Synthetic Variables in Bash (was: How to address a variable ceated with indirection)

Question

Sorry, that was all too confusing, (my bad)
Let me try to explain again, but with a more simple example:

#define a new var and export it to the "env":
export VAR_ONE=thisIsVarOne
# check if it was created:
env | grep VAR_ONE      #this displays: "VAR_ONE=thisIsVarOne"
# use it in a simple echo command:
echo VAR_ONE=$VAR_ONE  #this displays: "VAR_ONE=thisIsVarOne"
create a new var with the name that contains the value of VAR_ONE ("newvarthisIsVarOne") in its name:
export newvar${VAR_ONE}="somePrefix${VAR_ONE}"
verify if it has been created:
env | grep newvar  # this displays: "newvarthisIsVarOne=somePrefixthisIsVarOne"

Now my question:

How do I use that new, syntetic, variable?
Please note that: env | grep newvar shows that there has been created a variable with the name newvar${VAR_ONE}

But I can't address that new variable, e.g. in an echo command. I tried: echo ${newvar${VAR_ONE}}
But that gave me bash: ${newvar${VAR_ONE}}: bad substitution

Constraints:

1. The value of VAR_ONE can change, so I can't use that to address the newly created variable
2. I can't declare a new variable with a static name. For usage, further in the script, I need the keyword VAR_ONE as part of the name of the newly created variable. The name of the variable needs to be something like: newvar${VAR_ONE}

Thank you in advance ChriV

Answer 1

Do not use a $ to affect a variable in bash:

var_name=value
var_name1="value with spaces"
var_name2='value with $ or special characters'

To get the value of a variable, use a $ before its name:

echo "var_name=$var_name"
echo "var_name1=use_braces_to_protect_the variable_name_${var_name1}_"
my_new_var="$var_name2"

If your variable contains the name of another variable, you must use an exclamation point before the variable name into braces:

my_var=spam
the_var_name=my_var
echo "the value is '${!the_var_name}'"

So your program will be:

#! /bin/bash
VAR1=thisIsVar1
#VAR2=thisIsVar2
VAR3=thisIsVar3
VAR_NAMES=($VAR1 $VAR2 $VAR3)
echo "VAR_NAMES=" "${VAR_NAMES[@]}"
for VAR_NAME in "${VAR_NAMES[@]}"; do
    echo
    echo "VAR_NAME=$VAR_NAME"
NEW_VAR_NAME="newvar$VAR_NAME"
echo "NEW_VAR_NAME=$NEW_VAR_NAME"

export "${NEW_VAR_NAME}"="somevalue ${VAR_NAME}"
echo "${NEW_VAR_NAME}=${!NEW_VAR_NAME}"

done

UPDATE

You could also create a hash table (associative array) with new var_name/var_value as key/value:

#! /bin/bash
VAR1=thisIsVar1
#VAR2=thisIsVar2
VAR3=thisIsVar3
VAR_NAMES=($VAR1 $VAR2 $VAR3)
echo "VAR_NAMES=" "${VAR_NAMES[@]}"
declare -A NEW_VAR_NAMES=()
for VAR_NAME in "${VAR_NAMES[@]}"; do
    echo
    echo "VAR_NAME=$VAR_NAME"
NEW_VAR_NAME="newvar$VAR_NAME"
echo "NEW_VAR_NAME=$NEW_VAR_NAME"

export "${NEW_VAR_NAME}"="somevalue ${VAR_NAME}"
echo "${NEW_VAR_NAME}=${!NEW_VAR_NAME}"

NEW_VAR_NAMES+=(["${NEW_VAR_NAME}"]=${!NEW_VAR_NAME})

done
echo
echo "${NEW_VAR_NAMES[@]@A}"
for new_var_name in "${!NEW_VAR_NAMES[@]}"; do
    echo "new_var_name is ${new_var_name} and its content is '${NEW_VAR_NAMES[$new_var_name]}'"
done

Answer 2

Setup:

export VAR_ONE=thisIsVarOne

With bash 4.3+ you can use a nameref (declare -n):

$ declare -n _var="newvar${VAR_ONE}"            # nameref; define new var name and associate with nameref variable '_var' (can be any valid variable name)
$ _var="somePrefix${VAR_ONE}"                   # assign value to nameref
$ echo "${!_var}"                               # show nameref association
newvarthisIsVarOne
$ echo "${_var}"                                # show value
somePrefixthisIsVarOne
$ echo "${newvarthisIsVarOne}"                  # explicitly reference new variable by name
somePrefixthisIsVarOne

Answer 3

Setup:

export VAR_ONE=thisIsVarOne

An alternative using an associative array:

$ unset      newvar
$ declare -A newvar
$ newvar[${VAR_ONE}]="somePrefix${VAR_ONE}"
$ typeset -p newvar
declare -A newvar=([thisIsVarOne]="somePrefixthisIsVarOne" )
$ echo "${newvar[${VAR_ONE}]}"
somePrefixthisIsVarOne

---

In a comment OP has stated the need to reference multiple variables, eg, newvar${VAR_ONE}, newvar${OTHER_VAR} and newvar${ANOTHER_VAR}.

This approach with the associative array will work as long as the 3 variables (VAR_ONE, OTHER_VAR and ANOTHER_VAR) have different values, otherwise duplicate values will lead to a single array entry being created (with the latest assignment overwriting the previous assignment).

If the 3 variables could have the same value then another approach would be to use literals for the associative array's indices, eg:

#### instead of:
$ newvar[${VAR_ONE}]="somePrefix${VAR_ONE}"
$ typeset -p newvar
declare -A newvar=([thisIsVarOne]="somePrefixthisIsVarOne" )
we use:
$ newvar[VAR_ONE]="somePrefix${VAR_ONE}"
$ typeset -p newvar
declare -A newvar=([VAR_ONE]="somePrefixthisIsVarOne" )

Answer 4

To access a shell variable without using a separate variable (for either the ${!pointer} syntax or a nameref) the only solution is to use eval, for example

eval "echo \$newvar$VAR_ONE"

-- but this is dangerous it you ever make a mistake, which in practice all people who are humans do.

But for an environment variable which export creates, you can use a child process to do the lookup. In particular awk is POSIX and usually present on most systems that have bash:

    awk -v name=newvar$VAR_ONE 'BEGIN{printf "%s",ENVIRON[name]}'
    # or simpler {print ENVIRON[name]} adds newline which may be okay,
    # e.g. if you use this in $(...) or `...` which removes newline
    # or _instead of_ an echo command because echo also adds newline

Even simpler, many systems have printenv and you can just do printenv newvar$VAR_ONE -- which also adds newline.