private static void AddTags(HtmlDocument document, ScrappedCardModel scrappedCard)
{
var tags = document.DocumentNode.SelectSingleNode("//*[@id='articleCategories']/div[1]/ul");
var filteredTags = tags.ChildNodes.Where(x => !string.IsNullOrWhiteSpace(x.InnerText));
foreach (var tag in filteredTags)
{
var trimmedText = tag.InnerText.Trim();
if (trimmedText == "Cards" || trimmedText == "Add category")
{
continue;
}
scrappedCard.Tags.Add(trimmedText);
}
}
private static void SetExpansion(ScrappedCardModel scrappedCard, HtmlNodeCollection table)
{
var row = table.FirstOrDefault(x => x.ChildNodes[1].InnerText.Contains("Expansion"));
if (row != null) scrappedCard.Expansion = row.ChildNodes[2].InnerText.Trim();
}
private static void SetConvertedManaCost(ScrappedCardModel scrappedCard, HtmlNodeCollection table)
{
var row = table.FirstOrDefault(x => x.ChildNodes[1].InnerText.Contains("Converted Mana Cost"));
if (row != null) scrappedCard.ConvertedManaCost = int.Parse(row.ChildNodes[2].InnerText.Trim());
}
private void SetImageData(HtmlDocument document, ScrappedCardModel scrappedCard)
{
var image = document.DocumentNode.SelectSingleNode("//*[@id='mw-content-text']/p/img") ??
document.DocumentNode.SelectSingleNode("//*[@id='mw-content-text']/img");
scrappedCard.ImageUrl = image.GetAttributeValue("data-src", null) ?? image.GetAttributeValue("src", string.Empty);
scrappedCard.ImageName = image.GetAttributeValue("data-image-key", string.Empty);
}
private ScrappedCardModel ScrapCard(string page)
{
HtmlDocument document = new HtmlDocument();
document.LoadHtml(page);
var scrappedCard = new ScrappedCardModel();
var table = document.DocumentNode.SelectNodes("//*[@id='mw-content-text']/table/tr");
ExtractDataFromTable(scrappedCard, table);
SetImageData(document, scrappedCard);
AddTags(document, scrappedCard);
return scrappedCard;
}
private void ExtractDataFromTable(ScrappedCardModel scrappedCard, HtmlNodeCollection table)
{
SetName(scrappedCard, table);
SetManaCost(scrappedCard, table);
SetConvertedManaCost(scrappedCard, table);
SetTypes(scrappedCard, table);
SetCardText(scrappedCard, table);
SetFlavorText(scrappedCard, table);
SetPowerAndThoughness(scrappedCard, table);
SetExpansion(scrappedCard, table);
SetRarity(scrappedCard, table);
SetLoyalty(scrappedCard, table);
}
private static void SetTypes(ScrappedCardModel scrappedCard, HtmlNodeCollection table)
{
var row = table.FirstOrDefault(x => x.ChildNodes[1].InnerText.Contains("Types"));
if (row != null) scrappedCard.Types = row.ChildNodes[2].InnerText.Trim().Replace(" —", "").Split(' ').ToList();
}
private static void SetPowerAndThoughness(ScrappedCardModel scrappedCard, HtmlNodeCollection table)
{
var row = table.FirstOrDefault(x => x.ChildNodes[1].InnerText.Contains("P/T"));
if (row != null)
{
var pt = row.ChildNodes[2].InnerText.Trim().Trim('(', ')').Split('/');
scrappedCard.Power = pt[0];
scrappedCard.Thoughness = pt[1];
}
}
private static void SetManaCost(ScrappedCardModel scrappedCard, HtmlNodeCollection table)
{
var row = table.FirstOrDefault(x => x.ChildNodes[1].InnerText.Contains("Mana Cost"));
if (row != null)
{
foreach (var childNode in row.ChildNodes[2].ChildNodes)
{
if (childNode.Name == "#text")
continue;
var alt = childNode.GetAttributeValue("alt", "");
if (alt.Contains("CMC"))
{
if (alt.Length > 4)
{
scrappedCard.ColorlessMana = alt[3].ToString() + alt[4];
}
else
{
scrappedCard.ColorlessMana = alt[3].ToString();
}
}
else if (alt.Contains("Color R"))
{
scrappedCard.RedMana++;
}
else if (alt.Contains("Color U"))
{
scrappedCard.BlueMana++;
}
else if (alt.Contains("Color G"))
{
scrappedCard.GreenMana++;
}
else if (alt.Contains("Color W"))
{
scrappedCard.WhiteMana++;
}
else if (alt.Contains("Color B"))
{
scrappedCard.BlackMana++;
}
}
}
}
