peruse.py 文件源码

def generateTable(text, n=5):
    # Start by getting a frequency dictionary
    d = path.dirname(__file__)

    cloud_coloring = np.array(Image.open(path.join(d, "us-mask-white.png")))
    stopwords = set(STOPWORDS)
    stopwords.add("said")

    wc = WordCloud(background_color="black", max_words=2000, mask=cloud_coloring,
                   stopwords=stopwords, max_font_size=40, random_state=42)

    frequenciesDict = wc.process_text(text)

    words = frequenciesDict.keys()
    freq = frequenciesDict.values()

    frequencies = pd.DataFrame({ 'words' : words, 'frequencies' : freq })
    frequencies.sort_values('frequencies', ascending = False, inplace = True)

    print '\nTop 5 Terms\n'
    print frequencies.head(n = n).to_string(index = False)
    print '\n'