pa-tournament-stats/TournamentStats/scrapper/Downloader.swift

//
//  Downloader.swift
//  TournamentStats
//
//  Created by Laurent Morvillier on 02/10/2021.
//  Copyright © 2021 Stax River. All rights reserved.
//

import Foundation
import RealmSwift

class Downloader {
    
    struct TS {
        var id: String
        var date: String
    }
    
    static func downloadWSOPResults() {
        
        let infos = [
            TS(id: "19728", date: "02/10/2021"),
                     TS(id: "19729", date: "03/10/2021"),
                     TS(id: "19730", date: "02/10/2021"),
                     TS(id: "19732", date: "04/10/2021")]
        
        for info in infos {
            downloadWSOPResults(tournamentInfo: info)
        }
        
//        for i in 19728...19732 {
//            DispatchQueue.global(qos: .utility).async {
//
//                downloadWSOPResults(tournamentId: "\(i)")
//
//            }
//        }
        
    }
    
    fileprivate static func downloadWSOPResults(tournamentInfo: TS) {
        
        print("Importing tournament \(tournamentInfo.id)")
        
        guard let url = URL(string: "https://www.wsop.com/tournaments/results/?aid=2&grid=4865&tid=\(tournamentInfo.id)&rr=5") else {
            return
        }
        
        let realm = try! Realm()
        
        let nf = NumberFormatter()
        nf.numberStyle = .decimal
        nf.locale = Locale.init(identifier: "en_US")
        nf.groupingSeparator = ","
        
        do {
            let string = try String(contentsOf: url)
            
            let infos = string.slice(from: "<div id=\"eventinfo\">", to: "<div class='eventinfo-reports'>")
            ?? string.slice(from: "<div id=\"eventinfo\">", to: "eventinfo-progress")
            ?? string.slice(from: "<div id=\"eventinfo\">", to: "eventstats")
            
            if let infos = infos {
                
                try realm.write {
                    
                    let tournament = realm.create(Tournament.self)
                    
                    if let name = infos.slice(from: "<h1>", to: "</h1>") {
                        tournament.name = name
                        tournament.date = Formatter.basicDate.date(from: tournamentInfo.date) ?? Date()
                        
                        if let strNumber = name.slice(from: "Event #", to: ":"), let number = nf.number(from: strNumber) {
                            tournament.number = number.intValue
                        }
                        
                        if name.contains("6-Handed") {
                            tournament.tableSize = 6
                        } else if name.contains("8-Handed") {
                            tournament.tableSize = 8
                        } else if name.contains("Heads-Up") {
                            tournament.tableSize = 2
                        }

                    } else {
                        print("no name = \(infos)")
                    }
                    
                    if let stats = string.slice(from: "<ul id=\"eventstats\">", to: "<div id=\"nav-tabs\">") {
                        
                        if let strBuyin = stats.slice(from: "Buy-in:</b> &#36;", to: "</li>"), let buyin = nf.number(from: strBuyin) {
                            tournament.buyin = buyin.intValue
                        } else {
                            print("no buyin = \(stats)")
                        }
                        
                        if let strprizepool = stats.slice(from: "Prizepool:</b> &#36;", to: " </li>"), let pp = nf.number(from: strprizepool) {
                            tournament.prizepool = pp.doubleValue
                        } else {
                            print("no prizepool = \(stats)")
                        }
                        
                        if let strentries = stats.slice(from: "Entries:</b> ", to: " </li>"), let entries = nf.number(from: strentries) {
                            tournament.entries = entries.intValue
                        } else {
                            print("no entries = \(stats)")
                        }
                    }
                    
                    if let strresultsblock = string.slice(from: "Country</li>", to: "<div id=\"PagingNav\"") {
                        let strresults =  strresultsblock.components(separatedBy: "<li class='place")
                        
                        print("Results count = \(strresults.count)")
                        
                        for strresult in strresults {
                            
                            let cells = strresult.components(separatedBy: "<li")
                            if !(cells.count == 6 || cells.count == 7) {
                                print("cell count = \(cells.count)")
                                continue
                            }
                            
                            let hasPOY = cells.count == 7
                            
                            var rank: Int? = nil
                            var name: String? = nil
                            var earnings: Double? = nil
                            var city: String? = nil
                            var state: String? = nil
                            var country: String? = nil
                            var poy: Double? = nil
                            
                            if let strrank = cells[0].slice(from: ">", to: "<") {
                                rank = nf.number(from: strrank)?.intValue
                            }
                            
                            if let nameraw = cells[1].components(separatedBy: "'>").last {
                                name = nameraw.replacingOccurrences(of: "</a></li>", with: "")
                            }
                            
                            if let earningsraw = cells[2].components(separatedBy: "&#36;").last {
                                let strearnings = earningsraw.replacingOccurrences(of: "</li>", with: "")
                                earnings = nf.number(from: strearnings)?.doubleValue
                            }
                            
                            if hasPOY, let poyraw = cells[3].components(separatedBy: "\'>").last {
                                let strpoy = poyraw.replacingOccurrences(of: "</li>", with: "")
                                poy = nf.number(from: strpoy)?.doubleValue
                            }
                            
                            let cityIndex = hasPOY ? 4 : 3
                            if let cityraw = cells[cityIndex].components(separatedBy: "\'>").last {
                                city = cityraw.replacingOccurrences(of: "</li>", with: "")
                            }
                            
                            let stateIndex = hasPOY ? 5 : 4
                            if let stateraw
                                = cells[stateIndex].components(separatedBy: "\'>").last {
                                state = stateraw.replacingOccurrences(of: "</li>", with: "")
                            }
                            let countryIndex = hasPOY ? 6 : 5
                            if let countryraw
                                = cells[countryIndex].components(separatedBy: "\'>").last {
                                country = String(countryraw.prefix(2))
//                                country = countrytmp.prefix(2)
                            }

                            guard let rank = rank, let name = name, let earnings = earnings else {
                                print("Missing data: R: \(rank), N: \(name), E: \(earnings)")
                                continue
                            }
                            
                            let player = realm.getOrCreatePlayer(name: name, city: city ?? "", state: state ?? "", country: country ?? "")
                            
                            let result = realm.create(Result.self)
                            result.player = player
                            result.earnings = earnings
                            result.rank = rank
                            
                            tournament.results.append(result)
                            
                        }
                        
                    } else {
                        print("Bad page strresultsblock")
                    }
                }
            } else {
                print("Missing name block")
            }
            
        } catch {
            print(error)
        }
        
    }
    
}