public static string TryParseAndFixWebAddress(string webAddress)
{
if (!Uri.IsWellFormedUriString(webAddress, UriKind.RelativeOrAbsolute))
{
if (!Regex.IsMatch(webAddress, Constants.TextAnalysisConfiguration.BadOcrWebAddressRegex))
{
return(string.Empty);
}
for (int i = 2; i < 4; i++)
{
var webAddressTopLevelDomain = new string(
webAddress
.Reverse()
.Take(i)
.Reverse()
.ToArray()
);
if (FileReader.GetFileLines(Constants.TopLevelDomainListPath).Contains(webAddressTopLevelDomain.ToUpper()))
{
webAddress = webAddress.Insert(webAddress.IndexOf(webAddressTopLevelDomain), ".");
break;
}
}
}
if (Regex.IsMatch(webAddress, Constants.TextAnalysisConfiguration.WebAddressRegex))
{
try
{
var webAddressUri = new UriBuilder(webAddress).Uri;
if (Constants.ValidateWebAddresses)
{
return(HttpClientHelpers.IsWebsiteAvailable(webAddressUri)
? webAddress
: string.Empty);
}
return(webAddressUri.AbsoluteUri);
}
catch (UriFormatException)
{
return(string.Empty);
}
}
return(string.Empty);
}