Update mountinfo parsing

[SuperQ]

• Use a regexp to parse the mountinfo lines.
• Use util.ReadFileNoStat() to instead of a buffered reader.
• Add a test for hypens in the mountpoint.

Fixes: #225

Signed-off-by: Ben Kochie superq@gmail.com

[discordianfish]

Hrm not sure that using a regex is the best way. How do the usual linux tools parse this? I'd assume there is some regex-free way to do that. If not, this is fine.

[SuperQ]

The only code I've found so far was some random Perl code. It splits on spaces, which might be a reasonable way to go. But I thought the regexp would be a little better for validating the contents of the various sections.

[discordianfish]

I'd argue we don't want to validate this at all.

See: http://man7.org/linux/man-pages/man5/proc.5.html
I think the right way to parse this is:

• split by space
• store elements 1-6 individually
• start looping over element 7+ until you find a -
• store remaining elements individually

[SuperQ]

Ok, I'll re-work this.

[SuperQ]

@discordianfish Ok, now it's a bunch of much simpler offsets and loops. Thanks for the suggestion.

[SuperQ]

I cleaned up a couple more things, I feel like the optional field parsing is now easier to understand, and less buggy.

[pgier]

LGTM

[discordianfish]

Beside that, LGTM

[SuperQ]

Ok, I returned the optional field check to ignore not fail.