package Test::XPath;

use strict;
use 5.6.2;
use XML::LibXML '1.69';
use Test::Builder;

our $VERSION = '0.20';

sub new {
    my ($class, %p) = @_;
    my $doc = delete $p{doc} || _doc(\%p);
    my $xpc = XML::LibXML::XPathContext->new( $doc->documentElement );
    if (my $ns = $p{xmlns}) {
        while (my ($k, $v) = each %{ $ns }) {
            $xpc->registerNs( $k => $v );
    return bless {
        xpc    => $xpc,
        node   => $doc->documentElement,
        filter => do {
            if (my $f = $p{filter}) {
                if (ref $f eq 'CODE') {
                } elsif ($f eq 'css_selector') {
                    eval 'use HTML::Selector::XPath 0.06';
                    die 'Please install HTML::Selector::XPath to use CSS selectors'
                        if $@;
                    sub {
                        my $xpath = do {
                            my $xp = HTML::Selector::XPath->new(shift)->to_xpath(root => '//');
                            if (eval { $_->isa(__PACKAGE__) } && $_->node ne $doc->documentElement) {
                                # Make it relative to the current node.
                                $xp =~ s{^///[*]}{.};
                            } else {
                                # Start from the top.
                                $xp =~ s{^///[*]}{};
                        return $xpath;
                } else {
                    die "Unknown filter: $f\n";
            } else {
                sub { shift },
    }, $class;

sub ok {
    my ($self, $xpath, $code, $desc) = @_;
    my $xpc  = $self->{xpc};
    my $Test = Test::Builder->new;
    $xpath   = $self->{filter}->($xpath, $self);

    # Code and desc can be reversed, to support PerlX::MethodCallWithBlock.
    ($code, $desc) = ($desc, $code) if ref $desc eq 'CODE';

    if (ref $code eq 'CODE') {
        # Gonna do some recursive testing.
        my @nodes = $xpc->findnodes($xpath, $self->{node})
            or return $Test->ok(0, $desc);

        # Record the current test result.
        my $ret  = $Test->ok(1, $desc);

        # Call the code ref on each found node.
        local $_ = $self;
        for my $node (@nodes) {
            local $self->{node} = $node;
        return $ret;
    } else {
        # We're just testing for existence ($code is description).
        $Test->ok( $xpc->exists($xpath, $self->{node}), $code);


sub not_ok {
    my ($self, $xpath, $desc) = @_;
    $xpath = $self->{filter}->($xpath);
    my $Test = Test::Builder->new;
    $Test->ok( !$self->{xpc}->exists($xpath, $self->{node}), $desc);

sub is     { Test::Builder::new->is_eq(   shift->find_value(shift), @_) }
sub isnt   { Test::Builder::new->isnt_eq( shift->find_value(shift), @_) }
sub like   { Test::Builder::new->like(    shift->find_value(shift), @_) }
sub unlike { Test::Builder::new->unlike(  shift->find_value(shift), @_) }
sub cmp_ok { Test::Builder::new->cmp_ok(  shift->find_value(shift), @_) }

sub node   { shift->{node} }
sub xpc    { shift->{xpc}  }

sub find_value {
    my $self = shift;
    $self->{xpc}->findvalue( $self->{filter}->(shift), $self->{node} );

sub _doc {
    my $p = shift;

    # Create and configure the parser.
    my $parser = XML::LibXML->new;

    # Apply any parser options.
    if (my $opts = $p->{options}) {
        while (my ($k, $v) = each %{ $opts }) {
            if (my $meth = $parser->can($k)) {
            } else {
                $parser->set_option($k => $v);

    # Parse and return the document.
    if ($p->{xml}) {
        return $p->{is_html}
            ? $parser->parse_html_string($p->{xml})
            : $parser->parse_string($p->{xml});

    if ($p->{file}) {
        return $p->{is_html}
            ? $parser->parse_html_file($p->{file})
            : $parser->parse_file($p->{file});

    require Carp;
        'Test::XPath->new requires the "xml", "file", or "doc" parameter'

# Add Test::XML::XPath compatibility?
# sub like_xpath($$;$)   { __PACKAGE__->new( xml => shift )->ok(     @_ ) }
# sub unlike_xpath($$;$) { __PACKAGE__->new( xml => shift )->not_ok( @_ ) }
# sub is_xpath($$$;$)    { __PACKAGE__->new( xml => shift )->is(     @_ ) }


=head1 Name

Test::XPath - Test XML and HTML content and structure with XPath expressions

=head1 Synopsis

  use Test::More tests => 5;
  use Test::XPath;

  my $xml = <<'XML';
      <style type="text/css" src="foo.css"></style>
      <style type="text/css" src="bar.css"></style>
      <h1>Welcome to my lair.</h1>

  my $tx = Test::XPath->new( xml => $xml );

  $tx->ok( '/html/head', 'There should be a head' );
  $tx->is( '/html/head/title', 'Hello', 'The title should be correct' );

  # Recursing into a document:
  my @css = qw(foo.css bar.css);
  $tx->ok( '/html/head/style[@type="text/css"]', sub {
      my $css = shift @css;
      shift->is( './@src', $css, "Style src should be $css");
  }, 'Should have style' );

  # Better yet, use PerlX::MethodCallWithBlock:
  use PerlX::MethodCallWithBlock;
  my @css = qw(foo.css bar.css);
  use PerlX::MethodCallWithBlock;
  $tx->ok( '/html/head/style[@type="text/css"]', 'Should have style' ) {
      my $css = shift @css;
      shift->is( './@src', $css, "Style src should be $css");

  # Or use CSS Selectors:
  $tx = Test::XPath->new( xml => $xml, filter => 'css_selector' );
  $tx->ok( '> html > head', 'There should be a head' );

=head1 Description

Use the power of XPath expressions to validate the structure of your XML and
HTML documents.

=head2 About XPath

XPath is a powerful query language for XML documents. Test::XPath relies on
the libxml2 implementation provided by L<XML::LibXML>. libxml2 -- pretty much
the canonical library for XML processing -- provides an efficient and complete
implementation of the XPath spec.

XPath works by selecting nodes in an XML document. Nodes, in general,
correspond to the elements (a.k.a. tags) defined in the XML, text within those
elements, attribute values, and comments. The expressions for making such
selections use a URI-like syntax, the basics of which are:


=item C<$nodename>

Selects all child nodes with the name.

=item C</>

Selects the root node.

=item C<//>

Selects nodes from the current node that match the selection, regardless of
where they are in the node hierarchy.

=item C<.>

Selects the current node.

=item C<..>

Selects the parent of the current node.

=item C<@>

Selects attributes.


And some examples:


=item C<head>

Selects all of the child nodes of the "head" element.

=item C</html>

Selects the root "html" element.

=item C<body/p>

Selects all "p" elements that are children of the "body" element.

=item C<//p>

Selects all "p" elements no matter where they are in the document.

=item C<body//p>

Selects all "p" elements that are descendants of the "body" element, no matter
where they appear under the "body" element.

=item C<//@lang>

Selects all attributes named "lang".


There are also useful predicates to select certain nodes. Some examples:


=item C<body//p[1]>

Select the first paragraph under the body element.

=item C<body//p[last()]>

Select the last paragraph under the body element.

=item C<//script[@src]>

Select all "script" nodes that have a "src" attribute.

=item C<//script[@src='foo.js']>

Select all "script" nodes that have a "src" attribute set to "foo.js".

=item C<< //img[@height > 400] >>

Select all "img" nodes with a height attribute greater than 400.

=item C<head/*>

Select all child nodes below the "head" node.

=item C<p[@*]>

Select all "p" nodes that have any attribute.

=item C<count(//p)>

Select a count of all "p" nodes in the document.

=item C<contains(//title, "Welcome")>

Select true if the title node contains the string "Welcome", and false if it
does not.


There are a bunch of core functions in XPath. In addition to the (C<last()>
and C<count()>) examples above, there are functions for node sets, booleans,
numbers, and strings. See the
L<XPath 1.0 W3C Recommendation|>, for thorough (and
quite readable) documentation of XPath support, including syntax and the core
functions. The L<W3Schools tutorial|>
provides a nice overview of XPath.

=head2 Testing HTML

If you want to use XPath to test the content and structure of an HTML document,
be sure to pass the C<is_html> option to C<new()>, like so:

  my $tx = Test::XPath->new( xml => $html, is_html => 1 );

Test::XPath will then use XML::LibXML's HTML parser to parse the document,
rather than its XML parser. The upshot is that you won't have to worry about
namespace prefixes, and XML::LibXML won't try to fetch any DTD specified in
the DOCTYPE section of your HTML.

=head1 Class Interface

=head2 Constructor

=head3 C<new>

  my $tx = Test::XPath->new( xml => $xml );

Creates and returns an XML::XPath object. This object can be used to run XPath
tests on the XML passed to it. The supported parameters are:


=item C<xml>

  xml => '<foo><bar>hey</bar></foo>',

The XML to be parsed and tested. Required unless the C<file> or C<doc> option
is passed.

=item C<file>

  file => 'rss.xml',

Name of a file containing the XML to be parsed and tested. Required unless the
C<xml> or C<doc> option is passed.

=item C<doc>

  doc => XML::LibXML->new->parse_file($xml_file),

An XML::LibXML document object. Required unless the C<xml> or C<file> option
is passed.

=item C<is_html>

  is_html => 1,

If the XML you're testing is actually HTML, pass this option a true value and
XML::LibXML's HTML parser will be used instead of the XML parser. This is
especially useful if your HTML has a DOCTYPE declaration or an XML namespace
(xmlns attribute) and you don't want the parser grabbing the DTD over the
Internet and you don't want to mess with a namespace prefix in your XPath

=item C<xmlns>

  xmlns => {
      x => '',
      a => '',

Set up prefixes for XML namespaces. Required if your XML uses namespaces and
you want to write reasonable XPath expressions.

=item C<options>

  options => { recover_silently => 1, no_network => 1 },

Optional hash reference of
L<XML::LibXML::Parser options|XML::LibXML::Parser/"PARSER OPTIONS">, such as
"validation", "recover", "suppress_errors", and "no_network". These can be
useful for tweaking the behavior of the parser.

=item C<filter>

  filter => 'css_selector',
  filter => sub { my $xpath = shift; },

Pass a filter name or a code reference for Test::XPath to use to filter XPath
expressions before passing them on to XML::LibXML. The code reference argument
allows you to transform XPath expressions if, for example, you use a custom
XPath syntax that's more concise than XPath.

There is currently only one built-in filter, C<css_selector>. So if you pass

  filter => 'css_selector',

Then any paths passed to C<ok()>, C<is()>, etc., will be passed through
L<HTML::Selector::XPath>. This allows you to use CSS selector syntax, which
can be more compact for simple expressions. For example, this CSS selector:

    $tx->is('div#content div.article h1', '...')

Is equivalent to this XPath expression:

    $tx->is('//div[@id="content"]//div[@class="article"]//h1', '...')


=head1 Instance Interface

=head2 Assertions

=head3 C<ok>

  $tx->ok( $xpath, $description )
  $tx->ok( $xpath, $coderef, $description )

Test that an XPath expression evaluated against the XML document returns a
true value. If the XPath expression finds no nodes, the result will be false.
If it finds a value, the value must be a true value (in the Perl sense).

  $tx->ok( '//foo/bar', 'Should have bar element under foo element' );
  $tx->ok( 'contains(//title, "Welcome")', 'Title should "Welcome"' );

You can also run recursive tests against your document by passing a code
reference as the second argument to C<ok()>. Once the initial selection has
been completed, each selected node will be assigned to the C<node> attribute
and the XML::XPath object passed to the code reference. For example, if you
wanted to test for the presence of "story" elements in your document, and to
test that each such element had an incremented "id" attribute, you'd do
something like this:

  my $i = 0;
  $tx->ok( '//assets/story', sub {
      shift->is('./@id', ++$i, "ID should be $i in story $i");
  }, 'Should have story elements' );

Even better, use L<PerlX::MethodCallWithBlock> to pass a block to the method
instead of a code reference:

  use PerlX::MethodCallWithBlock;
  my $i = 0;
  $tx->ok( '//assets/story', 'Should have story elements' ) {
      shift->is('./@id', ++$i, "ID should be $i in story $i");

For convenience, the XML::XPath object is also assigned to C<$_> for the
duration of the call to the code reference. Either way, you can call C<ok()>
and pass code references anywhere in the hierarchy. For example, to ensure
that an Atom feed has entries and that each entry has a title, a link, and a
very specific author element with name, uri, and email subnodes, you can do

  $tx->ok( '/feed/entry', sub {
      $_->ok( './title', 'Should have a title' );
      $_->ok( './author', sub {
          $_->is( './name',  'Mark Pilgrim',        'Mark should be author' );
          $_->is( './uri',   '', 'URI should be correct' );
          $_->is( './email', '',    'Email should be right' );
      }, 'Should have author elements' );
  }, 'Should have entry elments' );

=head3 C<not_ok>

  $tx->not_ok( $xpath, $description )

The reverse of the non-recursive C<ok()>, the test succeeds if the XPath
expression matches no part of the document.

  $tx->not_ok( '//foo/bar[@id=0]', 'Should have no bar elements with Id 0' );

=head3 C<is>

=head3 C<isnt>

  $tx->is( $xpath, $want, $description );
  $tx->isnt( $xpath, $dont_want, $description );

C<is()> and C<isnt()> compare the value returned by evaluation of the XPath
expression against the document to a value using C<eq> and C<ne>,

  $tx->is( '/html/head/title', 'Welcome', 'Title should be welcoming' );
  $tx->isnt( '/html/head/link/@type', 'hello', 'Link type should not' );

As with C<Test::More::ok()>, a failing test will yield a useful diagnostic
message, something like:

  #   Failed test 'Title should be welcoming'
  #   at t/foo.t line 47.
  #          got: 'Bienvenidos'
  #     expected: 'Hello'

=head3 C<like>

=head3 C<unlike>

  $tx->like( $xpath, qr/want/, $description );
  $tx->unlike( $xpath, qr/dont_want/, $description );

Similar to C<is()> and C<isnt()>, but these methods match the value returned
by the XPath expression against a regular expression.

  $tx->like( '/html/head/title', qr/^Foobar Inc.: .+/, 'Title context' );
  $tx->unlike( '/html/head/title', qr/Error/, 'Should be no error in title' );

As with C<Test::More::like()>, a failing test will yield a useful diagnostic
message, something like:

  #   Failed test 'Title should, like, welcome'
  #   at t/foo.t line 62.
  #                   'Bye'
  #     doesn't match '(?-xism:^Howdy$)'

=head3 C<cmp_ok>

  $tx->cmp_ok( $xpath, $op, $want, $description );

Like C<Test::More::cmp_ok()>, this method allows you to compare the value
returned by an XPath expression to a value using any binary Perl operator.

  $tx->cmp_ok( '/html/head/title', 'eq', 'Welcome' );
  $tx->cmp_ok( '//story[1]/@id', '==', 1 );

As with C<Test::More::cmp_ok()>, a failing test will yield a useful diagnostic
message, something like:

  #   Failed test
  #   at t/foo.t line 104.
  #     '0'
  #         &&
  #     '1'

=head2 Accessors

=head3 C<node>

  my $node = $tx->node;

Returns the current context node. This will usually be the node for the entire
document, but in recursive tests run in code references passed to C<ok()>, the
node will be one of the nodes selected for the test.

=head3 C<xpc>

Returns the L<XML::LibXML::XPathContext> used to execute the XPath
expressions. It can be useful to access this object in order to create new
XPath functions to use in your tests. For example, say that you wanted to
define a C<grep()> XPath function that returns true for a node value that
matches a regular expression. You can define one like so:

  $tx->xpc->registerFunction( grep => sub {
      my ($nodelist, $regex) =  @_;
      my $result = XML::LibXML::NodeList->new;
      for my $node ($nodelist->get_nodelist) {
          $result->push($node) if $node->textContent =~ $regex;
      return $result;
  } );

You can then use C<grep()> like any other XPath function to select only those
nodes with content matching a regular expression. This example makes sure that
there are "email" nodes under "author" nodes that end in "" or

      'grep(//author/email, "@example[.](?:com|org)$")',
      'Should have example email'

=head2 Utilities

=head3 C<find_value>

  my $val = $tx->find_value($xpath);

Returns the value returned by evaluation of the XPath expression against the
document relative to the current node. This is the method used internally to
fetch the value to be compared by C<is>, C<isnt>, C<like>, C<unlike>, and
C<cmp_ok>. A simple example:

  my $val = $tx->find_value('/html/head/title');

=head1 See Also


=item *

L<XML Path Language (XPath) Version 1.0 W3C Recommendation|>.

=item *

L<W3Schools XPath Tutorial|>.

=item *

L<XML::LibXML::XPathContext> - The XML::LibXML XPath evaluation library.

=item *

L<Test::XML::XPath> - Another library for testing XPath assertions using a
functional interface. Ships with L<Test::XML>.

=item *

L<Test::HTML::Content> - Another module that that offers C<xpath_ok()> and
C<no_xpath()> test functions.


=head1 Support

This module is stored in an open L<GitHub
repository|>. Feel free to fork and

Please file bug reports via L<GitHub
Issues|> or by sending mail to

=head1 Author

David E. Wheeler <>

Currently maintained by Mohammad S Anwar <>

=head1 Copyright and License

Copyright (c) 2009-2010 David E. Wheeler. Some Rights Reserved.

This module is free software; you can redistribute it and/or modify it under
the same terms as Perl itself.