// ===========================================================================
// = Private Methods
// ===========================================================================
private static IEnumerable <Beer> LoadFromWeb()
{
var data = ScrapeHelper.FetchParseAsync("http://www.fynefest.com/?page_id=5", transform: TransformHtml).Result;
var next = data.QuerySelector("h1");
string breweryName = "";
while (next != null)
{
if (!next.Name.Equals("p"))
{
next = next.NextSibling;
continue;
}
if (next.Elements().Any(X => X.Name.Equals("strong", StringComparison.OrdinalIgnoreCase)))
{
breweryName = WebUtility.HtmlDecode(next.Element("strong").InnerText).Trim();
if (breweryName.Contains("–"))
{
breweryName = breweryName.Split('–').First().Trim();
}
next = next.NextSibling;
continue;
}
var record = WebUtility.HtmlDecode(next.InnerText);
var lines = record.Split(new[] { '\n' });
var regex = new Regex(@"^(?<beerName>[^–]+?)\s*–?\s*(?<abv>[0-9]+\.?[0-9]*)%?\s*\((?<caskOrKeg>[KC])\)\s*$");
var match = regex.Match(lines.First());
if (!match.Success)
{
throw new ApplicationException("Invalid beer: " + lines.First());
}
var beerName = match.Groups["beerName"].Value.Trim();
var abv = Decimal.Parse(match.Groups["abv"].Value.Trim());
var isCask = match.Groups["caskOrKeg"].Value.Trim() == "C";
var description = lines.Last().Trim();
var beer = new Beer
{
BreweryName = breweryName,
BeerName = beerName,
ABV = abv,
Id = $"{breweryName}$$${beerName}$$${abv}$$${isCask}",
StyleName = description
};
next = next.NextSibling;
yield return(beer);
}
}
public async Task <IList <Beer> > Fetch()
{
var beers = new List <Beer>();
Console.WriteLine("Scraping Cloudwater FFB...");
Console.WriteLine();
var client = new HttpClient();
var breweriesPage = await ScrapeHelper.FetchParseAsync("https://www.friendsandfamily.beer/family");
var breweryNodes = breweriesPage.QuerySelectorAll("h2 > a");
var breweryCount = breweryNodes.Count();
Console.WriteLine($"Found {breweryCount} breweries...");
Console.WriteLine();
foreach (var breweryNode in breweryNodes)
{
var href = breweryNode.Attributes["href"]?.Value;
if (string.IsNullOrWhiteSpace(href))
{
continue;
}
var breweryName = CultureInfo.CurrentCulture.TextInfo.ToTitleCase(
breweryNode.InnerText.Trim()?.ToLower()
);
Console.WriteLine(breweryName);
Console.WriteLine("------------------------------");
var beersPage = await ScrapeHelper.FetchParseAsync(href);
var beerNodes = beersPage.QuerySelectorAll("ul > li > p");
if (beerNodes.Count() == 0)
{
Console.WriteLine("No beers found (yet)");
}
else
{
foreach (var beerNode in beerNodes)
{
string beerName = null;
string description = null;
// Style: <strong>beer name</strong> description <strong>(v)</strong>
if (beerNode.ChildNodes[0].Name?.ToLower() == "strong")
{
beerName = beerNode.ChildNodes[0]?.InnerText?.Trim();
description = beerNode.ChildNodes[1]?.InnerText?.Trim();
}
// Style: beer name, description <strong>(v)</strong>
else
{
var parts = beerNode.ChildNodes[0]?.InnerText?.Trim()?.Split(',');
beerName = parts[0]?.Trim();
description = string.Join(',', parts.Skip(1));
}
description = description?.TrimStart(' ', ',');
Console.WriteLine($"{beerName} ----- {description}");
beers.Add(new Beer
{
BreweryName = breweryName != null ? HtmlEntity.DeEntitize(breweryName) : null,
BeerName = beerName != null ? HtmlEntity.DeEntitize(beerName) : null,
Description = description != null ? HtmlEntity.DeEntitize(description) : null
});
}
}
Console.WriteLine();
await Task.Delay(1000);
}
Console.WriteLine($"Found {beers.Count} beers.");
return(beers);
}