database-js-sqlparser v1.0.0

database-js-sqlparser

Common functionality for database-js drivers that operate on non-database backends

About

Database-js-sqlparser is a database-js driver that parses SQL statements and passes requests and commands to an underlying class which does the storage mechanism interaction. On its own it accomplishes nothing.

The sql parser supports the following SQL:

Tables

CREATE TABLE <table_name>(<column_name> <column_type>,...)

Where the column type can be one of:

• CHARACTER(n) - String of n length. Always padded or truncated to n length.
• VARCHAR(n) - String of up to n length. Always truncated to n length.
• BOOLEAN - Boolean (true or false)
• INTEGER, SMALLINT, BIGINT - Integer numeric values
• DECIMAL, NUMERIC, FLOAT, REAL, DOUBLE - Floating point numeric values, limited by Javascript's floating point implementation.
• DATE, TIME, TIMESTAMP - Date values
• TEXT - String values of arbitrary length

DROP TABLE <table_name>

Queries

SELECT [
    *,
    <column_name>[AS <column_label>],
    SUM|COUNT(<column_name)[AS <aggregate_label>]
] FROM <table_name>
[[INNER,LEFT,RIGHT] JOIN <table_name> ON <join_condition>]
[GROUP BY <column_name>]
[WHERE <where_condition>]
[ORDER BY <column_name>]
[LIMIT [row_offset,]<number_of_rows>]

Joins:

Inner, left and right joins are supported. Full or outer joins are not supported.

Aggregate Functions:

Sum and count are currently supported. Sum will not fail on non-numeric columns, but the return is undefined.

Inserts

INSERT INTO <table_name>(<column1>,<column2>,...) VALUES(<value1>,<value2>,...)

It's best to use paramaterized SQL:

INSERT INTO <table_name>(<column1>,<column2>,...) VALUES(?,?,...)

Updates

UPDATE <table_name> SET <column1> = <value1>, <column2> = <value2>,...
[WHERE <where_condition>]

Using parameterized SQL:

UPDATE <table_name> SET <column1> = ?, <column2> = ?,...
[WHERE <where_condition>]

Deletes

DELETE FROM <table_name> [WHERE <where_condition>]

Implementation in an extending class

A class extending the database-js-sqlparser class needs to override seven methods. Each method needs to return a Promise to allow for asynchronous implementations.

ready() : Promise<boolean>

Indicates that the underlying storage mechanism is loaded and ready to receive reads and writes.

To implement an always ready driver, use the following signature:

ready() {
    return Promise.resolve(true);
}

close() : Promse<boolean>

Allows the underlying storage mechanism to close if necessary.

load(table: string) : Promise<Array<{key:string:any}>>

Loads the rows from for a given table from the underlying storage and returns them via the Promise.

The resolved value of the Promise needs to be an array of table rows, where each row is a JSON like object with the column names as keys for the row values:

[
    {
        "id": 1,
        "name": "Me",
        "age": 32
    },
    {
        "id": 2,
        "name": "You",
        "age": 27
    }
]

store(table: string, index: string|number, row: any) : Promise<string|number>

Updates or inserts a row into the underlying storage system. If index is a string of number, then the action is an update, if index is null or undefined this is an insert. Resolves the promise with the updated or inserted index.

remove(table: string, index: string|number) : Promise<string|number>

Removes a row from the underlying storage system. Resolves the promise with the index that was removed.

create(table: string, definition: Array<column_definition>) : Promise<boolean>

Creates a new table according to the passed definition, resolves with true if successful.

The column definition is as follows:

{
    "name": string,    // The column name
    "index": number,   // The column index, can be ignored
    "type": "string"|"integer"|"float"|"date",
    "length"?: number, // For VARCHAR(n) or CHARACTER(n) the string length limit
    "pad"?: " ",       // For CHARACTER(n) the string to pad short strings with
}

drop(table: string) : Promise<boolean>

Drops the table from the underlying storage system. The user will expect the table data to be removed as well.