Пример #1
0
def createDocuments(directory):
    filenames = next(os.walk(directory))[2]
    trainDocs = []
    sampleDocs = []
    for filename in filenames:
        newDocument = Document.createDocument(directory, filename)
        try:
            if (newDocument.get_document_type() == DocumentType.SAMPLE):
                sampleDocs.append(newDocument)
            elif (newDocument.get_document_type() == DocumentType.TRAIN):
                trainDocs.append(newDocument)
            else:
                print "The following document could not be parsed: " + directory + filename
        except:
            print "The following document could not be parsed: " + directory + filename

    return [trainDocs, sampleDocs]
Пример #2
0
def createDocuments( directory  ):
    filenames = next(os.walk( directory ))[2]
    trainDocs = []
    sampleDocs = []
    for filename in filenames:
        newDocument = Document.createDocument( directory , filename )
        try:
            if ( newDocument.get_document_type() == DocumentType.SAMPLE ):
                sampleDocs.append( newDocument )
            elif ( newDocument.get_document_type() == DocumentType.TRAIN ):
                trainDocs.append( newDocument )
            else:
                print "The following document could not be parsed: " + directory + filename
        except:
            print "The following document could not be parsed: " + directory + filename
            
    return [ trainDocs , sampleDocs ]