/// <inheritdoc />
/// <summary>
/// Method for parse parameters in url.
/// </summary>
/// <param name="url">url</param>
/// <param name="urlElement">xml element for store data</param>
public override void Parse(string url, XElement urlElement)
{
var regex = new Regex("(\\?|&)(.+?)=(.+?)(&|$)");
var matches = regex.Matches(url);
var uriElement = new XElement("parameters");
foreach (Match match in matches)
{
XElement parametr = new XElement("parametr");
XAttribute key = new XAttribute("key", match.Groups[2].Value);
XAttribute value = new XAttribute("value", match.Groups[3].Value);
parametr.Add(key);
parametr.Add(value);
uriElement.Add(parametr);
}
if (matches.Count != 0)
{
urlElement.Add(uriElement);
}
NextParser?.Parse(url, urlElement);
}
public void ParseNextProductTest()
{
NextParser nextParser = new NextParser();
string url = "http://www.next.co.uk/g92236s7#322425";
Product product = nextParser.ParseNextProduct(url);
Assert.AreNotEqual(null, product.ThumbnailLink);
}
/// <inheritdoc />
/// <summary>
/// Method for parse url.
/// </summary>
/// <param name="url">url for parse</param>
/// <param name="urlElement">xml element to write data</param>
public override void Parse(string url, XElement urlElement)
{
var regex = new Regex("\\w+:\\/\\/[\\w@][\\w.:@]+\\/?[\\w\\.?=%&=\\-@/$,]*");
var matches = regex.Matches(url);
if (matches.Count == 0)
{
Exception e = new InvalidUrlException($"{url} - not a url");
this.Logger.Error(string.Empty, e);
throw e;
}
NextParser?.Parse(url, urlElement);
}
/// <inheritdoc />
/// <summary>
/// Method for parse host of url.
/// </summary>
/// <param name="url">url</param>
/// <param name="urlElement">xml element to write data</param>
public override void Parse(string url, XElement urlElement)
{
Regex regex = new Regex("://(.+?)/");
Match match = regex.Match(url);
string hostValue = match.Groups[1].Value;
XElement host = new XElement("host");
XAttribute attribute = new XAttribute("name", hostValue);
host.Add(attribute);
urlElement.Add(host);
NextParser?.Parse(url, urlElement);
}
public void ParseNextListTest()
{
NextParser nextParser = new NextParser();
IEnumerable <string> links = nextParser.ParseNextProductList("https://www.next.co.uk/shop/gender-newborngirls-gender-oldergirls-gender-youngergirls-category-dresses");
Assert.IsTrue(links.Count() > 650);
IList <Product> products = new List <Product>();
foreach (string link in links)
{
Console.WriteLine(@"Processing link {0}", link);
products.Add(nextParser.ParseNextProduct(link));
}
}
static void Main(string[] args)
{
string url = "https://www.next.co.uk/shop/gender-newborngirls-gender-oldergirls-gender-youngergirls-category-dresses";
NextParser nextParser = new NextParser();
// IEnumerable<string> links = nextParser.ParseNextProductList(url);
// Utilities.WriteToBinaryFile(@"temp.bin", links);
IEnumerable <string> links = (Utilities.ReadFromBinaryFile <IEnumerable <string> >(@"temp.bin"));
Console.WriteLine(@"Total of {0} products found in the page", links.Count());
IList <Product> products = new List <Product>();
foreach (string link in links.Take(10))
{
Console.WriteLine(@"Processing link {0}", link);
products.Add(nextParser.ParseNextProduct(link));
}
}
/// <inheritdoc />
/// <summary>
/// Method for parse uri of url.
/// </summary>
/// <param name="url">url</param>
/// <param name="urlElement">xml element for store data</param>
public override void Parse(string url, XElement urlElement)
{
Regex regex = new Regex("(?<=[^/]/)([^/]+?)(/|\\?|$)");
MatchCollection matches = regex.Matches(url);
XElement uriElement = new XElement("uri");
foreach (Match match in matches)
{
uriElement.Add(new XElement("segment", match.Groups[1].Value));
}
if (matches.Count != 0)
{
urlElement.Add(uriElement);
}
NextParser?.Parse(url, urlElement);
}
public AdminController(IProductRepository repo)
{
repository = repo;
nextParser = new NextParser();
bodenParser = new BodenParser();
}
