the pdf buffer generated by crawling the page in ubuntu is converted into base64 and then put to oss to generate a preview link to find that some Chinese fonts are displayed as blank, but the downloaded pdf file can be displayed normally after comparison, it is found that the encoding of the blank part of the Chinese font is different from that of normal Chinese character decode, but it can be decoded normally. For example, the word "day" is normally encoded as% e6% 97% a5 and captured by puppeteer as% e2% bd% 87, but it can be decoded normally. There is no such situation in mac computer, is there any god who can help?