/// <summary>
/// Creates the single vat return information.
/// </summary>
/// <param name="htmlUrl">The HTML URL.</param>
/// <param name="pdfUrl">The PDF URL.</param>
/// <param name="browser">The browser.</param>
/// <returns></returns>
private async Task <Optional <VatReturnInfo> > CreateSingleVatReturnInfo(string htmlUrl, string pdfUrl, IEzBobWebBrowser browser)
{
string html = await browser.DownloadPageAsyncAsString(htmlUrl);
Optional <VatReturnInfo> optional = this.ParseVatReturnHtml(html);
if (optional.HasValue)
{
byte[] pdfFile = await browser.DownloadPageAsyncAsByteArray(pdfUrl);
optional.GetValue()
.PdfFile = pdfFile;
}
return(optional);
}
/// <summary>
/// Scraps login details.
/// </summary>
/// <param name="baseUrl">The base URL.</param>
/// <param name="browser">The browser.</param>
/// <returns></returns>
/// <exception cref="System.IO.InvalidDataException">Login form not found or too many forms on the page.
/// or
/// User name field not found.
/// or
/// User name field's NAME attribute not specified.
/// or
/// Password field not found.
/// or
/// Password field's NAME attribute not specified.</exception>
public async Task <ScrapedLoginRequestDetails> ScrapLoginDetails(string baseUrl, IEzBobWebBrowser browser)
{
var html = await browser.DownloadPageAsyncAsString(baseUrl);
HtmlDocument htmlDocument = new HtmlDocument();
htmlDocument.LoadHtml(html);
HtmlNodeCollection forms = htmlDocument.DocumentNode.SelectNodes("//form[contains(@action,'login')]");
if (forms == null || forms.Count != 1)
{
throw new InvalidDataException("Login form not found or too many forms on the page.");
}
HtmlNode loginForm = forms[0];
string loginUrl = loginForm.Attributes["action"].Value;
string loginMethod = loginForm.Attributes.Contains("method") ? loginForm.Attributes["method"].Value : "GET";
HtmlNode userName = loginForm.SelectSingleNode("//input[@id=\"FieldUserID\"]");
if (userName == null)
{
throw new InvalidDataException("User name field not found.");
}
if (!userName.Attributes.Contains("name"))
{
throw new InvalidDataException("User name field's NAME attribute not specified.");
}
HtmlNode password = loginForm.SelectSingleNode("//input[@id=\"FieldPassword\"]");
if (password == null)
{
throw new InvalidDataException("Password field not found.");
}
if (!password.Attributes.Contains("name"))
{
throw new InvalidDataException("Password field's NAME attribute not specified.");
}
var loginDetails = new ScrapedLoginRequestDetails {
HttpMethod = loginMethod,
LoginPageUrl = loginUrl,
UserName = userName.Attributes["name"].Value,
Password = password.Attributes["name"].Value
};
Log.Info(loginDetails.ToString());
return(loginDetails);
}
/// <summary>
/// Obtains the vat returns.
/// </summary>
/// <param name="submittedReturns">The submitted returns.</param>
/// <param name="userVatId">The user vat identifier.</param>
/// <param name="baseUrl">The base URL.</param>
/// <param name="browser">The browser.</param>
/// <returns></returns>
private IEnumerable <Task <Optional <VatReturnInfo> > > ObtainVatReturns(IDictionary <string, string> submittedReturns, string userVatId, string baseUrl, IEzBobWebBrowser browser)
{
foreach (var submitedReturn in submittedReturns)
{
string htmlUrl = baseUrl + string.Format(
"/vat-file/trader/{0}{1}{2}",
userVatId,
userVatId.EndsWith("/") || submitedReturn.Key.StartsWith("/") ? "" : "/",
submitedReturn.Key);
string pdfUrl = htmlUrl + "?format=pdf";
string baseFileName = submitedReturn.Value.Replace(' ', '_');
Log.InfoFormat("VatReturns: requesting {0}.html <- {1}", baseFileName, htmlUrl);
yield return(CreateSingleVatReturnInfo(htmlUrl, pdfUrl, browser));
}
}
/// <summary>
/// Gets the vat returns.
/// </summary>
/// <param name="userVatId">The user vat identifier.</param>
/// <param name="baseUrl">The base URL.</param>
/// <param name="browser">The browser.</param>
/// <returns></returns>
public async Task <IEnumerable <VatReturnInfo> > GetVatReturns(string userVatId, string baseUrl, IEzBobWebBrowser browser)
{
InfoAccumulator info = new InfoAccumulator();
IDictionary <string, string> submittedReturns = await FetchSubmittedReturnsList(userVatId, baseUrl, info, browser);
if (info.HasErrors || submittedReturns == null || submittedReturns.Count == 0)
{
return(null);
}
var tasks = this.ObtainVatReturns(submittedReturns, userVatId, baseUrl, browser);
Optional <VatReturnInfo>[] optionals = await Task.WhenAll(tasks);
return(optionals.Where(o => o.HasValue)
.Select(o => o.GetValue()));
}
/// <summary>
/// Fetches the submitted returns list.
/// </summary>
/// <param name="vatId">The vat identifier.</param>
/// <param name="baseUrl">The base URL.</param>
/// <param name="info">The information.</param>
/// <param name="browser">The browser.</param>
/// <returns></returns>
private async Task <IDictionary <string, string> > FetchSubmittedReturnsList(string vatId, string baseUrl, InfoAccumulator info, IEzBobWebBrowser browser)
{
Log.Info("Loading list of submitted VAT returns...");
string html = await browser.DownloadPageAsyncAsString(baseUrl + "/vat-file/trader/" + vatId + "/periods");
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(html);
bool hasNoPrevious = doc.DocumentNode.InnerText.IndexOf(
"There are no returns previously submitted available to view.",
StringComparison.InvariantCulture
) >= 0;
if (hasNoPrevious)
{
Log.Info("There are no returns previously submitted available to view.");
return(null);
} // if
HtmlNode listTable = doc.DocumentNode.SelectSingleNode("//*[@id=\"VAT0011\"]/div[2]/table/tbody");
if (listTable == null)
{
info.AddError("Failed to find list of returns.");
return(null);
}
if (listTable.ChildNodes == null)
{
Log.Info("Loading list of submitted VAT returns complete, no files found.");
return(null);
} // if
var res = new Dictionary <string, string>();
foreach (HtmlNode tr in listTable.ChildNodes)
{
if (tr.Name.ToUpper() != "TR")
{
continue;
}
if (tr.ChildNodes == null)
{
continue;
}
HtmlNode td = tr.SelectSingleNode("td");
if (td == null)
{
continue;
}
HtmlNode link = td.SelectSingleNode("a");
if (link == null)
{
continue;
}
if (!link.Attributes.Contains("href"))
{
continue;
}
string href = link.Attributes["href"].Value;
if (string.IsNullOrWhiteSpace(href))
{
continue;
}
res[href] = link.InnerText;
} // for each row
Log.InfoFormat(
"Loading list of submitted VAT returns complete, {0} file{1} found.",
res.Count,
res.Count == 1 ? "" : "s"
);
return(res);
} // Load
/// <summary>
/// Gets the user vat identifier and tax office number.
/// </summary>
/// <param name="url">The URL.</param>
/// <returns></returns>
public async Task <TaxOfficeNumberAndVatId> GetUserVatIdAndTaxOfficeNumber(string url, IEzBobWebBrowser browser)
{
string html = await browser.DownloadPageAsyncAsString(url + "/home/services");
HtmlDocument document = new HtmlDocument();
document.LoadHtml(html);
InfoAccumulator info = new InfoAccumulator();
string vatId = GetUserVatIdOldFashionedWay(document, info);
if (string.IsNullOrEmpty(vatId))
{
vatId = GetUserVatIdNewFashionedWay(document, info);
if (string.IsNullOrEmpty(vatId))
{
LogErrors(info);
return(new TaxOfficeNumberAndVatId(null, null));
}
}
info = new InfoAccumulator();
string taxOfficeNumber = ExtractTaxOfficeNumber(document, info);
if (string.IsNullOrEmpty(taxOfficeNumber))
{
LogErrors(info);
}
return(new TaxOfficeNumberAndVatId(taxOfficeNumber, vatId));
}
/// <summary>
/// Gets the rti (real time information) tax years.
/// </summary>
/// <param name="taxOfficeNumber">The tax office number.</param>
/// <param name="baseUrl">The base URL.</param>
/// <param name="browser">The browser.</param>
/// <returns></returns>
/// <exception cref="System.IO.InvalidDataException">Failed to fetch PAYE account page.</exception>
public async Task <RtiTaxYearInfo> GetRtiTaxYears(string taxOfficeNumber, string baseUrl, IEzBobWebBrowser browser)
{
if (!ValidateParameters(taxOfficeNumber, baseUrl))
{
return(null);
}
string html = await browser.DownloadPageAsyncAsString(baseUrl + "/paye/org/" + taxOfficeNumber + "/account");
if (string.IsNullOrEmpty(html))
{
throw new InvalidDataException("Failed to fetch PAYE account page.");
}
HtmlDocument document = new HtmlDocument();
document.LoadHtml(html);
return(RtiTaxYearParser.Parse(document));
}
