public virtual IFilingParserResult Parse(IFilingParserParams parserParams)
{
ResetState();
SECParserParams secParams = parserParams as SECParserParams;
SECParserResult result = new SECParserResult();
try
{
ValidateFile(secParams, result);
if (result.Success)
{
var doc = OpenDocument(secParams);
if (doc != null)
{
InitNsManager(doc);
ExtractContexts(doc, result);
ExtractCompanyData(doc, result);
ExtractFilingData(doc, result);
ExtractValues(doc, secParams, result);
}
}
}
catch (Exception ex)
{
result.Success = false;
result.AddError(EErrorCodes.ParserError, EErrorType.Error, ex.Message);
}
return(result);
}
protected void ExtractValues(XmlDocument doc, SECParserParams secParams, SECParserResult secResult)
{
foreach (var s in _sections.Values)
{
ParseStatementSection(doc, secParams, secResult, s);
}
}
protected void ExtractFilingData(XmlDocument doc, SECParserResult secResult)
{
string[][] tags =
{
new string[] { "dei:DocumentPeriodEndDate", "DocumentPeriodEndDate" },
new string[] { "dei:DocumentType", "DocumentType" },
new string[] { "dei:DocumentFiscalYearFocus", "DocumentFiscalYearFocus" },
new string[] { "dei:DocumentFiscalPeriodFocus", "DocumentFiscalPeriodFocus" },
};
ExtractXmlData(doc, secResult, tags, secResult.FilingData);
// separately extracting end date - using contexts
if (secResult.FilingData.ContainsKey("DocumentPeriodEndDate") && secResult.FilingData.ContainsKey("DocumentType"))
{
DateTime endDate = DateTime.Parse(secResult.FilingData["DocumentPeriodEndDate"]);
string type = secResult.FilingData["DocumentType"];
foreach (var ctx in secResult.Contexts)
{
// TODO: WARNING! this supports only 10-K and 10-Q report types - need to change in future for other types of reports
if (ctx.StartDate != DateTime.MinValue &&
(endDate - ctx.StartDate).Days >= (type == "10-Q" ? 80 /*10-Q*/ : 350 /*10-K*/) &&
(endDate - ctx.StartDate).Days <= (type == "10-Q" ? 100 /*10-Q*/ : 370 /*10-K*/)
)
{
secResult.FilingData["DocumentPeriodStartDate"] = ctx.StartDate.ToShortDateString();
break;
}
}
}
}
protected void ValidateFile(SECParserParams secParams, SECParserResult secResult)
{
if (secParams.FileContent == null || !secParams.FileContent.Values.ElementAt(0).CanRead)
{
secResult.Success = false;
secResult.AddError(EErrorCodes.FileNotFound, EErrorType.Error, "Stream is unaccessable");
}
}
protected void ExtractXmlData(XmlDocument doc, SECParserResult secResult, string[][] tags, Dictionary <string, string> values)
{
foreach (var pair in tags)
{
XmlNodeList nodes = doc.GetElementsByTagName(pair[0]);
if (nodes != null && nodes.Count > 0)
{
values.Add(pair[1], nodes[0].InnerText);
}
}
}
protected void ExtractCompanyData(XmlDocument doc, SECParserResult secResult)
{
string[][] tags =
{
new string[] { "dei:EntityRegistrantName", "EntityRegistrantName" },
new string[] { "dei:TradingSymbol", "TradingSymbol" },
new string[] { "dei:EntityCentralIndexKey", "EntityCentralIndexKey" }
};
ExtractXmlData(doc, secResult, tags, secResult.CompanyData);
}
protected void ParseStatementSection(XmlDocument doc, SECParserParams secParams, SECParserResult result, Section section)
{
// preparing statements
Statement statementSection = new Statement(section.Name);
foreach (var value in section.ValueTags.Values)
{
foreach (var context in result.Contexts)
{
string xpath = "//" + value.Tag + "[@contextRef='" + (context.ID + (!string.IsNullOrEmpty(value.Suffix) ? value.Suffix : string.Empty)) + "']";
XmlNode valueTag = doc.SelectSingleNode(xpath, _nsmgr);
if (valueTag != null)
{
object valObject = null;
string innerText = valueTag.InnerText.Trim();
Decimal valDecimal;
DateTime valDateTime;
if (Decimal.TryParse(innerText, out valDecimal) && secParams.ExtractDates)
{
valObject = valDecimal;
}
else
{
if (DateTime.TryParse(innerText, out valDateTime) && secParams.ExtractDates)
{
valObject = valDateTime;
}
else
{
if (secParams.ExtractStrings)
{
valObject = innerText;
}
}
}
if (valObject != null)
{
StatementRecord record = new StatementRecord(
value.Code,
valObject,
valueTag.Attributes["unitRef"] != null ? valueTag.Attributes["unitRef"].Value : null,
context.StartDate,
context.EndDate,
context.Instant,
valueTag.Attributes["id"] != null ? valueTag.Attributes["id"].Value : null
);
if (!statementSection.Records.Contains(record))
{
statementSection.Records.Add(record);
}
}
}
}
}
result.Statements.Add(statementSection);
}
protected void ExtractContexts(XmlDocument doc, SECParserResult secResult)
{
XmlNamespaceManager nsmgr = new XmlNamespaceManager(doc.NameTable);
nsmgr.AddNamespace("xbrli", "http://www.xbrl.org/2003/instance");
nsmgr.AddNamespace("xbrll", "http://www.xbrl.org/2003/linkbase");
nsmgr.AddNamespace("df", doc.DocumentElement.NamespaceURI);
Dictionary <string, List <string> > tags = new Dictionary <string, List <string> >();
tags.Add("xbrli:context", new List <string>(new string[] { "xbrli:period", "xbrli:startDate", "xbrli:endDate", "xbrli:instant", "xbrli:entity/xbrli:segment" }));
tags.Add("context", new List <string>(new string[] { "df:period", "df:startDate", "df:endDate", "df:instant", "df:entity/df:segment" }));
string currContextTagName = string.Empty;
XmlNodeList contextTags = null;
int currKey = 0;
do
{
currContextTagName = tags.Keys.ElementAt(currKey);
contextTags = doc.GetElementsByTagName(currContextTagName); // "xbrli:context"
// TODO: This need to be rewritten with Xpath requests - currently Xpath doesn't work on expressions like period/startDate for unknown reason
if (contextTags != null)
{
foreach (XmlNode contextTag in contextTags)
{
if (contextTag.SelectSingleNode(tags[currContextTagName][4], nsmgr) == null)
{
string ID = contextTag.Attributes["id"].Value;
DateTime startDate = DateTime.MinValue;
DateTime endDate = DateTime.MinValue;
DateTime instant = DateTime.MinValue;
XmlNode tagStartDate = contextTag.SelectSingleNode(tags[currContextTagName][0] + "/" + tags[currContextTagName][1], nsmgr); // "xbrli:period/xbrli:startDate"
XmlNode tagEndDate = contextTag.SelectSingleNode(tags[currContextTagName][0] + "/" + tags[currContextTagName][2], nsmgr); // "xbrli:period/xbrli:endDate"
XmlNode tagInstant = contextTag.SelectSingleNode(tags[currContextTagName][0] + "/" + tags[currContextTagName][3], nsmgr); // "xbrli:period/xbrli:instant"
if (tagStartDate != null)
{
startDate = DateTime.Parse(tagStartDate.InnerText);
}
if (tagEndDate != null)
{
endDate = DateTime.Parse(tagEndDate.InnerText);
}
if (tagInstant != null)
{
instant = DateTime.Parse(tagInstant.InnerText);
}
FilingContex context = new FilingContex(ID, startDate, endDate, instant);
secResult.Contexts.Add(context);
}
}
}
++currKey;
}while (currKey < tags.Keys.Count);
}