monkeyiq / odf-2011-track-changes-tests Goto Github PK

The test suite uses abiword from the command line to convert from abw
format to odt format and inspect the result using Relax NG. Some tests
convert from odt back to abw again to see that the result of this
double conversion is what is expected. Both a relaxng validator and 
Perl Test::Class are needed for the test suite. 
# yum install perl-Test-Class

Some tools which are either used by the suite or are handy to know about
ar ethe jing / trang couple:
http://code.google.com/p/jing-trang/downloads/list

To get around path and installation issues, the tools expect trang and jing
to work as though the following borne script snippit is effectual:
alias trang="java -jar /usr/local/java/trang/trang.jar"
alias jing="java -jar /usr/local/java/jing/bin/jing.jar"

and libxml2's support for relax ng. As libxml2 only supports the xml
serialization of relax, you need to use a schema conversion tool in
order to validate schemas that use the compact syntax. Of course,
libxml2 might have gianed support for the compact rnc format by the
time you read this.

---------------

Note that the testing normally involves loading an abw file,
converting it to odt, then loading that odt and converting back to
abw, and perhaps converting the second abw to odt yet again. This is
because the coding involves both saving to and loading from odt files.
So the two round trips are required to ensure that nothing is lost
during the save/load process. A simple test here is to run a schmea
validation on the first and second odt file to make sure that the
first odt has the desired form and that nothing is lost during the
loading of the odt file again.

Luckily this can be automated using the abiword binary and jing to
verify the structure and content of the odf/content.xml file. Each
directory contains abw files, rnc (Relax NG Compact) schema files and
perl Text::Class code to use the former files and make sure the
abiword binary acts as expected during odt processing (load/save).

The perl code can be run by stepping into the directory and executing
the .t file from there. Temporary files will be placed into the
/tmp/`id -u` directory and might remain there. You will receive an
overview of which tests have run and which if any have failed (and
why).

$ cd para-add-delete
$ ./para_and_delete::Test.t

To test just a single thing from the unit test of a directory use
the following. The export line names the specific sub from the .pm
file you want to run. It is a regex, so .* will return to all tests.

$ cd para-add-delete
$ export TEST_METHOD=test5
$ ./para_and_delete::Test.t