lupdo-mssql
v2.3.0
Published
Sql Server Driver For Lupdo
Downloads
6
Maintainers
Readme
lupdo-mssql
Lupdo Driver For SQL Server.
Supported Databases
- sql-Server (v2017,v2019,v2022)
Third Party Library
Lupdo-mssql, under the hood, uses stable and performant npm packages:
Note Lupdo-mssql use forked tedious-better-data-types to preserve data integrity
Usage
Base Example
const { createMssqlPdo } = require('lupdo-mssql');
// ES6 or Typescrypt
import { createMssqlPdo } from 'ludpo-mssql';
const pdo = createMssqlPdo(
{
server: 'localhost',
options: {
port: 1433,
database: 'database'
},
authentication: {
type: 'default',
options: {
userName: 'user',
password: 'password'
}
}
},
{ min: 2, max: 3 }
);
const run = async () => {
const statement = await pdo.query('SELECT 2');
const res = statement.fetchArray().all();
console.log(res);
await pdo.disconnect();
};
run();
Driver Options
http://tediousjs.github.io/tedious/api-connection.html#function_newConnection
Lupdo-mssql patch for tedious expose 4 new options:
customParsers: CustomParsers | undefined
Custom Parsers For Values Types. (default:undefined
)returnDecimalAndNumericAsString?: boolean
If true, Numeric and Decimal will be serialized as string. (default: false)returnDateTimeAsObject?: boolean
If true, Dates and Times will be serialized as object. (default: false)returnMoneyAsString?: boolean
If true, Money and SmallMoney will be serialized as string. (default: false)
Note The
server
option also accepts a list ofhost:port
the pool will generate the connection using a random host from the list.
Tedious Overrides
By default Ludpo-mssql overrides user connection options with this:
{
tediousOptions.options.debug = undefined;
tediousOptions.options.rowCollectionOnDone = false;
tediousOptions.options.rowCollectionOnRequestCompletion = false;
tediousOptions.options.abortTransactionOnError = false;
tediousOptions.options.useColumnNames = false;
tediousOptions.options.useUTC = false;
tediousOptions.options.camelCaseColumns = false;
tediousOptions.options.debug = {};
tediousOptions.options.enableConcatNullYieldsNull = true;
tediousOptions.options.enableCursorCloseOnCommit = false;
tediousOptions.options.enableImplicitTransactions = false;
tediousOptions.options.enableNumericRoundabort = false;
tediousOptions.options.enableQuotedIdentifier = true;
tediousOptions.options.columnNameReplacer = undefined;
// new config after patch tedious
tediousOptions.options.returnDecimalAndNumericAsString = true;
tediousOptions.options.returnMoneyAsString = true;
tediousOptions.options.returnDateTimeAsObject = true;
tediousOptions.options.customParsers = mssqlParser;
}
Lupdo mssql has a custom type parser
boolean
are returned as number 1 or 0intN
are returned as number or BigInt when necessarybinary
are returned as Buffer- all others types are always returned as string
Constants
MSSQL_DATE_BINDING
Choose Mssql DateTime TYPE Binding between: [Default MSSQL_DATE_BINDING_TEMPORAL]MSSQL_DATE_BINDING_TEMPORAL
Mssql DateTime TYPE Binding use Temporal InstantMSSQL_DATE_BINDING_DATE
Mssql DateTime TYPE Binding use Date
MSSQL_PARAM_SMALLDATETIME
Param type for sql-server SmallDateTime
Lupdo Exec
Ludpo-mssql will return affected row number, even if a select will be executed.
const affected = await pdo.exec("INSERT INTO test (name) values ('Claudio')");
console.log(affected); // 1
const affected = await pdo.exec('SELECT * FROM test');
console.log(affected); // row retrieved
Parameters Binding
Lupdo-mssql does not support array of parameters.
The use of TypedBinding is strongly recommended for prepared Statement.
Warning Lupdo-mssql can only retrieve the correct type from the value in certain restricted circumstances.
Parameters Binding Map To DB Type
- PARAM_BINARY
Binary
- PARAM_BOOLEAN
Bit
- PARAM_CHAR
NChar
- PARAM_DATE
Date
- MSSQL_PARAM_SMALLDATETIME
SmallDateTime
- PARAM_DATETIME
DateTime
- PARAM_TIMESTAMP
DateTime2
- PARAM_DATETIMEZONE
DateTimeOffset
- PARAM_TEXT
NText;
- PARAM_DECIMAL
Decimal
- PARAM_FLOAT
Float
** - PARAM_INTEGER
Int
- PARAM_NUMERIC
Numeric
** - PARAM_TIME
Time
- PARAM_VARBINARY
VarBinary
- PARAM_BIGINT
NVarChar
** - PARAM_VARCHAR
NVarChar
- PARAM_GEOMETRY
NVarChar
** when value < Number.MIN_SAFE_INTEGER || value > Number.MAX_SAFE_INTEGER
Javascript can not safely generate a BigNumber so value will be sent to Database as NVarChar
Parameter Value Map To DB Type
When parameter are bound as primitive Lupdo-mssql try to generate the right column Type with this rule
- is Boolean
Bit
- is BigInt
NVarChar
orInt
- is Number
Numeric
orNvarChar
orInt
- is Date
DateTime
- is Buffer
VarBinary
- others
NVarChar
Warning null value will be always typed as NVarChar, but is not compatible with some database column types and sql-server will raise an error. Please prefer Lupdo TypedBinding
TypedBinding Options
option { scale: number }
is available for:
- PARAM_TIMESTAMP
- PARAM_DATETIMEZONE
- PARAM_TIME
when scale is provided nanoseconds will be rounded to provided scale even if db column has a larger scale.
Mssql Named Parameter
Lupdo-mssql support named parameter with syntax :name
, the support is guaranteed only if all placeholder have a binding.
Native @param
syntax is supported.
Mssql Numeric Parameter
Lupdo-mssql support numeric parameter with syntax ?
, you can escape it using syntax ??
.
Kill Connection
Lupdo-mssql support kill query.
Mssql Output
Lupdo-mssql support queries with OUTPUT
, results can be fetched from statement.
const stmt = pdo.query("INSERT INTO users (name, gender) OUTPUT INSERTED.* VALUES ('Claudio', 'All');");
console.log(stmt.fetchArray().all());
/*
[
[33, 'Claudio', 'All']
]
*/
Mssql lastInsertId
When pdo.query()
is executed, lupdo-mssql automatically try to fetch LastInsertId and if available it will return last id when stmt.lastInsertId()
is called.
Lupdo-mssql can fetch LastInsertId on real-time when called inside a transaction
or when statement is prepared through pdo.prepare()
.
If you pass sequence name as parameter, it should retrieve current value of sequence.
Warning Calling
stmt.lastInsertId()
without sequence name outside a transaction or a PreparedStatement should not works.
Note You can always get insert ID through
output syntax
.
Tedious Fork
Here you can find a discussion about Tedious Data Types.
Lupdo-mssql use a forked version tedious-better-data-types
Decimal and Numeric
Tedious by default read Decimal and Numeric bytes from Buffer as Number
, when data are big decimal or big integer precision and scale are lost.
Through the new options returnDecimalAndNumericAsString
it will read byte from Buffer as String
without loosing precision and scale.
Note credits to tedious-decimal fork.
Money
Tedious by default read Money bytes from Buffer as Number
, when data are big decimal or big integer precision and scale are lost.
Through the new options returnMoneyAsString
it will read byte from Buffer as Number
but it will cast high
value through BigInt
and merge low and high value as String
without loosing precision and scale.
Datetime Object
Tedious by default read Date and Time bytes from Buffer as Number
, but to generate a real date (sql server do not return a real date) it use javascript Date
. This cause a lot of problem due to Local System Timezone and it also lost precision due to microseconds and nanoseconds (tedious try to fix it with date.deltaNanoseconds).
Through the new options returnDateTimeAsObject
it will return db data as a DateObject.
interface DateTimeObject {
// year from which start adding
startingYear: number;
// day to add
days?: number;
// minutes to add
minutes?: number;
// milliseconds to add
milliseconds?: number;
// nanoseconds to add
nanoseconds?: number;
// timezone offset in minutes
offset?: number;
}
Here an example to generate a Temporal.ZoneDateTime.
// dto is DateTimeObject
const timezone = dto.offset ? offsetToTimezone(dto.offset) : '+00:00';
// console.log(dto);
const instant = Temporal.Instant.from(
`${dto.startingYear.toString().padStart(4, '0')}-01-01 00:00:00.000000000${timezone}`
);
const minutes = (dto.minutes ?? 0) + (dto.offset ?? 0);
let zdt = instant.toZonedDateTimeISO(timezone).add({
days: dto.days ?? 0,
milliseconds: dto.milliseconds ?? 0,
nanoseconds: dto.nanoseconds ?? 0
});
if (minutes !== 0) {
zdt = zdt[minutes > 0 ? 'add' : 'subtract']({
minutes: Math.abs(minutes)
});
}
Custom Parsers
Tedious by default doesn't Have a Custom Parser. This cause a lot of problem due to Complexity of Variant Type And TypeN
.
Through the new options customParsers
it will apply a custom Callback if type is found on customParsers
.
First Parameter is The original callback, second parameter is the value decoded from the buffer.
Before quit custom function must to call the original callback with the value, otherwise iterator will be stuck forever.
type CustomParserCallback = (parserCallback: (value: unknown) => void, ...value: any[]) => void;
interface CustomParsers {
BigInt?: CustomParserCallback;
Binary?: CustomParserCallback;
Bit?: CustomParserCallback;
Char?: CustomParserCallback;
Date?: CustomParserCallback;
DateTime?: CustomParserCallback;
DateTime2?: CustomParserCallback;
DateTimeOffset?: CustomParserCallback;
Decimal?: CustomParserCallback;
Float?: CustomParserCallback;
Image?: CustomParserCallback;
Int?: CustomParserCallback;
Money?: CustomParserCallback;
NChar?: CustomParserCallback;
NText?: CustomParserCallback;
Real?: CustomParserCallback;
SmallDateTime?: CustomParserCallback;
SmallInt?: CustomParserCallback;
SmallMoney?: CustomParserCallback;
Text?: CustomParserCallback;
Time?: CustomParserCallback;
TinyInt?: CustomParserCallback;
UDT?: CustomParserCallback;
UniqueIdentifier?: CustomParserCallback;
Xml?: CustomParserCallback;
}
Here An Example:
const customParser = {
BigInt: function (parserCallback: (value: unknown) => void, value: string | null): void {
if (value === null) {
return parserCallback(null);
}
const bigint = BigInt(value);
if (bigint > Number.MAX_SAFE_INTEGER || bigint < Number.MIN_SAFE_INTEGER) {
parserCallback(bigint);
} else {
parserCallback(Number(value));
}
}
};
Temporal Types
tedious-better-data-types add new Types to Manage Date And Time columns:
- TYPES.DateTemporal
- TYPES.DateTime2Temporal
- TYPES.DateTimeOffsetTemporal
- TYPES.DateTimeTemporal
- TYPES.SmallDateTimeTemporal
- TYPES.TimeTemporal
this new TYPES use an experimental @js-temporal/polyfill for javascript Temporal.
through these new types, full data consistency is ensured, when you bind a date string to a parameter.
SELECT
CAST('2007-05-17 23:59:59.999999999' AS time(7)) AS 'time'
,CAST('2007-05-08 23:59:29.999999999 -14:00' AS date) AS 'date'
,CAST('2007-05-13 23:59:29.998' AS smalldatetime) AS
'smalldatetime'
,CAST('2007-05-13 23:59:29.999' AS datetime) AS 'datetime'
,CAST('2007-05-13 23:59:59.998456789 -01:00' AS datetime2(7)) AS
'datetime2'
,CAST('2007-05-13 23:59:59.998456789 -01:00' AS datetimeoffset(7))AS
'datetimeoffset';
Output from these query will be always consistent no matter which is your timezone locale.
The options.useUTC
does not affect temporal types.
Valid string Dates are ISO 8601/RFC 3339 string like 2020-01-23T17:04:36.491865121-08:00
or 2020-01-24T01:04Z
.
[YYYY-MM-DD][T| ][hh:{mm:ss.sssssssss}][Z|±hh{:mm}]
Warning when string contains microseconds and/or timezone
TYPES.DateTimeTemporal
andTYPES.SmallDateTimeTemporal
will raise an error.
Output String Date will Be:
- TYPES.DateTemporal output
YYYY-MM-DD
- TYPES.DateTime2Temporal
YYYY-MM-DD hh:mm:ss.ssssssss
- TYPES.DateTimeOffsetTemporal
YYYY-MM-DD hh:mm:ss.ssssssss+hh:mm
- TYPES.DateTimeTemporal
YYYY-MM-DD hh:mm:ss.sss
- TYPES.SmallDateTimeTemporal
YYYY-MM-DD hh:mm:00.000
- TYPES.TimeTemporal
hh:mm:ss.sssssss
Warning Using a date with the new Temporal Types, do not guarantee consistence for datetimeoffset type. Use Strings.
Note Since the temporal library is still experimental, package has been locked to version 0.4.3