protected override async Task <Recipe> ScrapeInternalAsync(string url)
{
Console.WriteLine($"{nameof(ShefLavanScraper)} {nameof(ScrapeInternalAsync)} started [{nameof(url)}={url}]");
var web = new HtmlWeb();
var htmlDoc = await web.LoadWithRetryAsync(url);
var title = htmlDoc.DocumentNode.SelectSingleNode("//meta[@property='og:title']").GetAttributeValue("content", "");
var image = htmlDoc.DocumentNode.SelectSingleNode("//meta[@property='og:image']").GetAttributeValue("content", "");
var preparationTime = htmlDoc.DocumentNode.SelectSingleNode("//div[@class='short-properties bold']/span[@class='cooking-time']")?.InnerText;
var link = htmlDoc.DocumentNode.SelectSingleNode("//meta[@property='og:url']").GetAttributeValue("content", "");
var numberOfDishes = htmlDoc.DocumentNode.SelectSingleNode("//div[@class='properties']/span[@class='property']/strong")?.InnerText;
var ingredients = htmlDoc.DocumentNode.SelectNodes("//div[@class='ingredients col-lg-3 col-md-4 col-sm-6 ']/ul[@class='ingredients-list']/li");
var recipeToAdd = new Recipe
{
Id = GetIdFromUrl(link),
PreparationTime = preparationTime,
Link = link,
NumberOfDishes = GetNumberOfDishes(numberOfDishes),
Picture = image,
RecipeTitle = title,
IngredientsList = CreateIngredientsList(ingredients),
ValuesToSearch = TokanizationHelper.Tokenaize(title),
NormalaizedIngredients = CreatenormalaizedIngredientsList(ingredients)
};
Console.WriteLine($"{nameof(ShefLavanScraper)} {nameof(ScrapeInternalAsync)} finished.");
return(recipeToAdd);
}
private List <Recipe> SearchByQuery(string query)
{
var tokenaizedSearchValues = TokanizationHelper.Tokenaize(query);
var recipesBeforeSort =
_dbManager.
RecipesCollection.
Find(
recipe =>
tokenaizedSearchValues.
Any(value => recipe.ValuesToSearch.Contains(value))).
ToList();
var recipesWithRank = new List <RecipeWithRank>();
foreach (var recipe in recipesBeforeSort)
{
var recipeWithRank = new RecipeWithRank(recipe, CheckNumberOfMatchValues(recipe, tokenaizedSearchValues));
recipesWithRank.Add(recipeWithRank);
}
var matchToQueryAfterSort = recipesWithRank.OrderByDescending(recipeWithRank => recipeWithRank.Rank);
var recipesAfterSort = new List <Recipe>();
foreach (var recipeWithRank in matchToQueryAfterSort)
{
recipesAfterSort.Add(recipeWithRank.Recipe);
}
return(recipesAfterSort);
}
protected override async Task <Recipe> ScrapeInternalAsync(string url)
{
Console.WriteLine($"{nameof(MakoScraper)} {nameof(ScrapeInternalAsync)} started [{nameof(url)}={url}]");
var web = new HtmlWeb();
var htmlDoc = await web.LoadWithRetryAsync(url);
var title = htmlDoc.DocumentNode.SelectSingleNode("//meta[@property='og:title']").GetAttributeValue("content", "");
var image = htmlDoc.DocumentNode.SelectSingleNode("//meta[@property='og:image']").GetAttributeValue("content", "");
var prepTime = htmlDoc.DocumentNode.SelectSingleNode("//ul[@class='table_container']/li[@class='titleContainer']/div/span[@itemprop='totalTime']")?.InnerText;
var link = htmlDoc.DocumentNode.SelectSingleNode("//meta[@property='og:url']").GetAttributeValue("content", "");
var numOfDishes = htmlDoc.DocumentNode.SelectNodes("//div[@class='ingredients']/h3[@class='IngredientsTitle fontSize']")?.First().InnerText;
var ingredients = htmlDoc.DocumentNode.SelectNodes("//div[@class='ingredients']/ul[@class='recipeIngredients']/li/span");
var recipeToAdd = new Recipe
{
Id = GetIdFromUrl(link),
PreparationTime = prepTime,
Link = link,
NumberOfDishes = GetNumberOfDishes(numOfDishes),
Picture = image,
RecipeTitle = title,
IngredientsList = CreateIngredientsList(ingredients),
ValuesToSearch = TokanizationHelper.Tokenaize(title),
NormalaizedIngredients = CreatenormalaizedIngredientsList(ingredients)
};
Console.WriteLine($"{nameof(MakoScraper)} {nameof(ScrapeInternalAsync)} finished.");
return(recipeToAdd);
}
public List <string> CreatenormalaizedIngredientsList(HtmlNodeCollection ingredients)
{
List <string> normalaizedIngredients = new List <string>();
for (int currIngredientsIndex = 0; currIngredientsIndex < ingredients.Count; currIngredientsIndex++)
{
string ingredient = ingredients[currIngredientsIndex].InnerText;
List <string> ingredientValues = TokanizationHelper.Tokenaize(ingredient);
for (int ingredientValueIndex = 0; ingredientValueIndex < ingredientValues.Count; ingredientValueIndex++)
{
normalaizedIngredients.Add(ingredientValues[ingredientValueIndex]);
}
}
return(normalaizedIngredients);
}
private List <Recipe> SearchByIngredients(List <string> ingredientNames)
{
var tokenaizedSearchValues = new List <string>();
foreach (var ingredientName in ingredientNames)
{
tokenaizedSearchValues.Add(TokanizationHelper.TokenaizeForOneValue(ingredientName));
}
string queryToSearchByIngredients = null;
foreach (var ingredient in tokenaizedSearchValues)
{
queryToSearchByIngredients += ingredient;
queryToSearchByIngredients += " ";
}
return(SearchByQuery(queryToSearchByIngredients));
}