生成词云遇AttributeError：'ImageDraw'无'textbbox'属性如何解决？

问题描述 #

在文本挖掘项目中，完成新闻数据预处理后，使用WordCloud库生成词云时触发上述错误。

预处理代码 #

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from nltk.corpus import stopwords
from nltk.stem import PorterStemmer
from textblob import Word
from wordcloud import WordCloud 

newsData = pd.read_csv("data.txt", sep= '\t', header=None, 
                       names=["Description", "Category", "Tags"],on_bad_lines='skip', 
                       engine='python' , encoding='utf-8')

newsData['Description'] =  newsData['Description'].apply(lambda x:  " ".join(x.lower() for x in x.split()))
newsData['Category'] =  newsData['Category'].apply(lambda x:  " ".join(x.lower() for x in x.split()))
newsData['Tags'] =  newsData['Tags'].apply(lambda x:  " ".join(x.lower() for x in x.split()))

# 停用词过滤
stop = stopwords.words('english')
newsData['Description'] =  newsData['Description'].apply(lambda x: " ".join (x for x in x.split() if x not in stop))
# 词干提取
st = PorterStemmer()
newsData['Description'] =  newsData['Description'].apply(lambda x: " ".join ([st.stem(word) for word in x.split()]))
newsData['Category'] =  newsData['Category'].apply(lambda x: " ".join ([st.stem(word) for word in x.split()]))
newsData['Tags'] =  newsData['Tags'].apply(lambda x: " ".join ([st.stem(word) for word in x.split()]))

# 词形还原
newsData['Description'] =  newsData['Description'].apply(lambda x: " ".join ([Word(word).lemmatize() for word in x.split()]))
newsData['Category'] =  newsData['Category'].apply(lambda x: " ".join ([Word(word).lemmatize() for word in x.split()]))
newsData['Tags'] =  newsData['Tags'].apply(lambda x: " ".join ([Word(word).lemmatize() for word in x.split()]))

culture = newsData[newsData['Category'] == 'culture'].sample(n=200)
health = newsData[newsData['Category'] == 'health'].sample(n=200)
dataSample = pd.concat([culture, health],axis=0)

culturesmpl = culture[culture['Category'] == 'culture'].sample(n=200)
healthspml = health[health['Category'] == 'health'].sample(n=200)

cultureSTR = culturesmpl.Description.str.cat()
healthSTR = healthspml.Description.str.cat()

词云生成代码 #

wordcloud_culture =  WordCloud(collocations= False, background_color='white' ).generate(cultureSTR)

# 绘图
plt.imshow(wordcloud_culture, interpolation='bilinear')
plt.axis('off')
plt.show()

报错信息 #

File ~/anaconda3/lib/python3.9/site-packages/wordcloud/wordcloud.py:508 in generate_from_frequencies
    box_size = draw.textbbox((0, 0), word, font=transposed_font, anchor="lt")

AttributeError: 'ImageDraw' object has no attribute 'textbbox'

解决方案 #

这个错误的核心原因是WordCloud版本与Pillow版本不兼容：textbbox是Pillow 8.0.0及以上版本才新增的方法，当前使用的WordCloud新版本依赖这个方法，但你的Pillow版本低于8.0.0。

有两种解决方法：

1. 升级Pillow到8.0.0以上版本
   在终端执行命令：

   pip install --upgrade pillow>=8.0.0
   

2. 降级WordCloud到兼容旧版Pillow的版本
   如果不想升级Pillow，可以把WordCloud降级到1.8.1及以下版本，这个版本仍使用旧的textsize方法而非textbbox：

   pip install wordcloud<=1.8.1
   

执行完上述任一操作后，重新运行词云生成代码即可解决问题。

内容的提问来源于stack exchange，提问作者Sandra Moon