private void ParseFile(string filePath, string modelPath)
{
List <Dictionary <string, object> > pages;
Dictionary <string, object> cmeta;
// READ FILE
var provider = FileFormatRepository.GetMatchingProvider(filePath);
provider.ReadFile(filePath, out pages, out cmeta);
// DETECT LANGUAGE
cmeta.Add("LANGUAGE", LanguageDetectorHelper.DetectLanguage(ref pages));
AbstractCorpusAdapter corpus;
HashSet <string> list;
Dictionary <string, double> vecs;
ExecuteProcessingWorkflow(out corpus, out list, out vecs, pages, cmeta);
if (corpus == null)
{
return;
}
// SAVE MODEL
corpus.Save(modelPath, false);
Serializer.Serialize(list, modelPath + ".list", false);
Serializer.Serialize(vecs.ToArray(), modelPath + ".vecs", false);
}
public static void Inline(ref string inlineText, out AbstractCorpusAdapter corpus, out HashSet <string> list,
out Dictionary <string, double> vecs)
{
var pages = new List <Dictionary <string, object> >
{
new Dictionary <string, object>
{
{ "Text", inlineText },
{ "PAGE", 1 }
}
};
// DETECT LANGUAGE
var cmeta = new Dictionary <string, object> {
{ "LANGUAGE", LanguageDetectorHelper.DetectLanguage(ref pages) }
};
ExecuteProcessingWorkflow(out corpus, out list, out vecs, pages, cmeta);
}
