def __init__(self, questions, ngram=1): self.documents = {} self.index = {} Posting.index = self #questions is formatted as a list of dictionaries with attributes 'docID', 'text' and 'cluster' for d in range(len(questions)): #print questions[d] doc = Document(title=questions[d]['title'],body=questions[d]['body'], docID = questions[d]['qid'], gram=ngram) self.documents[doc.getName()] = doc #print 'loading', doc.getName(), '...' #doc.printPostingsList() pl = doc.getPostingsList() for term in pl: self.addTerm(term,pl[term]) self.setIDFForAll()
class NewDocumentDialog(QDialog): openElementSignal = pyqtSignal(str) def __init__(self, parent, project, document=None): super(NewDocumentDialog, self).__init__(parent) self.project = project self.document = document self.ui = Ui_NewDocumentDialog.Ui_NewDocumentDialog() self.ui.setupUi(self) self.accepted.connect(self.setDocument) def setDocument(self): title = unicode(self.ui.titleEdit.text()) prefix = unicode(self.ui.prefixEdit.text()) name = unicode(self.ui.nameEdit.text()) if (self.document is None): self.document = Document() self.document.setTitle(title) self.document.setPrefix(prefix) self.document.setName(name) self.project.addDocument(self.document) self.document.setProject(self.project) self.openElementSignal.emit("document:" + self.document.getName())