Solve the Chinese display problem of python matplotlib drawing in Linux system

Recently I wanted to learn some python data analysis content, so I made a crawler to crawl some data, and planned to use Anaconda's set of tools (pandas, numpy, scipy, matplotlib, jupyter) to conduct some preliminary data mining. and analysis.

When using matplotlib to draw, the abscissa is in Chinese, but the abscissa of the drawn bar chart always shows "frame", so I went to check the information to solve it. I feel that this should be a relatively common problem. There are indeed a lot of Chinese materials on the Internet, but none of them completely solved the problem I encountered. It took almost 3 hours of piecemeal work to finally get it done. I would like to share this in the hope that it can help other children who have the same problem.

Running environment:

• python2.7

• ##Linux Centos7

• Using matplotlib and pandas installed with conda

Problem:

• matplotlib draws pictures and cannot display Chinese

Cause of the problem:

• There is no Chinese font available in the Linux operating system and matplotlib's font library

• The matplotlib package only supports ASCII code by default , does not support unicode code

Summary of online information:

• Modify the resource configuration file of matplotlib, for example, add the "Simhei" font (this font does not Not all Linux systems have it! Modification has no effect)

• Install Chinese fonts for Linux and modify the resource configuration file of matplotlib. (Haha, it doesn’t work)

Solution:

is actually a combination of various solutions on the Internet. Generally speaking, there are the following steps:

1. Get the path where the matplotlibrc file is located. Get in jupyter notebook:

1 import matplotlib2 matplotlib.matplotlib_fname()

For example, my file is at:

u'~/miniconda2/lib/python2.7/site-packages/matplotlib/mpl-data/matplotlibrc'

The font parameter in this file will be modified in subsequent steps. .

2. Look at all the fonts in the system and the available Chinese fonts. Also in jupyter nb:

from matplotlib.font_manager import FontManagerimport subprocess
fm = FontManager()
mat_fonts = set(f.name for f in fm.ttflist)print mat_fonts
output = subprocess.check_output('fc-list :lang=zh -f "%{family}\n"', shell=True)print '*' * 10, '系统可用的中文字体', '*' * 10print output
zh_fonts = set(f.split(',', 1)[0] for f in output.split('\n'))
available = mat_fonts & zh_fontsprint '*' * 10, '可用的字体', '*' * 10for f in available:print f

After completing the above operation, you will find that the "Available Fonts" is empty. Because there are no Chinese fonts for matplotlib (that's why Chinese characters are all displayed as "boxes")

3. Assume that there are no Chinese fonts in the operating system. At this time, download a ttf Chinese font and install it in cenos. You need to install a system that can detect font-family, otherwise it will be invalid. I downloaded it from this website:

Unzip the rar file. Create the folder yourfontdir to store this font under the path

/usr/share/fonts, and copy the downloaded ttf file to yourfontdir (you can change the file to an English name for easier operation)

4. Install this font for cenos.

cd /usr/share/fonts/yourfontsdir#生成字体索引信息. 会显示字体的font-familysudo mkfontscale
sudo mkfontdir#更新字体缓存：fc-cache

5. Modify the matplotlibrc file

Modify the matplotlibrc file configuration obtained in step 1.

Remove the comments in the font.family part and add a Chinese font to the font.serif support font. Here, add the font-family of the Chinese font you just downloaded. You can search it through the

fc-list command (so it’s best to write it down before). What I added here is the "WenQuanYi Zen Hei Mono" font.

The following comment must be removed, otherwise the Chinese minus sign will also display a square:

axes.unicode_minus  : False

6. This step is the most important! Add Chinese fonts to matplotlib

After completing step 5, and then proceed to step 2, you will find that the "Available Chinese Fonts" already has the font you just installed, but the drawing still cannot display Chinese. This is because you installed the font in centos and told matplotlib to use this font, but matplotlib cannot find the ttf file of this font. . . . So I need to get one for it.

Copy the downloaded ttf font to the following path:

~/miniconda2/lib/python2.7/site-packages/matplotlib/mpl-data/fonts/ttf

and delete the relevant cache. In the following path:

~/.cache/matplotlib

Delete the font-related cache

7. Now try to draw a picture again. Done.

Reference:

# #

The above is the detailed content of Solve the Chinese display problem of python matplotlib drawing in Linux system. For more information, please follow other related articles on the PHP Chinese website!