Understanding the INSERT INTO Statement

This section describes how to use the INSERT INTO statement to insert or overwrite rows in nested HPE Ezmeral Data Fabric Database JSON tables, using the Hive connector.

Note: The output shown in these examples is for illustration only; actual Hive CLI output varies, depending on your specific situation.

Single-row insert

You can use the INSERT INTO statement to insert a single table row into a nested HPE Ezmeral Data Fabric Database table using one of two methods.

For example, imagine that you have the following Hive HPE Ezmeral Data Fabric Database JSON table, nested_data_insert:

 CREATE TABLE nested_data_insert 
 ( 
    entry STRING, 
    num INT, 
    postal_addresses MAP <STRING, 
    struct <USER_ID:STRING,ADDRESS:STRING,ZIP:STRING,COUNTRY:STRING>> 
 ) 
 stored BY 'org.apache.hadoop.hive.maprdb.json.MapRDBJsonStorageHandler' tblproperties
 ( 
    "maprdb.table.name" = "/nested_data_insert", 
    "maprdb.column.id" = "entry" 
 );

You can insert the new row into your table by using a dummy table:

WITH dummy_table AS
  (SELECT '001' AS KEY,
          '1' AS num,
          MAP ('Adam',
               Named_struct ('user_id', '1', 'address', '3205 Woodlake ct', 'zip', '45040', 'country', 'Usa'),
               'Wilfred',
               Named_struct ('user_id', '2', 'address', '777 Brockton Avenue', 'zip', '34000', 'country', 'Ita')) AS postal_addresses)
INSERT INTO nested_data_insert
SELECT *
FROM dummy_table;

Alternatively, you can insert the new row into your table by using a SELECT statement:

INSERT INTO TABLE nested_data_insert
SELECT '002',
       '2',
       MAP ('Bill',
            Named_struct ('user_id', '1', 'address', '328 Virginia Ave', 'zip', '54956', 'country', 'Bol'),
            'Stiv',
            Named_struct ('user_id', '2', 'address', 'Schererville', 'zip', '46375', 'country', 'Efi'));

After you insert data, you should verify that the data is inserted in both Hive and HPE Ezmeral Data Fabric Database JSON tables:

Verify the insertion into the Hive table by using the SELECT * FROM syntax.

SELECT * FROM nested_data_insert;

Sample output:

Table 1. Result of the insert queries 1.1 and 1.2:
entry	num	postal_address
entry	num		USER_ID	ADDRESS	ZIP	COUNTRY
001	1	Adam	1	3205 Woodlake ct	45040	Usa
		Wilfred	2	777 Brockton Avenue	34000	Ita
002	2	Bill	1	328 Virginia Ave	54956	Bol
		Stiv	2	Schererville	46375	Efi

Verify the insertion into the HPE Ezmeral Data Fabric Database JSON table data using the find statement:

find '/nested_data_insert'
              
{
  "Adam": {
      "user_id": "1",
      "address": "3205 Woodlake ct",
      "zip": "45040",
      "country": "Usa"
           },
  "Wilfred": {
      "user_id": "2",
      "address": "777 Brockton Avenue",
      "zip": "34000",
      "country": "Ita"
           }
        }
{
   "Bill": {
      "user_id": "1",
      "address": "328 Virginia Ave",
      "zip": "54956",
      "country": "Bol"
        },
   "Stiv": {
      "user_id": "2",
      "address": "Schererville",
      "zip": "46375",
      "country": "Efi"
        }
}

Multiple-row insert

Now imagine that you want to insert three rows of data into nested_data_insert.

You can insert the new rows into your table by using a dummy table:

WITH dummy_table AS
  (SELECT '003' AS KEY,
          '3' AS num,
          MAP ('Rony',
               Named_struct ('user_id', '1', 'address', '4333 Backer str', 'zip', '12311', 'country', 'Hun')) AS postal_addresses
   UNION ALL SELECT '004' AS KEY,
                    '4' AS num,
                    MAP ('Ivan',
                         Named_struct ('user_id', '1', 'address', '833 Bridle Avenue', 'zip', '95111', 'country', 'CA')) AS postal_addresses
   UNION ALL SELECT '005' AS KEY,
                    '5' AS num,
                    MAP ('Ivan',
                         Named_struct ('user_id', '1', 'address', '664 Devon Ave', 'zip', '92021', 'country', 'Tog')) AS postal_addresses)
INSERT INTO nested_data_insert
SELECT *
FROM dummy_table;

Alternatively, you can insert the new rows into your table by using a SELECT statement:

INSERT INTO TABLE nested_data_insert
SELECT '006',
       '6',
       MAP ('Rony',
            Named_struct ('user_id', '1', 'address', '150 National City', 'zip', '91950', 'country', 'Hun'))
UNION ALL
SELECT '007',
       '7',
       MAP ('Tomason',
            Named_struct ('user_id', '1', 'address', '272 Ocean Circle' , 'zip', '92801', 'country', 'CA'))
UNION ALL
SELECT '008',
       '8',
       MAP ('Davin',
            Named_struct ('user_id', '1', 'address', '81 Augusta Ave', 'zip', '93905', 'country', 'CA'));

After you insert data, you should verify that the data is inserted in both Hive and HPE Ezmeral Data Fabric Database JSON tables:

Verify the insertion into the Hive table by using the SELECT * FROM syntax.

SELECT * FROM nested_data_insert WHERE entry > '002' ;

Sample output:

Table 2. Result of the insert queries 2.1 and 2.2:
entry	num	postal_address
entry	num		USER_ID	ADDRESS	ZIP	COUNTRY
003	3	Rony	1	4333 Backer str	12311	Hun
004	4	Ivan	1	833 Bridle Avenue	95111	CA
005	5	Ivan	1	664 Devon Ave.	92021	Tog
006	6	Rony	1	150 National City	91950	Hun
007	7	Tomason	1	272 Ocean Circle	92801	CA
008	8	Davin	1	81 Augusta Ave	93905	CA

Verify the insertion into the HPE Ezmeral Data Fabric Database JSON table data using the find statement:

find '/nested_data_insert'
                
{
   "_id": "003",
   "num": {
     "$numberInt": 3
          },
   "postal_addresses": {
        "Rony": {
             "address": "4333 Backer str",
             "country": "Hun",
             "user_id": "1",
             "zip": "12311"
             }
         }
      }
{
   "_id": "004",
   "num": {
       "$numberInt": 4
          },
   "postal_addresses": {
          "Ivan": {
              "address": "833 Bridle Avenue",
              "country": "CA",
              "user_id": "1",
              "zip": "95111"
            }
       }
 }
 {
   "_id": "005",
   "num": {
           "$numberInt": 5
          },
   "postal_addresses": {
           "Ivan": {
               "address": "664 Devon Ave",
               "country": "Tog",
               "user_id": "1",
                "zip": "92021"
          }
       }
    }
{
     "_id": "006",
     "num": {
           "$numberInt": 6
            },
     "postal_addresses": {
          "Rony": {
             "address": "150 National City",
             "country": "Hun",
             "user_id": "1",
             "zip": "91950"
           }
        }
    }
{
    "_id": "007",
    "num": {
         "$numberInt": 7
           },
    "postal_addresses": {
          "Tomason": {
              "address": "272 Ocean Circle",
              "country": "CA",
              "user_id": "1",
              "zip": "92801"
           }
       }
    }
{
    "_id": "008",
    "num": {
         "$numberInt": 8
           },
    "postal_addresses": {
           "Davin": {
                "address": "81 Augusta Ave",
                "country": "CA",
                "user_id": "1",
                "zip": "93905"
            }
         }
}

Overwriting data

Still using sample table nested_data_insert, you can use the INSERT statement on a dummy table to overwrite one or more complete rows.

For example, to overwrite the first row in nested_data_insert (001) with new values, use the following syntax:

WITH dummy_table AS
(SELECT '001' AS KEY,
'1' AS num,
MAP ('newAdam',
Named_struct ('user_id', '1', 'address', 'newAdress', 'zip', 'newZip', 'country', 'newCountry')) AS postal_addresses)
INSERT INTO nested_data_insert
SELECT *
FROM dummy_table;

After you overwrite data, you should verify that the data is changed in both Hive and HPE Ezmeral Data Fabric Database JSON tables:

Verify the data into the Hive table by using the SELECT * FROM syntax.

hive> SELECT * FROM nested_data_insert WHERE entry = '001';

Sample output:

Table 3. Result of the insert query 3.1:
entry	num	postal_address
entry	num		USER_ID	ADDRESS	ZIP	COUNTRY
001	1	newAdam	1	newAddress	newZip	newCountry

Verify the data in the HPE Ezmeral Data Fabric Database JSON table data using the findbyid statement:

findbyid '/nested_data_insert' --id 001
              
{
   "_id": "001",
   "num": {
   "$numberInt": 1
           },
   "postal_addresses": {
        "newAdam": {
              "address": "newAdress",
              "country": "newCountry",
              "user_id": "1",
              "zip": "newZip"
           }
       }
}

For another example, imagine that you want to overwrite 003 and 004 rows in nested_data_insert with new values:

WITH dummy_table AS (
SELECT '003' AS KEY,
'3' AS num,
MAP ('newName1',
Named_struct ('user_id', '1', 'address', 'newAdress1', 'zip', 'newZip1', 'country', 'newCountry1')) AS postal_addresses
UNION ALL
SELECT '004' AS KEY,
'4' AS num,
MAP ('newName2',
Named_struct ('user_id', '1', 'address', 'newAdress2', 'zip', 'newZip2', 'country', 'newCountry2')) AS postal_addresses)
INSERT INTO nested_data_insert
SELECT * FROM dummy_table;

After you overwrite the data, you should verify that the data is changed in both Hive and HPE Ezmeral Data Fabric Database JSON tables.

Verify the data in the Hive table by using the SELECT * FROM syntax.

hive> SELECT * FROM nested_data_insert WHERE entry IN ('003', '004');

Sample output:

Table 4. Result of the insert query 3.3:
entry	num	postal_address
entry	num		USER_ID	ADDRESS	ZIP	COUNTRY
003	3	newName1	1	newAddress1	newZip1	newCountry1
004	4	newName2	1	newAddress2	newZip2	newCountry2

Verify the data in the HPE Ezmeral Data Fabric Database JSON table data using the findbyid statement:

findbyid '/nested_data_insert' --id 003
{
  "_id": "003",
  "num": {
       "$numberInt": 3
          },
  "postal_addresses": {
  "newName1": {
       "address": "newAdress1",
       "country": "newCountry1",
       "user_id": "1",
       "zip": "newZip1"
           }
        }
     }
findbyid '/nested_data_insert' --id 004
     {
       "_id": "004",
       "num": {
          "$numberInt": 4
          },
       "postal_addresses": {
           "newName2": {
              "address": "newAdress2",
              "country": "newCountry2",
              "user_id": "1",
              "zip": "newZip2"
           }
        }
    }

Warning: If you exclude columns both from the SELECT statement in your INSERT statement and from the table schema, the value of this column changes to NULL.

Finally, imagine that you want to overwrite the first row in

nested_data_insert
          (001)

with new values and overwrite the num column to NULL:

WITH dummy_table AS
(SELECT '001' AS KEY,
MAP ('newAdam',
Named_struct ('user_id', '1', 'address', 'newAdress', 'zip', 'newZip', 'country', 'newCountry')) AS postal_addresses)
INSERT INTO nested_data_insert (entry, postal_addresses)
SELECT * FROM dummy_table;

After you overwrite data, you should verify that the data is changed in both Hive and HPE Ezmeral Data Fabric Database JSON tables.

Verify the data in the Hive table by using the SELECT * FROM syntax.

hive> SELECT * FROM nested_data_insert WHERE entry = '001';

Sample output:

Table 5. Result of the insert query 3.5:
entry	num	postal_address
entry	num		USER_ID	ADDRESS	ZIP	COUNTRY
001	NULL	newAdam	1	newAddress	newZip	newCountry

Verify the data in the HPE Ezmeral Data Fabric Database JSON table (num row is not present):

findbyid '/nested_data_insert' --id 001
              
{
   "_id": "001",
   "postal_addresses": {
   "newAdam": {
         "address": "newAdress",
         "country": "newCountry",
         "user_id": "1",
         "zip": "newZip"
        }
     }
}