Provided by: libwww-search-perl_2.51.90+~cs6.78-2_all bug

NAME

       WWW::Search::Test - utilities to aid in testing WWW::Search backends

SYNOPSIS

         $oTest = new WWW::Search::Test('HotBot,Yahoo,Excite');
         $oTest->test('HotBot', 'Kingpin', 'one', $sQuery, $TEST_RANGE, 1, 10);

DESCRIPTION

       See file test.pl in the WWW-Search-HotBot distribution for a detailed "real-world"
       example.

METHODS AND FUNCTIONS

   find_websearch
       Returns the full path of an executable WebSearch program, or undef if none can be found.

   new
       Create a new WWW::Search::Test object.  All arguments are strings, names of backends that
       this object will be able to test.  If no arguments are given, will be able to test all
       backends.

   mode
       Set / get the test mode of this object.  If an argument is given, sets the mode to that
       value.  Returns the current (or newly set) value.

       There are three test modes available.  They are:

         $MODE_INTERNAL: parse URLs out of saved pages (as a sanity check or regression test);
         $MODE_EXTERNAL: send the query to the search engine "live", parse the results, and compare them to the previously saved results;
       and
         $MODE_UPDATE: send the query to the search engine "live", parse the results, and save them for future testing.

   relevant_test
       Given the name of a backend, returns true if this Test object is able to test that
       backend.

   eval_test
       Given the name of a backend, grabs the $TEST_CASES variable from that backend and
       evaluates it.

   test
       Run test(s) for a backend.  Arguments are, in order: name of a backend to test (string,
       required); name of backend maintainer (string, if undef $backend::MAINTAINER will be
       used); filename for results storage/comparison (string, required); query to be sent to
       backend (string, required); test method (required, one of the following).

       Several test methods are possible:

         $TEST_EXACTLY: list of URLs must match exactly (line for line, in order);
         $TEST_BY_COUNTING: test passes if number of resulting URLs is equal;
         $TEST_GREATER_THAN: test passes if we get more than N result URLs;
       and
         $TEST_RANGE: like $TEST_GREATER_THAN but constrained on both ends.

   no_test
       Prints a message stating that this backend does not have a test suite.  Takes two
       arguments, the backend name and the name of the maintainer.

   not_working
       Prints a message stating that this backend is known to be broken.  Takes two arguments,
       the backend name and the name of the maintainer.

   not_working_with_tests
       Prints a message stating that this backend is known to be broken even though it has a test
       suite.  Takes two arguments, the backend name and the name of the maintainer.

   not_working_and_abandoned
       Prints a message stating that this backend is known to be broken and is not being actively
       maintained.  Takes two arguments, the backend name and the name of the maintainer.

   reset_error_count
       Reset the counter of errors to zero.  You probably want to call this before each call to
       test() or eval_test().

   wc_l (private, not a method)
       Given a filename, count the number of lines of text contained within the file.  (I.e.
       simulate running UNIX command "wc -l" on a file)

   diff (private, not a method)
       Given two files, returns TRUE if contents are line-by-line different, or FALSE if contents
       are line-by-line same.  (I.e. like the UNIX command diff, but just reports true or false)

   Shortcuts for running backend tests
       WWW::Search::Test keeps its own count of test numbers, so if you want to mix-and-match
       these functions with your own tests, use the $WWW::Search::Test::iTest counter.

   new_engine
       One argument: the name of a backend suitable to be passed to WWW::Search::new().  Prints
       'ok' or 'not ok' and the test number.  Creates a WWW::Search object internally, to be used
       for all subsequent calls to run_test and run_gui_test (see below).

   tm_new_engine
       Same as new_engine(), but uses Test::More instead of just printing 'ok'.

   run_test
       Three arguments: a query string, NOT escaped; a minimum number of expected results; and a
       maximum number of expected results.  Optional fourth argument: integer value to be used as
       the search_debug.  Optional fifth argument: send any true value to dump the search
       results.  Optional sixth argument: reference to hash of search options (see backend
       documentation).  Optional seventh argument: send any true value to NOT escape the query
       string.

       If the minimum is undef, assumes zero.  If the maximum is undef, does not check.

       Prints 'ok' or 'not ok' and the test number.

   run_gui_test
       Same as run_test(), but calls gui_query() instead of native_query().

   tm_run_test
       Same as run_test(), but uses Test::More rather than just printing 'ok'.

       Note: If you use this function inside a TODO block, you must set global variable $TODO
       rather than a local $TODO, and you must set the global $TODO back to empty-string (or
       undef) at the end of your TODO block.  For example:

         TODO:
           {
           $TODO = 'I have not fixed this yet';
           tm_run_test(...);
           $TODO = '';
           } # end of TODO block

   tm_run_test_no_approx
       Same as tm_run_test, but does NOT check the approximate_result_count.

   count_results
       Run a query, and return the actual (not approximate) number of hits.  Required first
       argument determines which backend query method to call: 'gui' to call gui_query(),
       anything else to call native_query().  Remaining arguments are same as all the run_test()
       arguments.

   skip_test
       You can call this function instead of run_test() or run_gui_test() if the current test
       must be skipped for any reason.

   test_most_results
       Given an arrayref of things to test, runs all those things against all the results of the
       most-recently executed test search.