Reputation: 41
Is there a way to extract data from a text column containing json arrays with a varying number of json objects into a table?
For example if I...
CREATE TABLE tableWithJsonStr (location TEXT, jsonStr TEXT);
INSERT INTO tableWithJsonStr VALUES
('Home', '[{"animalId":"1","type":"dog", "color":"white","isPet":"1"},{"animalId":"2","type":"cat", "color":"brown","isPet":"1"}]'),
('Farm', '[{"animalId":"8","type":"cow", "color":"brown","isPet":"0"}, {"animalId":"33","type":"pig", "color":"pink","isPet":"0"}, {"animalId":"22","type":"horse", "color":"black","isPet":"1"}]'),
('Zoo', '[{"animalId":"5","type":"tiger", "color":"stripes","isPet":"0"}]');
and
CREATE TABLE animal (
location TEXT,
idx INT,
animalId INT,
type TEXT,
color TEXT,
isPet BOOLEAN
);
I am able to extract tableWithJsonStr.jsonStr by running:
INSERT INTO animal
SELECT location,
idx AS id,
TRIM(BOTH'"' FROM JSON_EXTRACT(jsonStr, CONCAT('$[', idx, '].animalId'))) AS animalId,
TRIM(BOTH'"' FROM JSON_EXTRACT(jsonStr, CONCAT('$[', idx, '].type'))) AS type,
TRIM(BOTH'"' FROM JSON_EXTRACT(jsonStr, CONCAT('$[', idx, '].color'))) AS color,
TRIM(BOTH'"' FROM JSON_EXTRACT(jsonStr, CONCAT('$[', idx, '].isPet'))) AS isPet
FROM tableWithJsonStr
JOIN(
SELECT 0 AS idx UNION
SELECT 1 AS idx UNION
SELECT 2 AS idx UNION
SELECT 3 AS idx
) AS indexes
WHERE JSON_EXTRACT(jsonStr, CONCAT('$[', idx, ']')) IS NOT NULL;
The result of the animal table is:
| location | idx | animalId | type | color | isPet |
|==========|=====|==========|=======|=========|=======|
| Farm | 0 | 8 | cow | brown | 0 |
| Farm | 1 | 33 | pig | pink | 0 |
| Farm | 2 | 22 | horse | black | 1 |
| Home | 0 | 1 | dog | white | 1 |
| Home | 1 | 2 | cat | brown | 1 |
| Zoo | 0 | 5 | tiger | stripes | 0 |
While the solution works, it is not extensible. If I have more than 3 objects in my json array, they will not be accounted for unless I add another SELECT 4 AS idx in my JOIN. Is there a better way to iterate over the objects in the array that doesn't require foreknowledge of the max number of objects that might be in each array?
Upvotes: 4
Views: 5904
Reputation: 147196
If you're using MySQL 8.0, you can use the JSON_TABLE
command to extract your data from each row of JSON
:
SELECT t1.location, farm.*
FROM tableWithJsonStr t1
JOIN JSON_TABLE(t1.jsonStr,
'$[*]'
COLUMNS (idx FOR ORDINALITY,
animalId INT PATH '$.animalId',
type TEXT PATH '$.type',
color TEXT PATH '$.color',
isPet BOOLEAN PATH '$.isPet')
) farm
ORDER BY location, idx
Output:
location idx animalId type color isPet
Farm 1 8 cow brown 0
Farm 2 33 pig pink 0
Farm 3 22 horse black 1
Home 1 1 dog white 1
Home 2 2 cat brown 1
Zoo 1 5 tiger stripes 0
If you are stuck with MySQL 5.7, you can use a stored procedure to extract the data:
DELIMITER $$
CREATE PROCEDURE extract_animals()
BEGIN
DECLARE idx INT;
DECLARE finished INT DEFAULT 0;
DECLARE loc, json VARCHAR(200);
DECLARE json_cursor CURSOR FOR SELECT location, jsonStr FROM tableWithJsonStr;
DECLARE CONTINUE HANDLER FOR NOT FOUND SET finished = 1;
DROP TABLE IF EXISTS animal;
CREATE TABLE animal (location TEXT, idx INT, animalId INT, type TEXT, color TEXT, isPet BOOLEAN);
OPEN json_cursor;
json_loop: LOOP
FETCH json_cursor INTO loc, json;
IF finished = 1 THEN
LEAVE json_loop;
END IF;
SET idx = 0;
WHILE JSON_CONTAINS_PATH(json, 'one', CONCAT('$[', idx, ']')) DO
INSERT INTO animal VALUES(loc,
idx,
JSON_UNQUOTE(JSON_EXTRACT(json, CONCAT('$[', idx, '].animalId'))),
JSON_UNQUOTE(JSON_EXTRACT(json, CONCAT('$[', idx, '].type'))),
JSON_UNQUOTE(JSON_EXTRACT(json, CONCAT('$[', idx, '].color'))),
JSON_UNQUOTE(JSON_EXTRACT(json, CONCAT('$[', idx, '].isPet'))));
SET idx = idx + 1;
END WHILE;
END LOOP json_loop;
END $$
Output:
location idx animalId type color isPet
Home 0 1 dog white 1
Home 1 2 cat brown 1
Farm 0 8 cow brown 0
Farm 1 33 pig pink 0
Farm 2 22 horse black 1
Zoo 0 5 tiger stripes 0
Upvotes: 5