fuelboard/Shared/StationAddress.swift

import Foundation

/// How a station's address should read underneath its name.
///
/// The official Fuel Finder `address` field is not a street address on its own.
/// It normally opens with the station's own name — 14% of stations start with
/// the exact name, and many more repeat it in the first segment ("CHIDSWELL
/// SERVICE STATION, LEEDS ROAD, DEWSBURY, WF12 7QW" for "Jet Chidswell") — and
/// 13% end with the postcode the row appends again. With the name already in the
/// row's header (and in the detail sheet's "Name" row), both repetitions read as
/// the name stated twice.
///
/// The rules are deliberately conservative: a leading segment carrying a house
/// number is a real street address and is left alone, and a partial-word match
/// ("Jet" against "JETSON FILLING STATION") never cuts into a word.
enum StationAddressLine {
    /// Words that carry no identity, so a leading segment made only of these
    /// plus words taken from the name still counts as the name.
    private static let filler: Set<String> = [
        "service", "services", "station", "stations", "filling", "petrol",
        "garage", "motors", "motor", "centre", "center", "store", "superstore",
        "ltd", "limited", "trading", "the", "and", "of", "at",
    ]

    /// The line to display: any leading repeat of the name removed, and the
    /// postcode written exactly once (whatever the address already carries).
    static func display(address: String, postcode: String, name: String) -> String {
        let postcodeText = postcode.trimmingCharacters(in: .whitespacesAndNewlines)
        var text = collapsingSpaces(address)
        if text.isEmpty { return postcodeText }
        text = removingLeadingName(from: text, name: name)
        text = removingTrailingPostcode(from: text, postcode: postcodeText)
        text = text.trimmingCharacters(in: CharacterSet(charactersIn: " ,;-"))
        if text.isEmpty { return postcodeText }
        if postcodeText.isEmpty { return text }
        return "\(text), \(postcodeText)"
    }

    // MARK: - Internals

    private static func collapsingSpaces(_ text: String) -> String {
        text.split(whereSeparator: { $0 == " " || $0 == "\t" || $0 == "\n" })
            .joined(separator: " ")
            .trimmingCharacters(in: .whitespaces)
    }

    /// Letters and digits only, lowercased — so punctuation and spacing cannot
    /// make two spellings of the same name look different.
    private static func key(_ text: String) -> String {
        text.lowercased().filter { $0.isLetter || $0.isNumber }
    }

    private static func removingLeadingName(from text: String, name: String) -> String {
        let nameKey = key(name)
        guard !nameKey.isEmpty else { return text }
        if key(text).hasPrefix(nameKey), endsAtWordBoundary(text, characters: nameKey.count) {
            return droppingFirst(text, characters: nameKey.count)
        }
        return removingLeadingNameSegment(from: text, nameKey: nameKey)
    }

    /// True when the first `count` letters/digits of `text` are followed by
    /// something that is not part of a word — a partial-word match must never
    /// cut a station out of the middle of a word.
    private static func endsAtWordBoundary(_ text: String, characters count: Int) -> Bool {
        var seen = 0
        var index = text.startIndex
        while index < text.endIndex, seen < count {
            let character = text[index]
            if character.isLetter || character.isNumber { seen += 1 }
            index = text.index(after: index)
        }
        guard index < text.endIndex else { return true }
        let next = text[index]
        return !(next.isLetter || next.isNumber)
    }

    /// Drops the first `count` letters/digits of `text` and any separator after
    /// them, leaving the rest of the original spelling untouched.
    private static func droppingFirst(_ text: String, characters count: Int) -> String {
        var seen = 0
        var index = text.startIndex
        while index < text.endIndex, seen < count {
            let character = text[index]
            if character.isLetter || character.isNumber { seen += 1 }
            index = text.index(after: index)
        }
        return String(text[index...]).trimmingCharacters(in: CharacterSet(charactersIn: " ,;-"))
    }

    /// Drops a trailing repeat of the postcode.
    private static func removingTrailingPostcode(from text: String, postcode: String) -> String {
        let postcodeKey = key(postcode)
        guard !postcodeKey.isEmpty else { return text }
        let textKey = key(text)
        guard textKey.hasSuffix(postcodeKey), textKey.count > postcodeKey.count else { return text }
        var count = postcodeKey.count
        var index = text.endIndex
        while index > text.startIndex, count > 0 {
            index = text.index(before: index)
            let character = text[index]
            if character.isLetter || character.isNumber { count -= 1 }
        }
        return String(text[text.startIndex..<index]).trimmingCharacters(in: CharacterSet(charactersIn: " ,;-"))
    }

    /// "CHIDSWELL SERVICE STATION, LEEDS ROAD…" holds the name in its first
    /// segment but never as a literal prefix. Drop that segment only when it
    /// carries no digits — a house number means it is really a street address —
    /// and every word in it is either filler or part of the name.
    private static func removingLeadingNameSegment(from text: String, nameKey: String) -> String {
        guard let comma = text.firstIndex(of: ",") else { return text }
        let segment = String(text[text.startIndex..<comma])
        if segment.contains(where: { $0.isNumber }) { return text }
        let words = segment.split(whereSeparator: { !$0.isLetter }).map { $0.lowercased() }
        let significant = words.filter { $0.count >= 3 && !filler.contains($0) }
        guard !significant.isEmpty, significant.allSatisfy({ nameKey.contains($0) }) else { return text }
        return String(text[text.index(after: comma)...])
            .trimmingCharacters(in: CharacterSet(charactersIn: " ,;-"))
    }
}