/// <summary>
/// Runs the content parsing part of the summarizing algorithm
/// </summary>
/// <param name="contentProvider"></param>
/// <param name="contentParser"></param>
/// <returns></returns>
public ParsedDocument ParseContent(IContentProvider contentProvider, IContentParser contentParser)
{
if (contentProvider == null)
{
throw new ArgumentNullException(nameof(contentProvider));
}
if (contentParser == null)
{
throw new ArgumentNullException(nameof(contentParser));
}
var resultingParsedDocument = new ParsedDocument
{
Sentences = contentParser.SplitContentIntoSentences(contentProvider.Content)
};
if (resultingParsedDocument.Sentences == null)
{
throw new InvalidOperationException(
$"{contentProvider.GetType().FullName}.SplitContentIntoSentences must not return null");
}
foreach (var workingSentence in resultingParsedDocument.Sentences)
{
workingSentence.TextUnits = contentParser.SplitSentenceIntoTextUnits(workingSentence.OriginalSentence);
if (workingSentence.TextUnits == null)
{
throw new InvalidOperationException(
$"{contentProvider.GetType().FullName}.SplitSentenceIntoTextUnits must not return null");
}
}
return resultingParsedDocument;
}
/// <summary>
/// Runs the content analyzis part of the summarizing algorithm
/// </summary>
/// <param name="parsedDocument"></param>
/// <param name="contentAnalyzer"></param>
/// <returns></returns>
public AnalyzedDocument AnalyzeParsedContent(ParsedDocument parsedDocument, IContentAnalyzer contentAnalyzer)
{
if (parsedDocument == null)
{
throw new ArgumentNullException(nameof(parsedDocument));
}
if (contentAnalyzer == null)
{
throw new ArgumentNullException(nameof(contentAnalyzer));
}
var importantTextUnits = contentAnalyzer.GetImportantTextUnits(parsedDocument.Sentences);
if (importantTextUnits == null)
{
throw new InvalidOperationException(
$"{contentAnalyzer.GetType().FullName}.GetImportantTextUnits must not return null");
}
var scoredSentences = contentAnalyzer.ScoreSentences(parsedDocument.Sentences, importantTextUnits);
if (scoredSentences == null)
{
throw new InvalidOperationException(
$"{contentAnalyzer.GetType().FullName}.ScoreSentences must not return null");
}
return new AnalyzedDocument() { ScoredTextUnits = importantTextUnits.OrderByDescending(tus => tus.Score).ToList(), ScoredSentences = scoredSentences.OrderByDescending(ss => ss.Score).ToList() };
}