I am new with Elasticsearch and I have spent several hours trying to solve this so, Thanks in advance if you will try to help me.
:) (Not too) Short explanation: (What I have so far and what I try to achieve):
I created a CouchDB database (spain_locales) that contains more then 8000 documents with Spanish Cities and Provinces. On the other hand I have a HTML Form with a jQuery Autocomplete and I show results as I type. I connect to the ElasticSearch from a PHP (Laravel Service Provider) that I created and I return results for the jQuery Autocomplete. I suppose that this can be made by connecting from client directly to the ElasticSearch but for security reasons I prefer it like this for now.
:( The Problem:
The results I get from ElasticSearch are not exactly what I expect and I don't know how to fix what I have or if it's the correct way to do it. I don't know if the bool query it's for what I need or if I should use other type of query.
I only get results if I type the words exactly like they are in the Database:
If I type Álava I obtain results but NOT for Alava ( The Á accent makes the difference)
I don't obtain results until I type the complete word:
If I type Albacete I obtain results but NOT for Albacet
I used CouchDB River Plugin for ElasticSearch to synchronise CouchDB with ElasticSearch >> https://github.com/elasticsearch/elasticsearch-river-couchdb and I made it with the following command trough terminal:
curl -XPUT 'localhost:9200/_river/spain_locales/_meta' -d '{
"type" : "couchdb",
"couchdb" : {
"host" : "localhost",
"port" : 5984,
"db" : "spain_locales",
"filter" : null
},
"index" : {
"index" : "spain_locales",
"type" : "spain_locales",
"bulk_size" : "100",
"bulk_timeout" : "10ms"
}
}'
I also tried with:
curl -XPUT 'localhost:9200/_river/spain_locales/_meta' -d '{
"type" : "couchdb",
"couchdb" : {
"host" : "localhost",
"port" : 5984,
"db" : "spain_locales",
"filter" : null
},
"index" : {
"number_of_shards" : 2,
"refresh_interval" : "1s",
"analysis": {
"analyzer": {
"folding": {
"tokenizer": "standard",
"filter": [ "lowercase", "asciifolding" ]
}
}
},
"index" : "spain_locales",
"type" : "spain_locales",
"bulk_size" : "100",
"bulk_timeout" : "10ms"
}
}'
None of above return any error and successfully create the _river synchronisation but still have the accents and whole words issue.
I also tried to somehow apply the needed filters with the following command trough the terminal:
curl -XPUT 'localhost:9200/spain_locales/' -d '
{
"settings": {
"analysis": {
"analyzer": {
"folding": {
"tokenizer": "standard",
"filter": [ "lowercase", "asciifolding" ]
}
}
}
},
"uuid":"KwKrBc3uQoG5Ld1nOdc5rQ"
}'
But I get the following error:
{"error":"IndexAlreadyExistsException[[spain_locales] already exists]","status":400}
CouchDB Documents Examples:
{
"_id": "1",
"_rev": "1-087ddbe8593f68f1d7d37a9c3f6de787",
"Provincia": "Álava",
"Poblacion": "Alegría-Dulantzi",
"helper": ""
}
{
"_id": "10",
"_rev": "1-ce38dcdabeb3b34d34d2296c6e2fdf24",
"Provincia": "Álava",
"Poblacion": "Ayala/Aiara",
"helper": ""
}
{
"_id": "100",
"_rev": "1-72e66601e378ee48519aa93601dc0717",
"Provincia": "Albacete",
"Poblacion": "Herrera (La)",
"helper": "La Herrera"
}
PHP Service Provider / Controller:
public function searchzones(){
$q = (Input::has('term')) ? Input::get('term') : 'null';
$params['index'] = 'spain_locales';
$params['type'] = 'spain_locales';
$params['body']['query']['bool']['should'] = array(
array('match' => array('Poblacion' => $q)),
array('match' => array('Provincia' => $q))
);
$query = $this->elasticsearch->search($params);
if ($query['hits']['total'] >= 1){
$results = $query['hits']['hits'];
foreach ($results as $zone) {
$databag[] = array( "value" => $zone['_source']['Poblacion'].', '.$zone['_source']['Provincia'],
"state" => $zone['_source']['Provincia'],
"city" => $zone['_source']['Poblacion'],
);
}
} else {
$results = ['res' => null];
$databag[] = array();
}
return $databag;
} // End Search Zones
jQuery (JavaScript):
// Sugest locations when user type in zones
$(document).ready(function() {
$('#zones').autocomplete({
source : applink + 'ajax/searchzones',
select : function(event, ui){
console.log(ui);
}
}); // End autocomplete
}); // End Document ready
HTML Form part (Twitter Bootstrap):
<div class="form-group">
<div class="input-group input-append dropdown">
<input type="text" class="form-control typeahead" placeholder="City name" name="zones" id="zones">
<div class="input-group-btn" >
<button type="button" class="btn btn-default dropdown-toggle" data-toggle="dropdown"><span class="caret"></span></button>
<ul class="dropdown-menu dropdown-menu-right" id="dropZonesAjax">
</ul>
</div>
</div>
<div id="zonesAjax"></div>
</div>
I found the following resource: http://www.elasticsearch.org/guide/en/elasticsearch/guide/current/asciifolding-token-filter.html but I don't know how to implement/achieve that.
Thanks a lot for your time and for try to help! Sorry for my English!
