/// <summary>
/// Constructor with specified dataset and default HTML extractor
/// </summary>
/// <param name="dataset">The query dataset</param>
/// <param name="defaultExtractor">The HTML extractor for articles</param>
public ArticleQuerier(QueryDataset dataset, IHtmlPlainTextExtractor defaultExtractor)
{
// Generate dataset path
switch (dataset)
{
case QueryDataset.TrainDataset:
this.datasetPath = Path.Combine(General.Constants.DatasetPath, General.Constants.TrainDatasetDir);
break;
case QueryDataset.TestDataset:
this.datasetPath = Path.Combine(General.Constants.DatasetPath, General.Constants.TestDatasetDir);
break;
default:
throw new NotImplementedException();
}
this.titlePath = Path.Combine(datasetPath, Constants.TitleDir);
this.summaryPath = Path.Combine(datasetPath, Constants.SummaryDir);
this.typePath = Path.Combine(datasetPath, Constants.TypeDir);
this.defaultExtractor = defaultExtractor;
}
/// <summary>
/// Constructor with specified dataset
/// </summary>
/// <param name="dataset">The query dataset</param>
public ArticleQuerier(QueryDataset dataset)
: this(dataset, new HtmlAgilityPackPlainTextExtractor())
{
}
