美文网首页iOS 收集
Swift网络爬虫

Swift网络爬虫

作者: angelababa | 来源:发表于2015-08-12 19:20 被阅读2743次
    
     NSLog("开始获取url")
            var URLARR = [String]()
            var count = 200
            
            
            for i in 1...count/20{
                println("\(i)page")
                let urlString = "http://b2b.huangye88.com/fuzhou/tuliao/pn\(i)"
                let request = NSURL(string: urlString)
                let data = NSData(contentsOfURL: request!)
                let dataString = NSString(data: data!, encoding: NSUTF8StringEncoding)!
                let bigScanner = NSScanner(string: dataString as String)
                var bigString:NSString?
                while !bigScanner.atEnd{
                    bigScanner.scanUpToString("<dt><h4><a href=\"", intoString: nil)
                    bigScanner.scanUpToString("\" ", intoString: &bigString)
                    let string = bigString!.stringByReplacingOccurrencesOfString("<dt><h4><a href=\"", withString: "")
                    URLARR.append(string)
                }
            }
            NSLog("获取url结束")
    
            var INFOS = ""
            
            NSLog("开始获取信息")
            let queue = dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0)
            let group = dispatch_group_create()
            if URLARR.count >= count{
    
                for num in 0..<20{
                    dispatch_group_async(group, queue, { () -> Void in
                        for i in num*(count/20)..<(num+1)*(count/20){
                            let request = NSURL(string: URLARR[i])
                            let data = NSData(contentsOfURL: request!)
                            if data != nil{
                                let dataString = NSString(data: data!, encoding: NSUTF8StringEncoding)!
                                //先截取信息体
                                let infoScanner = NSScanner(string: dataString as String)
                                var infoStr:NSString?
                                infoScanner.scanUpToString("<div class=\"Contact\">", intoString: nil)
                                infoScanner.scanUpToString("</div><!--Contact-->", intoString: &infoStr)
                                
                                var tempInfo = ""
                                if infoStr != nil{
                                    //公司名
                                    var componyScanner = NSScanner(string: infoStr! as String)
                                    var componyStr:NSString?
                                    componyScanner.scanUpToString("公司名称:", intoString: nil)
                                    componyScanner.scanUpToString("</a>", intoString: &componyStr)
                                    
                                    var componyScannerA = NSScanner(string:componyStr! as String)
                                    var componyScannerAStr:NSString?
                                    componyScannerA.scanUpToString("title=\"", intoString: nil)
                                    componyScannerA.scanUpToString("\">", intoString: &componyScannerAStr)
                                    let COMPONY = componyScannerAStr!.stringByReplacingOccurrencesOfString("title=\"", withString: "")
                                    tempInfo += "公司名称:" + COMPONY + "\n"
                                    //联系电话
                                    var phoneScanner = NSScanner(string: infoStr! as String)
                                    var phoneStr:NSString?
                                    phoneScanner.scanUpToString("联系人手机:", intoString: nil)
                                    phoneScanner.scanUpToString("/span>", intoString: &phoneStr)
                                    
                                    if phoneStr == nil{
                                        var phoneScannerB = NSScanner(string: infoStr! as String)
                                        var phoneStrB:NSString?
                                        phoneScannerB.scanUpToString("公司电话:", intoString: nil)
                                        phoneScannerB.scanUpToString("td>", intoString: &phoneStrB)
                                        
                                        if phoneStrB == nil {
                                            tempInfo += "公司电话:" + "无" + "\n"
                                        }else{
                                            var phoneScannerA = NSScanner(string: phoneStrB! as String)
                                            var phoneStrA:NSString?
                                            phoneScannerA.scanUpToString("boldred\">", intoString: nil)
                                            phoneScannerA.scanUpToString("<", intoString: &phoneStrA)
                                            let PHONE = phoneStrA!.stringByReplacingOccurrencesOfString("boldred\">", withString: "")
                                            tempInfo += "公司电话:" + PHONE + "\n"
                                        }
                                       
                                    }else{
                                        var phoneScannerA = NSScanner(string: phoneStr! as String)
                                        var phoneStrA:NSString?
                                        phoneScannerA.scanUpToString("boldred\">", intoString: nil)
                                        phoneScannerA.scanUpToString("<", intoString: &phoneStrA)
                                        let PHONE = phoneStrA!.stringByReplacingOccurrencesOfString("boldred\">", withString: "")
                                        tempInfo += "联系人手机:" + PHONE + "\n"
                                    }
                                    
                                    //联系人名字
                                    var nameScanner = NSScanner(string: infoStr! as String)
                                    var nameStr:NSString?
                                    nameScanner.scanUpToString("联系人:", intoString: nil)
                                    nameScanner.scanUpToString("/span>", intoString: &nameStr)
                                    if nameStr == nil {
                                        tempInfo += "联系人:" + "无" + "\n\n"
                                    }else{
                                        var nameScannerA = NSScanner(string: nameStr! as String)
                                        var nameStrA:NSString?
                                        nameScannerA.scanUpToString("span>", intoString: nil)
                                        nameScannerA.scanUpToString("<", intoString: &nameStrA)
                                        let NAME = nameStrA!.stringByReplacingOccurrencesOfString("span>", withString: "")
                                        tempInfo += "联系人:" + NAME + "\n\n"
                                    }
                                    
                                    INFOS += tempInfo
                                    println(tempInfo)
                                }
                                
                            }
                            
                        }
                    })
                }
                dispatch_group_notify(group, queue) { () -> Void in
                    NSLog("获取信息结束")
                    NSLog("开始写入文件")
                    //文件写入jinrong
                    let file = "Infos.txt"
                    if let dirs : [String] = NSSearchPathForDirectoriesInDomains(NSSearchPathDirectory.DocumentDirectory, NSSearchPathDomainMask.AllDomainsMask, true) as? [String] {
                        let dir = dirs[0]
                        let path = dir.stringByAppendingPathComponent(file);
                        println(path)
                        INFOS.writeToFile(path, atomically: false, encoding: NSUTF8StringEncoding, error: nil);
                        
                    }
                    NSLog("写入文件结束")
                }
                
            }else{
                println("URLARR.count->\(URLARR.count)")
            }
    

    相关文章

      网友评论

      本文标题:Swift网络爬虫

      本文链接:https://www.haomeiwen.com/subject/xugiqttx.html