TopolearnReshaping API output into exactly the fields you need is jq's everyday job.
A jq program is built from small filters combined with operators.
- The pipe
a | bfeeds every output ofaintob..users[] | .emailis the same as.users[].email, but pipes let you add any filter in between. - The comma
a, bruns both filters on the same input and outputs the results of both:.name, .age.
Object construction creates new JSON:
{name: .user.name, id: .user.id}builds an object with the given keys.- Shorthand:
{name, id}means{name: .name, id: .id}. - Keys can be computed with parentheses:
{(.key): .value}. - If a value produces several outputs, you get several objects, which can surprise you.
Array construction [ ... ] collects all outputs of the inner expression into one array: [.items[].price].
Putting it together, .users[] | {name, email: .contact.email} turns a verbose API response into a compact list of objects, one per user. Add -c for compact one-line output, handy for further processing or grep.
Parentheses group as you would expect, and | binds more loosely than ,, so .a, .b | f means (.a, .b) | f.
Example
cat > r.json <<'EOF'
{"users":[{"name":"Ann","id":1,"contact":{"email":"ann@x.io"}},
{"name":"Bob","id":2,"contact":{"email":"bob@x.io"}}]}
EOF
jq -c '.users[] | {name, email: .contact.email}' r.json
# {"name":"Ann","email":"ann@x.io"}
# {"name":"Bob","email":"bob@x.io"}
jq -c '[.users[].id]' r.json # [1,2]
jq -r '.users[0] | .name, .id' r.json # Ann 1
jq -c '.users | map({(.name): .id}) | add' r.json # {"Ann":1,"Bob":2}Remember
- a | b pipes each output of a into b
- a, b emits the outputs of both
- {name, id: .x} builds objects; [ ... ] collects outputs
Watch out for
- Writing {name: .tags[]} and getting one object per tag; a multi-output value multiplies objects.