node package manager


Concise SQL generation for Postgres and MySQL

Gesundheit - Concise SQL generation in CoffeeScript

Gesundheit generates SQL, it does this using a nice CoffeeScript friendly syntax.


npm install gesundheit


# SELECT chairs.* FROM chairs
q = select.from('chairs').visit -> # or Select.from, or SELECT.from   
    # SELECT chairs.type, chairs.size FROM chairs
    @fields "type", "size"

    # SELECT ... WHERE chairs.type = ? AND chairs.size < ?
    @where type: 'recliner', weight: {lt: 25}

    # SELECT ..., people.* FROM chairs INNER JOIN people ON people.chair_id = WHERE ...
    @join "people", on: {chair_id: @rel('chairs').field('id')}

    # SELECT ..., FROM ...
    @fields "name"
    # SELECT ... WHERE ... AND people.gender = ?
    @where gender: 'M'

# More concisely:
q = select.from('chairs', ['type', 'size']).where(type: 'recliner', weight: {lt: 25})
q.join("people", on: {chair_id: q.rel('chairs').field('id')}).fields("name").where(gender: 'M')

# Generate the SQL
q.toSql() # SELECT chairs.type, chairs.size, FROM chairs INNER JOIN people ...

# Execute the query with a client (must have a `query` method)
client = query: (sql, params, cb) -> cb null, [{col1: "Cool beans"}]

q.execute client, (err, res) ->
    throw err if err?
    # do something with result

# Execute the query with a connection pool (must have an 'acquire' method)
pool = 
    acquire: (cb) -> cb client
    release: ->

q.execute pool, (err, res) ->
    throw err if err?
    # do something with result

# Table aliasing
# SELECT ArtWTF.* FROM a_real_table_with_twenty_fields AS ArtWTF
q = select.from ArtWTF: "a_real_table_with_twenty_fields"

# Works with joins as well
q = q.join {so: 'some_other'}, on: {hard: q.rel('ArtWTF').field('is_it')}

# Field aliasing
# SELECT t.really_long_field AS 'rlf' FROM long_table AS t
q = select.from(t: 'long_table').field(rlf: 'really_long_field')

Gesundheit is not an ORM

But a pretty decent ORM could be built with it. The recommended approach is to provide a dialect (or extend an existing one) with render methods that understand your model and field types.


  • More testing
  • DBMS specific dialects




Stephen Sugden