Which MRT station has the most bicycle racks?

"""Bicycle racks at MRT and LRT stations: take the racks LTA names after a station and add them up for each station.
Run: python3 -I prep/prep_racks.py <data folder> <out json>
Reads <data folder>/BicycleParking_raw.json (LTA DataMall bicycle parking, downloaded 12 Oct 2026) and LTAMRTStationExitGEOJSON.geojson."""
import json, re, math, sys, collections
from pathlib import Path
RAW = Path(sys.argv[1]); OUT = Path(sys.argv[2])
STATION_TYPES = ('MRT_RACKS', 'LTA_RACKS', 'LTA_YELLOWBOX')   # the types LTA uses for its own racks; the others belong to HDB, parks, schools and so on
YELLOW = 'Yellow Box'   # LTA's yellow-box spaces (most are at bus stops or HDB blocks): kept only when the name has a station in it

def read_stations():
    ex = collections.defaultdict(list)
    for f in json.load(open(RAW / 'LTAMRTStationExitGEOJSON.geojson'))['features']:
        n = f['properties']['STATION_NA']
        if 'STATION' in n:                      # 7 exits carry only a station code; they are left out
            ex[n].append(tuple(f['geometry']['coordinates'][:2]))
    return ex

def metres(a, b):
    return math.hypot((a[0] - b[0]) * 111320 * math.cos(math.radians(1.35)), (a[1] - b[1]) * 110570)

def nice(n):
    kind = 'LRT' if ' LRT ' in n else 'MRT'
    base = re.sub(r' (MRT|LRT) STATION$', '', n)
    words = [w.capitalize() if w not in ('OF',) else w.lower() for w in base.split(' ')]
    return ' '.join(words), kind, base

def station_of(description, lookup):
    """'TAMPINES MRT C3' -> Tampines MRT; 'KADALOOR LRT A' -> Kadaloor LRT; 'HAVELOCK_MRT_1' -> Havelock; None if no station is named."""
    d = re.sub(r'\s+', ' ', description.upper().replace('_', ' ')).strip()
    m = re.match(r'^(?:OPP |OPPOSITE )?(.+?) (MRT|LRT|STATION)\b', d)
    if not m: return None
    base, word = m.group(1), m.group(2)
    return lookup.get((base, 'LRT' if word == 'LRT' else 'MRT'))

if __name__ == '__main__':
    feed = json.load(open(RAW / 'BicycleParking_raw.json'))
    ex = read_stations()
    lookup, info = {}, {}
    for k in ex:
        name, kind, base = nice(k)
        lookup[(base, kind)] = k
        info[k] = (name, kind)
    agg = {k: dict(n=0, sh=0, pts=0) for k in ex}
    used, left_out, far, missing = 0, collections.Counter(), [], []
    for r in feed:
        if r['RackType'] not in STATION_TYPES and r['RackType'] != YELLOW:
            left_out['other owners'] += 1; continue
        k = station_of(r['Description'], lookup)
        if not k and r['RackType'] == YELLOW:
            left_out['other owners'] += 1; continue
        if not k:
            left_out['no station named'] += 1
            if re.search(r'\b(MRT|LRT)\b', r['Description'].upper().replace('_', ' ')): missing.append((r['Description'], int(r['RackCount'])))
            continue
        a = agg[k]; n = int(r['RackCount'])
        a['n'] += n; a['pts'] += 1; used += 1
        if r['ShelterIndicator'] == 'Y': a['sh'] += n
        far.append(min(metres((r['Longitude'], r['Latitude']), p) for p in ex[k]))
    rows = []
    for k, a in agg.items():
        name, kind = info[k]
        rows.append(dict(key=k, name=name, kind=kind, n=a['n'], sh=a['sh'], pts=a['pts'],
                         lon=sum(p[0] for p in ex[k]) / len(ex[k]), lat=sum(p[1] for p in ex[k]) / len(ex[k]), exits=len(ex[k])))
    tot = sum(r['n'] for r in rows)
    json.dump(dict(rows=rows, total=tot, feed=len(feed), feed_racks=sum(int(r['RackCount']) for r in feed), records=used,
                   other=left_out['other owners'], unnamed=left_out['no station named'],
                   unnamed_racks=sum(int(r['RackCount']) for r in feed if r['RackType'] in STATION_TYPES and not station_of(r['Description'], lookup)),
                   missing=missing, within=dict((m, sum(1 for d in far if d <= m)) for m in (100, 200, 300, 500)), farthest=round(max(far))),
              open(OUT, 'w'))
    print('stations', len(rows), 'with racks', sum(1 for r in rows if r['n']), 'station rows', used, 'racks', tot, 'left out', dict(left_out), 'farthest m', round(max(far)))