ok

finder/scraper.py

@@ -14,6 +14,7 @@ import polars as pl
 from constants import (
     ARCGIS_PATH,
     CHANNELS,
+    NEW_HOMES_CHANNEL,
     DATA_DIR,
     DELAY_BETWEEN_OUTCODES,
     LONDON_OUTCODE_PREFIXES,
@@ -368,6 +369,34 @@ def _scrape_rightmove(
             except Exception as exc:
                 _record_error(errors, "rightmove", outcode, exc)
 
+            # Second pass: new-build developments (mustHave=newHome). These can
+            # rank low in the default resale sort, so a dedicated pass ensures
+            # they are captured; transform_property stamps their URL as RES_NEW.
+            # Overlap with the resale pass is removed by id in _merge_properties.
+            # Skipped if a stop was requested mid-outcode (don't start a new pass).
+            remaining = _source_remaining(
+                results, "rightmove", max_properties_per_source
+            )
+            if not shutdown.stop_requested() and remaining != 0:
+                try:
+                    new_props = rightmove_search_outcode(
+                        client,
+                        outcode_id,
+                        outcode,
+                        NEW_HOMES_CHANNEL,
+                        pc_index,
+                        max_properties=remaining,
+                    )
+                    added_new = _store_properties(
+                        results,
+                        "rightmove",
+                        new_props,
+                        max_properties_per_source,
+                    )
+                    log.info("Rightmove %s new-homes: +%d", outcode, added_new)
+                except Exception as exc:
+                    _record_error(errors, "rightmove", outcode, exc)
+
             shutdown.sleep(DELAY_BETWEEN_OUTCODES)
     finally:
         client.close()
@@ -378,7 +407,7 @@ class OutcodeTimeout(BaseException):
 
     Inherits BaseException (not Exception) so the SIGALRM-triggered raise can't
     be silently swallowed by any of the broad `except Exception:` handlers
-    inside zoopla.py — the signal may fire at any bytecode boundary, including
+    inside zoopla.py: the signal may fire at any bytecode boundary, including
     inside those handlers."""
 
 
@@ -431,7 +460,7 @@ def _wall_clock_timeout(seconds: int, label: str):
 
     Interrupts a hung Playwright IPC by delivering SIGALRM to the main thread;
     socket waits return EINTR and the handler raises into the caller. The
-    browser is presumed unhealthy afterwards — caller must relaunch it."""
+    browser is presumed unhealthy afterwards. Caller must relaunch it."""
     if seconds <= 0:
         yield
         return