Apache::TestUtil - Utility functions for writing tests

  use Apache::Test;
  use Apache::TestUtil;

  ok t_cmp("foo", "foo", "sanity check");
  t_write_file("filename", @content);
  my $fh = t_open_file($filename);
  t_mkdir("/foo/bar");
  t_rmtree("/foo/bar");
  t_is_equal($a, $b);

CWApache::TestUtil automatically exports a number of functions useful in writing tests.

All the files and directories created using the functions from this package will be automatically destroyed at the end of the program execution (via END block). You should not use these functions other than from within tests which should cleanup all the created directories and files at the end of the test.

t_cmp()

  t_cmp($received, $expected, $comment);

t_cmp() prints the values of $comment, $expected and $received. e.g.:

  t_cmp(1, 1, "1 == 1?");

prints:

  # testing : 1 == 1?
  # expected: 1
  # received: 1

then it returns the result of comparison of the $expected and the $received variables. Usually, the return value of this function is fed directly to the ok() function, like this:

  ok t_cmp(1, 1, "1 == 1?");

the third argument ($comment) is optional, mostly useful for telling what the comparison is trying to do. It is valid to use CWundef as an expected value. Therefore:

  my $foo;
  t_cmp(undef, $foo, "undef == undef?");

will return a true value. You can compare any two data-structures with t_cmp(). Just make sure that if you pass non-scalars, you have to pass their references. The datastructures can be deeply nested. For example you can compare:

  t_cmp({1 => [2..3,{5..8}], 4 => [5..6]},
        {1 => [2..3,{5..8}], 4 => [5..6]},
        "hash of array of hashes");

You can also compare the second argument against the first as a regex. Use the CWqr// function in the second argument. For example:

  t_cmp("abcd", qr/^abc/, "regex compare");

will do:

  "abcd" =~ /^abc/;

This function is exported by default.

t_filepath_cmp()

This function is used to compare two filepaths via t_cmp(). For non-Win32, it simply uses t_cmp() for the comparison, but for Win32, Win32::GetLongPathName() is invoked to convert the first two arguments to their DOS long pathname. This is useful when there is a possibility the two paths being compared are not both represented by their long or short pathname. This function is exported by default.

t_debug()

  t_debug("testing feature foo");
  t_debug("test", [1..3], 5, {a=>[1..5]});

t_debug() prints out any datastructure while prepending CW# at the beginning of each line, to make the debug printouts comply with CWTest::Harness's requirements. This function should be always used for debug prints, since if in the future the debug printing will change (e.g. redirected into a file) your tests won't need to be changed. the special global variable CW$Apache::TestUtil::DEBUG_OUTPUT can be used to redirect the output from t_debug() and related calls such as t_write_file(). for example, from a server-side test you would probably need to redirect it to STDERR:

  sub handler {
    plan $r, tests => 1;

    local $Apache::TestUtil::DEBUG_OUTPUT = \*STDERR;

    t_write_file('/tmp/foo', 'bar');
    ...
  }

left to its own devices, t_debug() will collide with the standard HTTP protocol during server-side tests, resulting in a situation both confusing difficult to debug. but STDOUT is left as the default, since you probably don't want debug output under normal circumstances unless running under verbose mode. This function is exported by default.

t_write_file()

  t_write_file($filename, @lines);

t_write_file() creates a new file at $filename or overwrites the existing file with the content passed in @lines. If only the $filename is passed, an empty file will be created. If parent directories of CW$filename don't exist they will be automagically created. The generated file will be automatically deleted at the end of the program's execution. This function is exported by default.

t_append_file()

  t_append_file($filename, @lines);

t_append_file() is similar to t_write_file(), but it doesn't clobber existing files and appends CW@lines to the end of the file. If the file doesn't exist it will create it. If parent directories of CW$filename don't exist they will be automagically created. The generated file will be registered to be automatically deleted at the end of the program's execution, only if the file was created by t_append_file(). This function is exported by default.

t_write_shell_script()

  Apache::TestUtil::t_write_shell_script($filename, @lines);

Similar to t_write_file() but creates a portable shell/batch script. The created filename is constructed from CW$filename and an appropriate extension automatically selected according to the platform the code is running under. It returns the extension of the created file.

t_write_perl_script()

  Apache::TestUtil::t_write_perl_script($filename, @lines);

Similar to t_write_file() but creates a executable Perl script with correctly set shebang line.

t_open_file()

  my $fh = t_open_file($filename);

t_open_file() opens a file $filename for writing and returns the file handle to the opened file. If parent directories of CW$filename don't exist they will be automagically created. The generated file will be automatically deleted at the end of the program's execution. This function is exported by default.

t_mkdir()

  t_mkdir($dirname);

t_mkdir() creates a directory $dirname. The operation will fail if the parent directory doesn't exist. If parent directories of CW$dirname don't exist they will be automagically created. The generated directory will be automatically deleted at the end of the program's execution. This function is exported by default.

t_rmtree()

  t_rmtree(@dirs);

t_rmtree() deletes the whole directories trees passed in @dirs. This function is exported by default.

t_chown()

  Apache::TestUtil::t_chown($file);

Change ownership of CW$file to the test's User/Group. This function is noop on platforms where chown(2) is unsupported (e.g. Win32).

t_is_equal()

  t_is_equal($a, $b);

t_is_equal() compares any two datastructures and returns 1 if they are exactly the same, otherwise 0. The datastructures can be nested hashes, arrays, scalars, undefs or a combination of any of these. See t_cmp() for an example. If CW$b is a regex reference, the regex comparison CW$a =~ $b is performed. For example:

  t_is_equal($server_version, qr{^Apache});

If comparing non-scalars make sure to pass the references to the datastructures. This function is exported by default.

t_server_log_error_is_expected()

If the handler's execution results in an error or a warning logged to the error_log file which is expected, it's a good idea to have a disclaimer printed before the error itself, so one can tell real problems with tests from expected errors. For example when testing how the package behaves under error conditions the error_log file might be loaded with errors, most of which are expected. For example if a handler is about to generate a run-time error, this function can be used as:

  use Apache::TestUtil;
  ...
  sub handler {
      my $r = shift;
      ...
      t_server_log_error_is_expected();
      die "failed because ...";
  }

After running this handler the error_log file will include:

  *** The following error entry is expected and harmless ***
  [Tue Apr 01 14:00:21 2003] [error] failed because ...

When more than one entry is expected, an optional numerical argument, indicating how many entries to expect, can be passed. For example:

  t_server_log_error_is_expected(2);

will generate:

  *** The following 2 error entries are expected and harmless ***

If the error is generated at compile time, the logging must be done in the BEGIN block at the very beginning of the file:

  BEGIN {
      use Apache::TestUtil;
      t_server_log_error_is_expected();
  }
  use DOES_NOT_exist;

After attempting to run this handler the error_log file will include:

  *** The following error entry is expected and harmless ***
  [Tue Apr 01 14:04:49 2003] [error] Can't locate "DOES_NOT_exist.pm"
  in @INC (@INC contains: ...

Also see CWt_server_log_warn_is_expected() which is similar but used for warnings. This function is exported by default.

t_server_log_warn_is_expected()

CWt_server_log_warn_is_expected() generates a disclaimer for expected warnings. See the explanation for CWt_server_log_error_is_expected() for more details. This function is exported by default.

t_client_log_error_is_expected()

CWt_client_log_error_is_expected() generates a disclaimer for expected errors. But in contrast to CWt_server_log_error_is_expected() called by the client side of the script. See the explanation for CWt_server_log_error_is_expected() for more details. For example the following client script fails to find the handler:

  use Apache::Test;
  use Apache::TestUtil;
  use Apache::TestRequest qw(GET);

  plan tests => 1;

  t_client_log_error_is_expected();
  my $url = "/error_document/cannot_be_found";
  my $res = GET($url);
  ok t_cmp(404, $res->code, "test 404");

After running this test the error_log file will include an entry similar to the following snippet:

  *** The following error entry is expected and harmless ***
  [Tue Apr 01 14:02:55 2003] [error] [client 127.0.0.1] 
  File does not exist: /tmp/test/t/htdocs/error

When more than one entry is expected, an optional numerical argument, indicating how many entries to expect, can be passed. For example:

  t_client_log_error_is_expected(2);

will generate:

  *** The following 2 error entries are expected and harmless ***

This function is exported by default.

t_client_log_warn_is_expected()

CWt_client_log_warn_is_expected() generates a disclaimer for expected warnings on the client side. See the explanation for CWt_client_log_error_is_expected() for more details. This function is exported by default.

t_catfile('a', 'b', 'c')

This function is essentially CWFile::Spec->catfile, but on Win32 will use CWWin32::GetLongpathName() to convert the result to a long path name (if the result is an absolute file). The function is not exported by default.

t_catfile_apache('a', 'b', 'c')

This function is essentially CWFile::Spec::Unix->catfile, but on Win32 will use CWWin32::GetLongpathName() to convert the result to a long path name (if the result is an absolute file). It is useful when comparing something to that returned by Apache, which uses a Unix-style specification with forward slashes for directory separators. The function is not exported by default.

Stas Bekman <stas@stason.org>

perl(1)