SPARRI

This repository holds a Scala-based implementation of the Static Program Analysis Result Reuse Index (SPARRI). In order to compile and run SPARRI, you will need:

• JDK 1.8
• SBT (Scala Build Tool)
• Docker

Overview

SPARRI consists of multiple components, each of which is realized as a separate Scala module:

• ./analysis-runner Component that executes built-in analyses on demand
• ./core Core functionality shared among SPARRI components
• ./evaluation Evaluation as reported in our paper
• ./maven-entity-miner Component that indexes programs (GAV-Triples) from Maven Central on demand
• ./maven-entity-name-publisher Component that requests indexing of all Maven Central programs
• ./webapi Implementation of an RESTful HTTP API

Furthermore, this repository contains:

• ./playground A Scala project for testing SPARRI functionality
• ./literature-review Directory holding all definitions and results of our literature survey on SPA implementations

Getting Started

For SPARRI to work you will need instances of RabbitMQ and Postgres available. They can be spawned by executing:

docker run -d --name sparri-mq -e RABBITMQ_DEFAULT_USER=<USER> -e RABBITMQ_DEFAULT_PASS=<PASS> -p 9000:5672 rabbitmq:3.10

docker run -d --name sparri-db -e POSTGRES_USER=<USER> -e POSTGRES_PASSWORD=<PASS> -e POSTGRES_DB=spa-results -p 9001:5432 postgres:14.2

Note that you will need to expose the RabbitMQ port 5672 on your machine (in this example the host's port 9000 is used). The same is true for port 5432 of the Postgres container (here: host port 9001).

Now you have to configure SPARRI to know where the message queue and database are located. To do so, open the file ./core/src/main/resources/application.conf and replace all occurrances of the term <CHANGEME> with the respective values. You will need to enter the host name (serverName and mq-host), the ports for the database (portNumber) and RabbitMQ (mq-port), as well as the credentials for the database (user and password) and RabbitMQ (mq-user and mq-pass).

pa-reuse.postgres {
    connectionPool = "HikariCP"
    initializationFailTimeout = 2000
    dataSourceClass = "org.postgresql.ds.PGSimpleDataSource"

    properties = {
        serverName="<CHANGEME>"
        portNumber="<CHANGEME>"
        databaseName="spa-results"
        user="<CHANGEME>"
        password="<CHANGEME>"
    }

    numThreads = 8
}

spa-reuse.mq {
    mq-user = "<CHANGEME>"
    mq-pass = "<CHANGEME>"
    mq-host = "<CHANGEME>"
    mq-port = 0 #ATTENTION: ALSO REPLACE THIS VALUE!

    analysis-queue-name = "mvn-analyses"
    analysis-queue-exchange-name = "analyses-exchange"
    analysis-queue-routing-key = "mvn"

    entity-queue-name = "mvn-library-names"
    entity-queue-exchange-name = "library-name-exchange"
    entity-queue-routing-key = "mvn"
    entity-queue-max-priority = 3
}

Now you are ready to build all SPARRI components by executing sbt clean docker in the root directory of this repository. This command will compile and assemble all SPARRI components, and will also publish them as Docker images to your local Docker registry.

Running SPARRI

You can execute the different components of SPARRI via Docker. You will need to host at least the WebAPI, EntityMiner and AnalysisRunner. This is done by executing

docker run -d --name sparri-api -p 9090:9090 spar-webapi:latest
docker run -d --name sparri-miner spar-maven-entity-miner:latest
docker run -d --name sparri-runner spar-analysis-runner:latest

Once all three containers are up and running, you can interact with SPARRI via the API hosted at port 9090. Refer to ./webapi/api_spec_openapi3.yaml for a complete specification of our HTTP API.