Optimize and modernize abbr

- Replace __fish_abbr_escape with `string escape`

- Don't double-parse the key

- Replace IFS magic with string

Together, this seems to speed it up by a factor of about 2.
This commit is contained in:
Fabian Homborg 2016-01-04 15:27:15 +01:00
parent 7a8cbcda06
commit ac545940de

View file

@ -76,13 +76,11 @@ function abbr --description "Manage abbreviations"
set -l value
__fish_abbr_parse_entry $mode_arg key value
# ensure the key contains at least one non-space character
set -l IFS \n\ \t
printf '%s' $key | read -lz key_ __
if test -z "$key_"
if not string match -qr "\w" -- $key
printf ( _ "%s: abbreviation must have a non-empty key\n" ) abbr >&2
return 1
end
if test -z "$value"
if not string match -qr "\w" -- $value
printf ( _ "%s: abbreviation must have a value\n" ) abbr >&2
return 1
end
@ -111,8 +109,6 @@ function abbr --description "Manage abbreviations"
case 'show'
for i in $fish_user_abbreviations
# Disable newline splitting
set -lx IFS ''
__fish_abbr_parse_entry $i key value
# Check to see if either key or value has a leading dash
@ -120,7 +116,7 @@ function abbr --description "Manage abbreviations"
set -l opt_double_dash ''
switch $key ; case '-*'; set opt_double_dash ' --'; end
switch $value ; case '-*'; set opt_double_dash ' --'; end
echo abbr$opt_double_dash (__fish_abbr_escape "$key") (__fish_abbr_escape "$value")
echo abbr$opt_double_dash (string escape -- $key $value)
end
return 0
@ -134,20 +130,6 @@ function abbr --description "Manage abbreviations"
end
end
function __fish_abbr_escape
# Prettify the common case: if everything is alphanumeric,
# we do not need escapes.
# Do this by deleting alnum characters, and check if there's anything left.
# Note we need to preserve spaces, so spaces are not considered alnum
if test -z (echo -n "$argv" | tr -d '[:alnum:]_')
echo $argv
else
# Escape via single quotes
# printf is nice for stripping the newline that sed outputs
printf "'%s'" (echo -n $argv | sed -e s,\\\\,\\\\\\\\,g -e s,\',\\\\\',g)
end
end
function __fish_abbr_get_by_key
if not set -q argv[1]
echo "__fish_abbr_get_by_key: expected one argument, got none" >&2
@ -155,9 +137,7 @@ function __fish_abbr_get_by_key
end
set -l count (count $fish_user_abbreviations)
if test $count -gt 0
set -l key
__fish_abbr_parse_entry $argv[1] key
set -l IFS \n # ensure newline splitting is enabled
set -l key $argv[1] # This assumes the key is valid and pre-parsed
for i in (seq $count)
set -l key_i
__fish_abbr_parse_entry $fish_user_abbreviations[$i] key_i
@ -177,19 +157,18 @@ function __fish_abbr_parse_entry -S -a __input __key __value
if test -z "$__value"
set __value __
end
set -l IFS '= '
switch $__input
case '=*'
# read will skip any leading ='s, but we don't want that
set __input " $__input"
set __key _
set IFS '='
case ' =*'
set __key _
set IFS '='
case "*=*"
# No need for bounds-checking because we already matched before
set -l KV (string split "=" -m 1 -- $__input)
set $__key $KV[1]
set $__value $KV[2]
case "* *"
set -l KV (string split " " -m 1 -- $__input)
set $__key $KV[1]
set $__value $KV[2]
case "*"
set $__key $__input
end
# use read -z to avoid splitting on newlines
# I think we can safely assume there will be no NULs in the input
printf "%s" $__input | read -z $__key $__value
return 0
end