public async Task GetWikipediaResources(string topic)
{
var res = new ResourceFinder();
var rawResultsPath = DirectoryManager.GetRawResultsPath(topic);
DirectoryManager.CreateDirectoryForTopic(topic);
await res.GetWikipediaRawText(topic, rawResultsPath);
}
public void GenerateQuestions(string topic)
{
topic = topic.Replace(" ", "_");
var outputJsonPath = DirectoryManager.GetOutputJsonPath(topic);
var tpr = new TextProcessing();
var resultList = tpr.GetSentencesInformationFromJson(outputJsonPath);
var questionList = new List <TopicQuestion>();
foreach (var sentence in resultList)
{
var dependencies = GetSentenceDependency(sentence);
var words = GetSentenceWords(sentence);
var sentenceInfo = new SentenceInformationDto(sentence.SentenceText, dependencies, words);
if (MustContinue(sentence, sentenceInfo))
{
continue;
}
GeneratedQuestion generatedQuestion;
try
{
generatedQuestion = QuestionGenerator.Generate(sentenceInfo);
}
catch
{
continue;
}
if (string.IsNullOrEmpty(generatedQuestion?.Question))
{
continue;
}
var cleanQuestion = QuestionCleaner.RemovePunctuationFromEnd(generatedQuestion.Question);
cleanQuestion = $"{cleanQuestion}?";
var question = new TopicQuestion
{
Topic = topic,
InitialSentence = sentence.SentenceText,
Question = cleanQuestion,
Answer = generatedQuestion.Answer
};
questionList.Add(question);
}
DirectoryManager.WriteQuestionsToFile(questionList, topic);
}
public async Task ProcessWikipediaText(string topic)
{
var rawResultsPath = DirectoryManager.GetRawResultsPath(topic);
var cleanTextPath = DirectoryManager.GetCleanResultsPath(topic);
var outputJsonPath = DirectoryManager.GetOutputJsonPath(topic);
var referencesPath = DirectoryManager.GetReferencesPath(topic);
var text = File.ReadAllText(rawResultsPath);
text = StringUtils.CleanText(text, referencesPath);
await DirectoryManager.WriteTextToFile(text, cleanTextPath);
var tpr = new TextProcessing();
tpr.ProcessText(text, outputJsonPath);
}
