Strict data sheet of new features of SQLite 3.37.0

Hello everyone , I only talk about technology, not haircut Tony teacher .

SQL 3.37.0 Added a new feature ： Strict field data types . This article introduces this function in detail .

If you think the article is useful , Welcome comments 、 give the thumbs-up 、 recommend

Introduction to data types

SQLite Flexible data types are used by default . for example , The field type of a table is INTEGER,SQLite When storing data, an attempt is made to convert the value inserted into the column to an integer . therefore , Insert string “123” The result is that integers are stored 123. however , If you cannot convert the inserted data to integers losslessly , for example , If the input is “xyz”, The original string... Will be inserted . More about it , You can refer to SQLite In document data type .

Some developers think SQLite Flexible type rules are good , But other developers are surprised , They think that SQL Strict data types in standards and other databases are better . For the latter ,SQLite 3.37.0 Start supporting this strict data type , This function can be specified based on the table level .

STRICT Data sheet

about CREATE TABLE sentence , If you specify STRICT Table options ,SQLite Strict data type rules will be applied to this table .STRICT The differences between keywords and ordinary tables are as follows ：

1. Each field must specify a data type , Fields without data types are no longer supported .

2. The data type of the field must be one of the following ：

   • INT
   • INTEGER
   • REAL
   • TEXT
   • BLOB
   • ANY

   Other data types are not allowed ,SQLite New data types may be added in the future .

3. except ANY Out of type , Other types of fields can only store content of the specified type or NULL（ Assume that there is no NOT NULL constraint ）.SQLite Try to convert the inserted data to the appropriate data type , Be similar to PostgreSQL、MySQL、SQL Server as well as Oracle. If the data cannot be converted losslessly to the specified data type , Will return SQLITE_CONSTRAINT_DATATYPE error .

4. ANY Fields of type can store any type of data （ If stored in NOT NULL constraint , Nature cannot store NULL）. Even if it's STRICT Data sheet ,ANY Type fields will not be type converted .

5. form PRIMARY KEY The field of... Will be added by default NOT NULL constraint . however , Even so , When we will NULL Value inserts a INTEGER PRIMARY KEY A field ,NULL The value is automatically converted to a unique integer , This is the same as the rule of ordinary tables .

6. PRAGMA integrity_check as well as PRAGMA quick_check The command will check STRICT Content type of data table field , And display any problems found .

for example ：

--  Fields without data types 
sqlite> CREATE TABLE t(id int, c1) STRICT;
Error: in prepare, missing datatype for t.c1 (1)

--  Unsupported field type 
sqlite> CREATE TABLE t(id int, c1 DATETIME) STRICT;
Error: in prepare, unknown datatype for t.c1: "DATETIME" (1)

--  Automatic data type conversion 
sqlite> CREATE TABLE t(id int, c1 text) STRICT;
sqlite> INSERT INTO t VALUES('01', 123);
sqlite> SELECT typeof(id), id, typeof(c1), c1 FROM t;
integer|1|text|123

-- ANY  data type 
sqlite> DROP TABLE t;
sqlite> CREATE TABLE t(id int, c1 any) STRICT;
sqlite> INSERT INTO t VALUES('01', '01');
sqlite> SELECT typeof(id), id, typeof(c1), c1 FROM t;
integer|1|text|01

STRICT Other behaviors of data tables are the same as those of ordinary data tables ：

• CHECK Same constraint .
• NOT NULL Same constraint .
• FOREIGN KEY Same constraint .
• UNIQUE Same constraint .
• DEFAULT The clauses are the same .
• COLLATE The clauses are the same .
• The generated columns are the same .
• ON CONFLICT The clauses are the same .
• The index is the same .
• AUTOINCREMENT identical .
• INTEGER PRIMARY KEY Field is equivalent to rowid, however INT PRIMARY KEY Fields are not equivalent .
• The disk storage format of table data is the same .

ANY data type

After years of verification , It is very useful for the same field to support different data types . In order to continue to support this function , Including the use of STRICT Data sheet ,SQLite New ANY data type . If the data type of a field is ANY, It means that it can store any type of data , Include integers 、 Floating point numbers 、 String or binary BLOB, And the data type and value will be saved as is . at present ,SQLite It is the only one that supports this function SQL database .

ANY The data type is STRICT The behavior in the data table is slightly different from that in the ordinary data table . about STRICT Data sheet ,ANY Fields of type always retain the original type of data ; For general data tables ,ANY A field of type will try to convert a number in string format to a number type , If the conversion is successful, it is stored as a digital value . for example ：

-- STRICT  Data sheet 
sqlite> CREATE TABLE t1(a ANY) STRICT;
sqlite> INSERT INTO t1 VALUES('000123');
sqlite> SELECT typeof(a), quote(a) FROM t1;
sqlite> text|'000123'

--  General data sheet 
sqlite> CREATE TABLE t1(a ANY);
sqlite> INSERT INTO t1 VALUES('000123');
sqlite> SELECT typeof(a), quote(a) FROM t1;
sqlite> integer|123