public async Task ScrapeRecipesAsync(IRecipeManager recipeManager)
{
var stopWatch = new Stopwatch();
stopWatch.Start();
var document = new HtmlDocument();
var pages = Config.RecipeCount / Config.PageItemCount;
Debug.WriteLine(pages);
for (var page = 0; page < pages; page++)
{
var httpResponseTask = Config.NextPage(page, document, _client);
var httpResponse = await httpResponseTask;
var html = await httpResponse.Content.ReadAsStringAsync();
document.LoadHtml(html);
var list = document.DocumentNode.SelectNodes(Config.RecipeItemXPath);
if (list != null)
{
var links = list.Select(x => x.Attributes["href"].Value).ToList();
foreach (var link in links)
{
Uri uri = null;
try
{
if (Uri.IsWellFormedUriString(link, UriKind.Absolute))
{
uri = new Uri(link);
}
else
{
uri = new Uri(Config.RootPage + link);
}
var recipe = await ScrapeRecipeAsync(uri.ToString());
await recipeManager.UpdateRecipeMetaAsync(recipe);
}
catch (Exception ex)
{
var trace = ex.StackTrace != null ? ex.StackTrace : string.Empty;
var failingLink = uri != null?uri.ToString() : string.Empty;
await recipeManager.ReportFailedRecipeAsync(failingLink, trace);
}
}
}
else
{
}
}
stopWatch.Stop();
Logger.Information("{0} took {1} min to rescrape.", Config.ProviderName, stopWatch.Elapsed.Minutes);
}