static IEnumerable <HtmlDataSet> ParseDocument(IDocument document, DataSetGeneratorSettings dataSetGeneratorSettings, int priceElementRepeatCount)
{
var htmlElements = document.QuerySelectorAll("*").Where(el => el.ChildElementCount == 0 && !String.IsNullOrEmpty(el.OuterHtml));
htmlElements = htmlElements.OfTypes(new Type[]
{
typeof(IHtmlSpanElement),
typeof(IHtmlDivElement),
typeof(IHtmlMetaElement),
typeof(IHtmlListItemElement)
});
var list = new List <HtmlDataSet>();
foreach (var element in htmlElements)
{
bool isContainsPrice = false;
foreach (var priceTag in dataSetGeneratorSettings.PriceTags)
{
if (element.OuterHtml.Contains(priceTag))
{
isContainsPrice = true;
}
}
foreach (var regex in dataSetGeneratorSettings.Regex)
{
if (Regex.IsMatch(element.OuterHtml, regex))
{
isContainsPrice = true;
}
}
var htmlElement = Transform(element.OuterHtml);
if (isContainsPrice)
{
for (int i = 0; i < priceElementRepeatCount; i++)
{
var randomGeneratedHtmlElement = GenerateRandomPriceElement(htmlElement);
list.Add(new HtmlDataSet {
IsContainsPrice = isContainsPrice, HtmlElement = randomGeneratedHtmlElement, ClassName = element.ClassName, HtmlElementName = element.LocalName
});
}
}
list.Add(new HtmlDataSet {
IsContainsPrice = isContainsPrice, HtmlElement = htmlElement, ClassName = element.ClassName, HtmlElementName = element.LocalName
});
}
return(list);
}
static async Task <IEnumerable <HtmlDataSet> > FileStorageDataSetGenerate(string folderPath, ServiceProvider serviceProvider, DataSetGeneratorSettings dataSetSettings)
{
var context = BrowsingContext.New(Configuration.Default);
var parser = context.GetService <IHtmlParser>();
List <HtmlDataSet> list = new List <HtmlDataSet>();
foreach (var filePath in Directory.GetFiles(folderPath, "*.txt"))
{
var document = parser.ParseDocument(await File.ReadAllTextAsync(filePath));
list.AddRange(ParseDocument(document, dataSetSettings, PriceElementRepeatCount));
}
return(list);
}