Save an array value to jsonb type column

[lifubang]

CREATE TABLE public.jsontest (
	test jsonb NULL
)
WITH (
	OIDS=FALSE
);

pool.connect(function(errp, client, done) {
  client.query('INSERT INTO jsontest(test1) VALUES($1::jsonb);', [[{"id":1, "name":"LFB"}, {"id":2, "name":"ZHM"}]], function(e,d) {
      console.log(e);
      console.log(d);
      done();
  });
});

* It returns:
{ error: invalid input syntax for type json
    at Connection.parseE (E:\develop\nodejs\mongoOPLog\node_modules\pg\lib\connection.js:546:11)
    at Connection.parseMessage (E:\develop\nodejs\mongoOPLog\node_modules\pg\lib\connection.js:371:19)
    at Socket.<anonymous> (E:\develop\nodejs\mongoOPLog\node_modules\pg\lib\connection.js:114:22)
    at emitOne (events.js:96:13)
    at Socket.emit (events.js:188:7)
    at readableAddChunk (_stream_readable.js:176:18)
    at Socket.Readable.push (_stream_readable.js:134:10)
    at TCP.onread (net.js:547:20)
  name: 'error',
  length: 174,
  severity: 'ERROR',
  code: '22P02',
  detail: 'Expected ":", but found ",".',
  hint: undefined,
  position: undefined,
  internalPosition: undefined,
  internalQuery: undefined,
  where: 'JSON data, line 1: {"{\\"id\\":1,\\"name\\":\\"LFB\\"}",...',
  schema: undefined,
  table: undefined,
  column: undefined,
  dataType: undefined,
  constraint: undefined,
  file: 'json.c',
  line: '1151',
  routine: 'report_parse_error' }

You can fix it by:

pool.connect(function(errp, client, done) {
  client.query('INSERT INTO jsontest(test1) VALUES($1::jsonb);', [JSON.stringify([{"id":1, "name":"LFB"}, {"id":2, "name":"ZHM"}])], function(e,d) {
      console.log(e);
      console.log(d);
      done();
  });
});

But if modify the arrayString function in utils.js, It will be more convenient.

[sehrope]

If I'm understanding this PR correctly, you'd checking to see if the array has non-scalar children to determine that the user's intent was to treat it as JSON (as opposed to an array):

if (val && val.length > 0 && typeof(val[0]) === 'object' && typeof val[0].toPostgres != 'function') {
  return JSON.stringify(val);
} else {
  // old logic for handling arrays goes here
}

I don't think this is a good idea as it doesn't handle every situation. None of the following would be covered because the first element of the array is not an object:

[null]
[null, 1, 2, 3]
[null, {foo: 'bar'}]
[1, {foo: 'bar'}]
[false, {foo: 'bar'}]

Each of those can be serialized to valid JSON but wouldn't pass the initial check in the PR. To use a value like that you'd still have to manually serialize it to JSON prior to use.

We could check all values to see if they're all non-scalars but then you'd still have divergent behavior if an empty array is specified. Should [] be sent as an empty Postgres array, {}, or an empty JSON array, []?

Arrays have been a part of Postgres long before JSON, I think it's fair for them to have priority here. If someone is saving a JSON array (or an arbitrary JSON value which could be a string, number, or null), then they need to explicitly handle it.

[abenhamdine]

I agree with @sehrope , I don't see any obvious way to distinguish arrays/json objects.
And you could also want to save a normal array in a jsonb column so it seems difficult to automatically apply the right format.
Personnaly, I indeed stringify specifically the object before passing it to the query.

Otherwise, it's better to open an issue when you think you enconter a bug : not everyone watchs the pull requests, so opening a bug allows users to easily find it & provide feedback if they want, thx !

[charmander]

As was said, this approach isn’t reliable. JSON.stringify is a workable workaround. See also #442.