pelias
/
api
mirror of https://github.com/pelias/api.git
1
0
Fork 0
Code Issues Releases Activity
Browse Source

refactor search

pull/229/head
Peter Johnson 9 years ago
parent
f5e182c63b
commit
6f817938bc
1 changed files with 130 additions and 157 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 285
      query/search.js

285
query/search.js
Unescape View File

@ -1,188 +1,161 @@
var queries = require('geopipes-elasticsearch-backend').queries, var peliasQuery = require('pelias-query'),
sort = require('../query/sort'), sort = require('../query/sort'),
adminFields = require('../helper/adminFields')(), adminFields = require('../helper/adminFields')();
addressWeights = require('../helper/address_weights');
//------------------------------
// general-purpose search query
//------------------------------
function generate( params ){ var query = new peliasQuery.layout.FilteredBooleanQuery();
var centroid = null;
if ( params.lat && params.lon ){ // mandatory matches
centroid = { query.score( peliasQuery.view.boundary_country, 'must' );
lat: params.lat, query.score( peliasQuery.view.ngrams, 'must' );
lon: params.lon
}; // scoring boost
} query.score( peliasQuery.view.phrase );
query.score( peliasQuery.view.focus );
// address components
query.score( peliasQuery.view.address('housenumber') );
query.score( peliasQuery.view.address('street') );
query.score( peliasQuery.view.address('postcode') );
// admin components
query.score( peliasQuery.view.admin('alpha3') );
query.score( peliasQuery.view.admin('admin0') );
query.score( peliasQuery.view.admin('admin1') );
query.score( peliasQuery.view.admin('admin1_abbr') );
query.score( peliasQuery.view.admin('admin2') );
query.score( peliasQuery.view.admin('local_admin') );
query.score( peliasQuery.view.admin('locality') );
query.score( peliasQuery.view.admin('neighborhood') );
// non-scoring hard filters
query.filter( peliasQuery.view.boundary_circle, 'must' );
query.filter( peliasQuery.view.boundary_rect, 'must' );
var query = queries.distance( centroid, { size: params.size } ); // --------------------------------
var input = params.input;
if (params.bbox) { function generate( clean ){
query = queries.bbox ( centroid, { size: params.size, bbox: params.bbox } );
var vs = new peliasQuery.Vars( peliasQuery.defaults );
// set input text
vs.var( 'input:name', clean.input );
// set size
if( clean.size ){
vs.var( 'size', clean.size );
} }
query.query.filtered.query = { // focus point
'bool': { if( clean.lat && clean.lon ){
'must': [], vs.set({
'should': [] 'focus:point:lat': clean.lat,
} 'focus:point:lon': clean.lon
}; });
}
if (params.parsed_input) {
// update input
if (params.parsed_input.number && params.parsed_input.street) {
input = params.parsed_input.number + ' ' + params.parsed_input.street;
} else if (params.parsed_input.admin_parts) {
input = params.parsed_input.name;
}
addParsedMatch(query, input, params.parsed_input); // bbox
if( clean.bbox ){
vs.set({
'boundary:rect:top': clean.bbox.top,
'boundary:rect:right': clean.bbox.right,
'boundary:rect:bottom': clean.bbox.bottom,
'boundary:rect:left': clean.bbox.left
});
} }
// add search condition to distance query // address parsing
query.query.filtered.query.bool.must.push({ if( clean.parsed_input ){
'match': {
'name.default': { // is it a street address?
'query': input, var isStreetAddress = clean.parsed_input.hasOwnProperty('number') && clean.parsed_input.hasOwnProperty('street');
'analyzer': 'peliasOneEdgeGram' if( isStreetAddress ){
} vs.var( 'input:name', clean.parsed_input.number + ' ' + clean.parsed_input.street );
} }
});
// I don't understand this
// add phrase matching query else if( clean.parsed_input.admin_parts ) {
// note: this is required for shingle/phrase matching vs.var( 'input:name', clean.parsed_input.name );
query.query.filtered.query.bool.should.push({
'match': {
'phrase.default': {
'query': input,
'analyzer': 'peliasPhrase',
'type': 'phrase',
'slop': 2
}
} }
});
query.sort = query.sort.concat( sort( params ) ); // or this..
else {
console.warn( 'chaos monkey asks: what happens now?' );
console.log( clean );
try{ throw new Error(); } catch(e){ console.error( e.stack ); } // print a stack trace
}
return query; // ==== add parsed matches [address components] ====
}
/** // house number
* Traverse the parsed input object, containing all the address parts detected in query string. if( clean.parsed_input.hasOwnProperty('number') ){
* Add matches to query for each identifiable component. vs.var( 'input:housenumber', clean.parsed_input.number );
* }
* @param {Object} query
* @param {string} defaultInput
* @param {Object} parsedInput
*/
function addParsedMatch(query, defaultInput, parsedInput) {
query.query.filtered.query.bool.should = query.query.filtered.query.bool.should || [];
// copy expected admin fields so we can remove them as we parse the address
var unmatchedAdminFields = adminFields.slice();
// address
// number, street, postalcode
addMatch(query, unmatchedAdminFields, 'address.number', parsedInput.number, addressWeights.number);
addMatch(query, unmatchedAdminFields, 'address.street', parsedInput.street, addressWeights.street);
addMatch(query, unmatchedAdminFields, 'address.zip', parsedInput.postalcode, addressWeights.zip);
// city
// admin2, locality, local_admin, neighborhood
addMatch(query, unmatchedAdminFields, 'admin2', parsedInput.city, addressWeights.admin2);
// state
// admin1, admin1_abbr
addMatch(query, unmatchedAdminFields, 'admin1_abbr', parsedInput.state, addressWeights.admin1_abbr);
// country
// admin0, alpha3
addMatch(query, unmatchedAdminFields, 'alpha3', parsedInput.country, addressWeights.alpha3);
addUnmatchedAdminFieldsToQuery(query, unmatchedAdminFields, parsedInput, defaultInput);
}
/** // street name
* Check for additional admin fields in the parsed input, and if any was found if( clean.parsed_input.hasOwnProperty('street') ){
* combine into single string and match against all unmatched admin fields. vs.var( 'input:street', clean.parsed_input.street );
* }
* @param {Object} query
* @param {Array} unmatchedAdminFields
* @param {Object} parsedInput
* @param {string} defaultInput
*/
function addUnmatchedAdminFieldsToQuery(query, unmatchedAdminFields, parsedInput, defaultInput) {
if (unmatchedAdminFields.length === 0 ) {
return;
}
var leftovers = []; // postal code
if( clean.parsed_input.hasOwnProperty('postalcode') ){
vs.var( 'input:postcode', clean.parsed_input.postalcode );
}
if (parsedInput.admin_parts) { // ==== add parsed matches [admin components] ====
leftovers.push(parsedInput.admin_parts);
}
else if (parsedInput.regions) {
leftovers.push(parsedInput.regions);
}
if (leftovers.length === 0) { // city
return; if( clean.parsed_input.hasOwnProperty('city') ){
} vs.var( 'input:admin2', clean.parsed_input.city );
}
leftovers = leftovers.join(' '); // state
if( clean.parsed_input.hasOwnProperty('state') ){
vs.var( 'input:admin1_abbr', clean.parsed_input.state );
}
// if there are additional regions/admin_parts found // country
if (leftovers !== defaultInput) { if( clean.parsed_input.hasOwnProperty('country') ){
unmatchedAdminFields.forEach(function (key) { vs.var( 'input:alpha3', clean.parsed_input.country );
// combine all the leftover parts into one string }
addMatch(query, [], key, leftovers);
});
}
}
/** // ==== deal with the 'leftover' components ====
* Add key:value match to query. Apply boost if specified. // @todo: clean up this code
*
* @param {Object} query
* @param {Array} unmatched
* @param {string} key
* @param {string|number|undefined} value
* @param {number|undefined} [boost] optional
*/
function addMatch(query, unmatched, key, value, boost) { // jshint ignore:line
if (typeof value === 'undefined') {
return;
}
var match = {}; // a concept called 'leftovers' which is just 'admin_parts' plus 'regions'.
var leftovers = [];
if( clean.parsed_input.hasOwnProperty('admin_parts') ){
leftovers.push( clean.parsed_input.admin_parts );
}
else if( clean.parsed_input.hasOwnProperty('regions') ){
leftovers.push( clean.parsed_input.regions );
}
if (boost) { // if we have 'leftovers' then assign them to any fields which
match[key] = { // currently don't have a value assigned.
query: value, if( leftovers.length ){
boost: boost var leftoversString = leftovers.join(' ');
}; var unmatchedAdminFields = adminFields.slice();
}
else { // cycle through fields and set fields which
match[key] = value; // are still currently unset
unmatchedAdminFields.forEach( function( key ){
if( !vs.isset( 'input:' + key ) ){
vs.var( 'input:' + key, leftoversString );
}
});
}
} }
query.query.filtered.query.bool.should.push({ 'match': match }); var result = query.render( vs );
result.sort = result.sort.concat( sort( clean ) );
removeFromUnmatched(unmatched, key);
}
/** // @todo: remove this hack
* If key is found in unmatched list, remove it from the array return JSON.parse( JSON.stringify( result ) );
*
* @param {Array} unmatched
* @param {string} key
*/
function removeFromUnmatched(unmatched, key) {
var index = unmatched.indexOf(key);
if (index !== -1) {
unmatched.splice(index, 1);
}
} }
module.exports = generate; module.exports = generate;

Write Preview
Loading…
Cancel
Save