private void ParseIngredients(ScannedRecipe scannedRecipe, ParserResult result)
{
var ingredients = new List <RecipeIngredient>();
var ingredientOrdinal = 1;
scannedRecipe.IngredientSection.Content.ForEach(c =>
{
try
{
var correctedIngredient = Regex.Replace(c.Trim(), @"^\d+(\)|\.)", string.Empty);
if (!string.IsNullOrEmpty(correctedIngredient.Trim()))
{
var ingredient = IngredientParser.Parse(correctedIngredient);
ingredient.Ordinal = ingredientOrdinal++;
ingredients.Add(ingredient);
}
}
catch (RecipeParseException ex)
{
result.Errors.Add(ParseError.FromException(ex, c));
}
});
result.Output.Ingredients = ingredients;
}
private void ParseDescription(ScannedRecipe scannedRecipe, ParserResult result)
{
var description = new StringBuilder();
scannedRecipe.DescriptionSection.Content.ForEach(c =>
{
if (!string.IsNullOrEmpty(c))
{
description.AppendLine(c); // we're appending line because that's how we found it.
}
});
result.Output.Description = description.ToString();
}
private bool ValidateScannedRecipe(ScannedRecipe scannedRecipe, ParserResult result)
{
if (!scannedRecipe.IngredientSection.Content.Any())
{
result.Errors.Add(new ParseError
{
Character = -1,
Line = -1,
Description = "Could Not find any ingredients",
ErrorCode = ParseErrorCode.NoIngredients,
ErrorType = ErrorType.MissingSection,
UnparsedLine = ""
});
}
if (!scannedRecipe.DescriptionSection.Content.Any())
{
result.Errors.Add(new ParseError
{
Character = -1,
Line = -1,
Description = "Could Not find any description",
ErrorCode = ParseErrorCode.NoDescription,
ErrorType = ErrorType.MissingSection,
UnparsedLine = ""
});
}
if (!scannedRecipe.InstructionSection.Content.Any())
{
result.Errors.Add(new ParseError
{
Character = -1,
Line = -1,
Description = "Could Not find any instructions",
ErrorCode = ParseErrorCode.NoInstructions,
ErrorType = ErrorType.MissingSection,
UnparsedLine = ""
});
}
if (result.Errors.Any(e => e.ErrorType == ErrorType.MissingSection))
{
return(false);
}
return(true);
}
private void ParseInstructions(ScannedRecipe scannedRecipe, ParserResult result)
{
var instructionList = new List <Step>();
var instructionOrdinal = 1;
scannedRecipe.InstructionSection.Content.ForEach(c =>
{
if (!string.IsNullOrEmpty(c))
{
// get rid of Numbers if they exist at start of line.
var correctedStep = Regex.Replace(c.Trim(), @"^\d+(\)|\.)", string.Empty);
instructionList.Add(new Step
{
Ordinal = instructionOrdinal++,
Instructions = correctedStep.Trim()
});
}
});
result.Output.Steps = instructionList;
}
public IEnumerable <ScannedRecipe> Scan()
{
var scanned = new List <ScannedRecipe>();
var web = new HtmlWeb();
if (Encoding != null)
{
web.OverrideEncoding = Encoding;
}
HtmlDocument doc = null;
try
{
doc = web.Load(Url);
}
catch (EncodingNotSupportedException)
{
if (Encoding != Encoding.UTF8)
{
Encoding = Encoding.UTF8;
return(Scan());
}
}
var ingredientsCaptionNodes = ScanIngredientsCaptionNodes(doc);
foreach (var ingredientsCaptionNode in ingredientsCaptionNodes)
{
Cancel?.ThrowIfCancellationRequested();
var ingredientsSectionNode = ScanIngredientsSectionNode(doc, ingredientsCaptionNode);
var recipe = new ScannedRecipe();
scanned.Add(recipe);
var dishCaptionNodes = ScanDishCaptionNodes(doc, ingredientsCaptionNode);
var dishes = new List <ScannedDish>();
var servingsList = new List <int>();
foreach (var dishCaptionNode in dishCaptionNodes)
{
dishes.Add(ScanDish(doc, dishCaptionNode));
servingsList.Add(ScanServings(doc, dishCaptionNode));
}
recipe.Dish = dishes.FirstOrDefault();
if (recipe.Dish == null)
{
recipe.Dish = new ScannedDish
{
Candidates = new List <Dish>(),
}
}
;
recipe.Servings = servingsList.Min();
var ingredientSectionNodes = ScanIngredientSectionNodes(doc, ingredientsSectionNode);
var ingredients = new List <ScannedIngredient>();
foreach (var subSectionNode in ingredientSectionNodes)
{
Cancel?.ThrowIfCancellationRequested();
var section = ScanIngredientSection(doc, subSectionNode);
var ingredientsBySection = ScanIngredientsFromIngredientSection(doc, subSectionNode);
ingredients.AddRange(ingredientsBySection.Select(i =>
{
var d = IngredientItemParser.Parse(i);
d.Section = section;
return(d);
}));
}
recipe.Ingredients = ingredients;
// refine data
var sections = recipe.Ingredients.Select(i => i.Section).Distinct();
var reliableSectionNames = sections.Where(s => s.Candidates.Count() == 1)
.SelectMany(s => s.Candidates.Select(c => c.Name)).ToList();
foreach (var s in sections.Where(s => s.Candidates.Count() > 1))
{
foreach (var sc in s.Candidates)
{
if (reliableSectionNames.Any(n => n == sc.Name))
{
s.Candidates = s.Candidates.Where(c => c != sc);
}
}
}
}
AppendTemporaryIndecies(scanned);
return(scanned);
}
