@repositive/query-parser

• Installation
• Available Serializers
• Usage

The purpose of this library is to transform a search string to a tokenized data structure that can be used to perform analysis on the search input or as an intermediate structure to translate the query into other DSLs or query structures.

Features:

• Basic boolean algebra operators (AND, OR, NOT)
  • Implicit intersections (AND) on non quoted spaces.
  • Logic grouping with parens

  (white or blue) flower not thistle

• Exact match between quoted strings on non predicated tokens.

  flower "white daisy"

• Predicated/Filtered search

  family:Asteraceae population:>100000 england

Installation

With npm:

$ npm i @repositive/query-parser

Available Serializers

• Natural Language String
• ElasticSearch 2.x

Usage

The library exposes the following functions:

Tree construction

• token: (str: string) => Token
• predicate: ({key: string, value: string}) => Predicate
• and: <L extends Node, R extends Node>({left: L, right: R}) => AND<L, R>
• or: <L extends Node, R extends Node>({left: L, right: R}) => OR<L, R>
• not: <N extends Node>(negated: N) => NOT<N>

Tools

• fold: <R>(node: Node, f: (node: Node, l: R, r: R) => R, R) => R
• filter: <R> (node: Node, f: (node: Node) => node is R) => R[]
• path: (node: Node, target: Node) => Node[]
• remove: (node: Node, target: Node) => Node
• replace: ({on: Node, target: Node, replacement: Node}) => Node

Parsing natural language string

• fromNatural: (str: string) => Node
  Parses the current string and returns a boolean binary tree representing the search.

Serialization

• toNatural: (tree: Node) => string
  Serializes a boolean binary tree into a string emulating how a human would write it.
• toElastic2: (tree: Node) => any
  Serializes the boolean binary tree into a elasticsearch 2.x query.

import QP from 'npm:@repositive/query-parser';
// var QP = require(`@repositive/query-parser`); non ES6
/**
* "fromNatural" will generate the following tree from "is:user Istar NOT profession:developer":
* {                            
*   "_id": "6bd6c61f-eab6-43bc-81d2-97f96c7c5f0a",           
*   "_type": "AND",            
*   "left": {                  
*     "_id": "081c058a-e8cc-4ede-9637-6fd6593d5388",         
*     "_type": "predicate",    
*     "key": "is",             
*     "relation": "=",         
*     "value": "user"          
*   },                         
*   "right": {                 
*     "_id": "4719c7d4-965a-45cc-8132-87ed3acdc560",         
*     "_type": "AND",          
*     "left": {                
*       "_id": "8c8fae87-3aeb-4298-8f29-baf4c761ca12",       
*       "_type": "token",      
*       "value": "Istar"       
*     },                       
*     "right": {               
*       "_id": "b0b49e4a-c2cc-4954-bc1d-b68bad17f441",       
*       "_type": "NOT",        
*       "negated": {           
*         "_id": "323596ca-24fa-4ec5-a7a5-0d4d1ed45645",     
*         "_type": "predicate",                              
*         "key": "profession", 
*         "relation": "=",     
*         "value": "developer" 
*       }                      
*     }                        
*   }                          
* }         
*/
const tree = QP.fromNatural('is:user Istar NOT profession:developer')

/*
* We can find the profession predicate using the filter function
*
*/
const profession = QP.filter(tree, (n) => n.key === 'profession')[0];

/**
* If we remove the profession filter with "remove" we expect to end with the following tree:
* {
*   value: 'AND',
*   right: { text: 'Istar' },
*   left: {predicate: 'is', text: 'user'}
* }
*/
const professional = QP.remove(tree, profession);

/**
* Adding a new filter (attribute:awesome) to the three will return a new tree with the attribute inserted in the leftmost position
* {
*   value: 'AND',
*   left: { predicate: 'attribute', text: 'awesome'},
*   right: {
*     value: 'AND',
*     right: { text: 'Istar' },
*     left: { predicate: 'is', text: 'user'}
*   }
* }
*/
const awesome = QP.and({left: professional, right: QP.predicate({key: 'attribute', value: 'awesome'}));

/**
* The serialization value of the new tree using "toBoolString" resembles the text as a human would write it:
* "is:user Istar attribute: awesome"
*/
const newQueryString = QP.toNatural(awesome);