Handling Other Types of Archives

Revapi understands JAR files. But java-based applications are not deployed just as JAR files. There are other types of artifacts like WAR files, EAR files, Spring Boot fat jars, etc. While these usually don’t have any Java API as such that would be reusable in other Java code (i.e. such non-jar archives usually cannot function as libraries) and thus performing API checks might not be that crucial on them, it still might be useful to perform the API check on such archives (for whatever reason, you, my dear user, know best).

Since revapi-java-spi-0.18.0 (used in revapi-java-0.19.0) the API check can be extended by JarExtractor implementations that can transform certain archives into "plain" JAR structure that Revapi will understand. See the SPI documentation for more details.

Handling WAR Archives

revapi-java by default contains an implementation of JarExtractor that makes it understand the structure of WAR files and automatically assume classes from WEB-INF/classes as the classes of the archive (i.e. If any archive contains a directory called /WEB-INF/classes/ it is automatically assumed that it is a WAR archive and only the contents of /WEB-INF/classes/ is considered for API checks.

The behavior of the WAR extractor is configurable. The below is the default configuration:

<revapi.java>
  <extract>
    <war>
      <disabled>false</disabled> (1)
      <scan>
        <item> (2)
          <archive>.*</archive> (3)
          <prefixes>
            <item>/WEB-INF/classes/</item> (4)
          </prefixes>
        </item>
      </scan>
    </war>
  </extract>
</revapi.java>

1	The WAR file detection and extraction can be disabled (it is enabled by default)
2	There can be many ways to extract classes from different WAR archives
3	To limit how and what is extracted from different archives, you can specify a regex on the archive name
4	And specify a list of path prefixes that are supposed to contain the "significant" classes for the archive

Given the above default configuration, you can see that the WAR extractor looks at all archives and if it finds any files under /WEB-INF/classes/ the classes from there are the only ones considered during the API analysis.