public async void ParseArticles(params string[] refs)
{
foreach (var url in refs)
{
try
{
Articles _article = AvitoDb.Articles.FirstOrDefault(art => art.Url == url);
if (_article != null)
{
continue;
}
//асинхронно загружаем содержание url
Task <HtmlDocument> _task = new Task <HtmlDocument>(() => { return(new HtmlWeb().Load(url)); }); _task.Start();
HtmlDocument _doc = await _task.ConfigureAwait(false);
//парсим заголовок
var _title = _doc.DocumentNode.SelectSingleNode("//header[@class='single-item-header b-with-padding']").InnerText.Remove(0, 5);
_title = _title.Remove(_title.Length - 2, 2);
//парсим цену
var _price = _doc.DocumentNode.SelectSingleNode("//span[@class='price-value']").InnerText.Remove(0, 3);
_price = _price.Split('&')[0].Replace(" ", "").Replace("\n", "");
//парсим имя хозяина
// var _person = _doc.DocumentNode.SelectSingleNode("//div[@class='person-name person-contact-name']").InnerText.Remove(0, 2).Replace("\n","");
// _person = _person.Remove(_person.Length - 21, 21);
//парсим адрес
var _address = _doc.DocumentNode.SelectSingleNode("//div[@class='person-address padding-bottom']").InnerText.Remove(0, 10).Replace("\n", "").Replace("\t", "").Replace(" ", "");
//парсим содержание
var _info = _doc.DocumentNode.SelectSingleNode("//div[@class='description-preview-wrapper']");
HtmlNode _info2 = null;
string _infoStr = "";
if (_info == null)
{
_info = _doc.DocumentNode.SelectSingleNode("//div[@class='description-preview-wrapper description-with-html']");
_info2 = _doc.DocumentNode.SelectSingleNode("//div[@class='shop-description']");
}
_infoStr = _info.InnerText.Replace("\n", "");//.Remove(0, 1).Remove(_infoStr.Length - 3, 3);
if (_info2 != null)
{
_infoStr += _info2.InnerText.Replace("\n", ""); //.Remove(0, 1).Remove(_infoStr.Length - 3, 3);
}
//парсим дату размещения
var _publicData = _doc.DocumentNode.SelectSingleNode("//div[@class='item-add-date']").InnerText;
if (_publicData.Contains("сегодня"))
{
_publicData = _publicData.Replace("сегодня", DateTime.Now.ToString("dd MMMM"));
}
if (_publicData.Contains("вчера"))
{
DateTime date = new DateTime(DateTime.Now.Year, DateTime.Now.Month, DateTime.Now.Day - 1);
_publicData = _publicData.Replace("вчера", date.ToString("dd MMMM"));
}
//парсим номер объявления
var _number = _doc.DocumentNode.SelectSingleNode("//div[@class='item-id']").InnerHtml.Replace("Объявление №", "");
//парсим номер телефона
var _phone = await _parsePhone(AvitoUrl + _doc.DocumentNode.SelectSingleNode("//a[@title='Телефон продавца']").Attributes["href"].Value + "?async", url);
_article = new Articles();
_article.Url = url;
_article.Numder = _number;
_article.Title = _title;
_article.Info = _infoStr;
_article.Price = _price;
_article.Phone = _phone;
_article.Address = _address;
_article.PublicDate = _publicData;
AvitoDb.Articles.Add(_article);
await AvitoDb.SaveChangesAsync();
}
catch (Exception ex)
{
Logger.Invoke(ex.ToString());
}
}
}
public async void ParseArticles(params string[] refs)
{
foreach (var url in refs)
{
try
{
Articles _article = AvitoDb.Articles.FirstOrDefault(art => art.Url == url);
if (_article != null)
{
continue;
}
//асинхронно загружаем содержание url
Task<HtmlDocument> _task = new Task<HtmlDocument>(() => { return new HtmlWeb().Load(url); }); _task.Start();
HtmlDocument _doc = await _task.ConfigureAwait(false);
//парсим заголовок
var _title = _doc.DocumentNode.SelectSingleNode("//header[@class='single-item-header b-with-padding']").InnerText.Remove(0, 5);
_title = _title.Remove(_title.Length - 2, 2);
//парсим цену
var _price = _doc.DocumentNode.SelectSingleNode("//span[@class='price-value']").InnerText.Remove(0, 3);
_price = _price.Split('&')[0].Replace(" ", "").Replace("\n", "");
//парсим имя хозяина
// var _person = _doc.DocumentNode.SelectSingleNode("//div[@class='person-name person-contact-name']").InnerText.Remove(0, 2).Replace("\n","");
// _person = _person.Remove(_person.Length - 21, 21);
//парсим адрес
var _address = _doc.DocumentNode.SelectSingleNode("//div[@class='person-address padding-bottom']").InnerText.Remove(0, 10).Replace("\n", "").Replace("\t", "").Replace(" ", "");
//парсим содержание
var _info = _doc.DocumentNode.SelectSingleNode("//div[@class='description-preview-wrapper']");
HtmlNode _info2 = null;
string _infoStr = "";
if (_info == null)
{
_info = _doc.DocumentNode.SelectSingleNode("//div[@class='description-preview-wrapper description-with-html']");
_info2 = _doc.DocumentNode.SelectSingleNode("//div[@class='shop-description']");
}
_infoStr = _info.InnerText.Replace("\n", "");//.Remove(0, 1).Remove(_infoStr.Length - 3, 3);
if (_info2 != null) _infoStr += _info2.InnerText.Replace("\n", "");//.Remove(0, 1).Remove(_infoStr.Length - 3, 3);
//парсим дату размещения
var _publicData = _doc.DocumentNode.SelectSingleNode("//div[@class='item-add-date']").InnerText;
if (_publicData.Contains("сегодня"))
{
_publicData = _publicData.Replace("сегодня", DateTime.Now.ToString("dd MMMM"));
}
if (_publicData.Contains("вчера"))
{
DateTime date = new DateTime(DateTime.Now.Year, DateTime.Now.Month, DateTime.Now.Day - 1);
_publicData = _publicData.Replace("вчера", date.ToString("dd MMMM"));
}
//парсим номер объявления
var _number = _doc.DocumentNode.SelectSingleNode("//div[@class='item-id']").InnerHtml.Replace("Объявление №", "");
//парсим номер телефона
var _phone = await _parsePhone(AvitoUrl + _doc.DocumentNode.SelectSingleNode("//a[@title='Телефон продавца']").Attributes["href"].Value + "?async", url);
_article = new Articles();
_article.Url = url;
_article.Numder = _number;
_article.Title = _title;
_article.Info = _infoStr;
_article.Price = _price;
_article.Phone = _phone;
_article.Address = _address;
_article.PublicDate = _publicData;
AvitoDb.Articles.Add(_article);
await AvitoDb.SaveChangesAsync();
}
catch (Exception ex)
{
Logger.Invoke(ex.ToString());
}
}
}