public Collection(string title, string url, Collection parentCollection)
{
this.Title = title;
if (url.Contains("http"))
this.Url = url.Substring(0, url.IndexOf("?"));
else
this.Url = "http://allrecipes.com" + url.Substring(0, url.IndexOf("?"));
ChildCollections = new List<Collection>();
this.ParentCollection = parentCollection;
}
private static List<Collection> getCollections(string url, Collection parentCollection)
{
List<Collection> childCollections = new List<Collection>();
var webGet = new HtmlWeb();
var document = webGet.Load(url);
var containerLists = from node in document.DocumentNode.Descendants()
where node.Name == "ul" &&
node.Attributes["id"] != null &&
node.Attributes["id"].Value == "subNavGroupContainer"
select node;
foreach (HtmlNode ulNode in containerLists)
{
string headerText = "";
foreach (HtmlNode recipeGroup in ulNode.Descendants())
{
if (recipeGroup.Attributes["id"] != null && recipeGroup.Attributes["id"].Value == "subNavHeaderContainer")
{
var title = from node in recipeGroup.Descendants()
where node.Name == "span" &&
node.Attributes["id"] != null &&
node.Attributes["id"].Value == "lblTitle"
select node;
foreach (HtmlNode headerNode in title)
{
headerText = headerNode.InnerText;
}
}
else
{
var links = from node in recipeGroup.Descendants()
where node.Name == "a" &&
node.Attributes["id"] != null &&
node.Attributes["id"].Value == "hlSubNavItem"
select node;
foreach (HtmlNode headerNode in links)
{
Collection col = new Collection(headerNode.InnerText, headerNode.Attributes["href"].Value, parentCollection);
childCollections.Add(col);
}
}
}
}
var resultsCount = from node in document.DocumentNode.Descendants()
where node.Name == "p" &&
node.Attributes["class"] != null &&
node.Attributes["class"].Value.Contains("searchResultsCount")
select node;
long recipeCount = 0;
foreach (HtmlNode countNode in resultsCount)
{
if (countNode.Attributes["class"] != null && countNode.Attributes["class"].Value.Contains("staff-picks"))
continue;
foreach (HtmlNode child in countNode.ChildNodes)
{
if (child.Name == "span")
{
if (child.Attributes["id"] != null && child.Attributes["id"].Value.Contains("Collections"))
continue;
recipeCount = long.Parse(child.InnerText.Replace(",", ""));
}
}
}
int numPages = 1;
if (recipeCount > 15)
{
recipeCount -= 15;
Double extraPages = (Double)recipeCount / 20.0;
numPages += (int)Math.Ceiling(extraPages);
}
for (int pageNum = 1; pageNum <= numPages; pageNum++)
{
getRecipes(url + "?Page=" + pageNum, parentCollection);
//if (recipesByUrl.Count > 10)
// break;
}
/*
foreach (Collection collection in childCollections)
{
if (!collectionsByUrl.ContainsKey(collection.Url))
{
collectionsByUrl.Add(collection.Url, collection);
collection.ChildCollections = getCollections(collection.Url, collection);
}
}
*/
return childCollections;
}
private static void getRecipes(string url, Collection parentCollection)
{
var webGet = new HtmlWeb();
var document = webGet.Load(url);
var itemWrapper = from node in document.DocumentNode.Descendants()
where node.Name == "div" &&
node.Attributes["id"] != null &&
node.Attributes["id"].Value == "divGridItemWrapper"
select node;
foreach (HtmlNode wrapper in itemWrapper)
{
var items = from node in wrapper.Descendants()
where node.Name == "a" &&
node.Attributes["class"] != null &&
node.Attributes["class"].Value == "title"
select node;
foreach (HtmlNode item in items)
{
Recipe recipe;
String recipeUrl = item.Attributes["href"].Value;
if (recipesByUrl.ContainsKey(recipeUrl))
recipe = recipesByUrl[recipeUrl];
else
{
recipe = new Recipe(item.InnerText, recipeUrl);
getRecipeContents(recipe);
recipesByUrl.Add(recipeUrl, recipe);
}
recipe.Collections.Add(parentCollection);
}
}
}
