Writing Tests
  =============

  Here's how to write tests.

  1. Create a new file in the t directory, called foo.t. Start with
     the following lines:

	require 't/test.pl';
	package OddMuse;
	use Test::More tests => $num
	clear_pages();

     This will load the testing library, make all its functions
     available to you, announce that you plan to make $num tests, and
     clear all the pages from the test wiki.

     The test wiki will be created in /tmp/oddmuse.

  2. The wiki is accessed via the command line only. You don't need to
     have your code installed on a webserver! Just run the test from
     the parent directory:

	perl t/foo.t

  3. Write $num tests. :)

  4. To examine the situation after having run some tests, you can
     also call the script from the command line. The only problem is
     that the tests use a specific data directory that you need to
     provide via an environment variable:

        WikiDataDir=test-data perl wiki.pl action=index raw=1

  add_module, remove_module, and remove_rule
  ------------------------------------------

  Load a module before you run any tests:

    add_module('usemod.pl');

  If the module has important setup code, you might have to add the
  following:

    InitVariables();

  The reason is this: If you add a module and the run the script
  again, you're fine. But if you run tests that don't invoke another
  copy of the script, then the init code will not have run.

  Modules and rules need rarely be removed, since every *.t file
  starts in a new process.  If you then want to run additional tests
  without the module you added (in the same *.t file!), then remove
  both the module and the rules it added. You'll have to do this
  manually, unfortunately.

    remove_module('usemod.pl');
    remove_rule(\&UsemodRule);
    


  update_page
  -----------

    $page = update_page($pagename, $content);
    update_page($pagename, $content, $summary, $minor, $admin, @rest);

  @rest is a list of parameter=value string pairs:

    @rest = ('username=joe', 'ns=Moore');

  If updating the page resultet in a redirect, the redirect is stored
  in the variable $redirect, and you still get the result page
  returned.

    test_page($redirect, split('\n',<<'EOT'));
    banned text
    wiki administrator
    matched
    See .*BannedContent.* for more information
    EOT

  You can even create pages containing file uploads directly:

    $page = update_page('alex pic', "#FILE image/png\niVBORw0KGgoAAAA");


  get_page
  --------

    $page = get_page('action=calendar');
    $page = get_page('action=rc all=1 showedit=1');
    $page = get_page('action=rc', 'all=1', 'showedit=1');
    test_page(get_page('action=all'),
	      'restricted to administrators');

  Return the text of the page. The parameters are the parameters
  available to you from the command line when using the CGI library:

    keyword1 keyword2 keyword3
    keyword1+keyword2+keyword3
    name1=value1 name2=value2
    name1=value1&name2=value2
    "name1='I am a long value'" "name2=two\ words"
    /your/path/here?name1=value1&name2=value2



  run_tests
  ---------

  Takes a list of alternating input and output strings, applies rules
  (and thus @MyRules) to input and compares it to the output. If you
  have html attributes in the output you want to test, use
  xpath_run_tests, because the order of the attributes is not
  guaranteed and varies with CGI library version.

    run_tests(q{"Get lost!", they say.},
	      q{&#x201c;Get lost!&#x201d;, they say.});

    run_tests(split('\n', <<'EOT'));
    input1
    output1
    input2
    output2
    EOT

  Newline excapes \n in the input and output are translated to real
  newlines when running the tests.

    run_tests(split('\n',<<'EOT'));
    * ''one\n** two
    <ul><li><em>one</em><ul><li>two</li></ul></li></ul>
    EOT



  test_page and test_page_negative
  --------------------------------

  Tests any string for regular expression matches:

    test_page($string, $regexp1, $regexp2, ...);
    test_page(update_page($pagename, $content), $re1, $re2);

  Or make sure that none of the regular expressions match:

    test_page_negative($page,
		       "rollback",
		       "Rollback",
		       "EvilPage",
		       "AnotherEvilPage",
		      );



  xpath_run_tests
  ---------------

  This is the equivalent of run_tests using XPath instead of simple
  string comparison.  It takes a list of alternating input and xpath
  tests, applies rules (and thus @MyRules) to the input and applies
  the test to the output.

    xpath_run_tests(split('\n',<<'EOT'));
    WikiWord
    //a[@class="edit"][@title="Click to edit this page"]
    This is a [:day for fun and laughter].
    //a[@class="anchor"][@name="day_for_fun_and_laughter"]
    EOT

  XPath is harder to write, but is ideal when the output contains tags
  with more than one attribute, since the order of attributes is
  undefined. And you don't even have to test for all the attributes.



  xpath_test and negative_xpath_test
  ----------------------------------

  The equivalent of test_page, but using xpath instead of exact
  matches.


    xpath_test(get_page('action=all pwd=foo'),
	     '//p/a[@href="#HomePage"][text()="HomePage"]',
	     '//h1/a[@name="foo"][text()="foo"]',
	     '//a[@class="local"][@href="#bar"][text()="bar"]',
	     '//h1/a[@name="bar"][text()="bar"]')

  And the same thing for negative matches, of course:

    negative_xpath_test($page, '//h1/a[not(text())]');


  run_macro_tests
  ---------------

    run_macro_tests(split('\n',<<'EOT'));
    $input1
    $output2
    $input2
    $output2
    EOT

  Tests @MyMacros.