.

pipeline/download/median_age.py

@@ -3,7 +3,7 @@
 Downloads five-year age band counts (TS007A) from the NOMIS API, then computes
 the median age per LSOA using linear interpolation within the median class.
 
-Source: NOMIS (ONS Census 2021 — TS007A dataset, NM_2020_1)
+Source: NOMIS (ONS Census 2021, TS007A dataset, NM_2020_1)
 License: Open Government Licence v3.0
 """
 
@@ -20,7 +20,7 @@ from pipeline.utils import ENGLAND_LSOA_COUNT_2021, download_nomis_csv
 BASE_URL = "https://www.nomisweb.co.uk/api/v01/dataset/NM_2020_1.data.csv?date=latest&geography=TYPE151&c2021_age_19=1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18&measures=20100&select=GEOGRAPHY_CODE,C2021_AGE_19_NAME,OBS_VALUE"
 
 # Five-year age bands in order, with lower bounds for interpolation.
-# The last band (85+) is open-ended — we treat it as 85-89 for median purposes.
+# The last band (85+) is open-ended. We treat it as 85-89 for median purposes.
 AGE_BANDS = [
     (0, 5),  # Aged 0 to 4 years
     (5, 5),  # Aged 5 to 9 years

pipeline/download/naptan.py

@@ -28,7 +28,7 @@ LONDON_UNDERGROUND_ATCO_PATTERN = r"(?i)^\d{3}[0G]ZZLU"
 # The Docklands Light Railway uses the analogous "ZZDL" system code (e.g.
 # "9400ZZDLBEC" for Beckton). Like ZZLU it is unique to one network, so a
 # TMU/MET stop carrying a ZZDL code is reclassified from the tram/metro family
-# to its own "DLR station" category — restoring DLR to the train/tube station
+# to its own "DLR station" category, restoring DLR to the train/tube station
 # list and giving it the DLR roundel rather than a generic tram icon.
 LONDON_DLR_ATCO_PATTERN = r"(?i)^\d{3}[0G]ZZDL"
 
@@ -337,7 +337,7 @@ class StationAccumulator:
         # A single node is never both (ZZLU vs ZZDL), but a co-located
         # interchange (Bank, Stratford, Canning Town, West Ham) merges its LU
         # and DLR halves into one group carrying both flags; Tube is checked
-        # first so these resolve to "Tube station" — their primary identity —
+        # first so these resolve to "Tube station" (their primary identity),
         # leaving "DLR station" for the DLR-only stops the fix targets.
         if self.category == TRAM_METRO_CATEGORY and self.is_lu:
             return TUBE_STATION_CATEGORY

pipeline/download/noise.py

@@ -85,24 +85,24 @@ RESOLUTION = NATIVE_RESOLUTION
 
 # Defra encodes TRUE "no data" with this sentinel (NOT 0.0). A 0.0 cell that is
 # otherwise inside the raster means "modelled below the lowest reporting band",
-# i.e. genuinely quiet — see noise_overlay_tiles.py:167.
+# i.e. genuinely quiet (see noise_overlay_tiles.py:167).
 NOISE_NODATA_SENTINEL = np.float32(-96.0)
 
 # Lowest modelled Defra Lden reporting band (dB). Verified against the actual
 # rasters: the minimum positive in-coverage value is 40.0 dB with NO values in
-# (0, 40) — below the band, cells are encoded as 0.0 (genuinely quiet). We floor
+# (0, 40). Below the band, cells are encoded as 0.0 (genuinely quiet). We floor
 # in-coverage cells to 40.0 so a below-band 0.0 surfaces as "we know it's quiet"
 # (~40 dB) instead of collapsing to null ("we don't know"), WITHOUT inflating the
 # ~35% of genuine 40-44.99 dB readings that a 45.0 floor would wrongly bump to 45.
 # NB: 45.0 is the overlay's lowest *paint* stop (noise_overlay_tiles.
-# NOISE_COLOR_STOPS[0]) — a rendering threshold, not the data's reporting floor.
+# NOISE_COLOR_STOPS[0]), a rendering threshold, not the data's reporting floor.
 NOISE_QUIET_FLOOR_DB = np.float32(40.0)
 
 # Sample noise at the postcode representative point itself (no neighbourhood
 # window). A 50m MAX-of-window grabbed the single loudest 10m cell within ~1.2 ha
 # of every postcode; because Defra road contours hug every modelled road and
 # representative points sit on/near streets, that inflated postcode noise by
-# roughly +9 dB (log scale) — making ~94% of England read >=55 dB Lden and
+# roughly +9 dB (log scale), making ~94% of England read >=55 dB Lden and
 # collapsing the metric's discrimination at the quiet end. Radius 0 ->
 # filter_size 1 -> the maximum_filter is skipped and each postcode reads the
 # 10m cell it actually sits in.
@@ -110,8 +110,8 @@ POSTCODE_NOISE_RADIUS_M = 0
 
 # Adjacent download tiles overlap by the sampling radius so every postcode's
 # sampling footprint is fully contained in at least one tile. With point
-# sampling (radius 0) this is 0 — a representative point falls inside exactly
-# one tile — but the relationship is kept so any future non-zero radius keeps
+# sampling (radius 0) this is 0 (a representative point falls inside exactly
+# one tile), but the relationship is kept so any future non-zero radius keeps
 # its window seam-safe.
 TILE_OVERLAP_M = POSTCODE_NOISE_RADIUS_M
 
@@ -273,7 +273,7 @@ def _download_tile(
             NoGeoTiffError,
             httpx.HTTPStatusError,
             # TransportError is the superset of TimeoutException, ConnectError,
-            # ReadError and ProtocolError — including RemoteProtocolError, raised
+            # ReadError and ProtocolError, including RemoteProtocolError, raised
             # when the WCS server closes the connection mid-stream ("incomplete
             # chunked read"). All are transient; retry/split rather than letting
             # one flaky tile crash the whole raster download.
@@ -446,7 +446,7 @@ def sample_noise_at_postcodes(
             # Defra rasters encode TRUE nodata as the -96.0 sentinel (and
             # occasionally non-finite / dataset.nodata); genuinely quiet ground
             # below the model's lowest reporting band is encoded as 0.0. Only
-            # the former is "we don't know" — the latter is a real "we know it's
+            # the former is "we don't know". The latter is a real "we know it's
             # quiet" reading and must not collapse to null. So treat ONLY true
             # nodata as -inf (it never wins a max and never counts as coverage),
             # and clamp every in-coverage cell up to NOISE_QUIET_FLOOR_DB so a
@@ -555,7 +555,7 @@ def main() -> None:
 
             if not tile_paths:
                 print(
-                    f"[{label}] WARNING: No tiles downloaded — column will be all null"
+                    f"[{label}] WARNING: No tiles downloaded; column will be all null"
                 )
                 series = pl.Series(col_name, [None] * len(lat), dtype=pl.Float32)
             else:

pipeline/download/os_greenspace.py

@@ -7,7 +7,7 @@ site id and the site's polygon centroid. Sites without access points fall
 back to polygon centroids.
 
 Using access points rather than polygon centroids gives much more accurate
-distance calculations — a property next to Hyde Park won't show 400m just
+distance calculations: a property next to Hyde Park won't show 400m just
 because the centroid is in the middle of the park. The site id / centroid
 columns let downstream consumers (poi_proximity) collapse the frame back to
 one row per SITE for counting, so a park with 30 gates counts as one park.
@@ -190,7 +190,7 @@ def download_greenspace(output: Path) -> None:
         print(f"Reading {site_shps[0].name} for function types...")
         site_funcs = _read_site_functions(site_shps[0])
 
-        # Step 2: Read access points (primary — park entrances)
+        # Step 2: Read access points (primary: park entrances)
         print(f"Reading {access_shps[0].name}...")
         ap_lats, ap_lngs, ap_cats, ap_site_ids = _read_access_points(
             access_shps[0], site_funcs

pipeline/download/places.py

@@ -930,7 +930,7 @@ def main() -> None:
         df.write_parquet(args.output)
         print(f"Saved to {args.output}")
     else:
-        print("No places found — skipping output")
+        print("No places found, skipping output")
 
 
 if __name__ == "__main__":

pipeline/download/rental_prices.py

@@ -35,7 +35,7 @@ def _data_rows(df: pl.DataFrame) -> pl.DataFrame:
 
     The preamble length varies (title, optional "This worksheet contains..."
     note, then the header row starting with "Time period"), so locate the
-    header by content instead of counting rows — a fixed slice leaves the
+    header by content instead of counting rows: a fixed slice leaves the
     header in the data whenever ONS adds or removes a note line.
     """
     header_marker = (
@@ -78,7 +78,7 @@ def _latest_rents_long(df: pl.DataFrame) -> pl.DataFrame:
     print(f"LAs in latest month: {df.height}")
 
     # Melt to long format: one row per area x bedroom count.
-    # PIPR has no Studio category — one-bed rent used as proxy for bedrooms=0.
+    # PIPR has no Studio category: one-bed rent used as proxy for bedrooms=0.
     frames = []
     for col, bedrooms in [
         ("rent_1bed", 0),  # Studio (proxy)

pipeline/download/tenure.py

@@ -2,8 +2,8 @@
 
 Downloads the household-tenure breakdown (TS054, classification C2021_TENURE_9)
 from the NOMIS API at LSOA 2021 granularity and folds the 8 detailed leaf
-categories into our 3 output buckets — Owner occupied / Social rent /
-Private rent — emitting one row per LSOA with the percentage of households in
+categories into our 3 output buckets (Owner occupied / Social rent /
+Private rent), emitting one row per LSOA with the percentage of households in
 each. The three buckets sum to 100%, so downstream they render as a
 composition/ratio (like the ethnicity and qualifications stacked bars) AND each
 percentage is independently filterable.
@@ -19,7 +19,7 @@ NOTE this table counts HOUSEHOLDS (not usual residents). The join key downstream
 (merge.py) is `lsoa21`, the same key used for ethnicity, qualifications,
 median age, and IoD.
 
-Source: NOMIS (ONS Census 2021 — TS054 dataset, NM_2072_1)
+Source: NOMIS (ONS Census 2021, TS054 dataset, NM_2072_1)
 License: Open Government Licence v3.0
 """
 

pipeline/download/test_ethnicity.py

@@ -72,7 +72,7 @@ def test_ethnicity_routes_chinese_to_east_and_other_asian_to_se():
 
 
 def test_ethnicity_percentages_independent_per_lsoa():
-    """Two LSOAs get independent profiles — the LSOA granularity is the point."""
+    """Two LSOAs get independent profiles: the LSOA granularity is the point."""
     df = pl.concat(
         [
             pl.DataFrame(

pipeline/download/test_noise.py

@@ -281,7 +281,7 @@ def test_sample_noise_preserves_genuine_reading_above_quiet_floor(
 
     # The lowest Defra reporting band is 40.0 dB; genuine readings populate
     # [40, ~80]. A genuine in-coverage reading at or just above the floor must be
-    # PRESERVED, not clamped UP to the floor — only true-quiet 0.0 is floored. A
+    # PRESERVED, not clamped UP to the floor. Only true-quiet 0.0 is floored. A
     # quiet floor set too high (e.g. 45) would inflate the ~35% of real 40-44.99
     # dB readings; this pins that they survive unchanged.
     floor = float(noise.NOISE_QUIET_FLOOR_DB)

pipeline/download/transit_network.py

@@ -50,7 +50,7 @@ ENGLAND_PBF_URL = (
     "https://download.geofabrik.de/europe/united-kingdom/england-latest.osm.pbf"
 )
 
-# Bus Open Data Service — pre-converted GTFS covering all England bus/tram/ferry
+# Bus Open Data Service: pre-converted GTFS covering all England bus/tram/ferry
 BODS_GTFS_URL = "https://data.bus-data.dft.gov.uk/timetable/download/gtfs-file/all/"
 
 # National Rail Open Data API
@@ -597,7 +597,7 @@ def validate_gtfs_feed(
             fail("has neither calendar.txt nor calendar_dates.txt")
         if not _calendar_active_in_window(z, names, window_start, window_end):
             fail(
-                f"no service active between {window_start} and {window_end} — "
+                f"no service active between {window_start} and {window_end}: "
                 "the feed's calendars are stale/expired and it would contribute "
                 "zero service to routing"
             )
@@ -697,7 +697,7 @@ def _iter_stop_time_trips(lines, trip_id_idx: int):
     dtd2mysql currently writes rows grouped by trip and ordered by
     stop_sequence, but neither is guaranteed by GTFS. Grouping is verified (a
     trip_id reappearing later raises instead of silently scrambling trips);
-    within-trip order is NOT assumed — callers sort each group by its original
+    within-trip order is NOT assumed: callers sort each group by its original
     stop_sequence.
     """
     current_trip: str | None = None
@@ -1175,7 +1175,7 @@ def main() -> None:
     cif = download_national_rail_cif(raw_dir)
     if cif is None:
         raise RuntimeError(
-            "National Rail timetable was not downloaded — set "
+            "National Rail timetable was not downloaded: set "
             "NATIONAL_RAIL_EMAIL / NATIONAL_RAIL_PASSWORD (register free at "
             "https://opendata.nationalrail.co.uk/). National Rail heavy rail is "
             "required; without it the transit network models every train journey "