To create a pure JSON table you need to provide a single field with the type JSONB:

CREATE TABLE mytable (data JSONB NOT NULL);

You should also create a basic index:

CREATE INDEX mytable_idx ON mytable USING gin (data jsonb_path_ops);

At this point you can insert data in to the table and query it efficiently.

Taking a complex JSON document in a table:

CREATE TABLE mytable (data JSONB NOT NULL);
CREATE INDEX mytable_idx ON mytable USING gin (data jsonb_path_ops);
INSERT INTO mytable VALUES($$
{
    "name": "Alice",
    "emails": [
        "[email protected]",
        "[email protected]"
    ],
    "events": [
        {
            "type": "birthday",
            "date": "1970-01-01"
        },
        {
            "type": "anniversary",
            "date": "2001-05-05"
        }
    ],
    "locations": {
        "home": {
            "city": "London",
            "country": "United Kingdom"
        },
        "work": {
            "city": "Edinburgh",
            "country": "United Kingdom"
        }
    }
}
$$);

Query for a top-level element:

SELECT data->>'name' FROM mytable WHERE data @> '{"name":"Alice"}';

Query for a simple item in an array:

SELECT data->>'name' FROM mytable WHERE data @> '{"emails":["[email protected]"]}';

Query for an object in an array:

SELECT data->>'name' FROM mytable WHERE data @> '{"events":[{"type":"anniversary"}]}';

Query for a nested object:

SELECT data->>'name' FROM mytable WHERE data @> '{"locations":{"home":{"city":"London"}}}';

Performance of @> compared to -> and ->>

It is important to understand the performance difference between using @>, -> and ->> in the WHERE part of the query. Although these two queries appear to be broadly equivalent:

SELECT data FROM mytable WHERE data @> '{"name":"Alice"}';
SELECT data FROM mytable WHERE data->'name' = '"Alice"';
SELECT data FROM mytable WHERE data->>'name' = 'Alice';

the first statement will use the index created above whereas the latter two will not, requiring a complete table scan.

It is still allowable to use the -> operator when obtaining resultant data, so the following queries will also use the index:

SELECT data->'locations'->'work' FROM mytable WHERE data @> '{"name":"Alice"}';
SELECT data->'locations'->'work'->>'city' FROM mytable WHERE data @> '{"name":"Alice"}';

Creating a DB and a Table

DROP DATABASE IF EXISTS books_db;
CREATE DATABASE books_db WITH ENCODING='UTF8' TEMPLATE template0;

DROP TABLE IF EXISTS books;

CREATE TABLE books (
  id SERIAL PRIMARY KEY,
  client TEXT NOT NULL,
  data JSONb NOT NULL
);

Populating the DB

INSERT INTO books(client, data) values (
    'Joe', 
    '{ "title": "Siddhartha", "author": { "first_name": "Herman", "last_name": "Hesse" } }'
),(
    'Jenny', 
    '{ "title": "Dharma Bums", "author": { "first_name": "Jack", "last_name": "Kerouac" } }'
),(
    'Jenny', 
    '{ "title": "100 años de soledad", "author": { "first_name": "Gabo", "last_name": "Marquéz" } }'
);

Lets see everything inside the table books:

SELECT * FROM books;

Output:

-> operator returns values out of JSON columns

Selecting 1 column:

SELECT client, 
    data->'title' AS title
    FROM books;

Output:

Selecting 2 columns:

SELECT client, 
   data->'title' AS title, data->'author' AS author
   FROM books;

Output:

-> vs ->>

The -> operator returns the original JSON type (which might be an object), whereas ->> returns text.

Return NESTED objects

You can use the -> to return a nested object and thus chain the operators:

SELECT client, 
   data->'author'->'last_name' AS author
   FROM books;

Output:

Filtering

Select rows based on a value inside your JSON:

 SELECT 
 client,
 data->'title' AS title
 FROM books
  WHERE data->'title' = '"Dharma Bums"';

Notice WHERE uses -> so we must compare to JSON '"Dharma Bums"'

Or we could use ->> and compare to 'Dharma Bums'

Output:

Nested filtering

Find rows based on the value of a nested JSON object:

SELECT 
 client,
 data->'title' AS title
 FROM books
  WHERE data->'author'->>'last_name' = 'Kerouac';

Output:

A real world example

CREATE TABLE events (
  name varchar(200),
  visitor_id varchar(200),
  properties json,
  browser json
);

We’re going to store events in this table, like pageviews. Each event has properties, which could be anything (e.g. current page) and also sends information about the browser (like OS, screen resolution, etc). Both of these are completely free form and could change over time (as we think of extra stuff to track).

INSERT INTO events (name, visitor_id, properties, browser) VALUES
(
  'pageview', '1',
  '{ "page": "/" }',
  '{ "name": "Chrome", "os": "Mac", "resolution": { "x": 1440, "y": 900 } }'
),(
  'pageview', '2',
  '{ "page": "/" }',
  '{ "name": "Firefox", "os": "Windows", "resolution": { "x": 1920, "y": 1200 } }'
),(
  'pageview', '1',
  '{ "page": "/account" }',
  '{ "name": "Chrome", "os": "Mac", "resolution": { "x": 1440, "y": 900 } }'
),(
  'purchase', '5',
  '{ "amount": 10 }',
  '{ "name": "Firefox", "os": "Windows", "resolution": { "x": 1024, "y": 768 } }'
),(
  'purchase', '15',
  '{ "amount": 200 }',
  '{ "name": "Firefox", "os": "Windows", "resolution": { "x": 1280, "y": 800 } }'
),(
  'purchase', '15',
  '{ "amount": 500 }',
  '{ "name": "Firefox", "os": "Windows", "resolution": { "x": 1280, "y": 800 } }'
);

Now lets select everything:

SELECT * FROM events;

Output:

JSON operators + PostgreSQL aggregate functions

Using the JSON operators, combined with traditional PostgreSQL aggregate functions, we can pull out whatever we want. You have the full might of an RDBMS at your disposal.

• Lets see browser usage:

    SELECT browser->>'name' AS browser, 
      count(browser)
      FROM events
      GROUP BY browser->>'name';
  

Output:

• Total revenue per visitor:

    SELECT visitor_id, SUM(CAST(properties->>'amount' AS integer)) AS total
    FROM events
    WHERE CAST(properties->>'amount' AS integer) > 0
    GROUP BY visitor_id;
  

Output:

• Average screen resolution

    SELECT AVG(CAST(browser->'resolution'->>'x' AS integer)) AS width,
      AVG(CAST(browser->'resolution'->>'y' AS integer)) AS height
    FROM events;
  

Output:

More examples and documentation here and here.