I'm trying to use associative arrays as a work around for Bash's poor function parameter passing. I can declare a global associative array and read/write to that but I would like to have the variable name passed to the function since many times I want to use the same function with different parameter blocks.
Various Stack Overflow posts have approaches for reading a passed array within a function but not writing to it to allow return values. Pseudo Bash for what I'm trying to do is thus:
TestFunc() {
local __PARMBLOCK__=${1} # Tried ${!1} as well
# Do something with incoming array
__PARMBLOCK__[__rc__]+=1 # Error occured
__PARMBLOCK__[__error__]+="Error in TestFunc"
}
declare -A FUNCPARM
# Populate FUNCPARM
TestFunc FUNCPARM
if [[ ${FUNCPARM[__rc__]} -ne 0 ]]; then
echo "ERROR : ${FUNCPARM[__error__]}
fi
Is this kind of thing possible or do I really need to abandon Bash for something like Python?
EDIT: Found the duplicate. This is basically the same answer as this one.
You can use a reference variable for that, see help declare:
declare [-aAfFgilnrtux] [-p] [name[=value] ...]
[...]
-n make NAME a reference to the variable named by its value
[...]
When used in a function, declare makes NAMEs local, as with the local command.
f() {
declare -n paramblock="$1"
# example for reading (print all keys and entries)
paste <(printf %s\\n "${!paramblock[#]}") <(printf %s\\n "${paramblock[#]}")
# example for writing
paramblock["key 1"]="changed"
paramblock["new key"]="new output"
}
Example usage:
$ declare -A a=(["key 1"]="input 1" ["key 2"]="input 2")
$ f a
key 2 input 2
key 1 input 1
$ declare -p a
declare -A a=(["key 2"]="input 2" ["key 1"]="changed" ["new key"]="new output" )
This works very well. The only difference to an actual associative array I found so far is, that you cannot print the referenced array using declare -p as that will only show the reference.
Related
How to return assciative arrays from zsh function?
I tried:
creatAARR() {
declare -A AARR=([k1]=2 [k2]=4)
return $AARR
}
creatAARR
But i get error:
creatAARR:return:2: too many arguments
What is the right way?
EDIT:
I captured output to standard output, like how #chepner suggests, but the new variable doesn't seem to behave like an associative array:
creatAARR() {
declare -A AARR=([k1]=2 [k2]=4)
echo "$AARR"
}
declare -A VALL
NEW_ARR=$(creatAARR)
echo "$NEW_ARR" # 2 4
echo "k1: $NEW_ARR[k1]" # prints just k1:
return
Any suggestions?
return accepts only an integer and sets the exit status of the function.
Shell commands cannot actually return values. If you want pass information to the caller of your function, you have a couple of options available to you:
You could print your return value, but this then relies on you to properly format your output and for the caller to correctly parse it. For associative arrays, there are so many ways that this can go wrong; I wouldn’t recommend doing this.
In Zsh, there is a convention that, to communicate a return value, a function can set $REPLY to a scalar value or $reply to an array. Unfortunately, there is no convention for passing associative arrays. You could, of course, put your key-value pairs simply as elements in the non-associative array $reply and then let the caller cast it to or wrap it in an associative array, but this would break the convention and thus might violate your caller's expectations.
The, in my opinion, best approach is to let the caller specify the name of an associative array, which you can then populate with values. This is also handy when you want to return multiple values of any type, since you can let the caller specify multiple variable names.
This last approach you can use as follows:
% creatAARR() {
# Restrict $name to function scope.
local name=$1
# Delete $1, so $# becomes the other args.
shift
# Assign elements to array.
set -A "$name" "$#"
}
% typeset -A AARR=() # Declare assoc. array
% creatAARR AARR k1 2 k2 4
% typeset -p1 AARR # Print details
typeset -A AARR=(
[k1]=2
[k2]=4
)
I've been able to get my requirement with this:
declare -A VAR1
declare -A VAR2
TYPE_VAR1="TYPE_VAR1"
TYPE_VAR2="TYPE_VAR2"
creatAARR() {
declare -A AARR=([k1]=2 [k2]=4)
case $1 in
"$TYPE_VAR1")
set -A VAR1 ${(kv)AARR}
;;
"$TYPE_VAR2")
set -A VAR2 ${(kv)AARR}
;;
esac
}
creatAARR $TYPE_VAR1 $VAR1
echo "${(kv)VAR1}"
return
There may be a better way, but this is what works for me now.
Please feel free to add in your methods.
I have a variable ($OUTPUT) that contains the following name / value pairs:
member_id=4611686018429783292
platform=Xbox
platform_id=1
character_id=2305843009264966985
period_dt=2020-11-25 20:31:14.923158 UTC
mode=all Crucible modes
mode_id=5
activities_entered=18
activities_won=10
activities_lost=8
assists=103
kills=233
average_kill_distance=15.729613
total_kill_distance=3665
seconds_played=8535
deaths=118
average_lifespan=71.72269
total_lifespan=8463.277
opponents_defeated=336
efficiency=2.8474576
kills_deaths_ratio=1.9745762
kills_deaths_assists=2.411017
suicides=1
precision_kills=76
best_single_game_kills=-1
Each line ends with \n.
I want to loop through them, and parse them into an associative array, and the access the values in the array by the variable names:
while read line
do
key=${line%%=*}
value=${line#*=}
echo $key=$value
data[$key]="$value"
done < <(echo "$OUTPUT")
#this always prints the last value
echo ${data['seconds_played']}
This seems to work, i.e. key/value print the right values, but when I try to pull any values from the array, it always returns the last value (in this case -1).
I feel like im missing something obvious, but have been banging my head against it for a couple of hours.
UPDATE: My particular issue is I'm running a version of bash (3.2.57 on OSX) that doesn't support associative arrays). I'll mark the correct answer below.
Without declare -A data, then data is a normal array. In normal arrays expressions in [here] first undergo expansions, then arithmetic expansion. Inside arithmetic expansion unset variables are expanded to 0. You are effectively only just setting data[0]=something, because data[$key] is data[seconds_played] -> variable seconds_played is not defined, so it expands to data[0]
Add declare -A data and it "should work". You could also just:
declare -A data
while IFS== read -r key value; do
data["$key"]="$value"
done <<<"$OUTPUT"
Try declaring data as an associative array before populating it, eg:
$ typeset -A data # declare as an associative array
$ while read line
do
key=${line%%=*}
value=${line#*=}
echo $key=$value
data[$key]="$value"
done <<< "${OUTPUT}"
$ typeset -p data
declare -A data=([mode]="all Crucible modes" [period_dt]="2020-11-25 20:31:14.923158 UTC" [deaths]="118" [best_single_game_kills]="-1" [efficiency]="2.8474576" [precision_kills]="76" [activities_entered]="18" [seconds_played]="8535" [total_lifespan]="8463.277" [average_lifespan]="71.72269" [character_id]="2305843009264966985" [kills]="233" [activities_won]="10" [average_kill_distance]="15.729613" [activities_lost]="8" [mode_id]="5" [assists]="103" [suicides]="1" [total_kill_distance]="3665" [platform]="Xbox" [kills_deaths_ratio]="1.9745762" [platform_id]="1" [kills_deaths_assists]="2.411017" [opponents_defeated]="336" [member_id]="4611686018429783292" )
$ echo "${data['seconds_played']}"
8535
I am creating dynamic arrays which all have different letters in their name. For the purpose of this question, my initial array of letters has been set at a fixed length. However, in my final implementation this letter array will be any length.
For each letter, I construct a string
I declare a new array with that string, making use of eval to evaluate the variable's value within the declare command.
I add some values to the array, again using eval to evaluate any variable values.
Here is the code:
declare -a LETTER_ARRAY=( "A" "B" "C" "D" "E" )
for i in "${LETTER_ARRAY[#]}"
do
name_string="apple${i}"
color0="red"
color1="green"
eval "declare -a ${name_string}_array"
eval "${name_string}_array[0]=$color0"
eval "${name_string}_array[1]=$color1"
done
So, how can I iterate through these dynamic arrays and echo what is in them? I have tried the following:
for i in "${LETTER_ARRAY[#]}"
do
eval "array_name='apple${i}_array'"
echo ${array_name[0]}
done
This has not worked for me. I can confirm that my dynamic arrays were successfully created and populated, as when I echo out a value manually, I get a result:
echo ${appleA_array[0]}
red
A perfect place to use a namereference:
letter_array=(A B C D E)
for i in "${letter_array[#]}"; do
declare -n var="apple${i}_array"
var[0]=red
var[1]=green
done
declare -p appleA_array
would output:
declare -a appleA_array=([0]="red" [1]="green")
how can I iterate through these dynamic arrays and echo what is in them?
With the above:
for i in "${letter_array[#]}"; do
declare -n var="apple${i}_array"
printf "%s\n" "${var[0]}"
done
Notes:
Do not use eval. Eval is evil.
Do not use upper case variables, by convention they are used for exported variables, like COLUMNS, PWD, UID, EUID, LINES. Use lower case variables in your scripts.
Check your scripts with http://shellcheck.net for most common mistakes
But if you are creating a 2d array, then an associative array might be better:
declare -A apple_arrays
letter_array=(A B C D E)
for i in "${letter_array[#]}"; do
apple_arrays[$i,0]=red
apple_arrays[$i,1]=green
done
for i in "${letter_array[#]}"; do
printf "one=%s two=%s\n" "${apple_arrays[$i,0]}" "${apple_arrays[$i,1]}"
done
how can I iterate through these dynamic arrays
echo ${array_name[0]} does not work because array_name is not the name of an array; $array_name is. Therefore, eval "echo \"\${${array_name}[0]}\"" would to the trick.
However, I'd recommend namerefs.
By The way: declare works without eval and is more reliable that way.
#! /usr/bin/env bash
letters=({A..E})
for i in "${letters[#]}"; do
declare -a "apple${i}_array=(red green)"
done
for i in "${letters[#]}"; do
declare -n array="apple${i}_array"
# now you can use `array` as if it was `appleA_array`, `appleB_array`, ...
echo "${array[0]}"
done
Your first line is not bash syntax. If I try the line
declare -a LETTER_ARRAY = [ "A" "B" "C" "D" "E" ]
I get:
bash: declare: `=': not a valid identifier
bash: declare: `[': not a valid identifier
bash: declare: `]': not a valid identifier
I think, you get similar error messages, but you ignored them
More errors:
Forgotten do
case mismatch: suffix on definition: _array, and for output: _ARRAY
Use always double quotes when using [#]
One correct syntax is:
declare -a LETTER_ARRAY=( "A" "B" "C" "D" "E" )
for i in "${LETTER_ARRAY[#]}"
do
name_string="apple${i}"
color0="red"
color1="green"
eval "declare -a ${name_string}_array"
echo "${name_string}_array[0]=$color0"
eval "${name_string}_array[0]=$color0"
eval "${name_string}_array[1]=$color1"
done
echo ${appleA_array[0]}
Your eval "array_name='AZ${i}_ARRAY'" makes array_name a scalar, not an array. Arrays in bash are usually created like this
arr=( your elements go here )
If you want to assign one array to another, you have to interpolate the elements between those parenthesis, for instance:
arr=( ${other_array[#]} )
Since you are using bash, this would perform word splitting the elements of other_array, if they contain spaces. Hence you would usually write it for the safe side as
arr=( "${other_array[#]}" )
Hence, for your case, you could do a
eval "array_name=( \${AZ${i}_ARRAY[#]} )"
This causes an array array_name to be created, with the elements of the respective AZi_ARRAY.
I omitted here for simplicity the prevention against word splitting, because in your example, the array elements contain single words only.
How I can have an access to the last key of the associative array in bash? In this example I need to have "lot" in the $last variable. I found a way described here: How to get the keys and values of an associative array indirectly in Bash?. But it doesn't work as expected in the example below and return this error:
./test.sh: line 9: keys2: ${!$addict[#]}: must use subscript when assigning associative array
Here are the contents of this test.sh:
declare -A addict=(
["foo"]="bar"
["few"]="baz"
["lot"]="pot"
)
index_last=$(( ${#addict[#]} - 1 ))
eval 'declare -A keys2=(${!$addict[#]})'
last="${keys2[$index_last]}"
echo "$keys2"
echo "$index_last"
echo "$last"
While Tom Fenech is absolutely right saying
The keys are unordered, so the concept of a "last key" doesn't really make sense, you can avoid the error by changing the line with eval to
keys2=( "${!addict[#]}" )
and see what you get. It may also be illuminating to look at declare -p addict. In order to get some key (first key that is returned from unordered keys list, not first key that was declared) you can do:
some_key="${keys2[0]}"
This way you could, for example, unset A[$some_key] and iterate over keys in such manner by picking first returned key each time.
Example:
$ declare -A A=( [a]=x [b]=y [c]=z )
$ echo "${!A[#]}"
c b a
$ keys=( "${!A[#]}" )
$ echo "${keys[0]}"
c
# you see that c was returned instead of a
# (on your computer order could be different)
$ unset A[$some_key]
$ echo "${!A[#]}"
b a
$ declare -p A
declare -A A=([b]="y" [a]="x" )
Further reading: GNU Bash - Arrays.
After thoroughly searching for a way to create an associative array in bash, I found that declare -A array will do the trick. But the problem is, it is only for bash version 4 and the bash version the server has in our system is 3.2.16.
How can I achieve some sort of associative array-like hack in bash 3? The values will be passed to a script like
ARG=array[key];
./script.sh ${ARG}
EDIT: I know that I can do this in awk, or other tools but strict bash is needed for the scenario I am trying to solve.
Bash 3 has no associative arrays, so you're going to have to use some other language feature(s) for your purpose. Note that even under bash 4, the code you wrote doesn't do what you claim it does: ./script.sh ${ARG} does not pass the associative array to the child script, because ${ARG} expands to nothing when ARG is an associative array. You cannot pass an associative array to a child process, you need to encode it anyway.
You need to define some argument passing protocol between the parent script and the child script. A common one is to pass arguments in the form key=value. This assumes that the character = does not appear in keys.
You also need to figure out how to represent the associative array in the parent script and in the child script. They need not use the same representation.
A common method to represent an associative array is to use separate variables for each element, with a common naming prefix. This requires that the key name only consists of ASCII letters (of either case), digits and underscores. For example, instead of ${myarray[key]}, write ${myarray__key}. If the key is determined at run time, you need a round of expansion first: instead of ${myarray[$key]}, write
n=myarray__${key}; echo ${!n}
For an assignment, use printf -v. Note the %s format to printf to use the specified value. Do not write printf -v "myarray__${key}" %s "$value" since that would treat $value as a format and perform printf % expansion on it.
printf -v "myarray__${key}" %s "$value"
If you need to pass an associative array represented like this to a child process with the key=value argument representation, you can use ${!myarray__*} to enumerate over all the variables whose name begins with myarray__.
args=()
for k in ${!myarray__*}; do
n=$k
args+=("$k=${!n}")
done
In the child process, to convert arguments of the form key=value to separate variables with a prefix:
for x; do
if [[ $x != *=* ]]; then echo 1>&2 "KEY=VALUE expected, but got $x"; exit 120; fi
printf -v "myarray__${x%%=*}" %s "${x#*=}"
done
By the way, are you sure that this is what you need? Instead of calling a bash script from another bash script, you might want to run the child script in a subshell instead. That way it would inherit from all the variables of the parent.
Here is another post/explanation on associative arrays in bash 3 and older using parameter expansion:
https://stackoverflow.com/a/4444841
Gilles' method has a nice if statement to catch delimiter issues, sanitize oddball input ...etc. Use that.
If you are somewhat familiar with parameter expansion:
http://www.gnu.org/software/bash/manual/html_node/Shell-Parameter-Expansion.html
To use in your scenario [ as stated: sending to script ]:
Script 1:
sending_array.sh
# A pretend Python dictionary with bash 3
ARRAY=( "cow:moo"
"dinosaur:roar"
"bird:chirp"
"bash:rock" )
bash ./receive_arr.sh "${ARRAY[#]}"
Script 2: receive_arr.sh
argAry1=("$#")
function process_arr () {
declare -a hash=("${!1}")
for animal in "${hash[#]}"; do
echo "Key: ${animal%%:*}"
echo "Value: ${animal#*:}"
done
}
process_arr argAry1[#]
exit 0
Method 2, sourcing the second script:
Script 1:
sending_array.sh
source ./receive_arr.sh
# A pretend Python dictionary with bash 3
ARRAY=( "cow:moo"
"dinosaur:roar"
"bird:chirp"
"bash:rock" )
process_arr ARRAY[#]
Script 2: receive_arr.sh
function process_arr () {
declare -a hash=("${!1}")
for animal in "${hash[#]}"; do
echo "Key: ${animal%%:*}"
echo "Value: ${animal#*:}"
done
}
References:
Passing arrays as parameters in bash
If you don't want to handle a lot of variables, or keys are simply invalid variable identifiers, and your array is guaranteed to have less than 256 items, you can abuse function return values. This solution does not require any subshell as the value is readily available as a variable, nor any iteration so that performance screams. Also it's very readable, almost like the Bash 4 version.
Here's the most basic version:
hash_index() {
case $1 in
'foo') return 0;;
'bar') return 1;;
'baz') return 2;;
esac
}
hash_vals=("foo_val"
"bar_val"
"baz_val");
hash_index "foo"
echo ${hash_vals[$?]}
More details and variants in this answer
You can write the key-value pairs to a file and then grep by key. If you use a pattern like
key=value
then you can egrep for ^key= which makes this pretty safe.
To "overwrite" a value, just append the new value at the end of the file and use tail -1 to get just the last result of egrep
Alternatively, you can put this information into a normal array using key=value as value for the array and then iterator over the array to find the value.
This turns out to be ridiculously easy. I had to convert a bash 4 script that used a bunch of associative arrays to bash 3. These two helper functions did it all:
array_exp() {
exp=${#//[/__}
eval "${exp//]}"
}
array_clear() {
unset $(array_exp "echo \${!$1__*}")
}
I'm flabbergasted that this actually works, but that's the beauty of bash.
E.g.
((all[ping_lo] += counts[ping_lo]))
becomes
array_exp '((all[ping_lo] += counts[ping_lo]))'
Or this print statement:
printf "%3d" ${counts[ping_lo]} >> $return
becomes
array_exp 'printf "%3d" ${counts[ping_lo]}' >> $return
The only syntax that changes is clearing. This:
counts=()
becomes
array_clear counts
and you're set. You could easily tell array_exp to recognize expressions like "=()" and handle them by rewriting them as array_clear expressions, but I prefer the simplicity of the above two functions.