—

              
package HTML::TreeBuilder::XPath;
use List::Util qw( first);
use Scalar::Util ();
use strict;
use warnings;
use vars qw($VERSION);
$VERSION = '0.14';
my %CHAR2DEFAULT_ENT= ( '&' => '&amp;', '<' => '&lt;', '>' => '&gt;', '"' => '&quot;');
my %NUM2DEFAULT_ENT= ( '38' => 'amp', '60' => 'lt', '62' => 'gt', '"' => '&quot;');
package HTML::TreeBuilder::XPath;
use base( 'HTML::TreeBuilder');
package HTML::TreeBuilder::XPath::Node;
sub isElementNode   { 0 }
sub isAttributeNode { 0 }
sub isNamespaceNode { 0 }
sub isTextNode      { 0 }
sub isProcessingInstructionNode { 0 }
sub isPINode        { 0 }
sub isCommentNode   { 0 }
sub getChildNodes { return wantarray ? () : []; }
sub getFirstChild { return undef; }
sub getLastChild { return undef; }
# need to do a complete look_down each time, as the id could have been changed 
# without any object being involved, hence without a potential cache being
# up to date
sub getElementById 
  { my ($self, $id) = @_;
    return scalar $self->look_down( id => $id);
  }
sub to_number { return XML::XPathEngine::Number->new( shift->getValue); }
sub cmp
  { my( $a, $b)=@_;
    # comparison with the root (in $b, or processed in HTML::TreeBuilder::XPath::Root)
    if( $b->isa( 'HTML::TreeBuilder::XPath::Root') ) { return -1; }
    # easy cases
    return  0 if( $a == $b);    
    return  1 if( $a->is_inside($b)); # a starts after b 
    return -1 if( $b->is_inside($a)); # a starts before b
    # lineage does not include the element itself
    my @a_pile= ($a, $a->lineage); 
    my @b_pile= ($b, $b->lineage);
     
    # the 2 elements are not in the same twig
    unless( $a_pile[-1] == $b_pile[-1]) 
      { warn "2 nodes not in the same pile: ", ref( $a), " - ", ref( $b), "\n"; 
        print "a: ", $a->string_value, "\nb: ", $b->string_value, "\n";
        return undef;
      }
    # find the first non common ancestors (they are siblings)
    my $a_anc= pop @a_pile;
    my $b_anc= pop @b_pile;
    while( $a_anc == $b_anc) 
      { $a_anc= pop @a_pile;
        $b_anc= pop @b_pile;
      }
    if( defined( $a_anc->{_rank}) && defined( $b_anc->{_rank}))
      { return $a_anc->{_rank} <=> $b_anc->{_rank}; }
    else
      {
        # from there move left and right and figure out the order
        my( $a_prev, $a_next, $b_prev, $b_next)= ($a_anc, $a_anc, $b_anc, $b_anc);
        while()
          { $a_prev= $a_prev->getPreviousSibling || return -1;
            return  1 if( $a_prev == $b_anc);
            $a_next= $a_next->getNextSibling     || return  1;
            return -1 if( $a_next == $b_anc);
            $b_prev= $b_prev->getPreviousSibling || return  1;
            return -1 if( $b_prev == $a_next);
            $b_next= $b_next->getNextSibling     || return -1;
            return  1 if( $b_next == $a_prev);
          }
      }
  }
# need to modify directly the HTML::Element package, because HTML::TreeBuilder won't let me
# change the class of the nodes it generates
package HTML::Element;
use Scalar::Util qw(weaken);
use vars qw(@ISA);
push @ISA, 'HTML::TreeBuilder::XPath::Node';
use XML::XPathEngine;
{ my $xp;
  sub xp
    { $xp ||=XML::XPathEngine->new();
      return $xp;
    }
}
sub findnodes            { my( $elt, $path)= @_; return xp->findnodes(            $path, $elt); }
sub findnodes_as_string  { my( $elt, $path)= @_; return xp->findnodes_as_string(  $path, $elt); }
sub findnodes_as_strings { my( $elt, $path)= @_; return xp->findnodes_as_strings( $path, $elt); }
sub findvalue            { my( $elt, $path)= @_; return xp->findvalue(            $path, $elt); }
sub findvalues           { my( $elt, $path)= @_; return xp->findvalues(           $path, $elt); }
sub exists               { my( $elt, $path)= @_; return xp->exists(               $path, $elt); }
sub find_xpath           { my( $elt, $path)= @_; return xp->find(                 $path, $elt); }
sub matches              { my( $elt, $path)= @_; return xp->matches( $elt, $path, $elt);        }
sub set_namespace        { my $elt= shift; xp->new->set_namespace( @_); }
sub getRootNode
  { my $elt= shift;
    # The parent of root is a HTML::TreeBuilder::XPath::Root
    # that helps getting the tree to mimic a DOM tree
    return $elt->root->getParentNode; # I like this one!
  }
sub getParentNode
  { my $elt= shift;
    return $elt->{_parent} || bless { _root => $elt }, 'HTML::TreeBuilder::XPath::Root';
  }
sub getName             { return shift->tag;   }
sub getLocalName        { (my $name= $_[0]->tag) =~ s{^.*:}{}; $name; }
sub getNextSibling      { my( $elt)= @_; 
                          my $parent= $elt->{_parent} || return undef;
                          return  $parent->_child_as_object( scalar $elt->right, ($elt->{_rank} || 0) + 1);
                        }
sub getPreviousSibling  { my( $elt)= @_; 
                          my $parent= $elt->{_parent} || return undef;
                          return undef unless $elt->{_rank};
                          return  $parent->_child_as_object( scalar $elt->left, $elt->{_rank} - 1); 
                        }
sub isElementNode       { return ref $_[0] && ($_[0]->{_tag}!~ m{^~}) ? 1 : 0; }
sub isCommentNode       { return ref $_[0] && ($_[0]->{_tag} eq '~comment') ? 1 : 0; }
sub isProcessingInstructionNode { return ref $_[0] && ($_[0]->{_tag} eq '~pi') ? 1 : 0; }
sub isTextNode          { return ref $_[0] ? 0 : 1; }
sub getValue 
  { my $elt= shift;
    if( $elt->isCommentNode) { return $elt->{text}; }
    return $elt->as_text;
  }
         
sub getChildNodes    
  { my $parent= shift;
    my $rank=0;
    my @children= map { $parent->_child_as_object( $_, $rank++) } $parent->content_list;
    return wantarray ? @children : \@children;
  }
sub getFirstChild
  { my $parent= shift;
    my @content= $parent->content_list;
    if( @content)
      { return $parent->_child_as_object( $content[0], 0); }
    else
      { return undef; }
  }
sub getLastChild
  { my $parent= shift;
    my @content= $parent->content_list;
    if( @content)
      { return $parent->_child_as_object( $content[-1], $#content); }
    else
      { return undef; }
  }
sub getAttributes
  { my $elt= shift;
    my %atts= $elt->all_external_attr;
    my $rank=0;
    my @atts= map { bless( { _name => $_, _value => $atts{$_}, 
                             _elt => $elt, _rank => $rank++, 
                           }, 
                               'HTML::TreeBuilder::XPath::Attribute'
                         )
                  } sort keys %atts;
    return wantarray ? @atts : \@atts;
  }
sub to_number { return XML::XPathEngine::Number->new( $_[0]->as_text); }
sub string_value 
  { my $elt= shift;
    if( $elt->isCommentNode) { return $elt->{text}; }
    return $elt->as_text;
  };
# called on a parent, with a child as second argument and its rank as third
# returns the child if it is already an element, or
# a new HTML::TreeBuilder::XPath::Text element if it is a plain string
sub _child_as_object
  { my( $elt, $elt_or_text, $rank)= @_;
    return undef unless( defined $elt_or_text);
    if( ! ref $elt_or_text)
      { # $elt_or_text is a string, turn it into a TextNode object
        $elt_or_text= bless { _content => $elt_or_text, _parent => $elt, }, 
                            'HTML::TreeBuilder::XPath::TextNode'
                      ;
        Scalar::Util::weaken($elt_or_text->{_parent});
      }
    if( ref $rank) { warn "rank is a ", ref( $rank), " elt_or_text is a ", ref( $elt_or_text); } 
    $elt_or_text->{_rank}= $rank; # used for sorting;
    return $elt_or_text;
  }
sub toString { return shift->as_XML( @_); }
# produces better looking XML
{
  no warnings 'redefine'; 
  sub as_XML_compact
    { my( $node, $opt)= @_;
      my $name = $node->{'_tag'};
      if( $name eq '~literal')     { return _xml_escape_text( $node->{text});                     }
      if( $name eq '~declaration') { return '<!'   . _xml_escape_text( $node->{text})    . ">";   }
      if( $name eq '~pi')          { return '<?'   . _xml_escape_text( $node->{text})    . '?>';  }
      if( $name eq '~comment')     { return '<!--' . _xml_escape_comment( $node->{text}) . '-->'; }
      my $lc_name= lc $name;
      my $xml= $node->_start_tag;
      if( $HTML::Tagset::isCDATA_Parent{$lc_name})
        { my $content= $node->{_content} || '';
          if( ref $content eq 'ARRAY' || $content->isa( 'ARRAY'))
            { $xml .= _xml_escape_cdata( join( '', @$content), $opt); }
          else
            { $xml .= $content; }
        }
      else
        { # start tag
          foreach my $child ($node->content_list) 
            { if( ref $child) { $xml .= $child->as_XML_compact(); }
              else            { $xml .=  _xml_escape_text( $child); }
            }
        }
          $xml.= "</$name>" unless $HTML::Tagset::emptyElement{$lc_name};
      return $xml;
    }
}
     
{ my %phrase_name;    # all phrase tags, + literals (those are not indented)
  my %extra_newline;  # tags that get an extra newline before the end tag
  my $default_indent; # 2 spaces, change with the 'indent' option
  BEGIN 
    { %phrase_name= %HTML::Tagset::isPhraseMarkup;
      $phrase_name{'~literal'}= 1;
      $default_indent= '  ';
      %extra_newline= map { $_ => 1 } qw(html head body script div table tbody thead tfoot tr form dl ol ul);
    }
  sub as_XML_indented
    { my( $node, $opt)= @_;
      my $name = $node->{'_tag'};
      my $lc_name= lc $name;
      if( $name eq '~literal')     { return _xml_escape_text( $node->{text});                     }
      if( $name eq '~declaration') { return '<!'   . _xml_escape_text( $node->{text})    . ">\n"; }
      if( $name eq '~pi')          { return '<?'   . _xml_escape_text( $node->{text})    . '?>';  }
      if( $name eq '~comment')     { return '<!--' . _xml_escape_comment( $node->{text}) . '-->'; }
       
      my $xml;
      my $pre_tag_indent='';
      if(!$phrase_name{$lc_name}) { $pre_tag_indent=  "\n" . ($opt->{indent} || $default_indent) x ($opt->{indent_level}||0); }
      if( $opt->{indent_level}) { $xml .= $pre_tag_indent; }
      $xml.= $node->_start_tag();
      my $content='';
      if( $HTML::Tagset::isCDATA_Parent{$lc_name})
        { my $content= $node->{_content} || '';
          if( ref $content && (ref $content eq 'ARRAY' || $content->isa( 'ARRAY') ))
            { $content= _xml_escape_cdata( join( '', @$content), $opt); }
        }
      else
        { 
          my %child_opt= %$opt;
          $child_opt{indent_level}++;
          foreach my $child ($node->content_list) 
            { if( ref $child) { $content .= $child->as_XML_indented( \%child_opt ); }
              else            { $content .=  _xml_escape_text( $child);             }
            }
        }
      $xml .= $content;
      if( $extra_newline{$lc_name} && $content ne '' ) { $xml.= $pre_tag_indent; }
      $xml.= "</$name>" unless $HTML::Tagset::emptyElement{$lc_name};
      $xml .="\n" if( !$opt->{indent_level});
      
      return $xml;
    }
}
sub _start_tag
  { my( $node)= @_;
    my $name = $node->{'_tag'};
    my $start_tag.= "<$name";
    foreach my $att_name (sort keys %$node) 
      { next if( (!length $att_name) ||  ($att_name=~ m{^_}) || ($att_name eq '/') );
        my $well_formed_att_name= well_formed_name( $att_name);
        $start_tag .= qq{ $well_formed_att_name="} . _xml_escape_attribute_value( $node->{$att_name}) . qq{"};
      }
    $start_tag.= $HTML::Tagset::emptyElement{lc $name} ? " />" : ">";
    return $start_tag;
  }
sub well_formed_name
  { my( $name)= @_;
    $name=~ s{[^\w:_-]+}{_}g;
    if( $name=~ m{^\d}) { $name= "a$name"; }
    return $name;
  }
sub _indent_level
  { my( $node)= @_;
    my $level= scalar grep { !$HTML::Tagset::isPhraseMarkup{lc $_->{_tag}} } $node->lineage;
    return $level;
  }
{ my( $indent, %extra_newline, $nl);
  BEGIN 
    { $indent= '  '; 
      $nl= "\n";
      %extra_newline= map { $_ => 1 } qw(html head body script div table tr form ol ul);
    }
    
  sub indents
    { my( $opt, $name)= @_;
      my $indents= { pre_start_tag => '', post_start_tag => '', pre_end_tag => '', post_end_tag => ''};
      if( $opt->{indented})
        { my $indent_level= $opt->{indent_level};
          my $wrapping_nl= $nl; 
          if( !defined( $indent_level)) { $indent_level = 0; $wrapping_nl= ''; }
          if( $HTML::Tagset::isKnown{lc $name} && !$HTML::Tagset::isPhraseMarkup{lc $name} && $indent_level > 0) 
            { $indents->{pre_start_tag}= $wrapping_nl . ($indent x $indent_level); }
          if( $extra_newline{lc $name})
            { $indents->{post_start_tag}= $nl; 
              $indents->{pre_end_tag}= $nl . ($indent x $indent_level);
            }
          if( $indent_level == 0) 
            { $indents->{post_end_tag} = $wrapping_nl; }
        }
      return $indents;
    }
}
     
sub _xml_escape_attribute_value
  { my( $text)= @_;
    $text=~ s{([&<>"])}{$CHAR2DEFAULT_ENT{$1}}g; # escape also quote, as it is the attribute separator
    return $text;
  }
sub _xml_escape_text
  { my( $text)= @_;
    $text=~ s{([&<>])}{$CHAR2DEFAULT_ENT{$1}}g;
    return $text;
  }
sub _xml_escape_comment
  { my( $text)= @_;
    $text=~ s{([&<>])}{$CHAR2DEFAULT_ENT{$1}}g;
    $text=~ s{--}{-&#45;}g; # can't have double --'s in XML comments
    return $text;
  }
sub _xml_escape_cdata
  { my( $text, $opt)= @_;
    if( $opt->{force_escape_cdata} || $text=~ m{[<&]})
      { $text=~ s{^\s*\Q<![CDATA[}{}s;
        $text=~ s{\Q]]>\E\s*$}{}s;
        $text=~ s{]]>}{]]&#62;}g; # can't have]]> in CDATA
        $text=  "<![CDATA[$text]]>";
      }
    return $text;
  }
package HTML::TreeBuilder::XPath::TextNode;
use base 'HTML::TreeBuilder::XPath::Node';
sub getParentNode { return shift->{_parent};    }
sub getValue      { return shift->{_content};   }
sub isTextNode    { return 1;                   }
sub getAttributes { return wantarray ? () : []; }
# similar to HTML::Element as_XML
sub as_XML
  { my( $node, $entities)= @_;
    my $content= $node->{_content};
    if( $node->{_parent} && $node->{_parent}->{_tag} eq 'script')
      { $content=~ s{(&\w+;)}{HTML::Entities::decode($1)}eg; }
    else
      { $content= HTML::Element::_xml_escape_text($content); }
    return $content;
  }
*as_XML_compact  = *as_XML;
*as_XML_indented = *as_XML;
sub getPreviousSibling
  { my $self= shift;
    my $rank= $self->{_rank}; 
    #unless( defined $self->{_rank})
    #  { warn "no rank for text node $self->{_content}, parent is $self->{_parent}->{_tag}\n"; }
    my $parent= $self->{_parent};
    return $rank ? $parent->_child_as_object( $parent->{_content}->[$rank-1], $rank-1) : undef;
  }
sub getNextSibling
  { my $self= shift;
    my $rank= $self->{_rank};
    #unless( defined $self->{_rank})
    #  { warn "no rank for text node $self->{_content}, parent is $self->{_parent}->{_tag}\n"; }
    my $parent= $self->{_parent};
    my $next_sibling= $parent->{_content}->[$rank+1];
    return defined( $next_sibling) ? $parent->_child_as_object( $next_sibling, $rank+1) : undef;
  }
sub getRootNode
  { return shift->{_parent}->getRootNode; }
sub string_value { return shift->{_content}; }
# added to provide element-like methods to text nodes, for use by cmp
sub lineage 
  { my( $node)= @_;
    my $parent= $node->{_parent};
    return( $parent, $parent->lineage);
  }
sub is_inside
  { my( $text, $node)= @_;
    return $text->{_parent}->is_inside( $node);
  }
1;
package HTML::TreeBuilder::XPath::Attribute;
use base 'HTML::TreeBuilder::XPath::Node';
sub getParentNode   { return $_[0]->{_elt}; }
sub getValue        { return $_[0]->{_value}; }
sub getName         { return $_[0]->{_name} ; }
sub getLocalName    { (my $name= $_[0]->{_name}) =~ s{^.*:}{}; $name; }
sub string_value    { return $_[0]->{_value}; }
sub to_number       { return XML::XPathEngine::Number->new( $_[0]->{_value}); }
sub isAttributeNode { 1 }
sub toString        { return qq{ $_[0]->{_name}="$_[0]->{_value}"}; }
sub getAttributes   {  return $_[0]->{_elt}->getAttributes; }
# awfully inefficient, but hopefully this is called only for weird (read test-case) queries
sub getPreviousSibling
  { my $self= shift;
    my $rank= $self->{_rank};
    return undef unless $rank;
    my %atts= $self->{_elt}->all_external_attr;
    my $previous_att_name= (sort keys %atts)[$rank-1]; 
    return bless( { _name => $previous_att_name, 
                             _value => $atts{$previous_att_name}, 
                             _elt => $self->{_elt}, _rank => $rank-1, 
                   }, 'HTML::TreeBuilder::XPath::Attribute'
                );
  }
sub getNextSibling
  { my $self= shift;
    my $rank= $self->{_rank};
    my %atts= $self->{_elt}->all_external_attr;
    my $next_att_name= (sort keys %atts)[$rank+1] || return undef; 
    return bless( { _name => $next_att_name, _value => $atts{$next_att_name}, 
                             _elt => $self->{_elt}, _rank => $rank+1, 
                   }, 'HTML::TreeBuilder::XPath::Attribute'
                );
     
  }
# added to provide element-like methods to attributes, for use by cmp
sub lineage 
  { my( $att)= @_;
    my $elt= $att->{_elt};
    return( $elt, $elt->lineage);
  }
sub is_inside
  { my( $att, $node)= @_;
    return ($att->{_elt} == $node) || $att->{_elt}->is_inside( $node);
  }
1;
package HTML::TreeBuilder::XPath::Root;
use base 'HTML::TreeBuilder::XPath::Node';
     
sub getParentNode   { return (); }
sub getChildNodes   { my @content= ( $_[0]->{_root}); return wantarray ? @content : \@content; }
sub getAttributes   { return []        }
sub isDocumentNode  { return 1         }
sub getRootNode     { return $_[0]     }
sub getName         { return           }
sub getNextSibling  { return           }
sub getPreviousSibling { return        }
# added to provide element-like methods to root, for use by cmp
sub lineage {  return ($_[0]); }
sub is_inside { return 0; }
sub cmp { return $_[1]->isa( ' HTML::TreeBuilder::XPath::Root') ? 0 : 1; }
1;
__END__
HideShow 113 lines of Pod
=head1 NAME
HTML::TreeBuilder::XPath - add XPath support to HTML::TreeBuilder
=head1 SYNOPSIS
  use HTML::TreeBuilder::XPath;
  my $tree= HTML::TreeBuilder::XPath->new;
  $tree->parse_file( "mypage.html");
  my $nb=$tree->findvalue( '/html/body//p[@class="section_title"]/span[@class="nb"]');
  my $id=$tree->findvalue( '/html/body//p[@class="section_title"]/@id');
  my $p= $html->findnodes( '//p[@id="toto"]')->[0];
  my $link_texts= $p->findvalue( './a'); # the texts of all a elements in $p
  $tree->delete; # to avoid memory leaks, if you parse many HTML documents 
   
   
=head1 DESCRIPTION
This module adds typical XPath methods to HTML::TreeBuilder, to make it
easy to query a document.
=head1 METHODS
Extra methods added both to the tree object and to each element:
=head2 findnodes ($path)
Returns a list of nodes found by C<$path>.
In scalar context returns an C<Tree::XPathEngine::NodeSet> object.
=head2 findnodes_as_string ($path)
Returns the text values of the nodes, as one string.
=head2 findnodes_as_strings ($path)
Returns a list of the values of the result nodes. 
=head2 findvalue ($path)
Returns either a C<Tree::XPathEngine::Literal>, a C<Tree::XPathEngine::Boolean>
or a C<Tree::XPathEngine::Number> object. If the path returns a NodeSet,
$nodeset->xpath_to_literal is called automatically for you (and thus a
C<Tree::XPathEngine::Literal> is returned). Note that
for each of the objects stringification is overloaded, so you can just
print the value found, or manipulate it in the ways you would a normal
perl value (e.g. using regular expressions).
=head2 findvalues ($path)
Returns the values of the matching nodes as a list. This is mostly the same
as findnodes_as_strings, except that the elements of the list are objects
(with overloaded stringification) instead of plain strings.
=head2 exists ($path)
Returns true if the given path exists.
=head2 matches($path)
Returns true if the element matches the path.
=head2 find ($path)
The find function takes an XPath expression (a string) and returns either a
Tree::XPathEngine::NodeSet object containing the nodes it found (or empty if
no nodes matched the path), or one of XML::XPathEngine::Literal (a string),
XML::XPathEngine::Number, or XML::XPathEngine::Boolean. It should always
return something - and you can use ->isa() to find out what it returned. If
you need to check how many nodes it found you should check $nodeset->size.
See L<XML::XPathEngine::NodeSet>.
=head2 as_XML_compact
HTML::TreeBuilder's C<as_XML> output is not really nice to look at, so
I added a new method, that can be used as a simple replacement for it. 
It escapes only the '<', '>' and '&' (plus '"' in attribute values), and
wraps CDATA elements in CDATA sections.
Note that the XML is actually not garanteed to be valid at this point. Nothing
is done about the encoding of the string. Patches or just ideas of how it could
work are welcome.
=head2 as_XML_indented 
Same as as_XML, except that the output is indented.
=head1 SEE ALSO
L<HTML::TreeBuilder>
L<XML::XPathEngine>
=head1 REPOSITORY
L<https://github.com/mirod/HTML--TreeBuilder--XPath>
=head1 AUTHOR
Michel Rodriguez, E<lt>mirod@cpan.orgE<gt>
=head1 COPYRIGHT AND LICENSE
Copyright (C) 2006-2011 by Michel Rodriguez
This library is free software; you can redistribute it and/or modify
it under the same terms as Perl itself, either Perl version 5.8.4 or,
at your option, any later version of Perl 5 you may have available.
=cut