public override HtmlAgilityPack.HtmlNode SelectElement(HtmlAgilityPack.HtmlNode element)
{
IList <HtmlAgilityPack.HtmlNode> elements = element.QuerySelectorAll(_selectorText);
if (elements != null && elements.Count > 0)
{
return(elements[0]);
}
return(null);
}
protected override ProductPageInfo ParsePage(HtmlNode document)
{
IEnumerable<HtmlNode> metaTags = document.QuerySelectorAll("meta");
//let if fire a NRE if some element is not found
string name = document.FindHiddenField("name").GetAttributeValue("value");
string description = metaTags.First(m => m.GetAttributeValue("property") == "og:description").GetAttributeValue("content");
string imageUrl = metaTags.First(m => m.GetAttributeValue("property") == "og:image").GetAttributeValue("content");
string priceText = document.FindHiddenField("price").GetAttributeValue("value");
return new ProductPageInfo(name, description, Decimal.Parse(priceText, NumberStyles.Currency), imageUrl);
}
public IEnumerable<Game> LoadTopGames(HtmlNode doc)
{
var gamesListItem = doc.QuerySelectorAll("ol.GameTiles li");
foreach (var listItem in gamesListItem)
{
var productBox = listItem.QuerySelector("img.ProductBox");
var title = listItem.QuerySelector("h2 a");
yield return new Game
{
ThumbUrl = productBox.Attributes["src"].Value,
Title = title.InnerText
};
}
}
protected override ProductPageInfo ParsePage(HtmlNode document)
{
var metaTags = document.QuerySelectorAll("meta");
//let if fire NRE if some element is not found
string name = metaTags.First(m => m.GetAttributeValue("name") == "title").GetAttributeValue("content");
string description = metaTags.First(m => m.GetAttributeValue("name") == "description").GetAttributeValue("content");
string imageUrl = metaTags.First(m => m.GetAttributeValue("property") == "og:image").GetAttributeValue("content");
var priceHidden = document.FindHiddenField("ADD_CART_ITEM<>salePriceAmt");
if (priceHidden == null)
{
priceHidden = document.FindHiddenField("ADD_CART_ITEM_ARRAY<>salePriceAmt");
}
string priceText = priceHidden.GetAttributeValue("value");
return new ProductPageInfo(name, description, Decimal.Parse(priceText, NumberStyles.Currency), imageUrl);
}
protected override ProductPageInfo ParsePage(HtmlNode document)
{
IEnumerable<HtmlNode> metaTags = document.QuerySelectorAll("meta");
string name = GetValueSomewhereInThePage(document, new[] { "name", "title" }, metaTags);
string description = GetValueSomewhereInThePage(document, new[] { "description" }, metaTags);
string priceText = GetValueSomewhereInThePage(document, new[] { "price" }, metaTags);
string imageUrl = GetValueSomewhereInThePage(document, new[] { "image" }, metaTags);
if (!String.IsNullOrWhiteSpace(name)
|| !String.IsNullOrWhiteSpace(description)
|| !String.IsNullOrWhiteSpace(priceText)
|| !String.IsNullOrWhiteSpace(imageUrl))
{
decimal price = String.IsNullOrWhiteSpace(priceText) ? 0 : Decimal.Parse(priceText, NumberStyles.Currency);
return new ProductPageInfo(name, description, price, imageUrl);
}
return null;
}
public static HtmlAgilityPack.HtmlNode QuerySelector(this HtmlAgilityPack.HtmlNode node, string cssSelector)
{
return(node.QuerySelectorAll(cssSelector).FirstOrDefault());
}
/// <summary>
/// Query selector collection
/// </summary>
/// <param name="node">node to query</param>
/// <param name="query">query string (css selector)</param>
/// <returns>HtmlNode results</returns>
protected IEnumerable<HtmlNode> QuerySelectorAll(HtmlNode node, string query)
{
return node.QuerySelectorAll(query);
}
public override IList <HtmlAgilityPack.HtmlNode> SelectElements(HtmlAgilityPack.HtmlNode element)
{
return(element.QuerySelectorAll(_selectorText));
}
/// <summary>
/// Create organized topic data using HtmlAgilityPack, with Fizzler query
/// selector extensions.
/// </summary>
/// <param name="document"></param>
/// <param name="initialTopic"></param>
/// <returns></returns>
private static Dictionary<string, string> BuildTopicData(HtmlNode document, string initialTopic)
{
var topicData = new Dictionary<string, string>();
string topic = initialTopic.ToLower();
var textAggregator = new StringBuilder();
// Get the contents of the main article content node
var nodes = document.QuerySelectorAll("#mw-content-text").First().ChildNodes;
foreach (var node in nodes)
{
// Header nodes contain headlines, which form subtopic names
if (node.Name.StartsWith("h"))
{
var headline = node.QuerySelector("span.mw-headline");
if (headline == null)
continue;
// If we have accumulated any text at this point, add it to the last
// subtopic and start a new one.
if (textAggregator.Length > 0)
AddTextData(topicData, topic, textAggregator);
textAggregator.Clear();
topic = headline.InnerText.ToLower();
continue;
}
// Paragraph nodes are regular subtopic content
if (node.Name == "p")
{
// As long as the paragraph contains data, add it to the subtopic.
if (textAggregator.Length > 0)
textAggregator.Append(" ");
textAggregator.Append(node.InnerText);
}
}
AddTextData(topicData, topic, textAggregator);
return topicData;
}
protected IEnumerable<HtmlNode> SelectItems(HtmlNode cq, string css)
{
return cq.QuerySelectorAll(css);
}
private static void RemoveAttributesWithName(HtmlNode document, params string[] attributes)
{
foreach (var attribute in attributes)
{
var elements = document.QuerySelectorAll(string.Format("[{0}]", attribute)).ToList();
foreach (var element in elements)
{
element.Attributes.Remove(attribute);
}
}
}
private static List<ProductDisadvantage> GetProductDisadvantages(HtmlNode reviewNode)
{
return reviewNode.QuerySelectorAll(".cons-cell li").Select(x => new ProductDisadvantage { Feature = WebUtility.HtmlDecode(x.InnerText.Trim()) }).ToList();
}
private RealtimeInfo ParseWebPage(HtmlNode document)
{
var buses = document.QuerySelectorAll(_searchPath);
return ConvertHtmlNodeToBuses(buses);
}