public decimal WebSelectByRegEx(Provider Provider)
{
IFfeWebParser webParser = null;
switch (WebParser)
{
case Parser.Auto:
webParser = FfeWebController.AutoWebParserSelection(Provider.Url);
break;
case Parser.HAP:
webParser = new FfeWebHap(Provider.Url);
break;
case Parser.AngleSharp:
webParser = new FfeWebAngleSharp(Provider.Url);
break;
case Parser.HttpClient:
webParser = new FfeWebHttpClient(Provider.Url);
break;
case Parser.WebClient:
webParser = new FfeWebClient(Provider.Url);
break;
default:
break;
}
string value = FfeRegEx.RegExByIndexAndGroup(webParser.GetHtml(), Provider.RegExPattern, Provider.RegExMatchIndex.Value, Provider.RegExGroupName);
decimal quote = decimal.Parse(value);
return(quote);
}
public decimal SelectByRegEx(Provider Provider)
{
string html = Resource.ResourceManager.GetString(Provider.Name);
IFfeWebParser webParser = null;
switch (WebParser)
{
case Parser.Auto:
throw new Exception("No parse method available.");
case Parser.HAP:
webParser = new FfeWebHap(html);
break;
case Parser.AngleSharp:
webParser = new FfeWebAngleSharp(html);
break;
case Parser.HttpClient:
throw new Exception("No parse method available.");
case Parser.WebClient:
throw new Exception("No parse method available.");
default:
break;
}
string value = FfeRegEx.RegExByIndexAndGroup(webParser.GetHtml(), Provider.RegExPattern, Provider.RegExMatchIndex.Value, Provider.RegExGroupName);
decimal price = decimal.Parse(value);
return(price);
}
public static (string value, string rawSource) GetValueFromWeb(string url,
string xPath = null,
string cssSelector = null,
string regExPattern = null, string regExGroup = null, int regExMatchIndex = 0,
string jsonPath = null,
Parser parser = Parser.Auto)
{
if (xPath == null &&
cssSelector == null &&
regExPattern == null &&
jsonPath == null)
{
throw new ArgumentException("No selection criteria were provided. Provide a least one of the following selection criteria: xPath, cssSelector, regExPattern or jsonPath.");
}
string value = null;
string rawSource = null;
Uri uri = new Uri(url);
if (parser != Parser.Newtonsoft)
{
IFfeWebParser ffeWebParser = null;
switch (parser)
{
case Parser.Auto:
ffeWebParser = AutoWebParserSelection(uri, xPath, cssSelector, regExPattern);
break;
case Parser.HAP:
ffeWebParser = new FfeWebHap(uri);
break;
case Parser.AngleSharp:
ffeWebParser = new FfeWebAngleSharp(uri);
break;
case Parser.HttpClient:
ffeWebParser = new FfeWebHttpClient(uri);
break;
case Parser.WebClient:
ffeWebParser = new FfeWebClient(uri);
break;
default:
ffeWebParser = new FfeWebHap(uri);
break;
}
rawSource = ffeWebParser.GetHtml();
if (log.IsEnabled(Serilog.Events.LogEventLevel.Debug))
{
rawSource.WriteToFile($"PageSource_{uri.Host}.html", parser.ToString());
}
// Input for RegEx (if set).
string input = null;
if (!String.IsNullOrEmpty(xPath))
{
value = ffeWebParser.SelectByXPath(xPath);
input = value;
}
else if (!String.IsNullOrEmpty(cssSelector))
{
value = ffeWebParser.SelectByCssSelector(cssSelector);
input = value;
}
else // Select by RegEx (HTML source code = RegEx input).
{
//HACK: AngelSharp does not provide full HTML source code.
if (parser == Parser.AngleSharp)
{
log.Warning("Regular Expression with AngleSharp WebParser does not work well. HttpClient is used instead.");
rawSource = FfeWeb.GetHttpResponseContent(uri);
}
input = rawSource;
}
if (!String.IsNullOrEmpty(regExPattern))
{
value = FfeRegEx.RegExByIndexAndGroup(input, regExPattern, regExMatchIndex, regExGroup);
if (String.IsNullOrEmpty(value))
{
if (log.IsEnabled(Serilog.Events.LogEventLevel.Debug))
{
input.WriteToFile("RegExInput.html", "RegEx");
}
throw new RegExException()
{
Input = input, Pattern = regExPattern
};
}
}
}
else
{
IFfeJsonParser ffeJsonParser = new FfeJsonNewtonsoft(uri);
value = ffeJsonParser.SelectByJsonPath(jsonPath);
rawSource = ffeJsonParser.GetJson();
if (log.IsEnabled(Serilog.Events.LogEventLevel.Debug))
{
rawSource.WriteToFile($"JsonSource_{uri.Host}.json", parser.ToString());
}
}
return(value, rawSource);
}