How to split blocks of terminal output into array elements that can be looped over - bash

Question

Just looking for the best practice method to split a multi line output such as you might get from a command or a curl request into an array of entries using Bash or Zsh
For example, if a command or printing a file gave an output of
"User A:
Name: Bob
Age: 36
User B:
Name: Jeff
Age: 42"
How would you create an array from that where each user were an entry in the array?
Or if you had an output of say devices similar to
"Computer A:
Name Bob's Computer
Serial 123456
Uptime 12hr
Computer B:
Name Jeff's Computer
Serial 789101
Uptime 8hr"
How would you split that into an array of computers, so that you could do things like see how many computers there were by the number of elements in the array, or pass them one by one to another command, etc?
I've been looking for ways to split strings and output, all the answers I find seem to target splitting a single line with a single character deliminator. I figure the way to do it is either to split by using "User" or "Computer" as the deliminator in the above examples, or to use those as a pattern to read from and to, but I'm not sure how to do that in Bash?
Thanks in advance

Answer 1

Assuming:
You want to split the lines into e.g. "Computer A" block and "Computer B" block then store the computer (or User) names into an array.
(May be optional) You want to parse the lines of attributes such as "Name", "Serial" ... and store them in array of arrays.
Then would you please try the following:
#!/bin/bash
str="Computer A:
Name Bob's Computer
Serial 123456
Uptime 12hr
Computer B:
Name Jeff's Computer
Serial 789101
Uptime 8hr"
keyword="Computer" # the keyword to split the lines into array
declare -A hash # create associative array
i=0 # index of the "real" array name
readarray -t a <<< "$str" # read the string splitting on newlines
for line in "${a[#]}"; do # loop over lines
if [[ $line =~ ^${keyword}[[:blank:]]+([^:]+): ]]; then
# if the line starts with the "keyword"
name="${BASH_REMATCH[1]}" # name is assigned to "A" or "B" ...
hash[$name]="array$i" # define "real" array name and assign the hash value to it
declare -A "array$i" # create a new associative array with the name above
declare -n ref="array$i" # "ref" is a reference to the newly created associative array
(( i++ )) # increment the index for new entry
else
read -r key val <<< "$line" # split on the 1st blank character
key=${key%:} # remove traiking ":"
key=${key## } # remove leading whitespace(s)
ref[$key]=$val # store the "key" and "val" pair
fi
done
print number of elements of the array
echo "The array has ${#hash[#]} elements"
# print the values of each array of the array
for i in "${!hash[#]}"; do # loop over "A" and "B"
echo "$i" # print it
declare -n ref=${hash[$i]} # assign ref to array name "array0" or "array1" ... then make it an indirect reference
for j in "${!ref[#]}"; do # loop over "${array0[#]}" or "${array1[#]}" ...
echo " $j => ${ref[$j]}"
done
done
Output:
The array has 2 elements
A
Name => Bob's Computer
Uptime => 12hr
Serial => 123456
B
Name => Jeff's Computer
Uptime => 8hr
Serial => 789101
Please note bash does not natively support array of arrays and we need to make use of declare -n statement to create a reference to a variable, which makes the code less readable. If Python is your option, please let me know. It will be much more suitable for this kind of task.

Answer 2

This will do the trick in Zsh:
split-blocks() {
local MATCH MBEGIN MEND
reply=(
${(0)1//(#m)$'\n'[^[:space:]]## [^[:space:]]##:[[:space:]]##/$'\0'$MATCH}
)
}
What this does:
Find each occurrence of <newline><text> <text>:<whitespace>.
Insert a null byte before each match.
Split the result on null bytes.
Assign the resulting elements to array $reply.