public void ContainsSections()
{
var data = new List <KnowlegeData>();
var inputFilePath = GetDataPath();
using (var reader = XmlReader.Create(inputFilePath))
{
while (reader.Read())
{
if (reader.NodeType == XmlNodeType.Element && reader.Name.Equals("page"))
{
var outerXml = reader.ReadOuterXml();
var infoboxes = outerXml.Split(WordUtils.Infobox, System.StringSplitOptions.RemoveEmptyEntries).Skip(1);
var geoboxes = outerXml.Split(WordUtils.Geobox, System.StringSplitOptions.RemoveEmptyEntries).Skip(1);
var citations = outerXml.Split(WordUtils.Citacia, System.StringSplitOptions.RemoveEmptyEntries).Skip(1);
if (infoboxes.Any() || geoboxes.Any() || citations.Any())
{
WordUtils.TrimBoxes <Infobox>(infoboxes, ref data);
WordUtils.TrimBoxes <Geobox>(geoboxes, ref data);
WordUtils.TrimBoxes <Citation>(citations, ref data);
break;
}
}
}
}
Assert.IsTrue(data.Any());
}
/// <summary>
/// Read XML - get pages structure
/// </summary>
/// <param name="inputFilePath"></param>
public void SetPagesFromInputFile(string inputFilePath)
{
using (var reader = XmlReader.Create(inputFilePath))
{
while (reader.Read())
{
if (reader.NodeType == XmlNodeType.Element && reader.Name.Equals("page"))
{
var outerXml = reader.ReadOuterXml();
//paralel work
Parallel.Invoke(() =>
{
//get all infoboxes from page
var infoboxes = outerXml.Split(WordUtils.Infobox, System.StringSplitOptions.RemoveEmptyEntries).Skip(1);
//get all geoboxes from page
var geoboxes = outerXml.Split(WordUtils.Geobox, System.StringSplitOptions.RemoveEmptyEntries).Skip(1);
//get all citations from page
var citations = outerXml.Split(WordUtils.Citacia, System.StringSplitOptions.RemoveEmptyEntries).Skip(1);
if (infoboxes.Any() || geoboxes.Any() || citations.Any())
{
var data = new List <KnowlegeData>();
WordUtils.TrimBoxes <Infobox>(infoboxes, ref data);
WordUtils.TrimBoxes <Geobox>(geoboxes, ref data);
WordUtils.TrimBoxes <Citation>(citations, ref data);
if (data.Any())
{
Pages.Add(new Page(data));
}
}
});
}
}
}
}
/// <summary>
/// Get issac newton page
/// </summary>
/// <returns></returns>
private Page GetPage()
{
var inputFilePath = GetDataPath();
Page page = null;
using (var reader = XmlReader.Create(inputFilePath))
{
while (reader.Read())
{
if (reader.NodeType == XmlNodeType.Element && reader.Name.Equals("page"))
{
var outerXml = reader.ReadOuterXml();
if (outerXml.Contains("<title>Isaac Newton</title>"))
{
var infobox = outerXml.Split(WordUtils.Infobox, System.StringSplitOptions.RemoveEmptyEntries).Skip(1);
var data = new List <KnowlegeData>();
WordUtils.TrimBoxes <Infobox>(infobox, ref data);
page = new Page(data);
page.Infoboxes.ForEach(x => x.SetRegexAttributes());
break;
}
}
}
}
if (page == null)
{
new AssertFailedException("Page not found");
}
return(page);
}
