private void ValidateParsedItems(HashSet <IBankovniPolozka> polozky, StatementOverview overview)
{
var currentCredit = polozky.Sum(p => p.Castka > 0 ? p.Castka : 0);
if (overview.CreditSum != currentCredit)
{
throw new ApplicationException(
$"Invalid total credit (expected {overview.CreditSum}, found {currentCredit}) - {Ucet.Url}");
}
var currentDebit = polozky.Sum(p => p.Castka < 0 ? p.Castka : 0);
if (overview.DebitSum != currentDebit)
{
throw new ApplicationException(
$"Invalid total debit (expected {overview.DebitSum}, found {currentDebit}) - {Ucet.Url}");
}
var currentFinalBalance = overview.OpeningBalance + currentCredit + currentDebit;
if (overview.FinalBalance != currentFinalBalance)
{
throw new ApplicationException(
$"Invalid final balance (expected {overview.FinalBalance}, found {currentFinalBalance}) - {Ucet.Url}");
}
}
private static void ValidateParsedItems(string sourceUrl, StatementOverview overview, List <IBankovniPolozka> statementItems)
{
var currentCreditItems = statementItems.Count(i => i.Castka > 0);
if (overview.CreditItems != currentCreditItems)
{
throw new ApplicationException(
$"Invalid count of credit items (expected {overview.CreditItems}, found {currentCreditItems}) - {sourceUrl}");
}
var currentDebitItems = statementItems.Count(i => i.Castka < 0);
if (overview.DebitItems != currentDebitItems)
{
throw new ApplicationException(
$"Invalid count of debit items (expected {overview.DebitItems}, found {currentDebitItems}) - {sourceUrl}");
}
var currentFinalBalance = overview.OpeningBalance + statementItems.Sum(i => i.Castka);
if (overview.FinalBalance != currentFinalBalance)
{
throw new ApplicationException(
$"Invalid final balance (expected {overview.FinalBalance}, found {currentFinalBalance}) - {sourceUrl}");
}
}
private List <IBankovniPolozka> ParseStatement(string textFile, DateTime date, string sourceUrl)
{
var statementItems = new List <IBankovniPolozka>();
var data = File.ReadAllLines(textFile);
var item = (StatementItemRecord)null;
var overview = new StatementOverview();
var positions = new StatementItemsPositions();
foreach (var line in data)
{
if (string.IsNullOrEmpty(line))
{
continue;
}
if (line.Contains("Změna úrokové sazby"))
{
continue;
}
if (line.StartsWith("Počet kreditních položek"))
{
overview.CreditItems = int.Parse(ItemsPattern.Match(GetValue(line, CreditItemsPosition)).Groups[1].Value);
if (!line.Contains(OpeningBalanceText))
{
continue;
}
}
if (line.StartsWith("Počet debetních položek"))
{
overview.DebitItems = int.Parse(ItemsPattern.Match(GetValue(line, DebitItemsPosition)).Groups[1].Value);
continue;
}
if (line.Contains(OpeningBalanceText))
{
overview.OpeningBalance = ParseTools.FromTextToDecimal(GetValue(line, new PositionIndex(line.IndexOf(OpeningBalanceText, SC) + OpeningBalanceText.Length + 1, default(int?)))) ?? 0;
continue;
}
if (line.Trim().StartsWith(FinalBalanceText))
{
overview.FinalBalance = ParseTools.FromTextToDecimal(GetValue(line, new PositionIndex(line.IndexOf(FinalBalanceText, SC) + FinalBalanceText.Length + 1, default(int?)))) ?? 0;
continue;
}
if (line.StartsWith("Datum"))
{
DefineItemsPositionsForFirstLine(positions, line);
continue;
}
if (line.StartsWith("Valuta"))
{
DefineItemsPositionsForSecondLine(positions, line);
continue;
}
if (item != null && (line[0] != ' ' || line.Trim().StartsWith("Převádí se") || line.Trim().StartsWith("Konec výpisu")))
{
statementItems.Add(item.ToBankPolozka(date, Ucet, sourceUrl));
item = null;
}
if (StatementItemPattern.IsMatch(line))
{
item = CreateStatementItem(line, positions);
}
else if (item != null && string.IsNullOrEmpty(item.ZpravaProPrijemce) && AccountNumberPattern.IsMatch(line))
{
UpdateStatementItem(item, line, positions);
}
else if (item != null)
{
item.ZpravaProPrijemce += (string.IsNullOrEmpty(item.ZpravaProPrijemce)
? string.Empty
: Environment.NewLine) + line.Trim();
}
}
ValidateParsedItems(sourceUrl, overview, statementItems);
return(statementItems);
}
private IEnumerable <IBankovniPolozka> ParseStatement(string url)
{
var polozky = new HashSet <IBankovniPolozka>();
using (var net = new Devmasters.Net.HttpClient.URLContent(url))
{
net.IgnoreHttpErrors = true;
var content = net.GetContent(Encoding.UTF8).Text;
if (content.Contains("Některé pohyby nemusí být zobrazeny. Zmenšete datumový rozsah."))
{
throw new StatementTooLongException();
}
var doc = new Devmasters.XPath(content);
var xoverviewRows = "//div[contains(@class, 'pohybySum')]/table/tbody/tr";
var overviewRows = doc.GetNodes(xoverviewRows)?.Count ?? 0;
if (overviewRows == 0)
{
TULogger.Warning($"FIO: Account statement page was not found for account {Ucet.CisloUctu}. Account has been probably canceled. Url: {url}");
return(new List <IBankovniPolozka>());
}
var overview = new StatementOverview
{
OpeningBalance = parseAmount(doc.GetNodeText(xoverviewRows + "/td[1]")),
FinalBalance = parseAmount(doc.GetNodeText(xoverviewRows + "/td[2]")),
CreditSum = parseAmount(doc.GetNodeText(xoverviewRows + "/td[3]")),
DebitSum = parseAmount(doc.GetNodeText(xoverviewRows + "/td[4]"))
};
var xrows = "//table[@class='table' and starts-with(@id,'id')]/tbody/tr";
var rows = doc.GetNodes(xrows)?.Count ?? 0;
for (var row = 1; row <= rows; row++)
{
var xroot = xrows + "[" + row + "]";
var p = new SimpleBankovniPolozka
{
CisloUctu = Ucet.CisloUctu,
Datum = Devmasters.DT.Util.ToDateTime(doc.GetNodeText(xroot + "/td[1]"), "dd.MM.yyyy").Value,
Castka = parseAmount(System.Net.WebUtility.HtmlDecode(doc.GetNodeText(xroot + "/td[2]"))),
PopisTransakce = System.Net.WebUtility.HtmlDecode(doc.GetNodeText(xroot + "/td[3]")),
NazevProtiuctu = System.Net.WebUtility.HtmlDecode(doc.GetNodeText(xroot + "/td[4]")),
ZpravaProPrijemce = Devmasters.TextUtil.NormalizeToBlockText(
System.Net.WebUtility.HtmlDecode(doc.GetNodeHtml(xroot + "/td[5]"))
?.Replace("<br>", " \n")
)
};
var poznamka = Devmasters.TextUtil.NormalizeToBlockText(
System.Net.WebUtility.HtmlDecode(doc.GetNodeHtml(xroot + "/td[9]"))
?.Replace("<br>", " \n")
);
if (poznamka != p.ZpravaProPrijemce)
{
p.ZpravaProPrijemce += " " + poznamka;
}
p.KS = doc.GetNodeText(xroot + "/td[6]");
p.VS = doc.GetNodeText(xroot + "/td[7]");
p.SS = doc.GetNodeText(xroot + "/td[8]");
p.ZdrojUrl = net.Url;
p.CisloProtiuctu = ""; //neni k dispozici
if (!polozky.Contains(p))
{
polozky.Add(p);
}
}
ValidateParsedItems(polozky, overview);
}
return(polozky);
}