Taipei Forcing Club

Computer science and contract bridge

How to set charset of all text responses on nginx

All text files on a site usually share the same character encoding. Especially UTF-8 is the modern de facto standard. However, the default charset_types does not contain text/css, let alone other non-plain text types like text/markdown.

The default charset_types should be text/* because most of them are parsed in ASCII (us-ascii) by default for backward compatibility. A text/xml response is parsed in ASCII even if BOM and XML declaration tells otherwise. Therefore, we should use application/xml for XML responses now.

Nevertheless, the charset_types setting checks complete matches only. Luckily, the map directive knows regex, and charset_types accepts a variable.

map $sent_http_content_type $charset {
    ~^text/   utf-8;
}

charset       $charset;
charset_types *;

This setting would make nginx specify UTF-8 for all text responses, such as text/css; charset=utf-8.

Author: Chen-Pang He

I’m Chen-Pang He (何震邦), M.D., Taipei Medical University. I’m a compiler engineer skilled in calculus and linear algebra. I’m also a bridge director in Taiwan.

Chen-Pang He
Uncategorized