Add text alignment

[zjp]

Let's try this again with a clean git log -- sorry about that.

Description

Adds the option to align the output to the space after the end of the separator string ":", "->", etc by padding all arguments to printf by the same amount regardless of length

Features

--align_output 'left', 'right', 'off'
left: Don't right-justify labels like "OS", but align what gets listed after
right: Align labels to make their contents aligned
off: Default behavior of neofetch

Issues

1. Addresses [Feature Request] Formatting #1241 and Is there padding/align option? #895

[dylanaraps]

Why was this removed?

[zjp]

The way things are printed in this PR makes concatenating arg2 to arg1 redundant, but arg1 is still formatted correctly.

[zjp]

I have a more complete explanation now.

Originally, this if statement joined the two arguments into the same string, but to properly align the info by label, the labels and their corresponding data had to stay separate. I've left the JSON logic in place, but set string to the first argument in both cases, so I've updated the code to be

# If $2 doesn't exist we format $1 as info.
if [[ "$(trim "$1")" && "$2" ]]; then
    [[ "$json" ]] && { printf '    %s\n' "\"${1}\": \"${2}\","; return; }
fi
string="${1}"
string="$(trim "${string//$'\e[0m'}")"

Now I'm not really sure what the lines

length="$(strip_sequences "$string")"
length="${#length}"

do, but removing them doesn't seem to change anything, so I've left them in.

Anyway, not concatenating $2 to $1 has several benefits:

First, we still strip all sequences from the labels, standardize them to bold, and get the length.

Second, now I can pad the first argument of calls to prin with an arbitrary number of \033[0m \033[0m ... and manually center text if I want to give my own short byline like this screenshot:

BUT DO NOTE that using sequences in prin -- for whatever reason -- means the string is not considered by the alignment logic. For this statement to be true the sequences have to be raw (e.g. you have to put \33[0m in your string instead of ${reset}). If you try to use substitutions the whole strong is considered and it throws the info way to the right. I consider this a feature, but I can see how someone might want this to be fixed later, so that they can choose whether to exclude something and use ${reset}${bold}${italic} etc.

Third, we can still properly detect automatically whether or not the user probably wants a separator at the end of the string with

string="${string}${reset}${colon_color}"
[[ "$2" ]] && string="${string}${separator}${separator:=:}"

Fourth: Now users can use arbitrary sequences to format text the way they like (italic, flashing, etc).

With the string formatting simplified we can do:

printf '%b' "${text_padding:+\e[${text_padding}C}${zws}"
case "$align_output" in
    "left")  printf '%-*b' $offset "${string}" ;;
    "right") printf '%+*b' $offset "${string}" ;;
    "off") printf '%b' "${string}" ;;
esac
[[ "$2" ]] && printf '%b' "${info_color}${2}"
printf '\n'

If any of this breaks existing behavior I haven't witnessed it, so I'll need feedback from people with more exotic setups.

[zjp]

IT'S ALIVE! Finally. Pending your review, I think it's ready -- at least, it's as good as it can be given my novice bash-fu.

[zjp]

I went ahead and rebased this on master to see if it would fix those failing checks from a few months ago, and it seems that it has.