public FundaResult ParsePage(string data)
{
FundaResult result = new FundaResult();
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(data);
IEnumerable<HtmlNode> houseNodes = doc.DocumentNode.GetNodesForClass("div", "search-result-content-inner");
foreach (HtmlNode house in houseNodes)
{
FundaObject fundaObject = new FundaObject();
var streetHouseNumber = house.GetNodesForClass("h3", "search-result-title").First();
fundaObject.StraatAdres = ParseAddress(streetHouseNumber.InnerText);
fundaObject.PostCodePlaats = ParsePostcodePlaats(streetHouseNumber);
result.FundaObjects.Add(fundaObject);
}
var nextPage = doc.DocumentNode.GetNodesForClass("li", "previous-next-page").ToList();
foreach (HtmlNode node in nextPage)
{
var link = node.Descendants("a").First();
if (link.InnerText.IndexOf("volgende", StringComparison.OrdinalIgnoreCase) > -1)
{
result.HasNextPage = true;
string urlPath = link.Attributes["href"].Value;
result.NextPage = new Uri(_baseUrl, urlPath);
}
}
return result;
}
public IEnumerable <FundaObject> GetAllHouses(Uri startUrl)
{
bool loop = true;
Uri urlToParse = startUrl;
while (loop)
{
string data = _webRepo.GetHtml(urlToParse);
FundaResult result = _parser.ParsePage(data);
foreach (var fundaObject in result.FundaObjects)
{
yield return(fundaObject);
}
loop = result.HasNextPage;
urlToParse = result.NextPage;
}
}
public FundaResult ParsePage(string data)
{
FundaResult result = new FundaResult();
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(data);
IEnumerable <HtmlNode> houseNodes = doc.DocumentNode.GetNodesForClass("div", "search-result-content-inner");
foreach (HtmlNode house in houseNodes)
{
FundaObject fundaObject = new FundaObject();
var streetHouseNumber = house.GetNodesForClass("h3", "search-result-title").First();
fundaObject.StraatAdres = ParseAddress(streetHouseNumber.InnerText);
fundaObject.PostCodePlaats = ParsePostcodePlaats(streetHouseNumber);
result.FundaObjects.Add(fundaObject);
}
var nextPage = doc.DocumentNode.GetNodesForClass("li", "previous-next-page").ToList();
foreach (HtmlNode node in nextPage)
{
var link = node.Descendants("a").First();
if (link.InnerText.IndexOf("volgende", StringComparison.OrdinalIgnoreCase) > -1)
{
result.HasNextPage = true;
string urlPath = link.Attributes["href"].Value;
result.NextPage = new Uri(_baseUrl, urlPath);
}
}
return(result);
}
public FundaResult ParsePage(string data)
{
FundaResult result = new FundaResult();
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(data);
IEnumerable <HtmlNode> houseNodes = doc.DocumentNode.GetNodesForClass("div", "specs");
foreach (HtmlNode house in houseNodes)
{
FundaObject fundaObject = new FundaObject();
var streetHouseNumber = house.Descendants("h3").First().Descendants("a").First();
fundaObject.StraatAdres = ParseAddress(streetHouseNumber.InnerText);
var liItems = house.GetNodesForThatDoesNotHaveClass("li", "object-tagline").ToList();
var surface = house.GetNodesForAttribute("span", "title", "Oppervlakte").FirstOrDefault();
string surfaceValue = surface?.InnerText.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries).FirstOrDefault();
if (!string.IsNullOrWhiteSpace(surfaceValue))
{
fundaObject.Surface = int.Parse(surfaceValue, NumberStyles.AllowThousands, nlCulture);
}
var pricesNode = house.GetNodesForClass("span", "price-wrapper").FirstOrDefault();
if (pricesNode != null)
{
var prices = pricesNode.InnerHtml.Split(new string[] { "<br>" }, StringSplitOptions.RemoveEmptyEntries);
foreach (var s in prices)
{
HtmlDocument node = new HtmlDocument();
node.LoadHtml(s);
var price = node.DocumentNode.GetNodesForClass("span", "price").FirstOrDefault();
var priceExt = node.DocumentNode.GetNodesForClass("abbr", "price-ext").FirstOrDefault();
if (priceExt != null && priceExt.InnerText.Contains("Huurprijs") ||
(price != null && (price.InnerText.Contains("/mnd") || price.InnerText.Contains("/jr"))))
{
if (price != null)
{
fundaObject.Huurprijs = ParsePrice(price.InnerText);
}
if (priceExt != null)
{
fundaObject.HuurprijsSpec = WebUtility.HtmlDecode(priceExt.InnerText);
}
}
else
{
if (price != null)
{
fundaObject.Koopprijs = ParsePrice(price.InnerText);
}
if (priceExt != null)
{
fundaObject.KoopprijsSpec = WebUtility.HtmlDecode(priceExt.InnerText);
}
}
}
}
if (liItems.Count > 1)
{
var address = liItems[0].InnerText;
fundaObject.PostCodePlaats = ParseZipCodeRow(address);
}
result.FundaObjects.Add(fundaObject);
}
var nextPage = doc.DocumentNode.GetNodesForClass("a", "paging next").ToList();
if (nextPage.Any())
{
result.HasNextPage = true;
string urlPath = nextPage[0].Attributes["href"].Value;
result.NextPage = new Uri(_baseUrl, urlPath);
}
return(result);
}
public FundaResult ParsePage(string data)
{
FundaResult result = new FundaResult();
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(data);
IEnumerable<HtmlNode> houseNodes = doc.DocumentNode.GetNodesForClass("div", "specs");
foreach (HtmlNode house in houseNodes)
{
FundaObject fundaObject = new FundaObject();
var streetHouseNumber = house.Descendants("h3").First().Descendants("a").First();
fundaObject.StraatAdres = ParseAddress(streetHouseNumber.InnerText);
var liItems = house.GetNodesForThatDoesNotHaveClass("li", "object-tagline").ToList();
var surface = house.GetNodesForAttribute("span", "title", "Oppervlakte").FirstOrDefault();
string surfaceValue = surface?.InnerText.Split(new[] {' '}, StringSplitOptions.RemoveEmptyEntries).FirstOrDefault();
if (!string.IsNullOrWhiteSpace(surfaceValue))
{
fundaObject.Surface = int.Parse(surfaceValue, NumberStyles.AllowThousands, nlCulture);
}
var pricesNode = house.GetNodesForClass("span", "price-wrapper").FirstOrDefault();
if (pricesNode != null)
{
var prices = pricesNode.InnerHtml.Split(new string[] {"<br>"}, StringSplitOptions.RemoveEmptyEntries);
foreach (var s in prices)
{
HtmlDocument node = new HtmlDocument();
node.LoadHtml(s);
var price = node.DocumentNode.GetNodesForClass("span", "price").FirstOrDefault();
var priceExt = node.DocumentNode.GetNodesForClass("abbr", "price-ext").FirstOrDefault();
if (priceExt != null && priceExt.InnerText.Contains("Huurprijs") ||
(price != null && (price.InnerText.Contains("/mnd") || price.InnerText.Contains("/jr"))))
{
if (price != null)
{
fundaObject.Huurprijs = ParsePrice(price.InnerText);
}
if (priceExt != null)
{
fundaObject.HuurprijsSpec = WebUtility.HtmlDecode(priceExt.InnerText);
}
}
else
{
if (price != null)
{
fundaObject.Koopprijs = ParsePrice(price.InnerText);
}
if (priceExt != null)
{
fundaObject.KoopprijsSpec = WebUtility.HtmlDecode(priceExt.InnerText);
}
}
}
}
if (liItems.Count > 1)
{
var address = liItems[0].InnerText;
fundaObject.PostCodePlaats = ParseZipCodeRow(address);
}
result.FundaObjects.Add(fundaObject);
}
var nextPage = doc.DocumentNode.GetNodesForClass("a", "paging next").ToList();
if (nextPage.Any())
{
result.HasNextPage = true;
string urlPath = nextPage[0].Attributes["href"].Value;
result.NextPage = new Uri(_baseUrl, urlPath);
}
return result;
}