Generate new array from sparse array of objects in JQ - jq

Question

I have a JSON file that I want to process with JQ. It has an array of objects inside another object, with a key that I want to use to populate a new array.
In my real use-case this is nested in with a lot of other fluff and there lots more arrays but take this as a simpler but representative example of the kind of thing:
{
"numbers": [
{
"numeral": 1,
"ordinal": "1st",
"word": "One"
},
{
"numeral": 2,
"ordinal": "2nd",
"word": "Two"
},
{
"numeral": 5,
"ordinal": "5th",
"word": "Five"
},
{
"some-other-fluff-i-want-to-ignore": true
}
]
}
I'd like to use JQ to get a new array based on the elements, ignoring some elements and handling the missing ones. e.g.
[
"The 1st word is One",
"The 2nd word is Two",
"Wot no number 3?",
"Wot no number 4?",
"The 5th word is Five"
]
Doing this in a loop for the elements that are there is simple, terse and elegant enough:
.numbers | map( . | select( .numeral) | [ "The", .ordinal, "word is", .word ] | join (" "))
But I can't find a way to cope with the missing entries. I have some code that sort-of works:
.numbers | [
( .[] | select(.numeral == 1) | ( [ "The", .ordinal, "word is", .word ] | join (" ")) ) // "Wot no number 1?",
( .[] | select(.numeral == 2) | ( [ "The", .ordinal, "word is", .word ] | join (" ")) ) // "Wot no number 2?",
( .[] | select(.numeral == 3) | ( [ "The", .ordinal, "word is", .word ] | join (" ")) ) // "Wot no number 3?",
( .[] | select(.numeral == 4) | ( [ "The", .ordinal, "word is", .word ] | join (" ")) ) // "Wot no number 4?",
( .[] | select(.numeral == 5) | ( [ "The", .ordinal, "word is", .word ] | join (" ")) ) // "Wot no number 5?"
]
It produces usable output, after a fashion:
richard#sophia:~$ jq -f make-array.jq < numbers.json
[
"The 1st word is One",
"The 2nd word is Two",
"Wot no number 3?",
"Wot no number 4?",
"The 5th word is Five"
]
richard#sophia:~$
However, whilst it produces the output, handles the missing elements and ignores the bits I don't want, it's obviously extremely naff code that cries out for a for-loop or something similar but I can't see a way in JQ to do this. Any ideas?

Answer 1

jq solution:
jq 'def print(o): "The \(o.ordinal) word is \(o.word)";
.numbers | (reduce map(select(.numeral))[] as $o ({}; .["\($o.numeral)"] = $o)) as $o
| [range(0; ($o | [keys[] | tonumber] | max))
| "\(.+1)" as $i
| if ($o[$i]) then print($o[$i]) else "Wot no number \($i)?" end
]' input.json
The output:
[
"The 1st word is One",
"The 2nd word is Two",
"Wot no number 3?",
"Wot no number 4?",
"The 5th word is Five"
]

Answer 2

Another solution !
jq '[
range(1; ( .numbers | max_by(.numeral)|.numeral ) +1 ) as $range_do_diplay |
.numbers as $thedata | $range_do_diplay |
. as $i |
if ([$thedata[]|contains( { numeral: $i })]|any )
then
($thedata|map(select( .numeral == $i )))|.[0]| "The \(.ordinal) word is \(.word) "
else
"Wot no number \($i)?"
end
] ' numbers.json
This solution use
max_by to find the max value of numeral
range to generate a list o values
use variables to store intermediate value