doormile_milderapp/lib/data/stop_area.dart

/// ─────────────────────────────────────────────────────────────────────────
/// WHERE A STOP IS, IN ONE WORD
///
/// A rider scanning twenty stops is deciding between *areas* — is Joe on the
/// way to Priya, or the opposite side of the city. He is not reading twenty
/// street strings, and he certainly is not reading twenty street strings cut
/// off mid-word: `SEQTEST Gandhipur…` answers nothing that a blank line would
/// not have answered, and costs a line to do it.
///
/// So the collapsed queue shows the locality and the full address lives where
/// the rare question is asked — the detail sheet, and the map at the door.
///
/// ── Why this is one function and not two ──
///
/// Home and Deliveries both needed it and both had their own. The rule is
/// subtle enough to get wrong in exactly the same way twice: the first version
/// skipped a component only when it was *entirely* digits, so
/// `12 SNS Colony, Peelamedu, Coimbatore 641004` returned `12 SNS Colony` — a
/// street with a door number on it, presented as an area. Two copies of that
/// bug is two places to find it.
/// ─────────────────────────────────────────────────────────────────────────
library;

/// The locality component of a stop's address, or `''` when it has none.
///
/// Prefers the drop, because on a delivery leg the area that matters is where
/// the bag is going. Falls back to the pickup for a stop that carries no drop,
/// and returns nothing rather than inventing a placeholder.
String areaOf(Map<String, dynamic> stop, {bool preferDrop = true}) {
  final keys = preferDrop
      ? const [
          'dropaddress',
          'DropAddress',
          'deliveryaddress',
          'pickupaddress',
          'PickupAddress',
        ]
      : const [
          'pickupaddress',
          'PickupAddress',
          'dropaddress',
          'DropAddress',
          'deliveryaddress',
        ];

  for (final key in keys) {
    final raw = (stop[key] ?? '').toString().trim();
    if (raw.isEmpty) continue;

    final parts = raw
        .split(',')
        .map((p) => p.trim())
        .where((p) => p.isNotEmpty)
        .toList();

    // ── An area is a NAME, not a description ──
    //
    // Three versions of this rule, each corrected by a real address:
    //
    // 1. "Skip a part only when it is entirely digits" printed `12 SNS
    //    Colony` — a door number presented as an area.
    // 2. "Skip a part containing any digit" fixed that, but then
    //    `12 Peelamedu, Coimbatore` lost `12 Peelamedu` wholesale and answered
    //    with the *city* — and every stop on a Coimbatore route is in
    //    Coimbatore, so the queue said nothing at all. Seen on a device.
    // 3. Now the digits are stripped from *inside* the part instead of taking
    //    the part with them: `12 Peelamedu` yields the name `Peelamedu`.
    //
    // What separates a locality from a landmark is length in words — names
    // are short (Gandhipuram, Peelamedu), landmarks and streets are phrases
    // (`Coimbatore Gandhipuram Mofussil Bus Stand`, verified live). And the
    // parts are scanned in order because an Indian address runs specific →
    // general: the earliest name is the most local one, which is exactly what
    // beats `Coimbatore` when both are one-word names.
    //
    // The word left when a number is removed is sometimes administrative
    // noise, not a place — `Ward 54` is not the ward's name — so those
    // residues are discarded by a small stop list.
    // `Ward 54` is not a place called Ward, and `4 Cross` — Indian street
    // numbering — is not a place called Cross. Both from live addresses.
    const noise = {
      'ward', 'zone', 'no', 'door', 'flat', 'floor', 'plot', 'block', //
      'cross', 'main', 'street', 'road',
    };
    String? nameOf(String part) {
      final words = [
        for (final w in part.split(RegExp(r'\s+')))
          if (!RegExp(r'[0-9]').hasMatch(w)) w,
      ];
      if (words.isEmpty) return null;
      // `RS Puram RS Puram` — the geocoder stammer [compactAddress] already
      // collapses — reaches here too, seen live as `Gandhipuram Gandhipuram`
      // on the queue. A name whose two halves are the same name is one name.
      if (words.length.isEven && words.length >= 2) {
        final half = words.length ~/ 2;
        final a = words.sublist(0, half).join(' ');
        final b = words.sublist(half).join(' ');
        if (a.toLowerCase() == b.toLowerCase()) {
          words.removeRange(half, words.length);
        }
      }
      final name = words.join(' ');
      if (words.length == 1 &&
          (name.length < 3 || noise.contains(name.toLowerCase()))) {
        return null;
      }
      return name;
    }

    // ── The city rung is named, because position cannot find it ──
    //
    // "Drop the last part" was tried and is structurally undecidable: an
    // address that ends at the city (`2 Saibaba Colony, Coimbatore`) and one
    // that ends at the locality (`12, Cross Cut Road, Gandhipuram`) look
    // identical from the end. Both are live addresses, and each broke the
    // heuristic the other needed.
    //
    // So the rung is knowledge, like the noise words above: the region this
    // tenant operates in, plus the state/country tails geocoders append. A
    // city name distinguishes nothing on a route that is entirely inside it,
    // so it ranks last — but it still stands when it is all the address has,
    // because an honest rung beats a blank.
    bool isCity(String n) => const {
      'coimbatore', 'coimbatore north', 'coimbatore south', //
      'tamil nadu', 'india', 'chennai', 'tiruppur', 'erode', 'salem',
      'madurai', 'pollachi', 'mettupalayam',
    }.contains(n.toLowerCase());

    // A two-word name ending in a thoroughfare word is a street, not a
    // locality — `Mettupalayam Road` must lose to the `RS Puram` behind it,
    // exactly as the longer landmark phrases already do.
    bool isStreet(String n) => const {
      'road', 'street', 'salai', 'lane', 'veedhi', 'highway', //
    }.contains(n.split(' ').last.toLowerCase());

    final names = [
      for (final p in parts)
        if (nameOf(p) case final n?) n,
    ];
    final local = [
      for (final n in names)
        if (!isCity(n)) n,
    ];

    // In rank order: the first one-word name — a locality is a name and
    // names are short; then the first short non-street name (RS Puram,
    // Saibaba Colony); then any phrase left; and only then the city.
    for (final n in local) {
      if (!n.contains(' ')) return n;
    }
    for (final n in local) {
      if (n.split(' ').length <= 2 && !isStreet(n)) return n;
    }
    if (local.isNotEmpty) return local.first;
    if (names.isNotEmpty) return names.first;
  }
  return '';
}

/// A full address, minus the boilerplate a geocoder appends.
///
/// The record page prints the address in full — it is the one screen whose
/// job is the specifics — and on live data that full address arrived as
/// `RS Puram RS Puram, Ward 24, West Zone, Perur, Coimbatore South,
/// Coimbatore, Tamil Nadu, 641002, India`, wrapped to two lines and cut with
/// an ellipsis. The trimmed tail was the useful half; the kept head began by
/// stammering.
///
/// This strips only what carries no information at a door in this region:
/// the state and country rungs, a part that is nothing but a pincode, an
/// immediate word-for-word repeat inside a part, and a part that repeats the
/// one before it. Everything that names a place stays, in order.
String compactAddress(String raw) {
  final parts = raw
      .split(',')
      .map((p) => p.trim())
      .where((p) => p.isNotEmpty)
      .toList();

  const tail = {'tamil nadu', 'india'};

  String dedupWords(String part) {
    final w = part.split(RegExp(r'\s+'));
    if (w.length.isEven && w.length >= 2) {
      final half = w.length ~/ 2;
      final a = w.sublist(0, half).join(' ');
      final b = w.sublist(half).join(' ');
      if (a.toLowerCase() == b.toLowerCase()) return a;
    }
    return part;
  }

  final kept = <String>[];
  for (final p in parts) {
    final cleaned = dedupWords(p);
    final lower = cleaned.toLowerCase();
    if (tail.contains(lower)) continue;
    if (RegExp(r'^[0-9]{6}$').hasMatch(cleaned)) continue;
    if (kept.isNotEmpty && kept.last.toLowerCase() == lower) continue;
    kept.add(cleaned);
  }
  return kept.join(', ');
}