What does it mean to echo multiple strings into a UNIX pipe?

Question

I just saw an example of using echo to put multiple things into the pipeline (for jq).

echo '{"A": {"a": 1}}' '{"A": {"b": 2}}' '{"B": 3}' |\
  jq --slurp 'reduce .[] as $item ({}; . * $item)'

If I understand it right, this seems to put three objects into the pipeline on the one end, and on the other, jq receives them and puts them into an array and reduce it with a merge operator.

This is surprising to me, and I am not sure how it works to have multiple things in the stdin.

More specifically, how does the system know the input is three files and where each file begins/ends?

If I just do the echo part echo '{"A":1}' '{"A": {"b": 2}}' '{"B": 3}', the output does not seem to have any visible delimiter between the three objects:

{"A":1} {"A": {"b": 2}} {"B": 3}

Does the single-quotes have some special effects that tells the next filter (jq) the boundaries? Or how does jq know the boundaries?

(This is with Ubuntu 24.04, jq 1.7.1)

Answer 1

It doesn't do any of that. There are no objects; there's just a "stream" of characters.

echo prints all arguments it gets, separated by spaces. If you do echo 'a' 'b' 'c' then the output will be a b c; a single stream

Similarly your echo statement just produces the output {"A": {"a": 1}} {"A": {"b": 2}} {"B": 3}. Again, just a single stream of characters.

The jq command knows enough about JSON to know those characters mean something (a set of separate JSON objects), but to echo and the pipeline it's just a stream of meaningless characters.

You'd get exactly the same result if you did echo '{"A": {"a": 1}} {"A": {"b": 2}} {"B": 3}'. Or even something like echo '{"A":' '{"a":' "1}}" '{"A": {"b": 2}} {"B": 3}'

Also note that you can produce output from multiple commands to the pipeline, and the result is still just the concatenation of all the bytes involved. E.g. here, foobar gets joined into a single line:

% (printf "foo"; printf "bar\nxxx\n") | head -1
foobar