private IList <Pages.PageComponents.ShopMenu> prepareMenu()
{
IList <Pages.PageComponents.ShopMenu> menuDTOArray = new List <Pages.PageComponents.ShopMenu>();
IElement menuList = document.GetElementsByClassName("cs-nav")[0];
int menuItemsCount = menuList.ChildElementCount;
for (int i = 1; i < menuItemsCount; i++)
{
IElement menuElement = (IElement)menuList.ChildNodes[i];
IElement aHrefelem = (IElement)menuElement.ChildNodes[0];
HTMLAnchorDetail ancdDetail = extractor.extractAchrorInfo(aHrefelem);
menuDTOArray.Add(new Pages.PageComponents.ShopMenu {
Href = invokedPage.ShopUrl + ancdDetail.Href, ItemName = ancdDetail.Name
});
}
menuDTOArray.Add(
new Pages.PageComponents.ShopMenu {
ItemName = "Auto", Href = "http://www.wartonlogo.com/Projectors/Gobos%20Auto-switch/"
}
);
return(menuDTOArray);
}
private static int GetNumberOfDisplayedResults(IHtmlDocument pageContent)
{
var regionResults = pageContent.GetElementsByClassName("directory-region-search-result-item");
var serviceResults = pageContent.GetElementsByClassName("directory-service-search-result-item");
var hospitalResults = pageContent.GetElementsByClassName("directory-hospital-search-result-item");
var userResults = pageContent.GetElementsByClassName("directory-user-search-result-item");
return(regionResults.Length + serviceResults.Length + hospitalResults.Length + userResults.Length);
}
GetDisplayedResultsTextContent(IHtmlDocument pageContent)
{
var regionResults = pageContent.GetElementsByClassName("directory-region-search-result-item")
.Select(e => e.TextContent);
var serviceResults = pageContent.GetElementsByClassName("directory-service-search-result-item")
.Select(e => e.TextContent);
var hospitalResults = pageContent.GetElementsByClassName("directory-hospital-search-result-item")
.Select(e => e.TextContent);
var userResults = pageContent.GetElementsByClassName("directory-user-search-result-item")
.Select(e => e.TextContent);
return(regionResults, serviceResults, hospitalResults, userResults);
}
private IHtmlCollection <IElement> GetOpcoes(IHtmlDocument document)
{
return(document.GetElementsByClassName("" +
"collection-item waves-effect waves-on-white-bg grey-text " +
"text-darken-4 d-flex flex-wrap flex-md-nowrap justify-between " +
"align-items-center p-relative"));
}
public string ScrapeMostRecentBooking(IHtmlDocument htmlDoc)
{
var lastBookingElement = htmlDoc.GetElementsByClassName(ScrapingConstants.LastBookingClass)
.FirstOrDefault();
return(lastBookingElement?.GetElementText() ?? string.Empty);
}
private IHtmlCollection <IElement> GetPageWithProduct(string url, out IHtmlDocument htmlDocument)
{
foreach (var proxy in proxyList)
{
try
{
WebRequest WR = WebRequest.Create(url);
WR.Method = "GET";
string[] fulladress = proxy.Split(":");
var(adress, port) = (fulladress[0], int.Parse(fulladress[1]));
WebProxy myproxy = new WebProxy(adress, port);
myproxy.BypassProxyOnLocal = false;
WR.Proxy = myproxy;
WebResponse response = WR.GetResponse();
string html;
using (Stream stream = response.GetResponseStream())
{
using (StreamReader reader = new StreamReader(stream))
{
html = reader.ReadToEnd();
}
}
HtmlParser parser = new HtmlParser();
htmlDocument = parser.ParseDocument(html);
return(htmlDocument.GetElementsByClassName("g-i-tile-catalog"));
}
catch (Exception ex)
{
}
}
throw new Exception("Proxy isn`t working");
}
public HabraModel Parse(IHtmlDocument document)
{
var post = document.GetElementsByClassName("post__wrapper").FirstOrDefault();
if (post == null)
{
return(null);
}
var author = post.GetElementsByClassName("user-info__nickname user-info__nickname_small").FirstOrDefault()?.TextContent?.Trim();
var dateStr = post.GetElementsByClassName("post__time").FirstOrDefault()?.GetAttribute("data-time_published")?.Trim();
var date = DateTime.MinValue;
if (!string.IsNullOrEmpty(dateStr))
{
DateTime.TryParse(dateStr, out date);
}
var title = post.GetElementsByClassName("post__title-text").FirstOrDefault()?.TextContent?.Trim();
var content = post.QuerySelectorAll("div").Where(m => m.LocalName == "div" &&
m.HasAttribute("id") &&
m.GetAttribute("id").StartsWith("post-content-body")).FirstOrDefault()?.TextContent?.Trim();
HabraModel model = new HabraModel()
{
Author = author,
Content = content,
PostDate = date,
Title = title
};
return(model);
}
private static string ParseSiteContent_Cinemagia(IHtmlDocument document)
{
var expandableSinopsis = document.GetElementsByClassName("expand_sinopsis").Count() > 0;
var synopsisWrapper = expandableSinopsis ? "body_sinopsis" : "short_body_sinopsis";
var synopsisParagraph = document.QuerySelectorAll("p").FirstOrDefault(x => x.ParentElement.Id == synopsisWrapper);
if (synopsisParagraph == null)
{
synopsisParagraph = document.GetElementById(synopsisWrapper);
if (synopsisParagraph == null)
{
throw new Exception("Element not found on page!");
}
}
var splitString = synopsisParagraph.InnerHtml.Split(new string[] { "<br>" }, StringSplitOptions.None);
var processedList = new List <string>();
foreach (var parag in splitString)
{
processedList.Add(parag.StripHtml().Trim());
}
return(string.Join(Environment.NewLine, processedList));
}
public HotelReview ScrapeHotelReview(IHtmlDocument htmlDoc)
{
var galleryReviewElement = htmlDoc.GetElementsByClassName(ScrapingConstants.GalleryReviewClass)
.FirstOrDefault();
if (galleryReviewElement == null)
{
return(null);
}
var scorewordElement = galleryReviewElement.GetElementsByClassName(ScrapingConstants.ReviewScorewordClass)
.FirstOrDefault();
var scoreValElement = galleryReviewElement.GetElementsByClassName(ScrapingConstants.ReviewScoreValClass)
.FirstOrDefault();
var outOfElement = galleryReviewElement.GetElementsByClassName(ScrapingConstants.ReviewBestScoreClass)
.FirstOrDefault();
var reviewCountElement = galleryReviewElement.GetElementsByClassName(ScrapingConstants.ReviewCountClass)
.FirstOrDefault();
var scoreword = scorewordElement?.GetElementText();
var scoreValParseResult = double.TryParse(scoreValElement?.GetElementText(), out var scoreVal);
var outOfParseResult = int.TryParse(outOfElement?.GetElementText(), out var outOf);
var reviewCountParseResult = int.TryParse(reviewCountElement?.GetElementText(), out var reviewCount);
return(new HotelReview {
Scoreword = scoreword,
Score = scoreValParseResult ? scoreVal : null,
ScoreOutOf = outOfParseResult ? outOf : null,
ReviewCount = reviewCountParseResult ? reviewCount : null
});
}
public int?ScrapeHotelRatingStars(IHtmlDocument htmlDoc)
{
var ratingStarsContainer = htmlDoc.GetElementsByClassName(ScrapingConstants.HotelRatingContainerClass)
.FirstOrDefault();
var starClass = ratingStarsContainer?.GetClassForChildWhereClassStartsWith(ScrapingConstants.StarRatingClassWildcard);
return(starClass?.GetFirstInteger());
}
private List <string> scrapeDoc(IHtmlDocument doc)
{
var rawResults = doc.GetElementsByClassName("g");
var links = rawResults.Select(z => z.GetElementsByTagName("a").First().GetAttribute("href"));
Console.WriteLine(links.Count());
return(new List <string>());
}
private static int GetTotalActualResults(IHtmlDocument firstPageContent)
{
return(Convert.ToInt32(
firstPageContent.GetElementsByClassName("case-manager-results-summary")
.Single()
.TextContent
.Split(" ", StringSplitOptions.RemoveEmptyEntries)[1]
));
}
public IEnumerable <string> ChapterUrls(IHtmlDocument doc)
{
if (doc is null)
{
return(Array.Empty <string>());
}
return(doc.GetElementsByClassName("chapterLink")
.Select(a => $"https://www.mangaeden.com{a.GetAttribute("href")}")
.ToList());
}
//Метод парсер HTML
public void Parser(string document, ref int countProduct)
{
string price = null;
string nameProduct = null;
string description = null;
string id = null;
List <string> listPhoto = null;
string sourse = null;
List <ModelDatePrice> modelDatePrices = null;
Product product = null;
HtmlParser htmlParser = new HtmlParser();
IHtmlDocument htmlDocument = htmlParser.Parse(document);
try
{
var elements = htmlDocument.GetElementsByClassName("plate-box").ToList();
for (int item = 0; item < elements.Count; item++)
{
if (countProduct >= 135)
{
return;
}
id = elements[item].GetElementsByClassName("id")
.ToList()[0].QuerySelector("span").TextContent;
if (ManagerShope.listProduct.Exists(e => e.id == id))
{
continue;
}
modelDatePrices = new List <ModelDatePrice>();
listPhoto = new List <string>();
IHtmlDocument document1 = null;
var el = elements[item].QuerySelectorAll("div")
.Where(elem => elem.ClassName == "title")
.ToList()[0].QuerySelector("a");
string url = el.OuterHtml.Remove(0, el.OuterHtml.IndexOf('\"') + 2);
url = urlShope + "/" + url.Remove(url.IndexOf('>') - 1);
sourse = ConnectorShope.GetContentSimplePage(url).GetAwaiter().GetResult();
document1 = htmlParser.Parse(sourse);
nameProduct = GetNameProductProduct(el);
price = GetPriceProduct(elements[item]);
description = GetNameDescriptionProduct(elements[item]);
listPhoto = GetNameListPhotoProduct(document1);
product = new Product(price, description, listPhoto, nameProduct, id);
product.dataTime = GetDatePrices(price);
ManagerShope.listProduct.Add(product);
countProduct++;
}
}
catch (Exception e)
{
throw new Exception(e.Message);
}
}
public JsonResult AllMusicRatings(string id)
{
string rymURL = "https://www.allmusic.com/album/";
string url = rymURL + id;
/* base URL */
HttpResponseMessage response = client.GetAsync(url).Result;
if (response.StatusCode == HttpStatusCode.OK) /* if return status is 200 */
{
string responseContent = response.Content.ReadAsStringAsync().Result;
/* Refer: https://stackoverflow.com/questions/7824138/how-to-grab-elements-by-class-or-id-in-html-source-in-c */
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(responseContent);
HttpResponseMessage request = client.GetAsync(url).Result;
Stream responses = request.Content.ReadAsStreamAsync().Result;
HtmlParser parser = new HtmlParser();
IHtmlDocument document = parser.ParseDocument(responses);
AngleSharp.Dom.IElement allMusicRateElement;
try {
allMusicRateElement = document.GetElementsByClassName("allmusic-rating")[0];
} catch (Exception e)
{
return(Json(new { success = false,
result = e.Message }));
}
string siteRateString = allMusicRateElement.TextContent.Trim();
/* check if the number is valid */
if (!float.TryParse(siteRateString, out float siteRate))
{
return(Json(new
{
success = false,
result = "invalid number" + siteRateString
}));
}
return(Json(new
{
success = true,
site_rating = siteRate,
max_rating = 10
}));
}
else
{
return(FailRetuenJson());
}
}
internal async Task <RankData> ScrapeProfileData(string siteUrl)
{
try
{
var response = await GetDataFromUrl(siteUrl);
HtmlParser parser = new HtmlParser();
IHtmlDocument document = parser.Parse(response);
int rank = -1;
int asp = -1;
var pic = document.GetElementById("character-portrait-image")?.Attributes?.FirstOrDefault(a => a.Name == "src")?.Value;
var job = AddSpacesAfterCapitals(document.GetElementsByClassName("allstar-header-icon")?.FirstOrDefault()?.ClassList?.FirstOrDefault(s => s.Contains("actor-sprite-"))?.Substring(13));
int.TryParse(document.GetElementsByClassName("header-zone-positions")?.FirstOrDefault()?.GetElementsByClassName("header-rank")?.FirstOrDefault()?.TextContent, out rank);
int.TryParse(document.GetElementsByClassName("header-zone-points")?.FirstOrDefault()?.GetElementsByClassName("header-rank")?.FirstOrDefault()?.TextContent, out asp);
var rankData = new RankData(pic, job, rank, asp);
return(rankData);
}
catch (OperationCanceledException e)
{
//Network related error
return(null);
}
}
private IReadOnlyList <(string Href, string TestId)> GetSubNavLinks(IHtmlDocument doc)
{
var results = new List <(string Href, string TestId)>();
foreach (var item in doc.GetElementsByClassName("pttcd-subnav__item"))
{
var anchor = item.GetElementsByTagName("a")[0];
var href = anchor.GetAttribute("href");
var testId = anchor.GetAttribute("data-testid");
results.Add((href, testId));
}
return(results);
}
public string ScrapeHotelSummary(IHtmlDocument htmlDoc)
{
var hotelSummaryContainer = htmlDoc.GetElementsByClassName(ScrapingConstants.HotelSummaryContainerClass)
.FirstOrDefault();
if (hotelSummaryContainer == null)
{
return(string.Empty);
}
var summaryTextParagraphs = hotelSummaryContainer.GetElementsByTagName("p");
var summary = summaryTextParagraphs.Aggregate(string.Empty,
(curr, next) => curr += next.GetElementText());
return(summary);
}
private static CaseBlobs GetCasesFromHtmlDoc(IHtmlDocument document)
{
CaseBlobs c = new CaseBlobs();
var stats = document.GetElementsByClassName("maincounter-number");
c.DateScraped = DateTime.Now;
c.TotalCasesBlob = stats[0].InnerHtml;
c.TotalDeathsBlob = stats[1].InnerHtml;
c.TotalRecoveriesBlob = stats[2].InnerHtml;
c.Processed = false;
return(c);
}
private void CheckPageError(IHtmlDocument dom)
{
var msg = "";
var mainContent = dom.GetElementsByClassName("feedbackPanelERROR");
if (mainContent.Any())
{
msg = mainContent.First().FirstElementChild.TextContent;
if (msg.Contains("查無訂位紀錄"))
{
throw new CritialPageErrorException(msg);
}
throw new ArgumentException(msg);
}
}
private void SetHeadInform(string sourse, ref Shipping shipping)
{
try
{
IHtmlDocument htmlDocument = htmlParser.Parse(sourse);
var element = htmlDocument.GetElementsByClassName("col-xs-12 col-sm-7 col-md-8")[0]
.GetElementsByTagName("p");
shipping.Id = element[0].TextContent.Remove(0, element[0].TextContent.IndexOf(": ") + 2);
shipping.idOrder = element[0].TextContent.Remove(0, element[0].TextContent.IndexOf(": ") + 2);
shipping.CurrentStatus = "NewLoad"; //element[1].TextContent.Remove(0, element[1].TextContent.IndexOf(": ") + 2);
shipping.LastUpdated = element[2].TextContent.Remove(0, element[2].TextContent.IndexOf(": ") + 2);
shipping.CDReference = element[3].TextContent.Remove(0, element[3].TextContent.IndexOf(": ") + 2);
}
catch (Exception)
{
}
}
private static string ParseSiteContent_Imdb(IHtmlDocument document)
{
var summaryDiv = document.GetElementsByClassName("summary_text").FirstOrDefault();
if (summaryDiv == null)
{
throw new Exception("Element not found on page!");
}
return
(summaryDiv.InnerHtml
.Replace("<br>", Environment.NewLine)
.Replace("<", "<")
.Replace(">", ">")
.StripHtml()
.Trim());
}
private static string ParseSiteContent_Filmvandaag(IHtmlDocument document)
{
var descriptionDiv = document.GetElementsByClassName("synopsis").FirstOrDefault();
if (descriptionDiv == null)
{
throw new Exception("Element not found on page!");
}
return
(descriptionDiv.InnerHtml
.Replace("<br>", Environment.NewLine)
.Replace("<", "<")
.Replace(">", ">")
.StripHtml()
.Trim());
}
public VisualCronModel Parse(IHtmlDocument document)
{
var post = document.GetElementsByClassName("post").FirstOrDefault();
if (post == null)
{
return(null);
}
// get title
var title = post.GetElementsByClassName("post-title").FirstOrDefault()?.TextContent?.Trim();
// get post date
var dateStr = post.GetElementsByClassName("post-date").FirstOrDefault()?.TextContent?.Trim();
var date = DateTime.MinValue;
if (!string.IsNullOrEmpty(dateStr))
{
DateTime.TryParse(dateStr, out date);
}
// get post author
var author = post.GetElementsByClassName("post-author").FirstOrDefault()?.TextContent?.Trim();
// get post category
var category = post.GetElementsByClassName("post-category").FirstOrDefault()?.TextContent?.Trim();
// get post content
var content = post.GetElementsByClassName("post-body text").FirstOrDefault()?.TextContent?.Trim();
// get post tags
var tags = post.GetElementsByClassName("post-tags").FirstOrDefault()?.QuerySelectorAll("a").Select(t => t.TextContent?.Trim()).ToArray();
VisualCronModel model = new VisualCronModel()
{
Author = author,
Content = content,
PostDate = date,
Title = title,
Category = category,
Tags = tags
};
return(model);
}
private string GetPage(IHtmlDocument parseElement)
{
var pagination = parseElement.GetElementsByClassName("pagination-number-list");
if (pagination.Length > 0)
{
var pages = pagination[0].GetElementsByTagName("li");
for (int i = 0; i < pages.Length; i++)
{
if (pages[i].ClassList.Contains("active"))
{
return(pages[++i].GetElementsByTagName("a")[0].GetElementsByTagName("span")[0].TextContent);
}
}
}
return("");
}
private void SetHeadInform(string sourse, ref Shipping shipping)
{
try
{
IHtmlDocument htmlDocument = htmlParser.Parse(sourse);
var element = htmlDocument.GetElementsByClassName("col-xs-12 col-sm-7 col-md-8")[0]
.GetElementsByTagName("p");
shipping.Id = element[0].TextContent.Remove(0, element[0].TextContent.IndexOf(": ") + 2);
shipping.idOrder = element[0].TextContent.Remove(0, element[0].TextContent.IndexOf(": ") + 2);
shipping.CurrentStatus = "NewLoad"; //element[1].TextContent.Remove(0, element[1].TextContent.IndexOf(": ") + 2);
shipping.LastUpdated = element[2].TextContent.Remove(0, element[2].TextContent.IndexOf(": ") + 2);
shipping.CDReference = element[3].TextContent.Remove(0, element[3].TextContent.IndexOf(": ") + 2);
}
catch (Exception)
{
LogEr.Logerr("Error", $"some data is not added, Load id {shipping.Id}, Url: {shipping.UrlReqvest}", "SetHeadInform", DateTime.Now.ToShortTimeString());
}
}
public JsonResult ScrapeAlbumChart()
{
string topAlbumsUrl = "https://www.billboard.com/charts/current-albums";
Billboard_Album[] topAlbums = new Billboard_Album[100];
HttpResponseMessage request = client.GetAsync(topAlbumsUrl).Result;
Stream response = request.Content.ReadAsStreamAsync().Result;
HtmlParser parser = new HtmlParser();
IHtmlDocument document = parser.ParseDocument(response);
AngleSharp.Dom.IHtmlCollection <AngleSharp.Dom.IElement> albums = document.GetElementsByClassName("chart-list-item__first-row chart-list-item__cursor-pointer");
for (int i = 0; i < albums.Length; i++)
{
string title = albums[i].GetElementsByClassName("chart-list-item__title-text")[0].TextContent.Trim();
string artist = "";
// Some albums have a link tag, some don't.
if (albums[i].GetElementsByClassName("chart-list-item__artist")[0].ChildElementCount > 0) //sometimes there is an <a> tag
{
artist = albums[i].GetElementsByClassName("chart-list-item__artist")[0].FirstElementChild.TextContent.Trim();
}
else
{
artist = albums[i].GetElementsByClassName("chart-list-item__artist")[0].TextContent.Trim();
}
var ab = new Billboard_Album
{
Title = title,
Artist = artist
};
topAlbums[i] = ab;
}
return(Json(new
{
success = true,
albums = topAlbums
}));
}
public bool CheckIsNextPage(string sourse)
{
bool isPageNext = false;
IHtmlDocument htmlDocument = htmlParser.Parse(sourse);
var elements = htmlDocument.GetElementsByClassName("col-xs-6 text-center");
if (elements != null)
{
string elementCountPageStr = elements[0].InnerHtml;
string countOrderStr = elementCountPageStr.Remove(0, elementCountPageStr.IndexOf("-") + 1);
countOrderStr = countOrderStr.Remove(countOrderStr.IndexOf(" "));
string fullCountOrderStr = elementCountPageStr.Remove(0, elementCountPageStr.IndexOf("of ") + 3);
fullCountOrderStr = fullCountOrderStr.Remove(2);
if (fullCountOrderStr != countOrderStr)
{
isPageNext = true;
}
}
return(isPageNext);
}
public string Scrap(string html)
{
HtmlParser parser = new HtmlParser();
IHtmlDocument doc = parser.Parse(html);
IHtmlCollection <IElement> docByPriceClasses = doc.GetElementsByClassName("price");
if (docByPriceClasses.Count() == 0)
{
return("");
}
String text = docByPriceClasses.First()
.GetElementsByTagName("span")
.First()
.TextContent;
return(String.Join("",
text.Where(c => char.IsNumber(c))
.Select(c => c.ToString())
));
}
private int AddMeal(IHtmlDocument doc)
{
var title = doc.GetElementsByClassName("recipeTitle").Select(item =>
{
return(item.TextContent.Trim());
}).First();
var ExistMeal = context.Meal.SingleOrDefault(m => m.Url == url);
if (ExistMeal != null)
{
return(ExistMeal.Id);
}
Meal meal = new Meal {
Name = title, Url = url
};
context.Add(meal);
context.SaveChanges();
return(context.Meal.Single(m => m.Url == url).Id);
}