of the requested outcodes are within the Greater London-ish areas "
+                f"({', '.join(requested)})."
+            )
+    else:
+        outcodes = load_outcodes()
+        if args.test and args.limit_outcodes is None:
+            preferred = [outcode for outcode in TEST_OUTCODES if outcode in set(outcodes)]
+            if preferred:
+                outcodes = preferred
+        if args.limit_outcodes is not None:
+            outcodes = outcodes[: args.limit_outcodes]
 
     if not outcodes:
         raise SystemExit("No Greater London-ish outcodes loaded; nothing to scrape.")