Get CentOS server capabilities in human-readable form using Python

Question

I am trying to analyze capabilities associated to my CentOS server. I have output of ps aufxwwwZ command and for each process the content of /proc/$pid/status. This is part of the dump of the configuration of the server I want to analyze.

I know I can get capabilities from /proc/$pid/status (lines beginning by ^Cap). I can use capsh --decode to get a human-readable form. However, I would like to implement this exclusively in python as part of a bigger script.

How is it possible to convert the capabilities mask to a human-readable list of capabilities without having to force users to install specific tools and launching subprocess stuff ?

I have tried to understand how the parsing is done reading linux/capability.h. Interesting parts :

#define CAP_TO_INDEX(x) ((x) >> 5)        /* 1 << 5 == bits in __u32 */
#define CAP_TO_MASK(x)  (1 << ((x) & 31)) /* mask for indexed __u32 */

So, for cap_net_admin, which is defined in this header file to 12, I use CAP_TO_MASK, ((12) & 31) gives 4096 :

printf("%d", 1 << ((12) & 31));

However, this is not what I observe for this specific capability in /proc/pid/status which is 1000.

Is there a more straightforward converting process ?