public override void SetUp()
{
base.SetUp();
myFormat = new PathSeriesFormat("test");
myFormat.Path = @"/BODY[0]/DIV[5]/DIV[0]/DIV[5]/DIV[0]/TABLE[0]/TBODY[0]";
myFormat.Anchor = Anchor.ForCell(new StringContainsLocator(1, @"xetra"), new AbsolutePositionLocator(2));
myFormat.TimeAxisPosition = 0;
myFormat.SeriesNamePosition = 0;
myPricesFile = Path.Combine(TestDataRoot, "Recognition", "ariva-prices.csv");
}
public DataTable ExtractTable(IFormat format)
{
PathSeriesFormat pathSeriesFormat = format as PathSeriesFormat;
PathTableFormat pathTableFormat = format as PathTableFormat;
if (pathSeriesFormat != null)
{
var htmlSettings = new HtmlExtractionSettings();
htmlSettings.ExtractLinkUrl = pathSeriesFormat.ExtractLinkUrl;
var result = Content.ExtractTable(HtmlPath.Parse(pathSeriesFormat.Path), pathSeriesFormat.ToExtractionSettings(), htmlSettings);
if (!result.Success)
{
throw new Exception("Failed to extract table from document: " + result.FailureReason);
}
return(pathSeriesFormat.ToFormattedTable(result.Value));
}
else if (pathTableFormat != null)
{
var result = Content.ExtractTable(HtmlPath.Parse(pathTableFormat.Path), true);
if (!result.Success)
{
throw new Exception("Failed to extract table from document: " + result.FailureReason);
}
return(pathTableFormat.ToFormattedTable(result.Value));
}
else if (format is PathSingleValueFormat)
{
var f = (PathSingleValueFormat)format;
var str = Content.GetTextByPath(HtmlPath.Parse(f.Path));
var value = f.ValueFormat.Convert(str);
// XXX: this is really ugly - i have to create a table just to satisfy the interface :(
return(CreateTableForScalar(f.ValueFormat.Type, value));
}
else
{
throw new NotSupportedException("Format not supported for Html documents: " + format.GetType());
}
}
private DataTable Parse(PathSeriesFormat format, string file)
{
var rawTable = CsvReader.Read(file, ";");
return(format.ToFormattedTable(rawTable));
}
