elasticsearch-phone

Elasticsearch Plugin for Phone and SIP Analysis

License	The Apache License, Version 2.0
Categories	Search Business Logic Libraries Elasticsearch
GroupId	com.inin.analytics
ArtifactId	elasticsearch-phone
Last Version	1.0.2
Release Date	Jan 26, 2017
Type	jar
Description	elasticsearch-phone Elasticsearch Plugin for Phone and SIP Analysis
Project URL	https://github.com/MyPureCloud/elasticsearch-phone
Source Code Management	https://github.com/MyPureCloud/elasticsearch-phone.git

Download elasticsearch-phone

Filename	Size
elasticsearch-phone-1.0.2.pom
elasticsearch-phone-1.0.2.zip	754 KB
elasticsearch-phone-1.0.2-sources.jar	5 KB
elasticsearch-phone-1.0.2-javadoc.jar	62 KB

How to add to project

Apache Maven

<!-- https://jarcasting.com/artifacts/com.inin.analytics/elasticsearch-phone/ -->
<dependency>
    <groupId>com.inin.analytics</groupId>
    <artifactId>elasticsearch-phone</artifactId>
    <version>1.0.2</version>
</dependency>

Gradle Groovy

// https://jarcasting.com/artifacts/com.inin.analytics/elasticsearch-phone/
implementation 'com.inin.analytics:elasticsearch-phone:1.0.2'

Gradle Kotlin

// https://jarcasting.com/artifacts/com.inin.analytics/elasticsearch-phone/
implementation ("com.inin.analytics:elasticsearch-phone:1.0.2")

Apache Buildr

'com.inin.analytics:elasticsearch-phone:jar:1.0.2'

Apache Ivy

<dependency org="com.inin.analytics" name="elasticsearch-phone" rev="1.0.2">
  <artifact name="elasticsearch-phone" type="jar" />
</dependency>

Groovy Grape

@Grapes(
@Grab(group='com.inin.analytics', module='elasticsearch-phone', version='1.0.2')
)

Scala SBT

libraryDependencies += "com.inin.analytics" % "elasticsearch-phone" % "1.0.2"

Leiningen

[com.inin.analytics/elasticsearch-phone "1.0.2"]

Dependencies

compile (4)

Group / Artifact	Type	Version
com.googlecode.libphonenumber : libphonenumber	jar	7.0.7
org.elasticsearch : elasticsearch	jar	1.6.0
org.apache.commons : commons-lang3	jar	3.4
commons-io : commons-io	jar	2.4

test (5)

Group / Artifact	Type	Version
org.apache.lucene : lucene-test-framework	jar	4.10.4
com.carrotsearch.randomizedtesting : randomizedtesting-runner	jar	2.1.11
org.elasticsearch : elasticsearch	test-jar	1.6.0
org.hamcrest : hamcrest-all	jar	1.3
junit : junit	jar	4.11

Project Modules

There are no modules declared in this project.

Elasticsearch-Phone

Indexing phone numbers & sip addresses in lucene is complicated. Most people use ngram tokenizers. We did that for a while with ngram min=3 & max=35, but the result was often 100s of tokens per sip address. Working in a call center focused company we quickly figured out how wasteful that is on the storage front. For us 6/7ths of our indexes were waisted on useless sip address tokens.

It's a hard problem to regex your way out of. An international phone number often includes a country code, but that can be 1, 2, or 3+ digits. A lot of people have requested elasticsearch integrate google's libphone library into a custom lucene analyzer. It hasn't happened yet, so here's a plugin that attempts to do just that.

Note: This is a young project. We'll improve as time goes on, but use at your own risk.

Building and installing the plugin

mvn package ./bin/plugin --url file:///....elasticsearch-phone/target/releases/elasticsearch-phone-1.0.0.zip --install elasticsearch-phone;

Analyzers

This project provides three analyzers that are intended for different contexts.

• The phone analyzer supports SIP URIs and other phone numbers and is intended to be used when indexing. It strips common prefixes such as sip: and tel: (and indexes those as separate tokens) and tokenizes the phone number with various prefix lengths.
• The phone-email analyzer extends the phone analyzer with additional tokenization for email addresses (e.g. generating tokens for the user part and the domain part of an email address).
• The phone-search analyzer is intended to be used as a search_analyzer with one of the other two analyzers used for indexing. It does minimal tokenization: If a term starts with sip: or tel: it strips this part and generates a token for it. The analyzer also strips a leading + from phone numbers.

Example inputs

Provide a telephone or sip address prefixed by tel: or sip: with no spaces or symbols.

Your indexing template will need to specify the analyzer for the field. EG

            "field": {
              "type": "string",
              "analyzer": "phone",
              "search_analyzer": "phone-search"
            }

Sample allowed inputs (see PhoneTokenizerIntegrationTest and PhoneSearchIntegrationTest for more):

• tel:[phone]
• tel:[phone]
• sip:abc@autosbcpc
• sip:[phone];[email protected]:8060

Example tokenization

SIP URI

Input (with country code): sip:[phone];[email protected]:8060

Tokens:

sip:[phone];[email protected]:8060
sip:
13169410766;[email protected]:8060
13169410766;ext=2233
1
2233
3169410766
3
13
31
131
316
1316
3169
13169
31694
131694
316941
1316941
3169410
13169410
31694107
131694107
316941076
1316941076
13169410766

Phone number

Input (without a country code): tel:[phone]

Tokens:

tel:[phone]
tel:
8177148350
8
81
817
8177
81771
817714
8177148
81771483
817714835

Email address

Input: [email protected]

Tokens:

[email protected]
user.name
user *
name *
domain.com *
domain *
com *

Tokens marked with * are only generated by the phone-email tokenizer.

Search examples

Term

Term queries will return exact matches without analyzing (without normalization as lowercase).

"query": {
  "term" : { "field" : "8177" }
}

"query": {
  "term" : { "field" : "domain" }
}

Match

Match queries use the configured analyzer (or search_analyzer). In this example, the query will be translated to a boolean and of two term queries for (tel: and 8177).

"query": {
  "match" : {
      "field" : {
          "query" : "tel:8177",
          "operator" : "and"
      }
  }
}

Genesys Cloud

Versions

Version
1.0.2 Jan 26, 2017
1.0.1 Jan 26, 2017
1.0.0 Aug 4, 2015