extract_phrase.py

def extract_phrase(test_sentence, tagger):
    test_features = sent2features(test_sentence)
    
    pred_labels, phrase_chunks = tagger.tag(test_features), []
    
    start, end = 0, 0
    for idx in range(len(pred_labels)):
        if pred_labels[idx] == 'B':
            start = idx
        elif pred_labels[idx] == 'E':
            end = idx
            phrase_chunks.append((start, end))
        
    if end < start:
        phrase_chunks.append((start, len(pred_labels)-1))
    
    return [test_sentence[start:end+1] for start, end in phrase_chunks]

def extract_phrase_chunks(test_sentences, tagger):
    output = [extract_phrase(sent, tagger) for sent in test_sentences]
    return output