you should have a permition write to destination.

PosgreSqlToCustomJSON.py

import psycopg2
import psycopg2.extras
import sys
import json
import uuid

def main():
	conn_string = "host='ec2-54-xxx-yyy-zzz.cn-north-1.compute.amazonaws.com.cn' " \
				  "dbname='musicbrainz' " \
				  "user='postgres' " \
				  "password=''"
	conn = psycopg2.connect(conn_string)

	cursor = conn.cursor('cursor_unique_name', cursor_factory=psycopg2.extras.DictCursor)
	SQL= """
			select recording.gid as in , artist.gid as musicGroupId 
			from musicbrainz.artist,musicbrainz.recording
			where recording.artist_credit = artist.id
			;
		"""
	cursor.execute(SQL)
	jsonfile = open('./musicGroupRelation.json', 'w')
	row_count = 0
	for row in cursor:
		row_count += 1
#		print row
		opt = {}
		opt[u"endDate"]		= 	" "
	    opt[u"description"]	=   " " 		
	    opt[u"beginDate"]	=   " " 	
		opt[u"label"]		=   "musicGroup"
	    opt[u"in"]			=   row[0]
	    opt[u"@id"]			=   str(uuid.uuid1()) 
	   	opt[u"@type"]		=   "Relation" 
	    opt[u"out"]			=   row[1] 

#	    	print opt

		json.dump(opt,jsonfile)
		jsonfile.write('\n')

if __name__ == "__main__":
	main()

postgreCSVAndJSON.sql

-- Backup to JSON
COPY ( 
  SELECT ROW_TO_JSON(a) 
  FROM (
    select 
    artist.id as artist_id, 
    artist_credit.id as artist_credit_id ,
    recording.id as recording_id ,
    artist.gid as artist_gid,
    recording.gid as recording_gid,
    artist.name as artist_name,
    artist_credit_name.artist as artist_credit_name_artist ,
    artist_credit_name.name as artist_credit_name_name
    from musicbrainz.artist,musicbrainz.artist_credit_name,musicbrainz.artist_credit,musicbrainz.recording 
    where artist.gid='a223958d-5c56-4b2c-a30a-87e357bc121b' limit 100
  )a
)
TO '/your/dist/path' with text;

-- Backup To CSV
Copy (Select * From musicbrainz.url) To '/tmp/url.csv' With CSV DELIMITER ',';