private void Validate(IStopwordsFile stopwords, ListingSource source)
{
if (stopwords == null && (source == ListingSource.Github))
{
throw new ArgumentNullException("Stopwords are required to preprocess listing: {0}.", source.ToString());
}
}
public IListingPreprocess GetPreprocess(ListingSource source, IStopwordsFile stopwords = null)
{
switch (source)
{
case ListingSource.Github:
return(new GitHubPreprocess(stopwords, procFact.GetTextProcessor(NLP.Language.English)));
default:
throw new Exception("Failed to instantiate listing preprocess.");
}
}
//Github
public Listing(GitHubJob job, IListingPreprocess preprocess)
{
source = ListingSource.Github;
words = new HashSet <string>();
Url = job.Url;
Title = job.Title;
Location = job.Location;
Body = job.Body;
Company = job.Company;
//TODO: figure out weird UTC format conversion - "Wed Oct 04 17:48:47 UTC 2017"
//Date = DateTime.ParseExact(job.Date, "ddd MMM dd hh:mm:ss zzz yyyy", new CultureInfo("en-US"), DateTimeStyles.None);
Date = new DateTime();
words = preprocess.Process(Body);
}