Test::HTML::Content(3) Perl extension for testing HTML output

SYNOPSIS


use Test::HTML::Content( tests => 13 );

  $HTML = "<html><title>A test page</title><body><p>Home page</p>
           <img src='http://www.perl.com/camel.png' alt='camel'>
           <a href='http://www.perl.com'>Perl</a>
           <img src='http://www.perl.com/camel.png' alt='more camel'>
           <!--Hidden message--></body></html>";
  link_ok($HTML,"http://www.perl.com","We link to Perl");
  no_link($HTML,"http://www.pearl.com","We have no embarassing typos");
  link_ok($HTML,qr"http://[a-z]+\.perl.com","We have a link to perl.com");
  title_count($HTML,1,"We have one title tag");
  title_ok($HTML,qr/test/);
  tag_ok($HTML,"img", {src => "http://www.perl.com/camel.png"},
                        "We have an image of a camel on the page");
  tag_count($HTML,"img", {src => "http://www.perl.com/camel.png"}, 2,
                        "In fact, we have exactly two camel images on the page");
  no_tag($HTML,"blink",{}, "No annoying blink tags ..." );
  # We can check the textual contents
  text_ok($HTML,"Perl");
  # We can also check the contents of comments
  comment_ok($HTML,"Hidden message");
  # Advanced stuff
  # Using a regular expression to match against
  # tag attributes - here checking there are no ugly styles
  no_tag($HTML,"p",{ style => qr'ugly$' }, "No ugly styles" );
  # REs also can be used for substrings in comments
  comment_ok($HTML,qr"[hH]idden\s+mess");
  # and if you have XML::LibXML or XML::XPath, you can
  # even do XPath queries yourself:
  xpath_ok($HTML,'/html/body/p','HTML is somewhat wellformed');
  no_xpath($HTML,'/html/head/p','HTML is somewhat wellformed');

DESCRIPTION

This is a module to test the HTML output of your programs in simple test scripts. It can test a scalar (presumably containing HTML) for the presence (or absence, or a specific number) of tags having (or lacking) specific attributes. Unspecified attributes are ignored, and the attribute values can be specified as either scalars (meaning a match succeeds if the strings are identical) or regular expressions (meaning that a match succeeds if the actual attribute value is matched by the given RE) or undef (meaning that the attribute must not be present).

If you want to specify or test the deeper structure of the HTML (for example, META tags within the BODY) or the (textual) content of tags, you will have to resort to "xpath_ok","xpath_count" and "no_xpath", which take an XPath expression. If you find yourself crafting very complex XPath expression to verify the structure of your output, it is time to rethink your testing process and maybe use a template based solution or simply compare against prefabricated files as a whole.

The used HTML parser is HTML::TokeParser, the used XPath module is XML::XPath or XML::LibXML. XML::XPath needs valid xHTML, XML::LibXML will try its best to force your code into xHTML, but it is best to supply valid xHTML (snippets) to the test functions.

If no XPath parsers/interpreters are available, the tests will automatically skip, so your users won't need to install XML::XPath or XML::LibXML. The module then falls back onto a crude implementation of the core functions for tags, links, comments and text, and the diagnostic output of the tests varies a bit.

The test functionality is derived from Test::Builder, and the export behaviour is the same. When you use Test::HTML::Content, a set of HTML testing functions is exported into the namespace of the caller.

EXPORT

Exports the bunch of test functions :

  link_ok() no_link() link_count()
  tag_ok() no_tag() tag_count()
  text_ok no_text() text_count()
  comment_ok() no_comment() comment_count()
  xpath_ok() no_xpath() xpath_count()
  has_declaration() no_declaration()

CONSIDERATIONS

The module reparses the HTML string every time a test function is called. This will make running many tests over the same, large HTML stream relatively slow. A possible speedup could be simple minded caching mechanism that keeps the most recent HTML stream in a cache.

CAVEATS

The test output differs between XPath and HTML parsing, because XML::XPath delivers the complete node including the content, where my HTML parser only delivers the start tag. So don't make your tests depend on the _exact_ output of my tests. It was a pain to do so in my test scripts for this module and if you really want to, take a look at the included test scripts.

The title functions "title_ok" and "no_title" rely on the XPath functionality and will thus skip if XPath functionality is unavailable.

BUGS

Currently, if there is text split up by comments, the text will be seen as two separate entities, so the following dosen't work :

  is_text( "Hello<!-- brave new--> World", "Hello World" );

Whether this is a real bug or not, I don't know at the moment - most likely, I'll modify text_ok() and siblings to ignore embedded comments.

TODO

My things on the todo list for this module. Patches are welcome !
  • Refactor the code to fold some of the internal routines
  • Implement a cache for the last parsed tree / token sequence
  • Possibly diag() the row/line number for failing tests
  • Allow RE instead of plain strings in the functions (for tags themselves). This one is most likely useless.

LICENSE

This code may be distributed under the same terms as Perl itself.

AUTHOR

Max Maischein <[email protected]>