public async Task <Document> InitDocument(string name = "cv000_29416.txt")
{
var result = await helper.GetTextSplitter().Process(new ParseRequest(File.ReadAllText(Path.Combine(path, name)))).ConfigureAwait(false);
var factory = helper.Resolve <IWordFactory>();
var document = result.Construct(factory);
var review = helper.Resolve <Func <Document, IParsedReviewManager> >()(document).Create();
var documentFromReview = new DocumentFromReviewFactory();
return(documentFromReview.ReparseDocument(new NullRatingAdjustment(review)));
}
private static IEnumerable <IParsedDocumentHolder> GetReview(ISessionContainer container, string path, double?stars)
{
if (File.Exists(path))
{
foreach (var line in File.ReadLines(path))
{
yield return(new ParsingDocumentHolder(container.GetTextSplitter(),
container.GetWordFactory(),
container.Resolve <IContextSentenceRepairHandler>(),
new Document(line.SanitizeXmlString())
{
Stars = stars
}));
}
}
else
{
foreach (var file in Directory.EnumerateFiles(path, "*.*", SearchOption.AllDirectories))
{
yield return(new AsyncParsingDocumentHolder(Task.Run(() =>
{
var fileInfo = new FileInfo(file);
var document = new Document(File.ReadAllText(file).SanitizeXmlString())
{
Id = $"{fileInfo.Directory.Name}_{Path.GetFileNameWithoutExtension(fileInfo.Name)}",
Stars = stars
};
return (IParsedDocumentHolder) new ParsingDocumentHolder(container.GetTextSplitter(),
container.GetWordFactory(),
container.Resolve <IContextSentenceRepairHandler>(),
document);
})));
}
}
}
private static async Task <IParsedDocumentHolder> ConstructHolder(ISessionContainer container, DataPair processingData)
{
var result = await processingData.Data.ConfigureAwait(false);
if (processingData.Sentiment == SentimentClass.Positive)
{
SetStars(result, 5);
}
else if (processingData.Sentiment == SentimentClass.Negative)
{
SetStars(result, 1);
}
return(new ParsingDocumentHolder(container.GetTextSplitter(),
container.GetWordFactory(),
container.Resolve <IContextSentenceRepairHandler>(),
result));
}
