pyramid

(Formerly called autonormal)

A library for storing graph data in a Clojure map that automatically normalizes nested data and allows querying via EQL, optimized for read (query) performance.

Use cases

The primary use case this library was developed for was to act as a client side cache for pathom APIs. However, you can imagine any time you might reach for DataScript to store data as entities, but where you need fast nested / recursive querying of many attributes and don't need the full expressive power of datalog, as being a good use case for pyramid.

Another common use case is like a select-keys on steroids: the ability to do nested selections on complex maps with nested collections pops up very often in code. Pyramid can take any non-normalized map and execute an EQL query on it, returning the result.

Project status

Pyramid has been used in production and it's core API is very stable.

Experiments are still on going in pyramid.query to provide datomic-style datalog query capabilities for maps.

Usage

Normalizing

A db is simply a map with a tabular structure of entities, potentially with references to other entities.

Pyramid currently makes a default conventional assumption: your entities are identified by a keyword whose name is "id", e.g. :id, :person/id, :my.corp.product/id, etc.

(require '[pyramid.core :as p])

(def data
  {:person/id 0 :person/name "Rachel"
   :friend/list [{:person/id 1 :person/name "Marco"}
                 {:person/id 2 :person/name "Cassie"}
                 {:person/id 3 :person/name "Jake"}
                 {:person/id 4 :person/name "Tobias"}
                 {:person/id 5 :person/name "Ax"}]})

;; you can pass in multiple entities to instantiate a db, so `p/db` gets a vector
(def animorphs (p/db [data]))
;; => {:person/id {0 {:person/id 0 
;;                    :person/name "Rachel"
;;                    :friend/list [[:person/id 1]
;;                                  [:person/id 2]
;;                                  [:person/id 3]
;;                                  [:person/id 4]
;;                                  [:person/id 5]]}
;;                 1 {:person/id 1 :person/name "Marco"}
;;                 2 {:person/id 2 :person/name "Cassie"}
;;                 3 {:person/id 3 :person/name "Jake"}
;;                 4 {:person/id 4 :person/name "Tobias"}
;;                 5 {:person/id 5 :person/name "Ax"}}}

The map structure of a db is very efficient for getting info about any particular entity; it's just a get-in away:

(get-in animorphs [:person/id 1])
;; => {:person/id 1 :person/name "Marco"}

You can assoc/dissoc/update/etc. this map in whatever way you would like. However, if you want to accrete more potentially nested data, there's a helpful add function to normalize it for you:

;; Marco and Jake are each others best friend
(def animorphs-2
  (p/add animorphs {:person/id 1
                    :friend/best {:person/id 3
                                  :friend/best {:person/id 1}}}))
;; => {:person/id {0 {:person/id 0 
;;                    :person/name "Rachel"
;;                    :friend/list [[:person/id 1]
;;                                  [:person/id 2]
;;                                  [:person/id 3]
;;                                  [:person/id 4]
;;                                  [:person/id 5]]}
;;                 1 {:person/id 1
;;                    :person/name "Marco" 
;;                    :friend/best [:person/id 3]}
;;                 2 {:person/id 2 :person/name "Cassie"}
;;                 3 {:person/id 3
;;                    :person/name "Jake"
;;                    :friend/best [:person/id 1]}
;;                 4 {:person/id 4 :person/name "Tobias"}
;;                 5 {:person/id 5 :person/name "Ax"}}}

Note that our animorphs db is an immutable hash map; add simply returns the new value. It's up to you to decide how to track its value and keep it up to date in your system, e.g. in an atom.

Adding non-entities

Maps that are added are typically entities, but you can also add arbitrary maps and add will merge any non-entities with the database, normalizing and referencing any nested entities.

Using this capability, you can create additional indexes on your entities. Example:

(def animorphs-3
  (p/add animorphs-2 {:species {:andalites [{:person/id 5
                                             :person/species "andalite"}]}}))
;; => {:person/id {,,,
;;                 5 {:person/id 5
;;                    :person/name "Ax"
;;                    :person/species "andalite"}}
;;     :species {:andalites [[:person/id 5]]}}

Pull queries

This library implements a fast EQL engine for Clojure data.

(p/pull animorphs-3 [[:person/id 1]])
;; => {[:person/id 1] {:person/id 1
;;                     :person/name "Macro"
;;                     :friend/best {:person/id 3}}}

You can join on idents and keys within entities, and it will resolve any references found in order to continue the query:

(p/pull animorphs-3 [{[:person/id 1] [:person/name
                                      {:friend/best [:person/name]}]}])
;; => {[:person/id 1] {:person/name "Marco"
;;                     :friend/best {:person/name "Jake"}}}

Top-level keys in the db can also be joined on.

(p/pull animorphs-3 [{:species [{:andalites [:person/name]}]}])
;; => {:species {:andalites [{:person/name "Ax"}]}}

Recursion is supported:

(def query '[{[:person/id 0] [:person/id
                              :person/name
                              {:friend/list ...}]}])

(= (-> (p/pull animorphs-3 query)
       (get [:person/id 0]))
   data)
;; => true

See the EQL docs and tests in this repo for more examples of what's possible!

More details

Collections like vectors, sets and lists should not mix entities and non-entities. Collections are recursively walked to find entities.

To get meta-information about what entities were added or queried, use the add-report and pull-report functions.

To delete an entity and all references to it, use the delete function.

Tips & Tricks

Replacing an entity

Data that is added about an existing entity are merged with whatever is in the db. To replace an entity, dissoc it first:

(-> (p/db [{:person/id 0 :foo "bar"}])
    (update :person/id dissoc 0)
    (p/add {:person/id 0 :bar "baz"}))
;; => {:person/id {0 {:person/id 0 :bar "baz"}}}

Getting data for a specific entity

Since a db is a simple map, you can always use get-in to get basic info regarding an entity. However, if your entity contains references, it will not resolve those for you. Enter EQL!

To write an EQL query to get info about a specific entity, you can use an ident to begin your query:

(p/pull animorphs-3 [[:person/id 1]])
;; => {[:person/id 1] 
;;     {:person/id 1, :person/name "Marco", :friend/best #:person{:id 3}}}

You can add to the query to resolve references and get information about, e.g. Marco's best friend:

(p/pull animorphs-3 [{[:person/id 1] [:person/name
                                      {:friend/best [:person/name]}]}])
;; => {[:person/id 1] {:person/name "Marco", :friend/best #:person{:name "Jake"}}}

Datalog queries

The latest version of pyramid includes an experimental datomic/datascript-style query engine in pyramid.query. It is not production ready, but is good enough to explore a pyramid db for developer inspection and troubleshooting.

(require '[pyramid.query :refer [q]])


;; find the names of people with best friends, and their best friends' name
(q [:find ?name ?best-friend
    :where
    [?e :friend/best ?bf]
    [?e :person/name ?name]
    [?bf :person/name ?best-friend]]
   animorphs-3)
;; => (["Marco" "Jake"] ["Jake" "Marco"])

Features

Prior art

Fulcro
@souenzzo's POC eql-refdb
MapGraph
EntityDB
DataScript and derivatives
Pathom
juxt/pull
ribelo/doxa

Cheap and reliable Node.js hosting starts at $3/month, and $1/month static HTML hosting

lilactown / pyramid

Programming Languages

Labels

Projects that are alternatives of or similar to pyramid