Kauko / specql

Automatic PostgreSQL CRUD queries

specql

An EXPERIMENTAL library for simple PostgreSQL queries with namespaced keys.

specql introspects your database at compile time and generates clojure.spec definitions for the rows and all the columns. You can then query for rows by giving a table, the keys to return and a map of keys to match against.

usage

See the docs for more examples.

(def db (make-my-db))

;; Define tables as [table-name :namespaced/table-keyword]
;; specql automatically creates specs for the columns of the table (in the same namespace as the
;; table keyword). For example if :employee/employees is the table and it has "id" and "name"
;; columns, specql will generate :employee/id and :employee/name specs with predicates
;; determined by the column data type
(define-tables define-db
  ["address" :address/address]
  ["employee" :employee/employees]
  ["department" :department/departments])

;; Querying

(fetch db
       ;; The table to query from
       :employee/employees
       ;; the columns to return
       #{:employee/id :employee/name :employee/title :employee/address}
       ;; the where clause
       {::employee/id 1})
;; => ({:employee/id 1
;;      :employee/name "Wile E. Coyote"
;;      :employee/title "Super genius"
;;      :employee/address {:address/street "Desert avenue 1"
;;                         :address/postal-code "31173"
;;                         :address/country "US"}})

Changes

0.6

• Support custom column transformations (eg. db enums to keywords)
• Better enum support (operators add type casts to parameters)
• Operators now take a third argument: column info
• Support "point" type as vector [x y] in composites
• New functions: columns and tables for querying the runtime registry
• Improved error handling and better error messages

work in progress

This is very much still work in progress.

Features I intend to implement:

• fetch
• update
• insert
• upsert
• delete
• JOIN navigation: has-one
• JOIN navigation: has-many
• standard operators for where ({:employee/name (like "%Smith%")})
• unpacking composite types (user defined record types as column values should be nested maps)

non-issues

This library is NOT intended to replace having to write SQL. The more complex SQL (like reporting queries and complex joins) queries are better written as SQL. Use yesql/jeesql/hugsql or the like.

This library intends to provide the most common case of CRUD queries and defer to SQL on the more difficult ones.

Specql can use VIEWs in the database just like tables, so you can write your complex queries as views and use them via specql.