public List <WorldOfficialGoldHoldingReport_Raw> Run()
{
List <string> textLines;
var t = typeof(WorldOfficialGoldHoldingReportsScraper_UpTo2000);
using (var stream = t.Assembly.GetManifestResourceStream(t, "annual_time_series_on_world_official_gold_reserves.pdf"))
using (var pdfReader = new PdfReader(stream))
{
var textExtractor = new TextExtractor();
for (int i = 1; i <= pdfReader.NumberOfPages; i++)
{
PdfTextExtractor.GetTextFromPage(pdfReader, i, textExtractor);
textExtractor.NextPage();
}
textExtractor.Lines.Sort((l1, l2) =>
{
int c = l1.PageNumber.CompareTo(l2.PageNumber);
if (c != 0)
{
return(c);
}
return(l2.Y.CompareTo(l1.Y));
});
textLines = textExtractor.Lines.Select(l => l.Value).ToList();
}
List <WorldOfficialGoldHoldingReport_Raw> reports1 = new List <WorldOfficialGoldHoldingReport_Raw>();
List <WorldOfficialGoldHoldingReport_Raw> reports2 = null;
int tableNo1 = 0;
List <string> textLinesNonTabular = new List <string>();
for (int i = 0; i < textLines.Count; i++)
{
var m1 = s_regexTableInfo.Match(textLines[i]);
if (m1.Success)
{
int tableNo2 = int.Parse(m1.Groups[1].Captures.Cast <Capture>().Single().Value, NumberStyles.None, NumberFormatInfo.InvariantInfo);
if (tableNo1 != tableNo2)
{
if (3 <= tableNo2)
{
goto winning;
}
tableNo1 = tableNo2;
}
if (++i == textLines.Count)
{
goto winning;
}
var yearPerColumn = textLines[i].Split(new char[] { ' ', }, StringSplitOptions.RemoveEmptyEntries)
.Select(s =>
{
var m2 = s_regexNote.Match(s);
return(int.Parse(m2.Success
? s.Substring(0, m2.Index)
: s, NumberStyles.AllowLeadingWhite, NumberFormatInfo.InvariantInfo));
}).ToList();
if (yearPerColumn.Any(year => 2000 <= year))
{
throw new Exception();
}
if (reports2 != null)
{
reports1.AddRange(reports2);
}
reports2 = yearPerColumn.Select(year => new WorldOfficialGoldHoldingReport_Raw()
{
DataTimePoint = new DateTime(year, 1, 1, 0, 0, 0, DateTimeKind.Unspecified),
PublishTimePoint = new DateTime(2011, 8, 10, 0, 0, 0, DateTimeKind.Unspecified),
Rows = new List <WorldOfficialGoldHoldingReportRow_Raw>(),
}).ToList();
continue;
}
var reportRowPerCell = TryGetCells(textLines[i], reports2.Count);
if (reportRowPerCell != null)
{
for (int j = 0; j < reports2.Count; j++)
{
reports2[j].Rows.Add(reportRowPerCell[j]);
}
continue;
}
textLinesNonTabular.Add(textLines[i]);
}
winning:
if (reports2 != null)
{
reports1.AddRange(reports2);
}
return(reports1);
}
