Check if an indexed array in bash is sparse or dense - arrays

I have a dynamically generated, indexed array in bash and want to know whether it is sparse or dense.
An array is sparse iff there are unset indices before the last entry. Otherwise the array is dense.
The check should work in every case, even for empty arrays, very big arrays (exceeding ARG_MAX when expanded), and of course arrays with arbitrary entries (for instance null entries or entries containing *, \, spaces, and linebreaks). The latter should be fairly easy, as you probably don't want to expand the values of the array anyways.
Ideally, the check should be efficient and portable.
Here are some basic test cases to check your solution.
Your check can use the hard-coded global variable name a for compatibility with older bash versions. For bash 4.3 and higher you may want to use local -n isDense_array="$1" instead, so that you can specify the array to be checked.
isDense() {
# INSERT YOUR CHECK HERE
# if array `a` is dense, return 0 (success)
# if array `a` is sparse, return any of 1-255 (failure)
}
test() {
isDense && result=dense || result=sparse
[[ "$result" = "$expected" ]] ||
echo "Test in line $BASH_LINENO failed: $expected array considered $result"
}
expected=dense
a=(); test
a=(''); test
a=(x x x); test
expected=sparse
a=([1]=x); test
a=([1]=); test
a=([0]=x [2]=x); test
a=([4]=x [5]=x [6]=x); test
a=([0]=x [3]=x [4]=x [13]=x); test
To benchmark your check, you can use
a=($(seq 9999999))
time {
isDense
unset 'a[0]'; isDense
a[0]=1; unset 'a[9999998]'; isDense
a=([0]=x [9999999999]=x); isDense
}

Approach
Non-empty, dense arrays have indices from 0 to ${#a[*]}-1. Due to the pigeonhole principle, the last index of a sparse array must be greater or equal ${#a[#]}.
Bash Script
To get the last index, we assume that the list of indices ${!a[#]} is in ascending order. Bash's manual does not specify any order, but (at least for bash 5 and below) the implementation guarantees this order (in the source code file array.c search for array_keys_to_word_list).
isDense() {
[[ "${#a[*]}" = 0 || " ${!a[*]}" == *" $((${#a[*]}-1))" ]]
}
For small arrays this works very well. For huge arrays the check is a bit slow because of the ${!a[*]}. The benchmark from the question took 9.8 seconds.
Loadable Bash Builtin
The approach in this answer only needs the last index. But bash only allows to extract all indices using ${!a[*]} which is unnecessary slow. Internally, bash knows what the last index is. So if you wanted, you could write a loadable builtin that has access to bash's internal data structures.
Of course this is not a really practical solution. If the performance really did matter that much, you shoudn't use a bash script. Nevertheless, I wrote such a builtin for the fun of it.
Loadable bash builtin
The space and time complexity of above builtin is indepent of the size and structure of the array. Checking isdense a should be as fast as something like b=1.

UPDATE: (re)ran tests in an Ubuntu 20 VM (much better times than previous tests in a cygwin/bash env; times are closer to those reported by Socowi)
NOTE: I populated my array with 10 million entries ({0..9999999}).
Using the same assumption as Socowi ...
To get the last index of an array, we assume that the list of indices ${!a[#]} is in ascending order. Bash's manual does not specify any order, but (at least for bash 5 and below) the implementation guarantees this order
... we can make the same assumption about the output from typeset -p a, namely, output is ordered by index.
Finding the last index:
$ a[9999999]='[abcdef]'
$ typeset -p a | tail -c 40
9999998]="9999998" [9999999]="[abcdef]")
1st attempt using awk to strip off the last index:
$ typeset -p a | awk '{delim="[][]"; split($NF,arr,delim); print arr[2]}'
9999999
This is surprisingly slow (> 3.5 minutes for cygwin/bash) while running at 100% (single) cpu utilization and eating up ~4 GB of memory.
2nd attempt using tail -c to limit the data fed to awk:
$ typeset -p a | tail -c 40 | awk '{delim="[][]"; split($NF,arr,delim); print arr[2]}'
9999999
This came in at a (relatively) blazing speed of ~1.6 seconds (ubuntu/bash).
3rd attempt saving last array value in local variable before parsing with awk
NOTES:
as Socowi's pointed out (in comments), the contents of the last element could contain several characters (spaces, brackets, single/double quotes, etc) that make parsing quite complicated
one workaround is to save the last array value in a variable, parse the typeset -p output, then put the value back
accessing the last array value can be accomplished via array[-1] (requires bash 4.3+)
This also comes in at a (relatively) blazing speed of ~1.6 seconds (ubuntu/bash):
$ lastv="${a[-1]}"
$ a[-1]=""
$ typeset -p a | tail -c 40 | awk -F'[][]' '{print $(NF-1)}'
9999999
$ a[-1]="${lastv}"
Function:
This gives us the following isDense() function:
initial function
unset -f isDense
isDense() {
local last=$(typeset -p a | tail -c 40 | awk '{delim="[][]"; split($NF,arr,delim); print arr[2]}')
[[ "${#a[#]}" = 0 || "${#a[#]}" -ne "${last}" ]]
}
latest function (variable=last array value)
unset -f isDense
isDense() {
local lastv="${a[-1]}"
a[-1]=""
local last=$(typeset -p a | tail -c 40 | awk -F'[][]' '{print $(NF-1)}')
[[ "${#a[#]}" = 0 || "${#a[#]}" -ne "${last}" ]]
rc=$?
a[-1]="${lastv}"
return "${rc}"
}
Benchmark:
from the question (minus the 4th test - I got tired of waiting to repopulate my 10-million entry array) which means ...
keep in mind the benchmark/test is calling isDense() 3x times
ran each function a few times in ubuntu/bash and these were the best times ...
Socowi's function:
real 0m11.717s
user 0m9.486s
sys 0m1.982s
oguz ismail's function:
real 0m10.450s
user 0m9.899s
sys 0m0.546s
My initial typeset|tail-c|awk function:
real 0m4.514s
user 0m3.574s
sys 0m1.442s
Latest test (variable=last array value) with Socowi's declare|tr|tail-n function:
real 0m5.306s
user 0m4.130s
sys 0m2.670s
Latest test (variable=last array value) with original typeset|tail-c|awk function:
real 0m4.305s
user 0m3.247s
sys 0m1.761s

isDense() {
local -rn array="$1"
((${#array[#]})) || return 0
local -ai indices=("${!array[#]}")
((indices[-1] + 1 == ${#array[#]}))
}
To follow the self-contained call convention:
test() {
local result
isDense "$1" && result=dense || result=sparse
[[ "$result" = "$2" ]] ||
echo "Test in line $BASH_LINENO failed: $2 array considered $result"
}
a=(); test a dense
a=(''); test a dense
a=(x x x); test a dense
a=([1]=x); test a sparse
a=([1]=); test a sparse
a=([0]=x [2]=x); test a sparse
a=([4]=x [5]=x [6]=x); test a sparse
a=([0]=x [3]=x [4]=x [13]=x); test a sparse

Related

BASH: Sorting Associative array without echoing it [duplicate]

I have an array in Bash, for example:
array=(a c b f 3 5)
I need to sort the array. Not just displaying the content in a sorted way, but to get a new array with the sorted elements. The new sorted array can be a completely new one or the old one.
You don't really need all that much code:
IFS=$'\n' sorted=($(sort <<<"${array[*]}"))
unset IFS
Supports whitespace in elements (as long as it's not a newline), and works in Bash 3.x.
e.g.:
$ array=("a c" b f "3 5")
$ IFS=$'\n' sorted=($(sort <<<"${array[*]}")); unset IFS
$ printf "[%s]\n" "${sorted[#]}"
[3 5]
[a c]
[b]
[f]
Note: #sorontar has pointed out that care is required if elements contain wildcards such as * or ?:
The sorted=($(...)) part is using the "split and glob" operator. You should turn glob off: set -f or set -o noglob or shopt -op noglob or an element of the array like * will be expanded to a list of files.
What's happening:
The result is a culmination six things that happen in this order:
IFS=$'\n'
"${array[*]}"
<<<
sort
sorted=($(...))
unset IFS
First, the IFS=$'\n'
This is an important part of our operation that affects the outcome of 2 and 5 in the following way:
Given:
"${array[*]}" expands to every element delimited by the first character of IFS
sorted=() creates elements by splitting on every character of IFS
IFS=$'\n' sets things up so that elements are expanded using a new line as the delimiter, and then later created in a way that each line becomes an element. (i.e. Splitting on a new line.)
Delimiting by a new line is important because that's how sort operates (sorting per line). Splitting by only a new line is not-as-important, but is needed preserve elements that contain spaces or tabs.
The default value of IFS is a space, a tab, followed by a new line, and would be unfit for our operation.
Next, the sort <<<"${array[*]}" part
<<<, called here strings, takes the expansion of "${array[*]}", as explained above, and feeds it into the standard input of sort.
With our example, sort is fed this following string:
a c
b
f
3 5
Since sort sorts, it produces:
3 5
a c
b
f
Next, the sorted=($(...)) part
The $(...) part, called command substitution, causes its content (sort <<<"${array[*]}) to run as a normal command, while taking the resulting standard output as the literal that goes where ever $(...) was.
In our example, this produces something similar to simply writing:
sorted=(3 5
a c
b
f
)
sorted then becomes an array that's created by splitting this literal on every new line.
Finally, the unset IFS
This resets the value of IFS to the default value, and is just good practice.
It's to ensure we don't cause trouble with anything that relies on IFS later in our script. (Otherwise we'd need to remember that we've switched things around--something that might be impractical for complex scripts.)
Original response:
array=(a c b "f f" 3 5)
readarray -t sorted < <(for a in "${array[#]}"; do echo "$a"; done | sort)
output:
$ for a in "${sorted[#]}"; do echo "$a"; done
3
5
a
b
c
f f
Note this version copes with values that contains special characters or whitespace (except newlines)
Note readarray is supported in bash 4+.
Edit Based on the suggestion by #Dimitre I had updated it to:
readarray -t sorted < <(printf '%s\0' "${array[#]}" | sort -z | xargs -0n1)
which has the benefit of even understanding sorting elements with newline characters embedded correctly. Unfortunately, as correctly signaled by #ruakh this didn't mean the the result of readarray would be correct, because readarray has no option to use NUL instead of regular newlines as line-separators.
If you don't need to handle special shell characters in the array elements:
array=(a c b f 3 5)
sorted=($(printf '%s\n' "${array[#]}"|sort))
With bash you'll need an external sorting program anyway.
With zsh no external programs are needed and special shell characters are easily handled:
% array=('a a' c b f 3 5); printf '%s\n' "${(o)array[#]}"
3
5
a a
b
c
f
ksh has set -s to sort ASCIIbetically.
Here's a pure Bash quicksort implementation:
#!/bin/bash
# quicksorts positional arguments
# return is in array qsort_ret
qsort() {
local pivot i smaller=() larger=()
qsort_ret=()
(($#==0)) && return 0
pivot=$1
shift
for i; do
# This sorts strings lexicographically.
if [[ $i < $pivot ]]; then
smaller+=( "$i" )
else
larger+=( "$i" )
fi
done
qsort "${smaller[#]}"
smaller=( "${qsort_ret[#]}" )
qsort "${larger[#]}"
larger=( "${qsort_ret[#]}" )
qsort_ret=( "${smaller[#]}" "$pivot" "${larger[#]}" )
}
Use as, e.g.,
$ array=(a c b f 3 5)
$ qsort "${array[#]}"
$ declare -p qsort_ret
declare -a qsort_ret='([0]="3" [1]="5" [2]="a" [3]="b" [4]="c" [5]="f")'
This implementation is recursive… so here's an iterative quicksort:
#!/bin/bash
# quicksorts positional arguments
# return is in array qsort_ret
# Note: iterative, NOT recursive! :)
qsort() {
(($#==0)) && return 0
local stack=( 0 $(($#-1)) ) beg end i pivot smaller larger
qsort_ret=("$#")
while ((${#stack[#]})); do
beg=${stack[0]}
end=${stack[1]}
stack=( "${stack[#]:2}" )
smaller=() larger=()
pivot=${qsort_ret[beg]}
for ((i=beg+1;i<=end;++i)); do
if [[ "${qsort_ret[i]}" < "$pivot" ]]; then
smaller+=( "${qsort_ret[i]}" )
else
larger+=( "${qsort_ret[i]}" )
fi
done
qsort_ret=( "${qsort_ret[#]:0:beg}" "${smaller[#]}" "$pivot" "${larger[#]}" "${qsort_ret[#]:end+1}" )
if ((${#smaller[#]}>=2)); then stack+=( "$beg" "$((beg+${#smaller[#]}-1))" ); fi
if ((${#larger[#]}>=2)); then stack+=( "$((end-${#larger[#]}+1))" "$end" ); fi
done
}
In both cases, you can change the order you use: I used string comparisons, but you can use arithmetic comparisons, compare wrt file modification time, etc. just use the appropriate test; you can even make it more generic and have it use a first argument that is the test function use, e.g.,
#!/bin/bash
# quicksorts positional arguments
# return is in array qsort_ret
# Note: iterative, NOT recursive! :)
# First argument is a function name that takes two arguments and compares them
qsort() {
(($#<=1)) && return 0
local compare_fun=$1
shift
local stack=( 0 $(($#-1)) ) beg end i pivot smaller larger
qsort_ret=("$#")
while ((${#stack[#]})); do
beg=${stack[0]}
end=${stack[1]}
stack=( "${stack[#]:2}" )
smaller=() larger=()
pivot=${qsort_ret[beg]}
for ((i=beg+1;i<=end;++i)); do
if "$compare_fun" "${qsort_ret[i]}" "$pivot"; then
smaller+=( "${qsort_ret[i]}" )
else
larger+=( "${qsort_ret[i]}" )
fi
done
qsort_ret=( "${qsort_ret[#]:0:beg}" "${smaller[#]}" "$pivot" "${larger[#]}" "${qsort_ret[#]:end+1}" )
if ((${#smaller[#]}>=2)); then stack+=( "$beg" "$((beg+${#smaller[#]}-1))" ); fi
if ((${#larger[#]}>=2)); then stack+=( "$((end-${#larger[#]}+1))" "$end" ); fi
done
}
Then you can have this comparison function:
compare_mtime() { [[ $1 -nt $2 ]]; }
and use:
$ qsort compare_mtime *
$ declare -p qsort_ret
to have the files in current folder sorted by modification time (newest first).
NOTE. These functions are pure Bash! no external utilities, and no subshells! they are safe wrt any funny symbols you may have (spaces, newline characters, glob characters, etc.).
NOTE2. The test [[ $i < $pivot ]] is correct. It uses the lexicographical string comparison. If your array only contains integers and you want to sort numerically, use ((i < pivot)) instead.
Please don't edit this answer to change that. It has already been edited (and rolled back) a couple of times. The test I gave here is correct and corresponds to the output given in the example: the example uses both strings and numbers, and the purpose is to sort it in lexicographical order. Using ((i < pivot)) in this case is wrong.
tl;dr:
Sort array a_in and store the result in a_out (elements must not have embedded newlines[1]
):
Bash v4+:
readarray -t a_out < <(printf '%s\n' "${a_in[#]}" | sort)
Bash v3:
IFS=$'\n' read -d '' -r -a a_out < <(printf '%s\n' "${a_in[#]}" | sort)
Advantages over antak's solution:
You needn't worry about accidental globbing (accidental interpretation of the array elements as filename patterns), so no extra command is needed to disable globbing (set -f, and set +f to restore it later).
You needn't worry about resetting IFS with unset IFS.[2]
Optional reading: explanation and sample code
The above combines Bash code with external utility sort for a solution that works with arbitrary single-line elements and either lexical or numerical sorting (optionally by field):
Performance: For around 20 elements or more, this will be faster than a pure Bash solution - significantly and increasingly so once you get beyond around 100 elements.
(The exact thresholds will depend on your specific input, machine, and platform.)
The reason it is fast is that it avoids Bash loops.
printf '%s\n' "${a_in[#]}" | sort performs the sorting (lexically, by default - see sort's POSIX spec):
"${a_in[#]}" safely expands to the elements of array a_in as individual arguments, whatever they contain (including whitespace).
printf '%s\n' then prints each argument - i.e., each array element - on its own line, as-is.
Note the use of a process substitution (<(...)) to provide the sorted output as input to read / readarray (via redirection to stdin, <), because read / readarray must run in the current shell (must not run in a subshell) in order for output variable a_out to be visible to the current shell (for the variable to remain defined in the remainder of the script).
Reading sort's output into an array variable:
Bash v4+: readarray -t a_out reads the individual lines output by sort into the elements of array variable a_out, without including the trailing \n in each element (-t).
Bash v3: readarray doesn't exist, so read must be used:
IFS=$'\n' read -d '' -r -a a_out tells read to read into array (-a) variable a_out, reading the entire input, across lines (-d ''), but splitting it into array elements by newlines (IFS=$'\n'. $'\n', which produces a literal newline (LF), is a so-called ANSI C-quoted string).
(-r, an option that should virtually always be used with read, disables unexpected handling of \ characters.)
Annotated sample code:
#!/usr/bin/env bash
# Define input array `a_in`:
# Note the element with embedded whitespace ('a c')and the element that looks like
# a glob ('*'), chosen to demonstrate that elements with line-internal whitespace
# and glob-like contents are correctly preserved.
a_in=( 'a c' b f 5 '*' 10 )
# Sort and store output in array `a_out`
# Saving back into `a_in` is also an option.
IFS=$'\n' read -d '' -r -a a_out < <(printf '%s\n' "${a_in[#]}" | sort)
# Bash 4.x: use the simpler `readarray -t`:
# readarray -t a_out < <(printf '%s\n' "${a_in[#]}" | sort)
# Print sorted output array, line by line:
printf '%s\n' "${a_out[#]}"
Due to use of sort without options, this yields lexical sorting (digits sort before letters, and digit sequences are treated lexically, not as numbers):
*
10
5
a c
b
f
If you wanted numerical sorting by the 1st field, you'd use sort -k1,1n instead of just sort, which yields (non-numbers sort before numbers, and numbers sort correctly):
*
a c
b
f
5
10
[1] To handle elements with embedded newlines, use the following variant (Bash v4+, with GNU sort):
readarray -d '' -t a_out < <(printf '%s\0' "${a_in[#]}" | sort -z).
Michał Górny's helpful answer has a Bash v3 solution.
[2] While IFS is set in the Bash v3 variant, the change is scoped to the command.
By contrast, what follows IFS=$'\n'  in antak's answer is an assignment rather than a command, in which case the IFS change is global.
In the 3-hour train trip from Munich to Frankfurt (which I had trouble to reach because Oktoberfest starts tomorrow) I was thinking about my first post. Employing a global array is a much better idea for a general sort function. The following function handles arbitary strings (newlines, blanks etc.):
declare BSORT=()
function bubble_sort()
{ #
# #param [ARGUMENTS]...
#
# Sort all positional arguments and store them in global array BSORT.
# Without arguments sort this array. Return the number of iterations made.
#
# Bubble sorting lets the heaviest element sink to the bottom.
#
(($# > 0)) && BSORT=("$#")
local j=0 ubound=$((${#BSORT[*]} - 1))
while ((ubound > 0))
do
local i=0
while ((i < ubound))
do
if [ "${BSORT[$i]}" \> "${BSORT[$((i + 1))]}" ]
then
local t="${BSORT[$i]}"
BSORT[$i]="${BSORT[$((i + 1))]}"
BSORT[$((i + 1))]="$t"
fi
((++i))
done
((++j))
((--ubound))
done
echo $j
}
bubble_sort a c b 'z y' 3 5
echo ${BSORT[#]}
This prints:
3 5 a b c z y
The same output is created from
BSORT=(a c b 'z y' 3 5)
bubble_sort
echo ${BSORT[#]}
Note that probably Bash internally uses smart-pointers, so the swap-operation could be cheap (although I doubt it). However, bubble_sort demonstrates that more advanced functions like merge_sort are also in the reach of the shell language.
Another solution that uses external sort and copes with any special characters (except for NULs :)). Should work with bash-3.2 and GNU or BSD sort (sadly, POSIX doesn't include -z).
local e new_array=()
while IFS= read -r -d '' e; do
new_array+=( "${e}" )
done < <(printf "%s\0" "${array[#]}" | LC_ALL=C sort -z)
First look at the input redirection at the end. We're using printf built-in to write out the array elements, zero-terminated. The quoting makes sure array elements are passed as-is, and specifics of shell printf cause it to reuse the last part of format string for each remaining parameter. That is, it's equivalent to something like:
for e in "${array[#]}"; do
printf "%s\0" "${e}"
done
The null-terminated element list is then passed to sort. The -z option causes it to read null-terminated elements, sort them and output null-terminated as well. If you needed to get only the unique elements, you can pass -u since it is more portable than uniq -z. The LC_ALL=C ensures stable sort order independently of locale — sometimes useful for scripts. If you want the sort to respect locale, remove that.
The <() construct obtains the descriptor to read from the spawned pipeline, and < redirects the standard input of the while loop to it. If you need to access the standard input inside the pipe, you may use another descriptor — exercise for the reader :).
Now, back to the beginning. The read built-in reads output from the redirected stdin. Setting empty IFS disables word splitting which is unnecessary here — as a result, read reads the whole 'line' of input to the single provided variable. -r option disables escape processing that is undesired here as well. Finally, -d '' sets the line delimiter to NUL — that is, tells read to read zero-terminated strings.
As a result, the loop is executed once for every successive zero-terminated array element, with the value being stored in e. The example just puts the items in another array but you may prefer to process them directly :).
Of course, that's just one of the many ways of achieving the same goal. As I see it, it is simpler than implementing complete sorting algorithm in bash and in some cases it will be faster. It handles all special characters including newlines and should work on most of the common systems. Most importantly, it may teach you something new and awesome about bash :).
Keep it simple ;)
In the following example, the array b is the sorted version of the array a!
The second line echos each item of the array a, then pipes them to the sort command, and the output is used to initiate the array b.
a=(2 3 1)
b=( $( for x in ${a[#]}; do echo $x; done | sort ) )
echo ${b[#]} # output: 1 2 3
min sort:
#!/bin/bash
array=(.....)
index_of_element1=0
while (( ${index_of_element1} < ${#array[#]} )); do
element_1="${array[${index_of_element1}]}"
index_of_element2=$((index_of_element1 + 1))
index_of_min=${index_of_element1}
min_element="${element_1}"
for element_2 in "${array[#]:$((index_of_element1 + 1))}"; do
min_element="`printf "%s\n%s" "${min_element}" "${element_2}" | sort | head -n+1`"
if [[ "${min_element}" == "${element_2}" ]]; then
index_of_min=${index_of_element2}
fi
let index_of_element2++
done
array[${index_of_element1}]="${min_element}"
array[${index_of_min}]="${element_1}"
let index_of_element1++
done
try this:
echo ${array[#]} | awk 'BEGIN{RS=" ";} {print $1}' | sort
Output will be:
3
5
a
b
c
f
Problem solved.
If you can compute a unique integer for each element in the array, like this:
tab='0123456789abcdefghijklmnopqrstuvwxyz'
# build the reversed ordinal map
for ((i = 0; i < ${#tab}; i++)); do
declare -g ord_${tab:i:1}=$i
done
function sexy_int() {
local sum=0
local i ch ref
for ((i = 0; i < ${#1}; i++)); do
ch="${1:i:1}"
ref="ord_$ch"
(( sum += ${!ref} ))
done
return $sum
}
sexy_int hello
echo "hello -> $?"
sexy_int world
echo "world -> $?"
then, you can use these integers as array indexes, because Bash always use sparse array, so no need to worry about unused indexes:
array=(a c b f 3 5)
for el in "${array[#]}"; do
sexy_int "$el"
sorted[$?]="$el"
done
echo "${sorted[#]}"
Pros. Fast.
Cons. Duplicated elements are merged, and it can be impossible to map contents to 32-bit unique integers.
array=(a c b f 3 5)
new_array=($(echo "${array[#]}" | sed 's/ /\n/g' | sort))
echo ${new_array[#]}
echo contents of new_array will be:
3 5 a b c f
There is a workaround for the usual problem of spaces and newlines:
Use a character that is not in the original array (like $'\1' or $'\4' or similar).
This function gets the job done:
# Sort an Array may have spaces or newlines with a workaround (wa=$'\4')
sortarray(){ local wa=$'\4' IFS=''
if [[ $* =~ [$wa] ]]; then
echo "$0: error: array contains the workaround char" >&2
exit 1
fi
set -f; local IFS=$'\n' x nl=$'\n'
set -- $(printf '%s\n' "${#//$nl/$wa}" | sort -n)
for x
do sorted+=("${x//$wa/$nl}")
done
}
This will sort the array:
$ array=( a b 'c d' $'e\nf' $'g\1h')
$ sortarray "${array[#]}"
$ printf '<%s>\n' "${sorted[#]}"
<a>
<b>
<c d>
<e
f>
<gh>
This will complain that the source array contains the workaround character:
$ array=( a b 'c d' $'e\nf' $'g\4h')
$ sortarray "${array[#]}"
./script: error: array contains the workaround char
description
We set two local variables wa (workaround char) and a null IFS
Then (with ifs null) we test that the whole array $*.
Does not contain any woraround char [[ $* =~ [$wa] ]].
If it does, raise a message and signal an error: exit 1
Avoid filename expansions: set -f
Set a new value of IFS (IFS=$'\n') a loop variable x and a newline var (nl=$'\n').
We print all values of the arguments received (the input array $#).
but we replace any new line by the workaround char "${#//$nl/$wa}".
send those values to be sorted sort -n.
and place back all the sorted values in the positional arguments set --.
Then we assign each argument one by one (to preserve newlines).
in a loop for x
to a new array: sorted+=(…)
inside quotes to preserve any existing newline.
restoring the workaround to a newline "${x//$wa/$nl}".
done
This question looks closely related. And BTW, here's a mergesort in Bash (without external processes):
mergesort() {
local -n -r input_reference="$1"
local -n output_reference="$2"
local -r -i size="${#input_reference[#]}"
local merge previous
local -a -i runs indices
local -i index previous_idx merged_idx \
run_a_idx run_a_stop \
run_b_idx run_b_stop
output_reference=("${input_reference[#]}")
if ((size == 0)); then return; fi
previous="${output_reference[0]}"
runs=(0)
for ((index = 0;;)) do
for ((++index;; ++index)); do
if ((index >= size)); then break 2; fi
if [[ "${output_reference[index]}" < "$previous" ]]; then break; fi
previous="${output_reference[index]}"
done
previous="${output_reference[index]}"
runs+=(index)
done
runs+=(size)
while (("${#runs[#]}" > 2)); do
indices=("${!runs[#]}")
merge=("${output_reference[#]}")
for ((index = 0; index < "${#indices[#]}" - 2; index += 2)); do
merged_idx=runs[indices[index]]
run_a_idx=merged_idx
previous_idx=indices[$((index + 1))]
run_a_stop=runs[previous_idx]
run_b_idx=runs[previous_idx]
run_b_stop=runs[indices[$((index + 2))]]
unset runs[previous_idx]
while ((run_a_idx < run_a_stop && run_b_idx < run_b_stop)); do
if [[ "${merge[run_a_idx]}" < "${merge[run_b_idx]}" ]]; then
output_reference[merged_idx++]="${merge[run_a_idx++]}"
else
output_reference[merged_idx++]="${merge[run_b_idx++]}"
fi
done
while ((run_a_idx < run_a_stop)); do
output_reference[merged_idx++]="${merge[run_a_idx++]}"
done
while ((run_b_idx < run_b_stop)); do
output_reference[merged_idx++]="${merge[run_b_idx++]}"
done
done
done
}
declare -ar input=({z..a}{z..a})
declare -a output
mergesort input output
echo "${input[#]}"
echo "${output[#]}"
Many thanks to the people that answered before me. Using their excellent input, bash documentation and ideas from other treads, this is what works perfectly for me without IFS change
array=("a \n c" b f "3 5")
Using process substitution and read array in bash > v4.4 WITH EOL character
readarray -t sorted < <(sort < <(printf '%s\n' "${array[#]}"))
Using process substitution and read array in bash > v4.4 WITH NULL character
readarray -td '' sorted < <(sort -z < <(printf '%s\0' "${array[#]}"))
Finally we verify with
printf "[%s]\n" "${sorted[#]}"
output is
[3 5]
[a \n c]
[b]
[f]
Please, let me know if that is a correct test for embedded /n as both solutions produce the same result, but the first one is not supposed to work properly with embedded /n
I am not convinced that you'll need an external sorting program in Bash.
Here is my implementation for the simple bubble-sort algorithm.
function bubble_sort()
{ #
# Sorts all positional arguments and echoes them back.
#
# Bubble sorting lets the heaviest (longest) element sink to the bottom.
#
local array=($#) max=$(($# - 1))
while ((max > 0))
do
local i=0
while ((i < max))
do
if [ ${array[$i]} \> ${array[$((i + 1))]} ]
then
local t=${array[$i]}
array[$i]=${array[$((i + 1))]}
array[$((i + 1))]=$t
fi
((i += 1))
done
((max -= 1))
done
echo ${array[#]}
}
array=(a c b f 3 5)
echo " input: ${array[#]}"
echo "output: $(bubble_sort ${array[#]})"
This shall print:
input: a c b f 3 5
output: 3 5 a b c f
a=(e b 'c d')
shuf -e "${a[#]}" | sort >/tmp/f
mapfile -t g </tmp/f
Great answers here. Learned a lot. After reading them all, I figure I'd throw my hat into the ring. I think this is the shortest method (and probably faster as it doesn't do much shell script parsing, though there is the matter of the spawning of printf and sort, but they're only called once each) and handles whitespace in the data:
a=(3 "2 a" 1) # Setup!
IFS=$'\n' b=( $(printf "%s\n" "${a[#]}" | sort) ); unset IFS # Sort!
printf "'%s' " "${b[#]}"; # Success!
Outputs:
'1' '2 a' '3'
Note that the IFS change is limited in scope to the line it is on. if you know that the array has no whitespace in it, you don't need the IFS modification.
Inspiration was from #yas's answer and #Alcamtar comments.
EDIT
Oh, I somehow missed the actually accepted answer which is even shorter than mine. Doh!
IFS=$'\n' sorted=($(sort <<<"${array[*]}")); unset IFS
Turns out that the unset is required because this is a variable assignment that has no command.
I'd recommend going to that answer because it has some interesting stuff on globbing which could be relevant if the array has wildcards in it. It also has a detailed description as to what is happening.
EDIT 2
GNU has an extension in which sort delimits records using \0 which is good if you have LFs in your data. However, when it gets returned to the shell to be assign to an array, I don't see a good way convert it so that the shell will delimit on \0, because even setting IFS=$'\0', the shell doesn't like it and doesn't properly break it up.
array=(z 'b c'); { set "${array[#]}"; printf '%s\n' "$#"; } \
| sort \
| mapfile -t array; declare -p array
declare -a array=([0]="b c" [1]="z")
Open an inline function {...} to get a fresh set of positional arguments (e.g. $1, $2, etc).
Copy the array to the positional arguments. (e.g. set "${array[#]}" will copy the nth array argument to the nth positional argument. Note the quotes preserve whitespace that may be contained in an array element).
Print each positional argument (e.g. printf '%s\n' "$#" will print each positional argument on its own line. Again, note the quotes preserve whitespace that may be contained in each positional argument).
Then sort does its thing.
Read the stream into an array with mapfile (e.g. mapfile -t array reads each line into the variable array and the -t ignores the \n in each line).
Dump the array to show its been sorted.
As a function:
set +m
shopt -s lastpipe
sort_array() {
declare -n ref=$1
set "${ref[#]}"
printf '%s\n' "$#"
| sort \
| mapfile -t $ref
}
then
array=(z y x); sort_array array; declare -p array
declare -a array=([0]="x" [1]="y" [2]="z")
I look forward to being ripped apart by all the UNIX gurus! :)
sorted=($(echo ${array[#]} | tr " " "\n" | sort))
In the spirit of bash / linux, I would pipe the best command-line tool for each step. sort does the main job but needs input separated by newline instead of space, so the very simple pipeline above simply does:
Echo array content --> replace space by newline --> sort
$() is to echo the result
($()) is to put the "echoed result" in an array
Note: as #sorontar mentioned in a comment to a different question:
The sorted=($(...)) part is using the "split and glob" operator. You should turn glob off: set -f or set -o noglob or shopt -op noglob or an element of the array like * will be expanded to a list of files.

How does bash array slicing work when start index is not provided?

I'm looking at a script, and I'm having trouble determining what is going on.
Here is an example:
# Command to get the last 4 occurrences of a pattern in a file
lsCommand="ls /my/directory | grep -i my_pattern | tail -4"
# Load the results of that command into an array
dirArray=($(echo $(eval $lsCommand) | tr ' ' '\n'))
# What is this doing?
yesterdaysFileArray=($(echo ${x[#]::$((${#x[#]} / 2))} | tr ' ' '\n'))
There is a lot going on here. I understand how arrays work, but I don't know how $x is getting referenced if it was never declared.
I see that the $((${#x[#]} / 2}} is taking the number of elements and dividing it in half, and the tr is used to create the array. But what else is going on?
I think the last line is an array slice pattern in bash of form ${array[#]:1:2}, where array[#] returns the contents of the array, :1:2 takes a slice of length 2, starting at index 1.
So for your case though you are taking the start index empty because you haven't specified any and length as half the count of array.
But there is a lot better way to do this in bash as below. Don't use eval and use the built-in globbing support from the shell itself
cd /my/directory
fileArray=()
for file in *my_pattern*; do
[[ -f "$file" ]] || { printf '%s\n' 'no file found'; return 1; }
fileArray+=( "$file" )
done
and do
printf '%s\n' "${fileArray[#]::${#fileArray[#]}/2}"

Finding elements in common between two ksh or bash arrays efficiently

I am writing a Korn shell script. I have two arrays (say, arr1 and arr2), both containing strings, and I need to check which elements from arr1 are present (as whole strings or substrings) in arr2. The most intuitive solution is having nested for loops, and checking if each element from arr1 can be found in arr2 (through grep) like this:
for arr1Element in ${arr1[*]}; do
for arr2Element in ${arr2[*]}; do
# using grep to check if arr1Element is present in arr2Element
echo $arr2Element | grep $arr1Element
done
done
The issue is that arr2 has around 3000 elements, so running a nested loop takes a long time. I am wondering if there is a better way to do this in Bash.
If I were doing this in Java, I could have calculated hashes for elements in one of the arrays, and then looked for those hashes in the other array, but I don't think Bash has any functionality for doing something like this (unless I was willing to write a hash calculating function in Bash).
Any suggestions?
Since version 4.0 Bash has associative arrays:
$ declare -A elements
$ elements[hello]=world
$ echo ${elements[hello]}
world
You can use this in the same way you would a Java Map.
declare -A map
for el in "${arr1[#]}"; do
map[$el]="x"
done
for el in "${arr2[#]}"; do
if [ -n "${map[$el]}" ] ; then
echo "${el}"
fi
done
Dealing with substrings is an altogether more weighty problem, and would be a challenge in any language, short of the brute-force algorithm you're already using. You could build a binary-tree index of character sequences, but I wouldn't try that in Bash!
BashFAQ #36 describes doing set arithmetic (unions, disjoint sets, etc) in bash with comm.
Assuming your values can't contain literal newlines, the following will emit a line per item in both arr1 and arr2:
comm -12 <(printf '%s\n' "${arr1[#]}" | sort -u) \
<(printf '%s\n' "${arr2[#]}" | sort -u)
If your arrays are pre-sorted, you can remove the sorts (which will make this extremely memory- and time-efficient with large arrays, moreso than the grep-based approach).
Since you're OK with using grep, and since you want to match substrings as well as full strings, one approach is to write:
printf '%s\n' "${arr2[#]}" \
| grep -o -F "$(printf '%s\n' "${arr1[#]}")
and let grep optimize as it sees fit.
Here's a bash/awk idea:
# some sample arrays
$ arr1=( my first string "hello wolrd")
$ arr2=( my last stringbean strings "well, hello world!)
# break array elements into separate lines
$ printf '%s\n' "${arr1[#]}"
my
first
string
hello world
$ printf '%s\n' "${arr2[#]}"
my
last
stringbean
strings
well, hello world!
# use the 'printf' command output as input to our awk command
$ awk '
NR==FNR { a[NR]=$0 ; next }
{ for (i in a)
if ($0 ~ a[i]) print "array1 string {"a[i]"} is a substring of array2 string {"$0"}" }
' <( printf '%s\n' "${arr1[#]}" ) \
<( printf '%s\n' "${arr2[#]}" )
array1 string {my} is a substring of array2 string {my}
array1 string {string} is a substring of array2 string {stringbean}
array1 string {string} is a substring of array2 string {strings}
array1 string {hello world} is a substring of array2 string {well, hello world!}
NR==FNR : for file #1 only: store elements into awk array named 'a'
next : process next line in file #1; at this point rest of awk script is ignored for file #1; the for each line in file #2 ...
for (i in a) : for each index 'i' in array 'a' ...
if ($0 ~ a[i] ) : see if a[i] is a substring of the current line ($0) from file #2 and if so ...
print "array1... : output info about the match
A test run using the following data:
arr1 == 3300 elements
arr2 == 500 elements
When all arr2 elements have a substring/pattern match in arr1 (ie, 500 matches), total time to run is ~27 seconds ... so the repetitive looping through the array takes a toll.
Obviously (?) need to reduce the volume of repetitive actions ...
for an exact string match the comm solution by Charles Duffy makes sense (it runs against the same 3300/500 test set in about 0.5 seconds)
for a substring/pattern match I was able to get a egrep solution to run in about 5 seconds (see my other answer/post)
An egrep solution for substring/pattern matching ...
egrep -f <(printf '.*%s.*\n' "${arr1[#]}") \
<(printf '%s\n' "${arr2[#]}")
egrep -f : take patterns to search from the file designated by the -f, which in this case is ...
<(printf '.*%s.*\n' "${arr1[#]}") : convert arr1 elements into 1 pattern per line, appending a regex wild card character (.*) for prefix and suffix
<(printf '%s\n' "${arr2[#]}") : convert arr2 elements into 1 string per line
When run against a sample data set like:
arr1 == 3300 elements
arr2 == 500 elements
... with 500 matches, total run time is ~5 seconds; there's still a good bit of repetitive processing going on with egrep but not as bad as seen with my other answer (bash/awk) ... and of course not as fast the comm solution which eliminates the repetitive processing.

Shell Array Cleared for Unknown Reason [duplicate]

This question already has answers here:
A variable modified inside a while loop is not remembered
(8 answers)
Closed 6 years ago.
I have a pretty simple sh script where I make a system cat call, collect the results and parse some relevant information before storing the information in an array, which seems to work just fine. But as soon as I exit the for loop where I store the information, the array seems to clear itself. I'm wondering if I am accessing the array incorrectly outside of the for loop. Relevant portion of my script:
#!/bin/sh
declare -a QSPI_ARRAY=()
cat /proc/mtd | while read mtd_instance
do
# split result into individiual words
words=($mtd_instance)
for word in "${words[#]}"
do
# check for uboot
if [[ $word == *"uboot"* ]]
then
mtd_num=${words[0]}
index=${mtd_num//[!0-9]/} # strip everything except the integers
QSPI_ARRAY[$index]="uboot"
echo "QSPI_ARRAY[] at index $index: ${QSPI_ARRAY[$index]}"
elif [[ $word == *"fpga_a"* ]]
then
echo "found it: "$word""
mtd_num=${words[0]}
index=${mtd_num//[!0-9]/} # strip everything except the integers
QSPI_ARRAY[$index]="fpga_a"
echo "QSPI_ARRAY[] at index $index: ${QSPI_ARRAY[$index]}"
# other items are added to the array, all successfully
fi
done
echo "length of array: ${#QSPI_ARRAY[#]}"
echo "----------------------"
done
My output is great until I exit the for loop. While within the for loop, the array size increments and I can check that the item has been added. After the for loop is complete I check the array like so:
echo "RESULTING ARRAY:"
echo "length of array: ${#QSPI_ARRAY[#]}"
for qspi in "${QSPI_ARRAY}"
do
echo "qspi instance: $qspi"
done
Here are my results, echod to my display:
dev: size erasesize name
length of array: 0
-------------
mtd0: 00100000 00001000 "qspi-fsbl-uboot"
QSPI_ARRAY[] at index 0: uboot
length of array: 1
-------------
mtd1: 00500000 00001000 "qspi-fpga_a"
QSPI_ARRAY[] at index 1: fpga_a
length of array: 2
-------------
RESULTING ARRAY:
length of array: 0
qspi instance:
EDIT: After some debugging, it seems I have two different arrays here somehow. I initialized the array like so: QSPI_ARRAY=("a" "b" "c" "d" "e" "f" "g"), and after my for-loop for parsing the array it is still a, b, c, etc. How do I have two different arrays of the same name here?
This structure:
cat /proc/mtd | while read mtd_instance
do
...
done
Means that whatever comes between do and done cannot have any effects inside the shell environment that are still there after the done.
The fact that the while loop is on the right hand side of a pipe (|) means that it runs in a subshell. Once the loop exits, so does the subshell. And all of its variable settings.
If you want a while loop which makes changes that stick around, don't use a pipe. Input redirection doesn't create a subshell, and in this case, you can just read from the file directly:
while read mtd_instance
do
...
done </proc/mtd
If you had a more complicated command than a cat, you might need to use process substitution. Still using cat as an example, that looks like this:
while read mtd_instance
do
...
done < <(cat /proc/mtd)
In the specific case of your example code, I think you could simplify it somewhat, perhaps like this:
#!/usr/bin/env bash
QSPI_ARRAY=()
while read -a words; do␣
declare -i mtd_num=${words[0]//[!0-9]/}
for word in "${words[#]}"; do
for type in uboot fpga_a; do
if [[ $word == *$type* ]]; then
QSPI_ARRAY[mtd_num]=$type
break 2
fi
done
done
done </proc/mtd
Is this potentially what you are seeing:
http://mywiki.wooledge.org/BashFAQ/024

How to sort an array in Bash

I have an array in Bash, for example:
array=(a c b f 3 5)
I need to sort the array. Not just displaying the content in a sorted way, but to get a new array with the sorted elements. The new sorted array can be a completely new one or the old one.
You don't really need all that much code:
IFS=$'\n' sorted=($(sort <<<"${array[*]}"))
unset IFS
Supports whitespace in elements (as long as it's not a newline), and works in Bash 3.x.
e.g.:
$ array=("a c" b f "3 5")
$ IFS=$'\n' sorted=($(sort <<<"${array[*]}")); unset IFS
$ printf "[%s]\n" "${sorted[#]}"
[3 5]
[a c]
[b]
[f]
Note: #sorontar has pointed out that care is required if elements contain wildcards such as * or ?:
The sorted=($(...)) part is using the "split and glob" operator. You should turn glob off: set -f or set -o noglob or shopt -op noglob or an element of the array like * will be expanded to a list of files.
What's happening:
The result is a culmination six things that happen in this order:
IFS=$'\n'
"${array[*]}"
<<<
sort
sorted=($(...))
unset IFS
First, the IFS=$'\n'
This is an important part of our operation that affects the outcome of 2 and 5 in the following way:
Given:
"${array[*]}" expands to every element delimited by the first character of IFS
sorted=() creates elements by splitting on every character of IFS
IFS=$'\n' sets things up so that elements are expanded using a new line as the delimiter, and then later created in a way that each line becomes an element. (i.e. Splitting on a new line.)
Delimiting by a new line is important because that's how sort operates (sorting per line). Splitting by only a new line is not-as-important, but is needed preserve elements that contain spaces or tabs.
The default value of IFS is a space, a tab, followed by a new line, and would be unfit for our operation.
Next, the sort <<<"${array[*]}" part
<<<, called here strings, takes the expansion of "${array[*]}", as explained above, and feeds it into the standard input of sort.
With our example, sort is fed this following string:
a c
b
f
3 5
Since sort sorts, it produces:
3 5
a c
b
f
Next, the sorted=($(...)) part
The $(...) part, called command substitution, causes its content (sort <<<"${array[*]}) to run as a normal command, while taking the resulting standard output as the literal that goes where ever $(...) was.
In our example, this produces something similar to simply writing:
sorted=(3 5
a c
b
f
)
sorted then becomes an array that's created by splitting this literal on every new line.
Finally, the unset IFS
This resets the value of IFS to the default value, and is just good practice.
It's to ensure we don't cause trouble with anything that relies on IFS later in our script. (Otherwise we'd need to remember that we've switched things around--something that might be impractical for complex scripts.)
Original response:
array=(a c b "f f" 3 5)
readarray -t sorted < <(for a in "${array[#]}"; do echo "$a"; done | sort)
output:
$ for a in "${sorted[#]}"; do echo "$a"; done
3
5
a
b
c
f f
Note this version copes with values that contains special characters or whitespace (except newlines)
Note readarray is supported in bash 4+.
Edit Based on the suggestion by #Dimitre I had updated it to:
readarray -t sorted < <(printf '%s\0' "${array[#]}" | sort -z | xargs -0n1)
which has the benefit of even understanding sorting elements with newline characters embedded correctly. Unfortunately, as correctly signaled by #ruakh this didn't mean the the result of readarray would be correct, because readarray has no option to use NUL instead of regular newlines as line-separators.
If you don't need to handle special shell characters in the array elements:
array=(a c b f 3 5)
sorted=($(printf '%s\n' "${array[#]}"|sort))
With bash you'll need an external sorting program anyway.
With zsh no external programs are needed and special shell characters are easily handled:
% array=('a a' c b f 3 5); printf '%s\n' "${(o)array[#]}"
3
5
a a
b
c
f
ksh has set -s to sort ASCIIbetically.
Here's a pure Bash quicksort implementation:
#!/bin/bash
# quicksorts positional arguments
# return is in array qsort_ret
qsort() {
local pivot i smaller=() larger=()
qsort_ret=()
(($#==0)) && return 0
pivot=$1
shift
for i; do
# This sorts strings lexicographically.
if [[ $i < $pivot ]]; then
smaller+=( "$i" )
else
larger+=( "$i" )
fi
done
qsort "${smaller[#]}"
smaller=( "${qsort_ret[#]}" )
qsort "${larger[#]}"
larger=( "${qsort_ret[#]}" )
qsort_ret=( "${smaller[#]}" "$pivot" "${larger[#]}" )
}
Use as, e.g.,
$ array=(a c b f 3 5)
$ qsort "${array[#]}"
$ declare -p qsort_ret
declare -a qsort_ret='([0]="3" [1]="5" [2]="a" [3]="b" [4]="c" [5]="f")'
This implementation is recursive… so here's an iterative quicksort:
#!/bin/bash
# quicksorts positional arguments
# return is in array qsort_ret
# Note: iterative, NOT recursive! :)
qsort() {
(($#==0)) && return 0
local stack=( 0 $(($#-1)) ) beg end i pivot smaller larger
qsort_ret=("$#")
while ((${#stack[#]})); do
beg=${stack[0]}
end=${stack[1]}
stack=( "${stack[#]:2}" )
smaller=() larger=()
pivot=${qsort_ret[beg]}
for ((i=beg+1;i<=end;++i)); do
if [[ "${qsort_ret[i]}" < "$pivot" ]]; then
smaller+=( "${qsort_ret[i]}" )
else
larger+=( "${qsort_ret[i]}" )
fi
done
qsort_ret=( "${qsort_ret[#]:0:beg}" "${smaller[#]}" "$pivot" "${larger[#]}" "${qsort_ret[#]:end+1}" )
if ((${#smaller[#]}>=2)); then stack+=( "$beg" "$((beg+${#smaller[#]}-1))" ); fi
if ((${#larger[#]}>=2)); then stack+=( "$((end-${#larger[#]}+1))" "$end" ); fi
done
}
In both cases, you can change the order you use: I used string comparisons, but you can use arithmetic comparisons, compare wrt file modification time, etc. just use the appropriate test; you can even make it more generic and have it use a first argument that is the test function use, e.g.,
#!/bin/bash
# quicksorts positional arguments
# return is in array qsort_ret
# Note: iterative, NOT recursive! :)
# First argument is a function name that takes two arguments and compares them
qsort() {
(($#<=1)) && return 0
local compare_fun=$1
shift
local stack=( 0 $(($#-1)) ) beg end i pivot smaller larger
qsort_ret=("$#")
while ((${#stack[#]})); do
beg=${stack[0]}
end=${stack[1]}
stack=( "${stack[#]:2}" )
smaller=() larger=()
pivot=${qsort_ret[beg]}
for ((i=beg+1;i<=end;++i)); do
if "$compare_fun" "${qsort_ret[i]}" "$pivot"; then
smaller+=( "${qsort_ret[i]}" )
else
larger+=( "${qsort_ret[i]}" )
fi
done
qsort_ret=( "${qsort_ret[#]:0:beg}" "${smaller[#]}" "$pivot" "${larger[#]}" "${qsort_ret[#]:end+1}" )
if ((${#smaller[#]}>=2)); then stack+=( "$beg" "$((beg+${#smaller[#]}-1))" ); fi
if ((${#larger[#]}>=2)); then stack+=( "$((end-${#larger[#]}+1))" "$end" ); fi
done
}
Then you can have this comparison function:
compare_mtime() { [[ $1 -nt $2 ]]; }
and use:
$ qsort compare_mtime *
$ declare -p qsort_ret
to have the files in current folder sorted by modification time (newest first).
NOTE. These functions are pure Bash! no external utilities, and no subshells! they are safe wrt any funny symbols you may have (spaces, newline characters, glob characters, etc.).
NOTE2. The test [[ $i < $pivot ]] is correct. It uses the lexicographical string comparison. If your array only contains integers and you want to sort numerically, use ((i < pivot)) instead.
Please don't edit this answer to change that. It has already been edited (and rolled back) a couple of times. The test I gave here is correct and corresponds to the output given in the example: the example uses both strings and numbers, and the purpose is to sort it in lexicographical order. Using ((i < pivot)) in this case is wrong.
tl;dr:
Sort array a_in and store the result in a_out (elements must not have embedded newlines[1]
):
Bash v4+:
readarray -t a_out < <(printf '%s\n' "${a_in[#]}" | sort)
Bash v3:
IFS=$'\n' read -d '' -r -a a_out < <(printf '%s\n' "${a_in[#]}" | sort)
Advantages over antak's solution:
You needn't worry about accidental globbing (accidental interpretation of the array elements as filename patterns), so no extra command is needed to disable globbing (set -f, and set +f to restore it later).
You needn't worry about resetting IFS with unset IFS.[2]
Optional reading: explanation and sample code
The above combines Bash code with external utility sort for a solution that works with arbitrary single-line elements and either lexical or numerical sorting (optionally by field):
Performance: For around 20 elements or more, this will be faster than a pure Bash solution - significantly and increasingly so once you get beyond around 100 elements.
(The exact thresholds will depend on your specific input, machine, and platform.)
The reason it is fast is that it avoids Bash loops.
printf '%s\n' "${a_in[#]}" | sort performs the sorting (lexically, by default - see sort's POSIX spec):
"${a_in[#]}" safely expands to the elements of array a_in as individual arguments, whatever they contain (including whitespace).
printf '%s\n' then prints each argument - i.e., each array element - on its own line, as-is.
Note the use of a process substitution (<(...)) to provide the sorted output as input to read / readarray (via redirection to stdin, <), because read / readarray must run in the current shell (must not run in a subshell) in order for output variable a_out to be visible to the current shell (for the variable to remain defined in the remainder of the script).
Reading sort's output into an array variable:
Bash v4+: readarray -t a_out reads the individual lines output by sort into the elements of array variable a_out, without including the trailing \n in each element (-t).
Bash v3: readarray doesn't exist, so read must be used:
IFS=$'\n' read -d '' -r -a a_out tells read to read into array (-a) variable a_out, reading the entire input, across lines (-d ''), but splitting it into array elements by newlines (IFS=$'\n'. $'\n', which produces a literal newline (LF), is a so-called ANSI C-quoted string).
(-r, an option that should virtually always be used with read, disables unexpected handling of \ characters.)
Annotated sample code:
#!/usr/bin/env bash
# Define input array `a_in`:
# Note the element with embedded whitespace ('a c')and the element that looks like
# a glob ('*'), chosen to demonstrate that elements with line-internal whitespace
# and glob-like contents are correctly preserved.
a_in=( 'a c' b f 5 '*' 10 )
# Sort and store output in array `a_out`
# Saving back into `a_in` is also an option.
IFS=$'\n' read -d '' -r -a a_out < <(printf '%s\n' "${a_in[#]}" | sort)
# Bash 4.x: use the simpler `readarray -t`:
# readarray -t a_out < <(printf '%s\n' "${a_in[#]}" | sort)
# Print sorted output array, line by line:
printf '%s\n' "${a_out[#]}"
Due to use of sort without options, this yields lexical sorting (digits sort before letters, and digit sequences are treated lexically, not as numbers):
*
10
5
a c
b
f
If you wanted numerical sorting by the 1st field, you'd use sort -k1,1n instead of just sort, which yields (non-numbers sort before numbers, and numbers sort correctly):
*
a c
b
f
5
10
[1] To handle elements with embedded newlines, use the following variant (Bash v4+, with GNU sort):
readarray -d '' -t a_out < <(printf '%s\0' "${a_in[#]}" | sort -z).
Michał Górny's helpful answer has a Bash v3 solution.
[2] While IFS is set in the Bash v3 variant, the change is scoped to the command.
By contrast, what follows IFS=$'\n'  in antak's answer is an assignment rather than a command, in which case the IFS change is global.
In the 3-hour train trip from Munich to Frankfurt (which I had trouble to reach because Oktoberfest starts tomorrow) I was thinking about my first post. Employing a global array is a much better idea for a general sort function. The following function handles arbitary strings (newlines, blanks etc.):
declare BSORT=()
function bubble_sort()
{ #
# #param [ARGUMENTS]...
#
# Sort all positional arguments and store them in global array BSORT.
# Without arguments sort this array. Return the number of iterations made.
#
# Bubble sorting lets the heaviest element sink to the bottom.
#
(($# > 0)) && BSORT=("$#")
local j=0 ubound=$((${#BSORT[*]} - 1))
while ((ubound > 0))
do
local i=0
while ((i < ubound))
do
if [ "${BSORT[$i]}" \> "${BSORT[$((i + 1))]}" ]
then
local t="${BSORT[$i]}"
BSORT[$i]="${BSORT[$((i + 1))]}"
BSORT[$((i + 1))]="$t"
fi
((++i))
done
((++j))
((--ubound))
done
echo $j
}
bubble_sort a c b 'z y' 3 5
echo ${BSORT[#]}
This prints:
3 5 a b c z y
The same output is created from
BSORT=(a c b 'z y' 3 5)
bubble_sort
echo ${BSORT[#]}
Note that probably Bash internally uses smart-pointers, so the swap-operation could be cheap (although I doubt it). However, bubble_sort demonstrates that more advanced functions like merge_sort are also in the reach of the shell language.
Another solution that uses external sort and copes with any special characters (except for NULs :)). Should work with bash-3.2 and GNU or BSD sort (sadly, POSIX doesn't include -z).
local e new_array=()
while IFS= read -r -d '' e; do
new_array+=( "${e}" )
done < <(printf "%s\0" "${array[#]}" | LC_ALL=C sort -z)
First look at the input redirection at the end. We're using printf built-in to write out the array elements, zero-terminated. The quoting makes sure array elements are passed as-is, and specifics of shell printf cause it to reuse the last part of format string for each remaining parameter. That is, it's equivalent to something like:
for e in "${array[#]}"; do
printf "%s\0" "${e}"
done
The null-terminated element list is then passed to sort. The -z option causes it to read null-terminated elements, sort them and output null-terminated as well. If you needed to get only the unique elements, you can pass -u since it is more portable than uniq -z. The LC_ALL=C ensures stable sort order independently of locale — sometimes useful for scripts. If you want the sort to respect locale, remove that.
The <() construct obtains the descriptor to read from the spawned pipeline, and < redirects the standard input of the while loop to it. If you need to access the standard input inside the pipe, you may use another descriptor — exercise for the reader :).
Now, back to the beginning. The read built-in reads output from the redirected stdin. Setting empty IFS disables word splitting which is unnecessary here — as a result, read reads the whole 'line' of input to the single provided variable. -r option disables escape processing that is undesired here as well. Finally, -d '' sets the line delimiter to NUL — that is, tells read to read zero-terminated strings.
As a result, the loop is executed once for every successive zero-terminated array element, with the value being stored in e. The example just puts the items in another array but you may prefer to process them directly :).
Of course, that's just one of the many ways of achieving the same goal. As I see it, it is simpler than implementing complete sorting algorithm in bash and in some cases it will be faster. It handles all special characters including newlines and should work on most of the common systems. Most importantly, it may teach you something new and awesome about bash :).
Keep it simple ;)
In the following example, the array b is the sorted version of the array a!
The second line echos each item of the array a, then pipes them to the sort command, and the output is used to initiate the array b.
a=(2 3 1)
b=( $( for x in ${a[#]}; do echo $x; done | sort ) )
echo ${b[#]} # output: 1 2 3
min sort:
#!/bin/bash
array=(.....)
index_of_element1=0
while (( ${index_of_element1} < ${#array[#]} )); do
element_1="${array[${index_of_element1}]}"
index_of_element2=$((index_of_element1 + 1))
index_of_min=${index_of_element1}
min_element="${element_1}"
for element_2 in "${array[#]:$((index_of_element1 + 1))}"; do
min_element="`printf "%s\n%s" "${min_element}" "${element_2}" | sort | head -n+1`"
if [[ "${min_element}" == "${element_2}" ]]; then
index_of_min=${index_of_element2}
fi
let index_of_element2++
done
array[${index_of_element1}]="${min_element}"
array[${index_of_min}]="${element_1}"
let index_of_element1++
done
try this:
echo ${array[#]} | awk 'BEGIN{RS=" ";} {print $1}' | sort
Output will be:
3
5
a
b
c
f
Problem solved.
If you can compute a unique integer for each element in the array, like this:
tab='0123456789abcdefghijklmnopqrstuvwxyz'
# build the reversed ordinal map
for ((i = 0; i < ${#tab}; i++)); do
declare -g ord_${tab:i:1}=$i
done
function sexy_int() {
local sum=0
local i ch ref
for ((i = 0; i < ${#1}; i++)); do
ch="${1:i:1}"
ref="ord_$ch"
(( sum += ${!ref} ))
done
return $sum
}
sexy_int hello
echo "hello -> $?"
sexy_int world
echo "world -> $?"
then, you can use these integers as array indexes, because Bash always use sparse array, so no need to worry about unused indexes:
array=(a c b f 3 5)
for el in "${array[#]}"; do
sexy_int "$el"
sorted[$?]="$el"
done
echo "${sorted[#]}"
Pros. Fast.
Cons. Duplicated elements are merged, and it can be impossible to map contents to 32-bit unique integers.
array=(a c b f 3 5)
new_array=($(echo "${array[#]}" | sed 's/ /\n/g' | sort))
echo ${new_array[#]}
echo contents of new_array will be:
3 5 a b c f
There is a workaround for the usual problem of spaces and newlines:
Use a character that is not in the original array (like $'\1' or $'\4' or similar).
This function gets the job done:
# Sort an Array may have spaces or newlines with a workaround (wa=$'\4')
sortarray(){ local wa=$'\4' IFS=''
if [[ $* =~ [$wa] ]]; then
echo "$0: error: array contains the workaround char" >&2
exit 1
fi
set -f; local IFS=$'\n' x nl=$'\n'
set -- $(printf '%s\n' "${#//$nl/$wa}" | sort -n)
for x
do sorted+=("${x//$wa/$nl}")
done
}
This will sort the array:
$ array=( a b 'c d' $'e\nf' $'g\1h')
$ sortarray "${array[#]}"
$ printf '<%s>\n' "${sorted[#]}"
<a>
<b>
<c d>
<e
f>
<gh>
This will complain that the source array contains the workaround character:
$ array=( a b 'c d' $'e\nf' $'g\4h')
$ sortarray "${array[#]}"
./script: error: array contains the workaround char
description
We set two local variables wa (workaround char) and a null IFS
Then (with ifs null) we test that the whole array $*.
Does not contain any woraround char [[ $* =~ [$wa] ]].
If it does, raise a message and signal an error: exit 1
Avoid filename expansions: set -f
Set a new value of IFS (IFS=$'\n') a loop variable x and a newline var (nl=$'\n').
We print all values of the arguments received (the input array $#).
but we replace any new line by the workaround char "${#//$nl/$wa}".
send those values to be sorted sort -n.
and place back all the sorted values in the positional arguments set --.
Then we assign each argument one by one (to preserve newlines).
in a loop for x
to a new array: sorted+=(…)
inside quotes to preserve any existing newline.
restoring the workaround to a newline "${x//$wa/$nl}".
done
This question looks closely related. And BTW, here's a mergesort in Bash (without external processes):
mergesort() {
local -n -r input_reference="$1"
local -n output_reference="$2"
local -r -i size="${#input_reference[#]}"
local merge previous
local -a -i runs indices
local -i index previous_idx merged_idx \
run_a_idx run_a_stop \
run_b_idx run_b_stop
output_reference=("${input_reference[#]}")
if ((size == 0)); then return; fi
previous="${output_reference[0]}"
runs=(0)
for ((index = 0;;)) do
for ((++index;; ++index)); do
if ((index >= size)); then break 2; fi
if [[ "${output_reference[index]}" < "$previous" ]]; then break; fi
previous="${output_reference[index]}"
done
previous="${output_reference[index]}"
runs+=(index)
done
runs+=(size)
while (("${#runs[#]}" > 2)); do
indices=("${!runs[#]}")
merge=("${output_reference[#]}")
for ((index = 0; index < "${#indices[#]}" - 2; index += 2)); do
merged_idx=runs[indices[index]]
run_a_idx=merged_idx
previous_idx=indices[$((index + 1))]
run_a_stop=runs[previous_idx]
run_b_idx=runs[previous_idx]
run_b_stop=runs[indices[$((index + 2))]]
unset runs[previous_idx]
while ((run_a_idx < run_a_stop && run_b_idx < run_b_stop)); do
if [[ "${merge[run_a_idx]}" < "${merge[run_b_idx]}" ]]; then
output_reference[merged_idx++]="${merge[run_a_idx++]}"
else
output_reference[merged_idx++]="${merge[run_b_idx++]}"
fi
done
while ((run_a_idx < run_a_stop)); do
output_reference[merged_idx++]="${merge[run_a_idx++]}"
done
while ((run_b_idx < run_b_stop)); do
output_reference[merged_idx++]="${merge[run_b_idx++]}"
done
done
done
}
declare -ar input=({z..a}{z..a})
declare -a output
mergesort input output
echo "${input[#]}"
echo "${output[#]}"
Many thanks to the people that answered before me. Using their excellent input, bash documentation and ideas from other treads, this is what works perfectly for me without IFS change
array=("a \n c" b f "3 5")
Using process substitution and read array in bash > v4.4 WITH EOL character
readarray -t sorted < <(sort < <(printf '%s\n' "${array[#]}"))
Using process substitution and read array in bash > v4.4 WITH NULL character
readarray -td '' sorted < <(sort -z < <(printf '%s\0' "${array[#]}"))
Finally we verify with
printf "[%s]\n" "${sorted[#]}"
output is
[3 5]
[a \n c]
[b]
[f]
Please, let me know if that is a correct test for embedded /n as both solutions produce the same result, but the first one is not supposed to work properly with embedded /n
I am not convinced that you'll need an external sorting program in Bash.
Here is my implementation for the simple bubble-sort algorithm.
function bubble_sort()
{ #
# Sorts all positional arguments and echoes them back.
#
# Bubble sorting lets the heaviest (longest) element sink to the bottom.
#
local array=($#) max=$(($# - 1))
while ((max > 0))
do
local i=0
while ((i < max))
do
if [ ${array[$i]} \> ${array[$((i + 1))]} ]
then
local t=${array[$i]}
array[$i]=${array[$((i + 1))]}
array[$((i + 1))]=$t
fi
((i += 1))
done
((max -= 1))
done
echo ${array[#]}
}
array=(a c b f 3 5)
echo " input: ${array[#]}"
echo "output: $(bubble_sort ${array[#]})"
This shall print:
input: a c b f 3 5
output: 3 5 a b c f
a=(e b 'c d')
shuf -e "${a[#]}" | sort >/tmp/f
mapfile -t g </tmp/f
Great answers here. Learned a lot. After reading them all, I figure I'd throw my hat into the ring. I think this is the shortest method (and probably faster as it doesn't do much shell script parsing, though there is the matter of the spawning of printf and sort, but they're only called once each) and handles whitespace in the data:
a=(3 "2 a" 1) # Setup!
IFS=$'\n' b=( $(printf "%s\n" "${a[#]}" | sort) ); unset IFS # Sort!
printf "'%s' " "${b[#]}"; # Success!
Outputs:
'1' '2 a' '3'
Note that the IFS change is limited in scope to the line it is on. if you know that the array has no whitespace in it, you don't need the IFS modification.
Inspiration was from #yas's answer and #Alcamtar comments.
EDIT
Oh, I somehow missed the actually accepted answer which is even shorter than mine. Doh!
IFS=$'\n' sorted=($(sort <<<"${array[*]}")); unset IFS
Turns out that the unset is required because this is a variable assignment that has no command.
I'd recommend going to that answer because it has some interesting stuff on globbing which could be relevant if the array has wildcards in it. It also has a detailed description as to what is happening.
EDIT 2
GNU has an extension in which sort delimits records using \0 which is good if you have LFs in your data. However, when it gets returned to the shell to be assign to an array, I don't see a good way convert it so that the shell will delimit on \0, because even setting IFS=$'\0', the shell doesn't like it and doesn't properly break it up.
array=(z 'b c'); { set "${array[#]}"; printf '%s\n' "$#"; } \
| sort \
| mapfile -t array; declare -p array
declare -a array=([0]="b c" [1]="z")
Open an inline function {...} to get a fresh set of positional arguments (e.g. $1, $2, etc).
Copy the array to the positional arguments. (e.g. set "${array[#]}" will copy the nth array argument to the nth positional argument. Note the quotes preserve whitespace that may be contained in an array element).
Print each positional argument (e.g. printf '%s\n' "$#" will print each positional argument on its own line. Again, note the quotes preserve whitespace that may be contained in each positional argument).
Then sort does its thing.
Read the stream into an array with mapfile (e.g. mapfile -t array reads each line into the variable array and the -t ignores the \n in each line).
Dump the array to show its been sorted.
As a function:
set +m
shopt -s lastpipe
sort_array() {
declare -n ref=$1
set "${ref[#]}"
printf '%s\n' "$#"
| sort \
| mapfile -t $ref
}
then
array=(z y x); sort_array array; declare -p array
declare -a array=([0]="x" [1]="y" [2]="z")
I look forward to being ripped apart by all the UNIX gurus! :)
sorted=($(echo ${array[#]} | tr " " "\n" | sort))
In the spirit of bash / linux, I would pipe the best command-line tool for each step. sort does the main job but needs input separated by newline instead of space, so the very simple pipeline above simply does:
Echo array content --> replace space by newline --> sort
$() is to echo the result
($()) is to put the "echoed result" in an array
Note: as #sorontar mentioned in a comment to a different question:
The sorted=($(...)) part is using the "split and glob" operator. You should turn glob off: set -f or set -o noglob or shopt -op noglob or an element of the array like * will be expanded to a list of files.

Resources