private void ParseAdvs()
{
if (!_db.Connect())
{
MessageBox.Show(General.ERROR_DB_CONNECT);
Invoke(new Action(() =>
{
statusLabel.Text = STATUS_BAR_DEFAULT;
uploadButton.Enabled = true;
exploreButton.Enabled = true;
parseButton.Enabled = true;
exportButton.Enabled = true;
textBox2.Enabled = true;
fromBox.Enabled = true;
toBox.Enabled = true;
}));
_db.Close();
return;
}
var targets = textBox2.Text.Split(';');
var def = (int)(toBox.Value - fromBox.Value) + 1;
uint i = 0;
foreach (var target in targets)
{
var source = (target.Contains("_arc")) ? 2 : 1;
var dirs = Directory.GetDirectories(target);
var pathElements = target.Split('\\');
var comma = pathElements[pathElements.Length - 1].Split('_');
var type = comma[2] + '_' + comma[3];
foreach (var dir in dirs)
{
if (_stopFlag)
{
break;
}
var dsp = dir.Split('\\');
var desp = dsp[dsp.Length - 2];
var idStr = dsp[dsp.Length - 1].Split('_')[0];
if (!int.TryParse(idStr, out int id))
{
i++;
continue;
}
if (id < fromBox.Value || id > toBox.Value)
{
i++;
continue;
}
char status = '-';
var singleName = dir.Split('\\')[dir.Split('\\').Length - 1];
var fileName = target + "\\" + singleName + ".html";
if (!File.Exists(fileName))
{
fileName = target + "\\" + singleName + "_arch.html";
if (!File.Exists(fileName))
{
fileName = target + "\\" + singleName + "_udalen.html";
if (!File.Exists(fileName))
{
i++;
continue;
}
}
}
if (fileName.Contains("udalen"))
{
status = 'u';
_db.UpdateStatus(id, status, source);
continue;
}
FileInfo fileInfo = new FileInfo(fileName);
string newname = fileName;
if (fileInfo.Length <= 25600)
{
//if(!fileName.Contains("udalen"))
newname = target + "\\" + singleName + "_udalen.html";
File.Move(fileName, newname);
status = 'u';
_db.UpdateStatus(id, status, source);
continue;
}
if (fileName.Contains("arch"))
{
status = 'a';
}
_db.UpdateStatus(id, status, source);
var doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(File.ReadAllText(fileName));
var restrictions = General.ExtractValueFromDoc(SiteType.Farpost, "restrictions", doc);
var agency = General.ExtractValueFromDoc(SiteType.Farpost, "agency", doc);
var rooms = General.ExtractValueFromDoc(SiteType.Farpost, "rooms", doc);
var description = General.ExtractValueFromDoc(SiteType.Farpost, "description", doc);
_db.UpdateAdvData(id, restrictions, string.Empty, agency, rooms, description);
var added = General.ExtractValueFromDoc(SiteType.Farpost, "added", doc);
if (!String.IsNullOrEmpty(added))
{
var underlines = singleName.Split('_');
/*var savedDateTime = DateTime.MinValue;
* DateTime.TryParseExact(underlines[2], "yyyyddMM", new CultureInfo("ru-RU"), DateTimeStyles.None, out savedDateTime);*/
var savedDateTime = File.GetCreationTime(fileName);
var hour = 0;
var minute = 0;
var dateTime = DateTime.MinValue;
var addedArr = added.Split(',');
if (addedArr.Length >= 2)
{
var time = addedArr[0].Trim();
var timeArr = time.Split(':');
if (timeArr.Length >= 2)
{
int.TryParse(timeArr[0], out hour);
int.TryParse(timeArr[1], out minute);
}
var date = addedArr[1].Trim().ToLower();
if (date == "сегодня")
{
dateTime = new DateTime(savedDateTime.Year, savedDateTime.Month, savedDateTime.Day, hour, minute, 0);
date = "Сегодня";
}
else if (date == "вчера")
{
dateTime = new DateTime(savedDateTime.Year, savedDateTime.Month, savedDateTime.Day, hour, minute, 0);
dateTime.AddDays(-1);
date = "Вчера";
}
else
{
var dateArr = date.Split(' ');
var format = dateArr.Length == 2 ? "d MMMM" : "d MMMM yyyy";
DateTime.TryParseExact(date, format, new CultureInfo("ru-RU"), DateTimeStyles.None, out dateTime);
}
_db.UpdateDate(id, dateTime, date);
}
}
var username = General.ExtractValueFromDoc(SiteType.Farpost, "username", doc);
var phone = General.ExtractValueFromDoc(SiteType.Farpost, "phone", doc);
var email = General.ExtractValueFromDoc(SiteType.Farpost, "email", doc);
var advsStr = General.GetNumbers(General.ExtractValueFromDoc(SiteType.Farpost, "advs", doc));
int.TryParse(advsStr, out int advs);
_db.UpdateUserInfo(id, username, phone, email, advs);
var address = General.ExtractValueFromDoc(SiteType.Farpost, "address", doc);
var hood = General.ExtractValueFromDoc(SiteType.Farpost, "district", doc);
var cadastre = General.ExtractValueFromDoc(SiteType.Farpost, "cadastreNumber", doc);
var city = General.ExtractValueFromDoc(SiteType.Farpost, "city", doc);
_db.UpdateAddr(id, address, string.Empty, string.Empty, string.Empty, string.Empty, cadastre, city);
var advName = fileName.Split('\\')[fileName.Split('\\').Length - 1];
var advNumber = advName.Split('_')[1];
var ID = _db.GetIdByNumber(advNumber);
if (ID != -1) // Если объява с таким внутренним идом уже существует
{
var adv = _db.GetAdvPdata(ID);
var advOld = _db.GetAdvPdata(id);
if ((General.GetNumbers(adv.Price) != General.GetNumbers(advOld.Price) || General.GetNumbers(adv.Square) != General.GetNumbers(advOld.Square)))
{
_db.UpdateIsNew(id, 0);
}
}
else
{
_db.UpdateIsNew(id, 1);
}
if (type == "КН_ар")
{
var url = _db.GetUrlById(id);
var comType = "n/a";
if (url.Contains("/market/"))
{
comType = "Торговая точка";
}
else if (url.Contains("/terminal/"))
{
comType = "Складское помещение";
}
else if (url.Contains("/workroom/"))
{
comType = "Производственное помещение";
}
else if (url.Contains("/office/"))
{
comType = "Офисное помещенине";
}
_db.UpdateComTypeSec(advNumber, comType);
try { stopButton.Enabled = false; } catch { }
}
else if (type == "КН_пр" && !string.IsNullOrEmpty(rooms))
{
_db.UpdateComTypeSec(advNumber, char.ToUpper(rooms[0]) + rooms.Substring(1));
}
Invoke(new Action(() =>
{
statusLabel.Text = "Обработка объявления #" + id + " [" + i + "/" + def + "]";
}));
i++;
}
}
Invoke(new Action(() =>
{
statusLabel.Text = STATUS_BAR_DEFAULT;
uploadButton.Enabled = true;
exploreButton.Enabled = true;
parseButton.Enabled = true;
exportButton.Enabled = true;
textBox2.Enabled = true;
fromBox.Enabled = true;
toBox.Enabled = true;
stopButton.Enabled = false;
}));
_db.Close();
}
private void ParsePage(string url)
{
if (!_db.Connect())
{
MessageBox.Show(General.ERROR_DB_CONNECT);
_db.Close();
Invoke(new Action(() =>
{
startButton.Text = "Начать";
startButton.Enabled = true;
statusLabel.Text = General.STATUS_DEFAULT;
}));
return;
}
/*
* Трём '#' и всё после него, если таковой есть
*/
var shard_idx = url.IndexOf("#");
if (shard_idx > -1)
{
url = url.Remove(shard_idx);
}
//
Invoke(new Action(() =>
{
textBox1.Text = url;
}));
//
var lastId = _db.GetLastID() + 1;
var siteType = General.GetSiteType(url);
uint cnt = 0;
if (siteType == SiteType.Farpost)
{
if (url.Contains("status=archive"))
{
_archiveFlag = true;
}
else
{
_archiveFlag = false;
}
var singleType = Farpost.GetTypeByLink(url);
var dateTime = DateTime.Now.ToString("yyyyMMdd_HHmmss");
var folderName = dateTime + "_" + singleType + (_archiveFlag ? "_arc" : string.Empty);
var workingDir = General.FARPOST_FOLDER + folderName + "\\";
var Comment = (CommentBox.Text != String.Empty ? CommentBox.Text.Replace(" ", "_") : string.Empty);
Directory.CreateDirectory(workingDir);
List <Advertisement> advsData;
var saveTo = workingDir + "L_" + dateTime + "_" + (_archiveFlag ? "_arc" : string.Empty) + singleType + ".html";
var typeInt = _db.GetAdvTypeByString(singleType);
if (url.Contains("?page=") || url.Contains("&page="))
{
Invoke(new Action(() =>
{
statusLabel.Text = "Собираем информацию с ленты...";
}));
advsData = Farpost.GetDataFromFeed(url, saveTo);
_db.InsertPageInfo(saveTo, typeInt, lastId, lastId + advsData.Count - 1, DateTime.Now.ToString("yyyy.dd.MM hh:mm:ss"));
}
else
{
var pages = Farpost.GetItemsCount(url);
pages = (int)Math.Ceiling((float)(pages / 50)) + 1;
if (pages != 1)
{
Invoke(new Action(() =>
{
statusLabel.Text = "Собираем информацию со всех страниц ленты (0/" + pages + ")...";
}));
advsData = new List <Advertisement>();
for (var i = 1; i <= pages; i++)
{
if (_stopFlag)
{
break;
}
saveTo = workingDir + "L_" + dateTime + "_" + (_archiveFlag ? "_arc" : string.Empty) + singleType + "_" + i + ".html";
var urlp = url + (url.Contains("?") ? "&" : "?") + "page=" + i;
var temp = Farpost.GetDataFromFeed(urlp, saveTo);
advsData.AddRange(temp);
_db.InsertPageInfo(saveTo, typeInt, lastId, lastId + advsData.Count - 1, DateTime.Now.ToString("yyyy.dd.MM hh:mm:ss"));
Invoke(new Action(() =>
{
statusLabel.Text = "Собираем информацию со всех страниц ленты (" + i + "/" + pages + ")...";
}));
if (pages != i)
{
General.SmallDelay();
}
}
}
else
{
Invoke(new Action(() =>
{
statusLabel.Text = "Собираем информацию с ленты...";
}));
advsData = Farpost.GetDataFromFeed(url, saveTo);
_db.InsertPageInfo(saveTo, typeInt, lastId, lastId + advsData.Count - 1, DateTime.Now.ToString("yyyy.dd.MM hh:mm:ss"));
}
}
if (_stopFlag)
{
_db.Close();
Invoke(new Action(() =>
{
startButton.Text = "Начать";
startButton.Enabled = true;
statusLabel.Text = General.STATUS_DEFAULT;
}));
return;
}
if (advsData == null || advsData.Count == 0)
{
MessageBox.Show("Не удалось собрать объявления с указаной страницы.");
_db.Close();
Invoke(new Action(() =>
{
startButton.Text = "Начать";
startButton.Enabled = true;
statusLabel.Text = General.STATUS_DEFAULT;
}));
return;
}
cnt = 0;
Invoke(new Action(() =>
{
listBox1.Items.Clear();
}));
Invoke(new Action(() =>
{
statusLabel.Text = "Обработка данных...";
}));
var todel = new List <Advertisement>();
var updated = new List <int>();
var trigger = checkBox1.Checked;
int newId = lastId;
for (var i = 0; i < advsData.Count; i++)
{
if (_stopFlag)
{
break;
}
var id = _db.GetIdByNumber(advsData[i].Number);
if (id != -1) // Если объява с таким внутренним идом уже существует, то
{
// Получаем информацию о ней. И если цена и площадь остались прежними - удаляем из списка на инсерты данную великолепную запись.
var adv = _db.GetAdvPdata(id);
if ((!trigger && General.GetNumbers(adv.Price) == General.GetNumbers(advsData[i].Price) && General.GetNumbers(adv.Square) == General.GetNumbers(advsData[i].Square) ||
(trigger && General.GetNumbers(adv.Price) == General.GetNumbers(advsData[i].Price) && General.GetNumbers(adv.Square) == General.GetNumbers(advsData[i].Square) && _db.GetTownByMainId(id) == -1 && _db.GetTownByMainId(id) == 0)))
{
todel.Add(advsData[i]);
continue;
}
updated.Add(newId);
}
_db.InsertRecord();
newId++;
}
foreach (var item in todel)
{
advsData.Remove(item);
}
string distr = String.Empty;
if (advsData.Count > 0)
{
foreach (var advData in advsData)
{
if (_stopFlag)
{
break;
}
Invoke(new Action(() =>
{
listBox1.Items.Add(advData.Link);
countLabel.Text = "Кол-во объявлений: " + ++cnt;
}));
if (advData.Annotations != null)
{
foreach (string item in advData.Annotations)
{
if (item.Contains("р-н"))
{
distr = item;
break;
}
}
}
var dir = workingDir + lastId + "_" + advData.Number + "_" + DateTime.Now.ToString("yyyyddMM_hhmmss") + (_archiveFlag ? "_arc" : string.Empty);
var newdir = workingDir.Trim('\\') + "_" + advsData.Count + Comment;
int is_new = 1;
if (updated.Contains(lastId))
{
is_new = 0;
}
int source = 1;
if (_archiveFlag)
{
source = 2;
}
_db.UpdateAdvData(lastId, is_new, source, advData.Number, advData.Link, "www.farpost.ru", newdir, advData.CurLink, advData.Title, advData.Price, advData.Square, advData.Geo, advData.City, distr, advData.Views, typeInt);
_db.InsertSecData(lastId);
Directory.CreateDirectory(dir);
lastId++;
}
if (singleType.Contains("КН_") && !(url.Contains("&flatType") || url.Contains("?flatType")))
{
var commercialTypes = new string[][] {
new [] { "outlet", "Торговая точка" },
new [] { "manufacture", "Производство" },
new [] { "office", "Офис" },
new [] { "storage", "Склад" },
new [] { "etc", "Другое" },
};
var advsDataAd = new List <Advertisement>();
foreach (var commercialType in commercialTypes)
{
var urlp = url + (url.Contains("?") ? "&" : "?") + "flatType%5B%5D=" + commercialType[0];
var pages = Farpost.GetItemsCount(urlp);
pages = (int)Math.Ceiling((float)(pages / 50)) + 1;
if (pages != 1)
{
for (var i = 1; i <= pages; i++)
{
if (_stopFlag)
{
break;
}
var paged = urlp + (urlp.Contains("?") ? "&" : "?") + "page=" + i;
var temp = Farpost.GetDataFromFeed(paged, null);
advsDataAd.AddRange(temp);
if (pages != i)
{
General.SmallDelay();
}
}
}
else
{
advsDataAd = Farpost.GetDataFromFeed(urlp, null);
}
foreach (var advDataAd in advsDataAd)
{
_db.UpdateComType(advDataAd.Number, commercialType[1]);
}
}
}
}
var newPath = workingDir.Trim('\\') + "_" + advsData.Count + "_" + Comment;
Directory.Move(workingDir, newPath);
_lastSaveToPath = newPath;
}
else if (siteType == SiteType.Cian)
{
var links = Cian.GetLinksFromFeed(url);
if (links == null)
{
MessageBox.Show("Не удалось собрать ссылки на объявления с указаной страницы.");
_db.Close();
Invoke(new Action(() =>
{
startButton.Text = "Начать";
startButton.Enabled = true;
statusLabel.Text = General.STATUS_DEFAULT;
}));
return;
}
cnt = 0;
Invoke(new Action(() =>
{
listBox1.Items.Clear();
}));
foreach (var link in links)
{
if (string.IsNullOrEmpty(link))
{
continue;
}
Invoke(new Action(() =>
{
listBox1.Items.Add(link);
countLabel.Text = "Кол-во объявлений: " + ++cnt;
}));
General.Delay();
}
}
_db.Close();
Invoke(new Action(() =>
{
startButton.Text = "Начать";
startButton.Enabled = true;
statusLabel.Text = General.STATUS_DEFAULT;
}));
}
private void UploadAdvs()
{
if (!_db.Connect())
{
MessageBox.Show(General.ERROR_DB_CONNECT);
_db.Close();
Invoke(new Action(() =>
{
statusLabel.Text = STATUS_BAR_DEFAULT;
uploadButton.Enabled = true;
exploreButton.Enabled = true;
parseButton.Enabled = true;
exportButton.Enabled = true;
textBox2.Enabled = true;
stopButton.Enabled = false;
fromBox.Enabled = true;
toBox.Enabled = true;
}));
return;
}
var targets = textBox2.Text.Split(';');
var i = 0;
foreach (var target in targets)
{
var paths = Directory.GetDirectories(target);
Invoke(new Action(() =>
{
statusLabel.Text = "Инициализация Selenium...";
}));
IWebDriver browser;
var option = new ChromeOptions();
option.AddArgument("--headless");
option.AddArgument("--user-agent=" + General.DEFAULT_USER_AGENT);
//option.AddArgument("--proxy-server=socks5://109.234.35.41:8888");
var service = ChromeDriverService.CreateDefaultService();
service.HideCommandPromptWindow = true;
browser = new ChromeDriver(service, option);
// Нужно сделать проверку является-ли фарпостовской сессией
var defaultPageAddress = "https://www.farpost.ru/vladivostok/realty/sell_flats/?page=1";
browser.Navigate().GoToUrl(defaultPageAddress);
var def = (int)(toBox.Value - fromBox.Value) + 1;
foreach (var path in paths)
{
if (_stopFlag)
{
break;
}
var pathSplit = path.Split('\\');
var advName = pathSplit[pathSplit.Length - 1];
var nlink = target + "\\" + advName;
if (File.Exists(nlink + ".html") || File.Exists(nlink + "_arch.html"))
{
i++;
continue;
}
var idStr = pathSplit[pathSplit.Length - 1].Split('_')[0];
if (!int.TryParse(idStr, out int id))
{
i++;
continue;
}
if (id < fromBox.Value || id > toBox.Value)
{
i++;
continue;
}
Invoke(new Action(() =>
{
statusLabel.Text = "Загрузка объявления #" + idStr + " [" + i + "/" + def + "]";
}));
var link = _db.GetLinkById(id);
while (true)
{
if (_stopFlag)
{
break;
}
try
{
browser.Navigate().GoToUrl(link);
if (browser.PageSource.Contains("Из вашей подсети наблюдается подозрительная активность"))
{
_db.Close();
Invoke(new Action(() =>
{
statusLabel.Text = STATUS_BAR_DEFAULT;
uploadButton.Enabled = true;
exploreButton.Enabled = true;
parseButton.Enabled = true;
exportButton.Enabled = true;
textBox2.Enabled = true;
stopButton.Enabled = false;
fromBox.Enabled = true;
toBox.Enabled = true;
}));
service.Dispose();
browser.Dispose();
MessageBox.Show("Капча! Завершаем работу.");
return;
}
break;
}
catch (Exception ex)
{
General.WriteLog(ex.Message);
Thread.Sleep(10000);
}
}
if (!browser.PageSource.Contains("Объявление находится в архиве и может быть неактуальным."))
{
for (var k = 0; k < 5; k++)
{
if (_stopFlag)
{
break;
}
try
{
var element = browser.FindElement(By.PartialLinkText("Показать контакты"));
element.Click();
var isf = false;
for (var ti = 0; ti < 20 && !_stopFlag; ti++)
{
try
{
browser.FindElement(By.ClassName("dummy-listener_new-contacts"));
isf = true;
break;
}
catch
{
Thread.Sleep(500);
}
}
if (isf)
{
break;
}
}
catch (Exception ex)
{
General.WriteLog(ex.Message);
Thread.Sleep(1000);
}
}
}
else
{
nlink += "_arch";
}
try
{
var element = browser.FindElement(By.ClassName("expand--button"));
element.Click();
for (var ti = 0; ti < 20 && !_stopFlag; ti++)
{
try
{
browser.FindElement(By.ClassName("mod__active"));
break;
}
catch
{
}
Thread.Sleep(1000);
}
}
catch (Exception)
{
}
var content = browser.PageSource;
var images = new List <string>();
var doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(content);
var nodes = doc.DocumentNode.SelectNodes("//img");
if (nodes != null)
{
var nodesArray = nodes.ToArray();
foreach (var node in nodesArray)
{
images.Add(node.Attributes["src"].Value);
}
}
int img_name = 0;
using (var wc = new WebClient())
{
foreach (var imageSrc in images)
{
if (_stopFlag)
{
break;
}
try
{
var ext = imageSrc;
var idx = ext.LastIndexOf("/");
if (idx == -1)
{
continue;
}
ext = ext.Remove(0, idx + 1);
idx = ext.LastIndexOf(".");
ext = idx != -1 ? ext.Remove(0, idx) : ".jpg";
var pictureName = img_name.ToString() + ext;
content = content.Replace(imageSrc, Path.Combine(advName, pictureName));
wc.DownloadFile(imageSrc, Path.Combine(path, pictureName));
img_name++;
}
catch (Exception ex)
{
General.WriteLog(ex.Message);
}
}
}
nodes = doc.DocumentNode.SelectNodes("//*[href]");
if (nodes != null)
{
var nodesArray = nodes.ToArray();
foreach (var node in nodesArray)
{
var value = node.Attributes["href"].Value;
if (!value.StartsWith("http://") && !value.StartsWith("https://"))
{
var newValue = value;
if (newValue.StartsWith("/"))
{
newValue.Remove(0, 1);
}
content = content.Replace(value, "https://www.farpost.ru/" + newValue);
}
}
}
var htmled = nlink + ".html";
File.WriteAllText(htmled, content, Encoding.UTF8);
FileInfo fileInfo = new FileInfo(htmled);
string newname;
if (fileInfo.Length <= 25600)
{
newname = nlink + "_udalen.html";
File.Move(htmled, newname);
}
_db.ChangeAdvNLink(id, htmled);
General.Delay();
i++;
}
_db.Close();
Invoke(new Action(() =>
{
statusLabel.Text = STATUS_BAR_DEFAULT;
uploadButton.Enabled = true;
exploreButton.Enabled = true;
parseButton.Enabled = true;
exportButton.Enabled = true;
textBox2.Enabled = true;
stopButton.Enabled = false;
fromBox.Enabled = true;
toBox.Enabled = true;
}));
service.Dispose();
browser.Dispose();
}
}
