geirha / bash-builtins Goto Github PK

Pointed out by mknod on irc.

Currently, an illegal line like this

foo",bar",baz

is treated as the two fields foo,bar and baz, but should rather be treated as foo", bar", and baz. The main argument for this is that writing the same data back out could then produce the same csv record.

It would be nice to be able to sort on wcslen() of each element.

It looks like you're using 8-bit strcoll to sort in asort. It would be better to only do this when shopt -s globasciiranges is in effect, so the "natural" collating order of IL8N languages can be used.
Link to some source material: https://github.com/geirha/bash-tagged/blob/master/lib/glob/smatch.c
Aside:
Thanks for the work you've done. I've always missed a built-in hash function and sort function that's a bit faster than what can be cobbled together with bash scripts. Now make a stat and od built-in :-)