testing an elasticsearch custom analyzer - pipe delimited keywords - elasticsearch

Question

I have this index with pipe as custom analyzer. When I am trying to test it, it returns every char, and not pipe delimited words.
I am trying to construct for an use case where my input line keywords looks like: crockpot refried beans|corningware replacement|crockpot lids|recipe refried beans and EL will return matches after it has been exploded.
{
"keywords": {
"aliases": {
},
"mappings": {
"cloud": {
"properties": {
"keywords": {
"type": "text",
"analyzer": "pipe"
}
}
}
},
"settings": {
"index": {
"number_of_shards": "5",
"provided_name": "keywords",
"creation_date": "1513890909384",
"analysis": {
"analyzer": {
"pipe": {
"type": "custom",
"tokenizer": "pipe"
}
},
"tokenizer": {
"pipe": {
"pattern": "|",
"type": "pattern"
}
}
},
"number_of_replicas": "1",
"uuid": "DOLV_FBbSC2CBU4p7oT3yw",
"version": {
"created": "6000099"
}
}
}
}
}
When I am trying to test it following this guide.
curl -XPOST 'http://localhost:9200/keywords/_analyze' -d '{
"analyzer": "pipe",
"text": "pipe|pipe2"
}'
I get back char-by-char results.
{
"tokens": [
{
"token": "p",
"start_offset": 0,
"end_offset": 1,
"type": "word",
"position": 0
},
{
"token": "i",
"start_offset": 1,
"end_offset": 2,
"type": "word",
"position": 1
},
{
"token": "p",
"start_offset": 2,
"end_offset": 3,
"type": "word",
"position": 2
},
{
"token": "e",
"start_offset": 3,
"end_offset": 4,
"type": "word",
"position": 3
},

Answer 1

Good work, you're almost there. Since the pipe | character is a reserved character in regular expressions, you need to escape it like this:
"tokenizer": {
"pipe": {
"pattern": "\\|", <--- change this
"type": "pattern"
}
}
And then your analyzer will work and produce this:
{
"tokens": [
{
"token": "pipe",
"start_offset": 0,
"end_offset": 4,
"type": "word",
"position": 0
},
{
"token": "pipe2",
"start_offset": 5,
"end_offset": 10,
"type": "word",
"position": 1
}
]
}