|author||Hugo Hirsch <email@example.com>||Fri Aug 14 22:27:21 2020 +0200|
|committer||Hugo Hirsch <firstname.lastname@example.org>||Fri Aug 14 22:27:21 2020 +0200|
RAT-271: Upgrade animal sniffer to comply with JDK8
The tool will download all the archives from a staging repo, unpack them and create a little report of what is there.
java -ea -jar apache-tentacles-0.1-SNAPSHOT.jar https://repository.apache.org/content/repositories/orgapacheopenejb-090
Assertions must be enabled.
The tool is not specific to maven and will simply recursively walk the provided URL and download all files matching the following pattern:
Tar.gz files are downloaded though there is currently no support for unpacking them.
Once the tool has run, the following files directories will exist:
repo/ content/ archives.html licenses.html notices.html style.css org.apache.openejb.openejb-core.3.0.4.openejb-core-3.0.4.jar.licenses.html org.apache.openejb.openejb-core.3.0.4.openejb-core-3.0.4.jar.notices.html org.apache.openejb.openejb-standalone.3.0.4.openejb-standalone-3.0.4.zip.licenses.html org.apache.openejb.openejb-standalone.3.0.4.openejb-standalone-3.0.4.zip.notices.html org.apache.openejb.openejb-tomcat-webapp.3.0.4.openejb-tomcat-webapp-3.0.4.war.licenses.html org.apache.openejb.openejb-tomcat-webapp.3.0.4.openejb-tomcat-webapp-3.0.4.war.notices.html ...
The repo directory will contain the full set of binaries, unmodified. Theoretically, this tool could also download and check signatures though it does not do that now.
The content directory will contain the unpacked version of the downloaded binaries
So this file for example:
Will be unpacked at the following location:
content/foo.zip.contents/ content/foo.zip.contents/LICENSE content/foo.zip.contents/NOTICE content/foo.zip.contents/README.txt content/foo.zip.contents/lib/bar.jar
Unpacking is recursive, so any binaries contained in foo.zip will also be unpacked.
content/foo.zip.contents/lib/bar.jar content/foo.zip.contents/lib/bar.jar.contents/ content/foo.zip.contents/lib/bar.jar.contents/LICENSE content/foo.zip.contents/lib/bar.jar.contents/NOTICE content/foo.zip.contents/lib/bar.jar.contents/README.txt content/foo.zip.contents/lib/bar.jar.contents/org/ content/foo.zip.contents/lib/bar.jar.contents/org/bar/ content/foo.zip.contents/lib/bar.jar.contents/org/bar/Some.class
The “main” report is currently called
archives.html and will list all of the top-level binaires, their LICENSE and NOTICE files and any LICENSE and NOTICE files of any binaries they may contain.
Validation of the output at this point is all still manual. One of the first improvements would be to automatically flag any binaries that:
In this report, each binary will have three links listed after its name ‘(licenses, notices, contents)’
This page will display the full text of the LICENSE files included in the binary. There will be two sections Declared and Undeclared
The Declared section lists the single LICENSE file that was supplied by the binary itself. As the tool works recursively, it will also collect any LICENSE file text from any binaries contained in the foo.zip. Well call these “sub” LICENSES for simplicity.
Some attempt is made to figure out if the text from sub LICENSE files are contained in the declared LICENSE file. If the sub license text is contained in the declared LICENSE file it is not listed as Undeclared.
The matching is not complete or perfect, but does help in more quickly seeing where there might be a missing LICENSE text that should be declared.
Functions identical to the previously described LICENSE page with identical matching.
Note on the code, this all could probably be abstracted. We probably don't need separate License and Notice classes.
The unpacked contents of the foo.zip as described above. Can be nice to be able to browse around the zip and look for any jars that might have LICENSE or NOTICE requirements but were overlooked.
Overall it would be great if this tool could perform some validation
Existence of LICENSE/NOTICE files:
Contents of LICENSE/NOTICE files: