After 2025, the problem of garbled characters is gone. There is still a problem with parsing the Zhihu text. What is saved on the web can't be parsed at all; only half of what is saved on the phone can be parsed (probably until the full text is expanded).

It's November, and every time I collect Zhihu, I still have to manually update the analysis. I'm curious; I can't automatically parse the article, but why can I successfully parse it by clicking on it manually?

There is a similar project called freezhihu. There was also a problem with parsing garbled codes in August. Now I see that the scraping problem has been solved. Maybe the author can contact them and ask how they solved it

It hasn't been fixed yet?

I'm also running into the garbled code problem, haven't it been fixed yet?

Has this been fixed recently?

The problem of garbled characters is particularly serious, especially after sharing from the Zhihu app to Cubox. Almost every analysis and snapshot is garbled, and the text content of the snapshot is incomplete. Most of it is obscured by “expand to read the full text”, and it is still impossible to click. To be honest, these issues cannot be solved; I am very confused about payment, which can be said to be a huge obstacle. Hope this gets fixed soon.

Recently, Zhihu's analysis has always randomly displayed garbled characters

After bookmarking on the website, I opened a different article on the client, but the respondent's ID was correct