Page History

When

For use when developing an application with Saga Library embedded, and connecting to

the

ElasticSearch

manage

managed by the Saga Server.

Note

title	Prerequisites

This tutorial assumes:

The reader has the ability to create a project with Maven Framework support.
The data that Saga will use is
manage
managed through the Saga
's
user interface.
Java 11+ is installed in the machine.

Step-by-

step guide

Step Guide

Table of Contents
maxLevel 3
minLevel 3
outline true

Panel

title	On this page

Table of Contents

Configure pom.xml

In the pom.xml of the project, a basic example of the minimum configuration can be found below. But The elementary section of this configuration is the dependencies section, where we need two main libraries: saga-library and saga-elastic-provider.

saga-library

saga-elastic-provider

The core library of Saga, this dependency includes the Engine, Stages, Tag Manager, Pipeline Manager and Resource Manager which are all of the parts necessary to use Saga in any application.

This dependency will grant us access to ElasticSearch as a provider for Saga, which means our Stages and Managers will be able to fetch the data directly from this provider.

Info
More providers will be available in the future, but to use Saga full functionality, we recommend the use of the saga-elastic-provider.

Other Configuration Considerations

Other important configuration to notice is considerations of note are the use of Java 11, for for the compilation of the code, and the encoding UTF-8, as you can see as shown in the lines 36-38.

Code Block

language	xml
theme	RDark
firstline	0
title	pom.xml
linenumbers	true

<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
    <modelVersion>4.0.0</modelVersion>

    <groupId>com.accenture.saga</groupId>
    <artifactId>saga-howto</artifactId>
    <version>1.0.0-SNAPSHOT</version>

    <properties>
        <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
        <saga.version>1.0.0-SNAPSHOT</saga.version>
    </properties>

    <build>
        <plugins>
            <plugin>
                <groupId>org.apache.maven.plugins</groupId>
                <artifactId>maven-compiler-plugin</artifactId>
                <version>3.8.0</version>
                <executions>
                    <execution>
                        <id>compile</id>
                        <phase>compile</phase>
                        <goals>
                            <goal>compile</goal>
                        </goals>
                    </execution>
                    <execution>
                        <id>testCompile</id>
                        <phase>test-compile</phase>
                        <goals>
                            <goal>testCompile</goal>
                        </goals>
                    </execution>
                </executions>
                <configuration>
                    <source>11</source>
                    <target>11</target>
                    <encoding>${project.build.sourceEncoding}</encoding>
                </configuration>
            </plugin>
            <plugin>
                <artifactId>maven-assembly-plugin</artifactId>
                <version>3.1.0</version>
                <configuration>
                    <archive>
                        <manifest>
                            <mainClass>com.accenture.saga.server.SagaServer</mainClass>
                        </manifest>
                    </archive>
                    <descriptorRefs>
                        <descriptorRef>jar-with-dependencies</descriptorRef>
                    </descriptorRefs>
                    <appendAssemblyId>false</appendAssemblyId>
                </configuration>
                <executions>
                    <execution>
                        <id>make-assembly</id> <!-- this is used for inheritance merges -->
                        <phase>package</phase> <!-- bind to the packaging phase -->
                        <goals>
                            <goal>single</goal>
                        </goals>
                    </execution>
                </executions>
            </plugin>
        </plugins>
    </build>

    <dependencies>
        <dependency>
            <groupId>com.accenture.saga</groupId>
            <artifactId>saga-library</artifactId>
            <version>${saga.version}</version>
            <scope>compile</scope>
        </dependency>

        <dependency>
            <groupId>com.accenture.saga</groupId>
            <artifactId>saga-elastic-indexer</artifactId>
            <version>${saga.version}</version>
            <scope>compile</scope>
        </dependency>

        <dependency>
            <groupId>com.accenture.saga</groupId>
            <artifactId>saga-elastic-provider</artifactId>
            <version>${saga.version}</version>
            <scope>compile</scope>
        </dependency>
    </dependencies>
</project>

Initializing

The

the Saga Components

For starters, we will create 1. Begin by creating a main class , which which will hold a SagaEngine, ResourceManager, TagManager and PipelineManager.

Code Block

language	java
theme	Midnight
linenumbers	true

package com.accenture.saga;

import com.accenture.saga.engine.PipelineManager;
import com.accenture.saga.engine.SagaEngine;
import com.accenture.saga.resourcemgr.ResourceManager;
import com.accenture.saga.tags.TagManager;

public class Main {

    SagaEngine engine;
    ResourceManager resourceManager;
    TagManager tagManager;
    PipelineManager pipelineManager;

    /**
     * Constructor
     */
    public Main() {
        
        
    }
    
    public static void main(String[] args) {
        Main _instance = new Main();
    }
}

2. Create and configure the ResourceManager.

3. Add First we start by creating an configuring the ResourceManager, and adding a provider to it, but this . (This configuration will be hard-coded, so we need to add )

4. Add the SagaJsonFactory class, which allow allows us to create SagaJson objects (the standard document of Saga) from text, files or readers.

The configuration we are going to use for the provider is the following:

Saga_json
"name": "saga-provider", "type": "Elastic", "scheme": "http", "hostname": "localhost", "port": 9200, "timestamp": "updatedAt", "exclude": [ "updatedAt", "createdAt" ]

Each field from the top, starting with the common:

Parameter
summary The name we are going to use for the provider. It doesn't which name you use, but our take is "saga-provider"
name name
required true
Parameter
summary Indicates the type of provider we are using, in this case since we are using saga-elastic-provider, it's type would be "Elastic"
name type
required true

from From here on, all of the properties are specific to saga-elastic-indexer:

Parameter
summary Schema for the url to Elasticsearch
default http
name shema
Parameter
summary Name of the hosting server
default localhost
name hostname
Parameter
summary Port of ElasticSearch
default 9200
name 9200
type integer
Parameter
summary Name of the field reflecting any change done to the data
default updatedAt
name timestamp
Parameter
summary Name of the fields omitted (when possible) from the response of ElasticSearch
name exclude
type string array

Our code should look this this:

Code Block

language	java
theme	Midnight
linenumbers	true

public Main(String text, List<String> tags) throws SagaException {

    resourceManager = new ResourceManager();

    resourceManager.registerProvider(
            SagaJsonFactory.getInstance(
                    "{" +
                            "    \"name\": \"saga-provider\"," +
                            "    \"type\": \"Elastic\"," +
                            "    \"scheme\": \"http\"," +
                            "    \"hostname\": \"localhost\"," +
                            "    \"port\": 9200,\n" +
                            "    \"timestamp\": \"updatedAt\"," +
                            "    \"exclude\": [" +
                            "      \"updatedAt\"," +
                            "      \"createdAt\"" +
                            "    ]" +
                            "}"
            )
    );

}

5. Next we proceed with the configuration of the TagManager below the ResourceManager, once again we will hard. (Hard-code the configuration for this one.)

Saga_json
"resource": "saga-provider:saga_tags"

In the configuration above, saga-provider is representing the provider we add added to the ResourceManager in the previous configuration, then . Then the colon (:) indicates the division between the provider and the actual resource; since .

Since we are using a saga-elastic-provider, the resources will be index names.

Since we indexes names, and since were are connecting to a Saga index , created by the Saga serverServer, all of the indexes will be a combination between of the solution's name (usually will be saga), and an underscore (_) and the type of data the index holds , (in this case, tags , forming the name saga_tags ).

The code should look now like this now:

Code Block

language	java
theme	Midnight
linenumbers	true

public Main(String text, List<String> tags) throws SagaException {

    resourceManager = new ResourceManager();

    resourceManager.registerProvider(
            SagaJsonFactory.getInstance(
                    "{" +
                            "    \"name\": \"saga-provider\"," +
                            "    \"type\": \"Elastic\"," +
                            "    \"scheme\": \"http\"," +
                            "    \"hostname\": \"localhost\"," +
                            "    \"port\": 9200,\n" +
                            "    \"timestamp\": \"updatedAt\"," +
                            "    \"exclude\": [" +
                            "      \"updatedAt\"," +
                            "      \"createdAt\"" +
                            "    ]" +
                            "}"
            )
    );

    tagManager = new TagManager(resourceManager, SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_tags\"}"));
}

As you can see, the TagManager , receives the ResourceManager as a parameter the ResourceManager, which grant grants us access to the resource saga-provider:saga_tags.

In the same way6. Similarly, we will proceed to configure the PipelinesManager using the following configuration:

Saga_json
"resource": "saga-provider:saga_pipelines"

Once again, saga-provider , does reference to references the provider added in the ResourceManager, and saga_pipelines is a combination between of the name of the solution's name and the type of data, data (in this case, pipelines).

Code Block

language	java
theme	Midnight
linenumbers	true

public Main(String text, List<String> tags) throws SagaException {

    resourceManager = new ResourceManager();

    resourceManager.registerProvider(
            SagaJsonFactory.getInstance(
                    "{" +
                            "    \"name\": \"saga-provider\"," +
                            "    \"type\": \"Elastic\"," +
                            "    \"scheme\": \"http\"," +
                            "    \"hostname\": \"localhost\"," +
                            "    \"port\": 9200,\n" +
                            "    \"timestamp\": \"updatedAt\"," +
                            "    \"exclude\": [" +
                            "      \"updatedAt\"," +
                            "      \"createdAt\"" +
                            "    ]" +
                            "}"
            )
    );

    tagManager = new TagManager(resourceManager, SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_tags\"}"));

    pipelineManager = new PipelineManager(resourceManager,  SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_pipelines\"}"));
}

Setting Up

The

the Engine

So one we are at the fun part, once we got the After we have set up the Resource, Tag and Pipeline Manager all set up, we can assign the ResourceManager and the TagManager to the Engine.

Code Block

language	java
theme	Midnight
linenumbers	true

public Main(String text, List<String> tags) throws SagaException {

    resourceManager = new ResourceManager();

    resourceManager.registerProvider(
            SagaJsonFactory.getInstance(
                    "{" +
                            "    \"name\": \"saga-provider\"," +
                            "    \"type\": \"Elastic\"," +
                            "    \"scheme\": \"http\"," +
                            "    \"hostname\": \"localhost\"," +
                            "    \"port\": 9200,\n" +
                            "    \"timestamp\": \"updatedAt\"," +
                            "    \"exclude\": [" +
                            "      \"updatedAt\"," +
                            "      \"createdAt\"" +
                            "    ]" +
                            "}"
            )
    );

    tagManager = new TagManager(resourceManager, SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_tags\"}"));

    pipelineManager = new PipelineManager(resourceManager,  SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_pipelines\"}"));

    engine = new SagaEngine();

    engine.setResourceManager(resourceManager);
    engine.setTagManager(tagManager);
}

Working

With The

with the Pipeline Manager

From here we got 2 options, first option is letting

We have two options:*

Option 1: Let the PipelineManager build the pipeline for us

,

using a set of tags (that we will provide)

; the second option is to manually

.

Option 2: Manually provide a complete pipeline configuration to the PipelineManager.

Pros & Cons

Option 1: Automatic Pipeline

Option 2: Manual Pipeline

Pros

Uses configuration set up through Saga's UI
Loads only the necessary from the resource (tags, stages, ...)
Builds pipeline based on tag dependency
Can generate multiple and different pipelines

Each Recognizer can have a base pipeline as dependency

Pros

Complete control over the flow of the data

Cons

Pipelines not always the most efficient (...yet)
Each base pipeline must be configure manually (... for the moment)

Cons

Configuration of every stage must be done manually
Relies strongly in the knowledge of the user for each possible stage configuration
Lack of flexibly when changing to another pipeline

Tie

Needs a stage of type TextBlockReader configure manually

Tie

Needs a stage of type TextBlockReader configure manually

*Since the first one is the most Option 1 is more flexible and the one that makes use of the configuration in ElasticSearch, we will use that one.

Request

A

a Pipeline

Before asking the PipelineManager for a pipeline, we need to provide a stage of type TextBlockReader, at . At the moment, we only have one stage of that type, the SimpleReaderStage, which requires a splitRegex in the configuration , as a SagaJson object. So let

1. Let's add that to the code shall we.

Code Block

language	java
theme	Midnight
linenumbers	true

public Main(String text, List<String> tags) throws SagaException {

    resourceManager = new ResourceManager();

    resourceManager.registerProvider(
            SagaJsonFactory.getInstance(
                    "{" +
                            "    \"name\": \"saga-provider\"," +
                            "    \"type\": \"Elastic\"," +
                            "    \"scheme\": \"http\"," +
                            "    \"hostname\": \"localhost\"," +
                            "    \"port\": 9200,\n" +
                            "    \"timestamp\": \"updatedAt\"," +
                            "    \"exclude\": [" +
                            "      \"updatedAt\"," +
                            "      \"createdAt\"" +
                            "    ]" +
                            "}"
            )
    );

    tagManager = new TagManager(resourceManager, SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_tags\"}"));

    pipelineManager = new PipelineManager(resourceManager,  SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_pipelines\"}"));

    //The Fun Part

    engine = new SagaEngine();

    engine.setResourceManager(resourceManager);
    engine.setTagManager(tagManager);


    SimpleReaderStage simpleReaderStage = new SimpleReaderStage(engine, SagaJsonFactory.getInstance("{ \"splitRegex\": \"[\\r|\\n]+\"}"));
}

Parameter
summary asdas
name splitRegex

With the regex [\r\n]+ we are indicating the character signaling a break line; also , Also note that the SimpleReaderStage receives the engine as the first parameter.

Tip
The regex [\r\n]+, is the standard for mostly all the text you will be processing

Now we can ask the PipelineManager to build a pipeline for the tags... Which we We still don't know where they came from, but let's fix that; first we add .

2. Add the building of the pipeline.

Code Block

language	java
theme	RDark
linenumbers	true

pipelineManager.buildPipelineFor(engine, tags, simpleReaderStage);

Now we just add 3. Add the tags as a parameter of the constructor, our . Our code should look like this:

Code Block

language	java
theme	Midnight
linenumbers	true

package com.accenture.saga;

import com.accenture.saga.engine.PipelineManager;
import com.accenture.saga.engine.SagaEngine;
import com.accenture.saga.engine.stages.SimpleReaderStage;
import com.accenture.saga.exception.SagaException;
import com.accenture.saga.json.SagaJsonFactory;
import com.accenture.saga.resourcemgr.ResourceManager;
import com.accenture.saga.tags.TagManager;

import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;

public class Main {

    SagaEngine engine;
    ResourceManager resourceManager;
    TagManager tagManager;
    PipelineManager pipelineManager;

    /**
     * Constructor
     */
    public Main(List<String> tags) throws SagaException {

        resourceManager = new ResourceManager();

        resourceManager.registerProvider(
                SagaJsonFactory.getInstance(
                        "{" +
                                "    \"name\": \"saga-provider\"," +
                                "    \"type\": \"Elastic\"," +
                                "    \"scheme\": \"http\"," +
                                "    \"hostname\": \"localhost\"," +
                                "    \"port\": 9200,\n" +
                                "    \"timestamp\": \"updatedAt\"," +
                                "    \"exclude\": [" +
                                "      \"updatedAt\"," +
                                "      \"createdAt\"" +
                                "    ]" +
                                "}"
                )
        );

        tagManager = new TagManager(resourceManager, SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_tags\"}"));

        pipelineManager = new PipelineManager(resourceManager,  SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_pipelines\"}"));

        engine = new SagaEngine();

        engine.setResourceManager(resourceManager);
        engine.setTagManager(tagManager);

        SimpleReaderStage simpleReaderStage = new SimpleReaderStage(engine, SagaJsonFactory.getInstance("{ \"splitRegex\": \"[\\r|\\n]+\"}"));

        pipelineManager.buildPipelineFor(engine, tags, simpleReaderStage);
    }

    public static void main(String[] args) throws SagaException {

        List<String> tags = new ArrayList<>(Arrays.asList("tag1", "tag2", "tag3"));

        Main _instance = new Main(tags);
    }
}

Here we just build built a list with the tags "tag1", "tag2", and "tag3", which we passed as a parameter of the constructor, so that the PipelineManager can have them to build the pipeline.

At this point pipelineManager.buildPipelineFor uses the tags we specified to

identified

identify the stages

which

that recognize these tags

, from this stages

.
From these stages, build a dependency hierarchy

, which adds

to add any necessary tags and stages in order to

found

find the specified tags

; once

.
After it has identified all of the stages and tags necessary, the PipelineManger adds them to the Engine we provided.

Now , which means our Engine is ready to receive text and build a graph.

Process

A

a Text

Let's also add a String parameter to the constructor , for the text we want to process, this . This text will be added to the Engine using the method reset.

Code Block

language	java
theme	Midnight

engine.reset(new ByteArrayInputStream(text.getBytes(StandardCharsets.UTF_8)));

This method accepts an InputStream, this so we can specify the encoding in UTF-8; so in . In order to pass the text to Saga, we get the bytes from the text in UTF-8 encoding and create a ByteArrayInputStream with them.

Note
All of the process processing done by Saga is with the encoding UTF-8.

At the moment, we only told Saga which is the text, now . Now we need to process it. For this we , use the method advance, which returns a Vertex, this This will be the first Vertex of the text block.

Code Block

language	java
theme	Midnight

Vertex start = engine.advance();

Currently our code should look like this:

Code Block

language	java
theme	Midnight
linenumbers	true

package com.accenture.saga;

import com.accenture.saga.engine.PipelineManager;
import com.accenture.saga.engine.SagaEngine;
import com.accenture.saga.engine.Vertex;
import com.accenture.saga.engine.stages.SimpleReaderStage;
import com.accenture.saga.exception.SagaException;
import com.accenture.saga.json.SagaJsonFactory;
import com.accenture.saga.resourcemgr.ResourceManager;
import com.accenture.saga.tags.TagManager;

import java.io.ByteArrayInputStream;
import java.nio.charset.StandardCharsets;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;

public class Main {

    SagaEngine engine;
    ResourceManager resourceManager;
    TagManager tagManager;
    PipelineManager pipelineManager;

    /**
     * Constructor
     */
    public Main(String text, List<String> tags) throws SagaException {

        resourceManager = new ResourceManager();

        resourceManager.registerProvider(
                SagaJsonFactory.getInstance(
                        "{" +
                                "    \"name\": \"saga-provider\"," +
                                "    \"type\": \"Elastic\"," +
                                "    \"scheme\": \"http\"," +
                                "    \"hostname\": \"localhost\"," +
                                "    \"port\": 9200,\n" +
                                "    \"timestamp\": \"updatedAt\"," +
                                "    \"exclude\": [" +
                                "      \"updatedAt\"," +
                                "      \"createdAt\"" +
                                "    ]" +
                                "}"
                )
        );

        tagManager = new TagManager(resourceManager, SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_tags\"}"));

        pipelineManager = new PipelineManager(resourceManager,  SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_pipelines\"}"));

        engine = new SagaEngine();

        engine.setResourceManager(resourceManager);
        engine.setTagManager(tagManager);

        SimpleReaderStage simpleReaderStage = new SimpleReaderStage(engine, SagaJsonFactory.getInstance("{ \"splitRegex\": \"[\\r|\\n]+\"}"));

        pipelineManager.buildPipelineFor(engine, tags, simpleReaderStage);

        engine.reset(new ByteArrayInputStream(text.getBytes(StandardCharsets.UTF_8)));

        Vertex start = engine.advance();
    }

    public static void main(String[] args) throws SagaException {

        List<String> tags = new ArrayList<>(Arrays.asList("tag1", "tag2", "tag3"));

        String text = "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed egestas orci eu mauris luctus consequat.";

        Main _instance = new Main(text, tags);
    }
}

Working

With The

with the Graph

Tip
For more specific information on how to navigate through the graph, please visit Understanding Interpretation Graphs

Before we start with this section, just a clarification, the graph has a width range of applications that varies depending of the content, the tags, the stages and the user final goal. For this example we will do 3 of the most primordial cases, print the graph, search for a specific type of tag, and getting the highest value route.

For this examples we will asume that tag1 has an Entity Recognizer identifying "adipiscing elit", and the baseline-pipeline is the following

Saga_json

"stages": [
        {
            "language": "en",
            "type": "TextBreakerStage"
        },
        {
            "requiredFlags": [
                "SENTENCE"
            ],
            "type": "WhitespaceTokenizerStage"
        },
        {
            "type": "StopWordsStage"
        },
        {
            "type": "CaseAnalysis"
        },
        {
            "type": "CharChangeSplitter"
        }
    ]

Printing The Graph

Printing the graph it's the most basic of all the cases, since it allow us to have a more graphical picture on how Saga processed the text.

For this case we will be using the class GraphPrinter and the static method printOnce, which needs the engine, the start vertex and the last vertex, so basically we would be printing a section of the graph, which in this case is the entire text block section.

We have the engine, and the start vertex; for the last vertex, the engine has a method getAllVertex, which returns a LinkedList of all the vertex processed by the engine, so we can get the last element (vertex) from the list.

Code Block

language	java
theme	Midnight

Vertex start = engine.advance();

Vertex last = engine.getAllVertex().getLast();

// Print Graph

GraphPrinter.printOnce(engine, start, last);

This will show a text version of the graph in the console, much like this one

Code Block

language	text
theme	FadeToGrey

INFO | {GraphPrinter} |  V---------------------------------[Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed egestas orci eu mauris luctus consequat.]---------------------------------V 
INFO | {GraphPrinter} |  ^-----------------[Lorem ipsum dolor sit amet, consectetur adipiscing elit]------------------V-------------[Sed egestas orci eu mauris luctus consequat]-------------V-[]-^ 
INFO | {GraphPrinter} |  ^-[Lorem]-V-[ipsum]-V-[dolor]-V-[sit]-V---[amet,]----V-[consectetur]-V-[adipiscing]-V-[elit]-^-[Sed]-V-[egestas]-V-[orci]-V-[eu]-V-[mauris]-V-[luctus]-V-[consequat]-^ 
INFO | {GraphPrinter} |  ^-[lorem]-^                           ^-[amet]-V-[,]-^               ^-------[{tag1}]--------^-[sed]-^

Where V represents the first appearance of a vertex, ^ is a connection to an existent vertex, everything between [ ] is a token, and inside a token if it appears between { }, is a semantic tag

Tip
If you want reduce the vertex returned by the getAllVertex, you can also specify only the first vertex, or the first and the last vertex

Searching For a Tag

In this case we will search for all the tokens that are semantic tags.

First obstacle, How do we identify a semantic tag?

That's easy, all the tokens that are semantic tags, have a flag called SEMANTIC_TAG, so we need to first get that flag; these flags are define by the engine, so we will ask it.

Code Block

language	java
theme	Midnight

Flag SEMANTIC_TAG = engine.getFlagForRead(LexObject.LexObjectType.LEX_ITEM, "SEMANTIC_TAG");

We define the object flag (We recommend to name the object just like the flag, just good practice)
From the engine we called the method getFlagForRead, which needs the type of flag and the name of the flag
We indicate the type of flag LEX_ITEM, and the name of the flag SEMANTIC_TAG

Tip
There are 2 types of flags LEX_ITEM and VERTEX, they are enums of type LexObjectType, which is in the class LexObject

Tip
You can know the flags by looking into every Stage and see what are the flags it sets for the items it creates.

Second obstacle, How do we get the tokens?

This can also be done in one line,

Code Block

language	java
theme	Midnight

List<LexItem> items = SagaGraph.getTokens(start, last, lexItem -> lexItem.hasFlag(SEMANTIC_TAG));

We use the class SagaGraph, and call the method getTokens
We specify the start and end from where we want to find the tokens
We add a filter function which check the item has the flag SEMANTIC_TAG
Return the result as a list of LexItems

Now our code should look like this

Code Block

language	java
theme	Midnight
linenumbers	true

package com.accenture.saga;

import com.accenture.saga.engine.*;
import com.accenture.saga.engine.stages.SimpleReaderStage;
import com.accenture.saga.exception.SagaException;
import com.accenture.saga.json.SagaJsonFactory;
import com.accenture.saga.resourcemgr.ResourceManager;
import com.accenture.saga.tags.TagManager;
import com.accenture.saga.utilities.SagaGraph;

import java.io.ByteArrayInputStream;
import java.nio.charset.StandardCharsets;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;

public class Main {

    SagaEngine engine;
    ResourceManager resourceManager;
    TagManager tagManager;
    PipelineManager pipelineManager;

    /**
     * Constructor
     */
    public Main(String text, List<String> tags) throws SagaException {

        resourceManager = new ResourceManager();

        resourceManager.registerProvider(
                SagaJsonFactory.getInstance(
                        "{" +
                                "    \"name\": \"saga-provider\"," +
                                "    \"type\": \"Elastic\"," +
                                "    \"scheme\": \"http\"," +
                                "    \"hostname\": \"localhost\"," +
                                "    \"port\": 9200,\n" +
                                "    \"timestamp\": \"updatedAt\"," +
                                "    \"exclude\": [" +
                                "      \"updatedAt\"," +
                                "      \"createdAt\"" +
                                "    ]" +
                                "}"
                )
        );

        tagManager = new TagManager(resourceManager, SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_tags\"}"));

        pipelineManager = new PipelineManager(resourceManager, SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_pipelines\"}"));

        engine = new SagaEngine();

        engine.setResourceManager(resourceManager);
        engine.setTagManager(tagManager);

        SimpleReaderStage simpleReaderStage = new SimpleReaderStage(engine, SagaJsonFactory.getInstance("{ \"splitRegex\": \"[\\r|\\n]+\"}"));

        pipelineManager.buildPipelineFor(engine, tags, simpleReaderStage);

        engine.reset(new ByteArrayInputStream(text.getBytes(StandardCharsets.UTF_8)));

        Vertex start = engine.advance();

        Vertex last = engine.getAllVertex(start).getLast();

        // Print Graph

        GraphPrinter.printOnce(engine, start, last);

        // Get specific items

        Flag SEMANTIC_TAG = engine.getFlagForRead(LexObject.LexObjectType.LEX_ITEM, "SEMANTIC_TAG");

        List<LexItem> items = SagaGraph.getTokens(start, last, lexItem -> lexItem.hasFlag(SEMANTIC_TAG));

        items.forEach(item -> System.out.println(item.toStringForDebug()));
    }

    public static void main(String[] args) throws SagaException {

        List<String> tags = new ArrayList<>(Arrays.asList("tag1", "tag2", "tag3"));

        String text = "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed egestas orci eu mauris luctus consequat.";

        Main _instance = new Main(text, tags);

        System.exit(0);
    }
}

At the end a println was added at the end to see the token we got

Code Block

language	text
theme	FadeToGrey

I"{tag1}"(40:55)

I indicates Item, the text inside the quotes is the text of the item, and the numbers between parenthesis is the position in characters of the token (start:end)

Getting The Highest Route

The highest route is the interpretation with the highest confidence, amount other factors (e.g. largest tokens, more complex patterns, ...), and is the interpretation that most like we want. Let's start we the code and then explained

Code Block

language	java
theme	Midnight

Flag TEXT_BLOCK = engine.getFlagForRead(LexObject.LexObjectType.LEX_ITEM, "TEXT_BLOCK");

List<LexItem> route = SagaGraph.getRoute(start, last, lexItem -> !lexItem.hasFlag(TEXT_BLOCK));

Normally we would want to ignore any token with the TEXT_BLOCK flag in it
We use again the class SagaGraph, and call the function getRoute
Add the start and end vertex
We add a filter function which check the item doesn't have TEXT_BLOCK flag

Code Block

language	java
theme	Midnight
linenumbers	true

package com.accenture.saga;

import com.accenture.saga.engine.*;
import com.accenture.saga.engine.stages.SimpleReaderStage;
import com.accenture.saga.exception.SagaException;
import com.accenture.saga.json.SagaJsonFactory;
import com.accenture.saga.resourcemgr.ResourceManager;
import com.accenture.saga.tags.TagManager;
import com.accenture.saga.utilities.SagaGraph;

import java.io.ByteArrayInputStream;
import java.nio.charset.StandardCharsets;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;

public class Main {

    SagaEngine engine;
    ResourceManager resourceManager;
    TagManager tagManager;
    PipelineManager pipelineManager;

    /**
     * Constructor
     */
    public Main(String text, List<String> tags) throws SagaException {

        resourceManager = new ResourceManager();

        resourceManager.registerProvider(
                SagaJsonFactory.getInstance(
                        "{" +
                                "    \"name\": \"saga-provider\"," +
                                "    \"type\": \"Elastic\"," +
                                "    \"scheme\": \"http\"," +
                                "    \"hostname\": \"localhost\"," +
                                "    \"port\": 9200,\n" +
                                "    \"timestamp\": \"updatedAt\"," +
                                "    \"exclude\": [" +
                                "      \"updatedAt\"," +
                                "      \"createdAt\"" +
                                "    ]" +
                                "}"
                )
        );

        tagManager = new TagManager(resourceManager, SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_tags\"}"));

        pipelineManager = new PipelineManager(resourceManager, SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_pipelines\"}"));

        engine = new SagaEngine();

        engine.setResourceManager(resourceManager);
        engine.setTagManager(tagManager);

        SimpleReaderStage simpleReaderStage = new SimpleReaderStage(engine, SagaJsonFactory.getInstance("{ \"splitRegex\": \"[\\r|\\n]+\"}"));

        pipelineManager.buildPipelineFor(engine, tags, simpleReaderStage);

        engine.reset(new ByteArrayInputStream(text.getBytes(StandardCharsets.UTF_8)));

        Vertex start = engine.advance();

        Vertex last = engine.getAllVertex(start).getLast();

        // Print Graph

        GraphPrinter.printOnce(engine, start, last);

        // Get specific items

        Flag SEMANTIC_TAG = engine.getFlagForRead(LexObject.LexObjectType.LEX_ITEM, "SEMANTIC_TAG");

        List<LexItem> items = SagaGraph.getTokens(start, last, lexItem -> lexItem.hasFlag(SEMANTIC_TAG));

        items.forEach(item -> System.out.println(item.toStringForDebug()));

        // Get the highest route

        Flag TEXT_BLOCK = engine.getFlagForRead(LexObject.LexObjectType.LEX_ITEM, "TEXT_BLOCK");

        List<LexItem> route = SagaGraph.getRoute(start, last, lexItem -> !lexItem.hasFlag(TEXT_BLOCK));

        route.forEach(item -> System.out.println(item.toStringForDebug()));
    }

    public static void main(String[] args) throws SagaException {

        List<String> tags = new ArrayList<>(Arrays.asList("tag1", "tag2", "tag3"));

        String text = "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed egestas orci eu mauris luctus consequat.";

        Main _instance = new Main(text, tags);

        System.exit(0);
    }
}

Once again a println was added at the end to see the tokens we got, from the route

Code Block

language	text
theme	FadeToGrey

I"Lorem"(0:5)
I"ipsum"(6:11)
I"dolor"(12:17)
I"sit"(18:21)
I"amet"(22:26)
I","(26:27)
I"consectetur"(28:39)
I"{tag1}"(40:55)
I"Sed"(57:60)
I"egestas"(61:68)
I"orci"(69:73)
I"eu"(74:76)
I"mauris"(77:83)
I"luctus"(84:90)
I"consequat"(91:100)

As you can see in the eighth token is actually the tag tag1, since it'sconfidence is higher than the tokens adipiscing and elit.

Important Note

Something important to keep in mind is that advanced function returns the first vertex of a text block, and that text block can represent the entirety of the text, but many times will only represent a fraction of the text; so we need to keep calling advanced again and again until we reach the end, meaning the vertex comes null

Code Block

language	java
theme	Midnight
linenumbers	true

package com.accenture.saga;

import com.accenture.saga.engine.*;
import com.accenture.saga.engine.stages.SimpleReaderStage;
import com.accenture.saga.exception.SagaException;
import com.accenture.saga.json.SagaJsonFactory;
import com.accenture.saga.resourcemgr.ResourceManager;
import com.accenture.saga.tags.TagManager;
import com.accenture.saga.utilities.SagaGraph;

import java.io.ByteArrayInputStream;
import java.nio.charset.StandardCharsets;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;

public class Main {

    SagaEngine engine;
    ResourceManager resourceManager;
    TagManager tagManager;
    PipelineManager pipelineManager;

    /**
     * Constructor
     */
    public Main(String text, List<String> tags) throws SagaException {

        resourceManager = new ResourceManager();

        resourceManager.registerProvider(
                SagaJsonFactory.getInstance(
                        "{" +
                                "    \"name\": \"saga-provider\"," +
                                "    \"type\": \"Elastic\"," +
                                "    \"scheme\": \"http\"," +
                                "    \"hostname\": \"localhost\"," +
                                "    \"port\": 9200,\n" +
                                "    \"timestamp\": \"updatedAt\"," +
                                "    \"exclude\": [" +
                                "      \"updatedAt\"," +
                                "      \"createdAt\"" +
                                "    ]" +
                                "}"
                )
        );

        tagManager = new TagManager(resourceManager, SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_tags\"}"));

        pipelineManager = new PipelineManager(resourceManager, SagaJsonFactory.getInstance("{ \"resource\": \"saga-provider:saga_pipelines\"}"));

        engine = new SagaEngine();

        engine.setResourceManager(resourceManager);
        engine.setTagManager(tagManager);

        SimpleReaderStage simpleReaderStage = new SimpleReaderStage(engine, SagaJsonFactory.getInstance("{ \"splitRegex\": \"[\\r|\\n]+\"}"));

        pipelineManager.buildPipelineFor(engine, tags, simpleReaderStage);

        engine.reset(new ByteArrayInputStream(text.getBytes(StandardCharsets.UTF_8)));

        Vertex start = engine.advance();

        do {

            Vertex last = engine.getAllVertex(start).getLast();

            // Print Graph

            GraphPrinter.printOnce(engine, start, last);

            // Get specific items

            Flag SEMANTIC_TAG = engine.getFlagForRead(LexObject.LexObjectType.LEX_ITEM, "SEMANTIC_TAG");

            List<LexItem> items = SagaGraph.getTokens(start, last, lexItem -> lexItem.hasFlag(SEMANTIC_TAG));

            items.forEach(item -> System.out.println(item.toStringForDebug()));

            // Get the highest route

            Flag TEXT_BLOCK = engine.getFlagForRead(LexObject.LexObjectType.LEX_ITEM, "TEXT_BLOCK");

            List<LexItem> route = SagaGraph.getRoute(start, last, lexItem -> !lexItem.hasFlag(TEXT_BLOCK));

            route.forEach(item -> System.out.println(item.toStringForDebug()));

            start = engine.advance();

        }while (start != null);
    }

    public static void main(String[] args) throws SagaException {

        List<String> tags = new ArrayList<>(Arrays.asList("tag1", "tag2", "tag3"));

        String text = "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed egestas orci eu mauris luctus consequat.";

        Main _instance = new Main(text, tags);

        System.exit(0);
    }
}

Page tree

Versions Compared

Old Version 28

New Version 29

Key

For use when developing an application with Saga Library embedded, and connecting to

ElasticSearch

managed by the Saga Server.

This tutorial assumes:

The reader has the ability to create a project with Maven Framework support.

The data that Saga will use is

managed through the Saga

user interface.

Java 11+ is installed in the machine.

Step-by-

Step Guide

Table of Contents
maxLevel 3
minLevel 3
outline true

Configure pom.xml

Other Configuration Considerations

Initializing

the Saga Components

Setting Up

the Engine

Working

with the Pipeline Manager

Pros & Cons

Request

a Pipeline

Process

a Text

Working

with the Graph

Printing The Graph

Searching For a Tag

Getting The Highest Route

Important Note

Related articles

Related articles

Page tree

Page History

Versions Compared

Old Version 28

New Version 29

Key

For use when developing an application with Saga Library embedded, and connecting to

ElasticSearch

managed by the Saga Server.

This tutorial assumes:

The reader has the ability to create a project with Maven Framework support.

The data that Saga will use is

managed through the Saga

user interface.

Java 11+ is installed in the machine.

Step-by-

Step Guide

Table of ContentsmaxLevel3minLevel3outlinetrue

Configure pom.xml

Other Configuration Considerations

Initializing

the Saga Components

Setting Up

the Engine

Working

with the Pipeline Manager

Pros & Cons

Request

a Pipeline

Process

a Text

Working

with the Graph

Printing The Graph

Searching For a Tag

Getting The Highest Route

Important Note

Related articles

Related articles

Table of Contents
maxLevel 3
minLevel 3
outline true