LISA is a highly generic multi-revision graph database and computation engine designed for analyzing software artifacts that exist in many revisions of a project.
It avoids redundancy when analyzing multiple revisions by loading artifacts from different revisions into a shared representation where only the smallest delta between revisions requires extra space. Computations also run on entire ranges of revisions where no change occured at the sub-graph level.
For more information, please refer to this paper
LISA works on Linux and Mac.
Table of Contents
The lisa-quickstart repository contains information on how to try out LISA or to use it as a library. If you're not interested in the implementation or extension of LISA itself, it provides a better starting point.
LISA is split into multiple sub-projects. The
lisa-core contains the generic
framework components which exist independently of any language or analysis.
lisa-antlr project contains grammars used to create ANTLR-based parsers
lisa-module project contains various parsers and pre-composed
The following tools need to be installed in order to build and use LISA core:
- sbt >= 0.13.12
- JRE 8 (JRE 7 is not supported)
If you want to use the native JDK JavaParser instead of the ANTLR generated parser you also need to install JDK 8 and not just JRE.
Building & Testing
To build LISA, you can use sbt:
To run all default tests, run
To build self-contained jar files for each sub-project, run
Using LISA as a library
On its own, lisa-core does not provide any parsers or analyses. It's possible to either use existing ones from lisa-module or provide new, custom built parsers and analyses.
To use only
lisa-core as a library, place the following in your
libraryDependencies += "lisa-core" % "lisa-core" % "0.2.2" from "https://files.ifi.uzh.ch/seal/lisa/jar/lisa-core-assembly-0.2.2.jar"
To use the parsers and analyses contained in
lisa-module, add the following instead:
libraryDependencies += "lisa-module" % "lisa-module" % "0.2.2" from "https://files.ifi.uzh.ch/seal/lisa/jar/lisa-module-assembly-0.2.2.jar"
sbt ";project lisa-module ;test:runMain ch.uzh.ifi.seal.lisa.misc.ReadmeExample"
Writing new analyses
To formulate new analyses, it's best to first familiarize with the Signal/Collect computation paradigm: https://uzh.github.io/signal-collect/
Refer to existing analyses contained in
lisa-module, for example MccAnalysis.scala to observe how they are implemented.
In general, one needs to define a case class to hold the analysis data (e.g.
Mcc on line 15 of
MccAnalysis), a function that determines how the analysis
gets started (like on line 19 of
MccAnalysis), and a class which represents
the message which is transmitted between vertices (
MccPacket on line 25). This
class needs to specify how the message is processed (
collect on line 26).
Depending on the size of the projects to be analyzed, LISA requires a large
amount of memory. Be sure to provide enough memory to the JVM by editing
.sbtopts and setting suitable values for
Copyright 2017 Carol V. Alexandru
Licensed under the Apache License, Version 2.0 (the "License"); You may obtain a copy of the License at
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.