biblicit 2.0.3 → 2.0.4

data/parscit/lib/Omni/Omniline.pm

@@ -1,423 +0,0 @@
-package Omni::Omniline;
-
-# Configuration
-use strict;
-
-# Local libraries
-use Omni::Config;
-use Omni::Omniword;
-use Omni::Omnirun;
-
-# Extern libraries
-use XML::Twig;
-use XML::Parser;
-use XML::Writer;
-use XML::Writer::String;
-
-# Global variables
-my $tag_list = $Omni::Config::tag_list;
-my $att_list = $Omni::Config::att_list;
-my $obj_list = $Omni::Config::obj_list;
-
-# Temporary variables
-my $tmp_content 	= undef;
-my $tmp_baseline	= undef;
-my $tmp_bottom		= undef;
-my $tmp_top			= undef;
-my $tmp_left		= undef;
-my $tmp_right		= undef;
-my @tmp_objs		= ();
-
-###
-# A line object in Omnipage xml: a line can contain one or many runs
-#
-# Do Hoang Nhat Huy, 09 Jan 2011
-###
-# Initialization
-sub new
-{
-	my ($class) = @_;
-
-	# A line can have multiple runs or words
-	my @objs	= ();
-
-	# Class members
-	my $self = {	'_self'			=> $obj_list->{ 'OMNILINE' },
-					'_raw'			=> undef,
-					'_content'		=> undef,
-					'_baseline'		=> undef,
-					'_bottom'		=> undef,
-					'_top'			=> undef,
-					'_left'			=> undef,
-					'_right'		=> undef,
-					'_bullet'		=> undef,
-					'_objs'			=> \@objs	};
-
-	bless $self, $class;
-	return $self;
-}
-
-# 
-sub set_raw
-{
-	my ($self, $raw) = @_;
-	
-	# Save the raw xml <ln> ... </ln>
-	$self->{ '_raw' }	= $raw;
-
-	# Parse the raw string
-	my $twig_roots		= { $tag_list->{ 'LINE' }	=> 1 };
-	my $twig_handlers 	= { $tag_list->{ 'LINE' }	=> \&parse};
-
-	# XML::Twig 
-	my $twig = new XML::Twig(	twig_roots 		=> $twig_roots,
-						 	 	twig_handlers	=> $twig_handlers,
-						 	 	pretty_print 	=> 'indented'	);
-
-	# Start the XML parsing
-	$twig->parse($raw);
-	$twig->purge;
-
-	# Copy information from temporary variables to class members
-	$self->{ '_baseline' }	= $tmp_baseline;
-	$self->{ '_bottom' }	= $tmp_bottom;
-	$self->{ '_top' }		= $tmp_top;
-	$self->{ '_left' }		= $tmp_left;
-	$self->{ '_right' } 	= $tmp_right;
-	
-	# Copy all objects
-	@{ $self->{ '_objs' } }	= @tmp_objs;
-
-	# Copy content
-	$self->{ '_content' } 	= $tmp_content;
-}
-
-sub get_raw
-{
-	my ($self) = @_;
-	return $self->{ '_raw' };
-}
-
-sub parse
-{
-	my ($twig, $node) = @_;
-
-	# At first, content is blank
-	$tmp_content 	= "";
-	# because there's no run
-	@tmp_objs		= ();
-
-	# Get <line> node attributes
-	$tmp_bottom		= GetNodeAttr($node, $att_list->{ 'BOTTOM' });
-	$tmp_top		= GetNodeAttr($node, $att_list->{ 'TOP' });
-	$tmp_left		= GetNodeAttr($node, $att_list->{ 'LEFT' });
-	$tmp_right		= GetNodeAttr($node, $att_list->{ 'RIGHT' });
-	$tmp_baseline	= GetNodeAttr($node, $att_list->{ 'BASELINE' });
-	
-	# Get <line> node possible attributes
-	my $tmp_font_face	= GetNodeAttr($node, $att_list->{ 'FONTFACE' });
-	my $tmp_font_family	= GetNodeAttr($node, $att_list->{ 'FONTFAMILY' });
-	my $tmp_font_pitch	= GetNodeAttr($node, $att_list->{ 'FONTPITCH' });
-	my $tmp_font_size	= GetNodeAttr($node, $att_list->{ 'FONTSIZE' });
-	my $tmp_spacing		= GetNodeAttr($node, $att_list->{ 'SPACING' });
-	my $tmp_su_script	= GetNodeAttr($node, $att_list->{ 'SUSCRIPT' });	# sub-script or super-script
-	my $tmp_underline	= GetNodeAttr($node, $att_list->{ 'UNDERLINE' });
-	my $tmp_bold		= GetNodeAttr($node, $att_list->{ 'BOLD' });
-	my $tmp_italic		= GetNodeAttr($node, $att_list->{ 'ITALIC' });
-
-	# Check if there's any run
-	my @all_runs = $node->descendants( $tag_list->{ 'RUN' });
-	# There is not
-	if (scalar(@all_runs) == 0)
-	{
-		my $output = XML::Writer::String->new();
-		my $writer = new XML::Writer(OUTPUT => $output, UNSAFE => 'true');
-
-		# Form the fake <run>
-		$writer->startTag(	"run", 
-							$att_list->{ 'FONTFACE' } 	=> $tmp_font_face,
-							$att_list->{ 'FONTFAMILY' }	=> $tmp_font_family,
-							$att_list->{ 'FONTPITCH' } 	=> $tmp_font_pitch,
-							$att_list->{ 'FONTSIZE' } 	=> $tmp_font_size,
-							$att_list->{ 'SPACING' } 	=> $tmp_spacing,
-							$att_list->{ 'SUSCRIPT' } 	=> $tmp_su_script,
-							$att_list->{ 'UNDERLINE' } 	=> $tmp_underline,
-							$att_list->{ 'BOLD' }		=> $tmp_bold,
-							$att_list->{ 'ITALIC' }		=> $tmp_italic	);
-
-		# Get the inner line content
-		$writer->raw( $node->xml_string() );
-		$writer->endTag("run");
-		$writer->end();
-
-		# Fake run
-		my $run = new Omni::Omnirun();
-		
-		# Set raw content
-		$run->set_raw($output->value());
-
-		# Update run list
-		push @tmp_objs, $run;
-
-		# Update content
-		$tmp_content = $tmp_content . $run->get_content();
-	}
-	else
-	{
-		# Get the first child in the line
-		my $child = $node->first_child();
-
-		# Some type of child
-		my $space_tag	= $tag_list->{ 'SPACE' };
-		my $tab_tag		= $tag_list->{ 'TAB' };
-		my $newline_tag	= $tag_list->{ 'NEWLINE' };
-		my $word_tag	= $tag_list->{ 'WORD' };
-		my $run_tag		= $tag_list->{ 'RUN' };
-
-		# A damn line can contain both <run> and <wd>
-		while (defined $child)
-		{
-			my $xpath = $child->path();
-
-			# if this child is <run>
-			if ($xpath =~ m/\/$run_tag$/)
-			{
-				my $run = new Omni::Omnirun();
-
-				# Set raw content
-				$run->set_raw($child->sprint());
-				
-				# Update object list
-				push @tmp_objs, $run;
-
-				# Update content
-				$tmp_content = $tmp_content . $run->get_content();
-			}
-			# if this child is <wd>
-			elsif ($xpath =~ m/\/$word_tag$/)
-			{
-				# One word can contain many <run>
-				my $grand_child = $child->first_child( $tag_list->{ 'RUN' } );
-
-				# The first <run> is a special child
-				if (defined $grand_child)
-				{
-					while (defined $grand_child)
-					{
-						# NOTE: The following code is controvery. Consider the case when a word
-						# contain two runs, which means that this word has two parts in two different
-						# format.
-						#
-						# If I want to keep all these format info, I need to consider that this word
-						# is actually two words with no space between them
-						#
-						# But for compatible with Thang's code, there must be only one word here and
-						# subsequently only one run. So I only keep the first run
-
-						# NOTE: The first <run> in <wd> will has the word position information
-						my $output = XML::Writer::String->new();
-						my $writer = new XML::Writer(OUTPUT => $output, UNSAFE => 'true');					
-				
-						# Form the fake <run>
-						$writer->startTag(	"run", 
-											$att_list->{ 'FONTFACE' } 	=> GetNodeAttr($grand_child, $att_list->{ 'FONTFACE' }),
-											$att_list->{ 'FONTFAMILY' }	=> GetNodeAttr($grand_child, $att_list->{ 'FONTFAMILY' }),
-											$att_list->{ 'FONTPITCH' } 	=> GetNodeAttr($grand_child, $att_list->{ 'FONTPITCH' }),
-											$att_list->{ 'FONTSIZE' } 	=> GetNodeAttr($grand_child, $att_list->{ 'FONTSIZE' }),
-											$att_list->{ 'SPACING' } 	=> GetNodeAttr($grand_child, $att_list->{ 'SPACING' }),
-											$att_list->{ 'SUSCRIPT' } 	=> GetNodeAttr($grand_child, $att_list->{ 'SUSCRIPT' }),
-											$att_list->{ 'UNDERLINE' } 	=> GetNodeAttr($grand_child, $att_list->{ 'UNDERLINE' }),
-											$att_list->{ 'BOLD' }		=> GetNodeAttr($grand_child, $att_list->{ 'BOLD' }),
-											$att_list->{ 'ITALIC' }		=> GetNodeAttr($grand_child, $att_list->{ 'ITALIC' })	);
-						# Form the fake <wd>
-						$writer->startTag(	"wd", 
-											$att_list->{ 'BOTTOM' } 	=> GetNodeAttr($child, $att_list->{ 'BOTTOM' }),
-											$att_list->{ 'TOP' }		=> GetNodeAttr($child, $att_list->{ 'TOP' }),
-											$att_list->{ 'LEFT' } 		=> GetNodeAttr($child, $att_list->{ 'LEFT' }),
-											$att_list->{ 'RIGHT' } 		=> GetNodeAttr($child, $att_list->{ 'RIGHT' })	);
-		
-						$writer->raw( $grand_child->xml_string() );
-						$writer->endTag("wd");
-						$writer->endTag("run");
-						$writer->end();
-
-						# Fake run 
-						my $run = new Omni::Omnirun();
-	
-						# Set raw content
-						$run->set_raw($output->value());
-
-						# Update run list
-						push @tmp_objs, $run;
-
-						# Update content
-						$tmp_content = $tmp_content . $run->get_content();
-
-						# Little brother
-						if ($grand_child->is_last_child) 
-						{ 
-							last; 
-						}
-						else
-						{
-							$grand_child = $grand_child->next_sibling( $tag_list->{ 'RUN' } );
-						}
-					}	
-				}
-				# Special case: <wd> contains no <run> but stores the format itself
-				else
-				{
-					my $output = XML::Writer::String->new();
-					my $writer = new XML::Writer(OUTPUT => $output, UNSAFE => 'true');
-
-					# Form the fake <run>
-					$writer->startTag(	"run", 
-										$att_list->{ 'FONTFACE' } 	=> GetNodeAttr($child, $att_list->{ 'FONTFACE' }),
-										$att_list->{ 'FONTFAMILY' }	=> GetNodeAttr($child, $att_list->{ 'FONTFAMILY' }),
-										$att_list->{ 'FONTPITCH' } 	=> GetNodeAttr($child, $att_list->{ 'FONTPITCH' }),
-										$att_list->{ 'FONTSIZE' } 	=> GetNodeAttr($child, $att_list->{ 'FONTSIZE' }),
-										$att_list->{ 'SPACING' } 	=> GetNodeAttr($child, $att_list->{ 'SPACING' }),
-										$att_list->{ 'SUSCRIPT' } 	=> GetNodeAttr($child, $att_list->{ 'SUSCRIPT' }),
-										$att_list->{ 'UNDERLINE' } 	=> GetNodeAttr($child, $att_list->{ 'UNDERLINE' }),
-										$att_list->{ 'BOLD' }		=> GetNodeAttr($child, $att_list->{ 'BOLD' }),
-										$att_list->{ 'ITALIC' }		=> GetNodeAttr($child, $att_list->{ 'ITALIC' })	);
-					# Form the fake <wd>
-					$writer->startTag(	"wd",
-										$att_list->{ 'BOTTOM' } 	=> GetNodeAttr($child, $att_list->{ 'BOTTOM' }),
-										$att_list->{ 'TOP' }		=> GetNodeAttr($child, $att_list->{ 'TOP' }),
-										$att_list->{ 'LEFT' } 		=> GetNodeAttr($child, $att_list->{ 'LEFT' }),
-										$att_list->{ 'RIGHT' } 		=> GetNodeAttr($child, $att_list->{ 'RIGHT' })	);
-					# Get the inner <wd> content
-					$writer->raw( $child->xml_string() );
-					$writer->endTag("wd");
-					$writer->endTag("run");
-					$writer->end();
-
-					# Fake run
-					my $run = new Omni::Omnirun();
-		
-					# Set raw content
-					$run->set_raw($output->value());
-					
-					# Update run list
-					push @tmp_objs, $run;
-
-					# Update content
-					$tmp_content = $tmp_content . $run->get_content();
-				}
-			}
-			elsif  ($xpath =~ m/\/$space_tag$/)
-			{
-				# Update content
-				$tmp_content = $tmp_content . " ";
-			}
-			elsif  ($xpath =~ m/\/$tab_tag$/)
-			{
-				# Update content
-				$tmp_content = $tmp_content . "\t";
-			}
-			elsif  ($xpath =~ m/\/$newline_tag$/)
-			{
-				# Update content
-				$tmp_content = $tmp_content . "\n";
-			}
-
-			# Little brother
-			if ($child->is_last_child) 
-			{ 
-				last; 
-			}
-			else
-			{
-				$child = $child->next_sibling();
-			}
-		}
-	}
-}
-
-sub get_bullet
-{
-	my ($self) = @_;
-	return $self->{ '_bullet' };
-}
-
-sub set_bullet
-{
-	my ($self, $bullet) = @_;
-	$self->{ '_bullet' } = $bullet;		
-}
-
-sub get_name
-{
-	my ($self) = @_;
-	return $self->{ '_self' };
-}
-
-sub get_objs_ref
-{
-	my ($self) = @_;
-	return $self->{ '_objs' };
-}
-
-sub get_content
-{
-	my ($self) = @_;
-	return $self->{ '_content' };
-}
-
-sub get_baseline
-{
-	my ($self) = @_;
-	return $self->{ '_baseline' };
-}
-
-sub get_bottom_pos
-{
-	my ($self) = @_;
-	return $self->{ '_bottom' };
-}
-
-sub get_top_pos
-{
-	my ($self) = @_;
-	return $self->{ '_top' };
-}
-
-sub get_left_pos
-{
-	my ($self) = @_;
-	return $self->{ '_left' };
-}
-
-sub get_right_pos
-{
-	my ($self) = @_;
-	return $self->{ '_right' };
-}
-
-# Support functions
-sub GetNodeAttr 
-{
-	my ($node, $attr) = @_;
-	return ($node->att($attr) ? $node->att($attr) : "");
-}
-
-sub SetNodeAttr 
-{
-	my ($node, $attr, $value) = @_;
-	$node->set_att($attr, $value);
-}
-
-sub GetNodeText
-{
-	my ($node) = @_;
-	return $node->text;
-}
-
-sub SetNodeText
-{
-	my ($node, $value) = @_;
-	$node->set_text($value);
-}
-
-1;

data/parscit/lib/Omni/Omnipage.pm

@@ -1,282 +0,0 @@
-package Omni::Omnipage;
-
-# Configuration
-use strict;
-
-# Local libraries
-use Omni::Config;
-use Omni::Omnidd;
-use Omni::Omnicol;
-use Omni::Omniframe;
-
-# Extern libraries
-use XML::Twig;
-use XML::Parser;
-
-# Global variables
-my $tag_list = $Omni::Config::tag_list;
-my $att_list = $Omni::Config::att_list;
-my $obj_list = $Omni::Config::obj_list;
-
-# Temporary variables
-my $tmp_content 	= undef;
-my @tmp_objs		= ();
-
-###
-# A page object in Omnipage xml: a page contains zero or many collums
-#
-# Do Hoang Nhat Huy, 09 Jan 2011
-###
-# Initialization
-sub new
-{
-	my ($class) = @_;
-
-	# Page: a page can have many columns, many tables, or many images
-	my @objs	= ();
-
-	# Class members
-	my $self = {	'_self'			=> $obj_list->{ 'OMNIPAGE' },
-					'_raw'			=> undef,
-					'_content'		=> undef,
-					'_objs'			=> \@objs	};
-
-	bless $self, $class;
-	return $self;
-}
-
-# 
-sub set_raw
-{
-	my ($self, $raw) = @_;
-
-	# Save the raw xml <page> ... </page>
-	$self->{ '_raw' }	= $raw;
-
-	# Parse the raw string
-	my $twig_roots		= { $tag_list->{ 'PAGE' }	=> 1 };
-	my $twig_handlers 	= { $tag_list->{ 'PAGE' }	=> \&parse};
-
-	# XML::Twig 
-	my $twig = new XML::Twig(	twig_roots 		=> $twig_roots,
-						 	 	twig_handlers	=> $twig_handlers,
-						 	 	pretty_print 	=> 'indented'	);
-
-	# Start the XML parsing
-	$twig->parse($raw);
-	$twig->purge;
-
-	# Copy information from temporary variables to class members
-
-	# Copy all columns 
-	@{$self->{ '_objs' } }	= @tmp_objs;
-	
-	# Copy content
-	$self->{ '_content' }	= $tmp_content;
-}
-
-sub get_raw
-{
-	my ($self) = @_;
-	return $self->{ '_raw' };
-}
-
-sub parse
-{
-	my ($twig, $node) = @_;
-
-	# At first, content is blank
-	$tmp_content	= "";
-	# because there's no columnm, table or image
-	@tmp_objs		= ();
-
-	# Get <page> node attributes
-	# At version 16, Omnipage page does not have any interesting atribute
-
-	my $child = undef;
-	# Get the body text
-	$child = $node->first_child( $tag_list->{ 'BODY' } );
-	# Page with no body, return
-	if (! defined $child) { return; }
-
-	# Get the first child in the body text
-	$child = $child->first_child();
-
-	# The child of <page> is usually <section> but it's not always the case
-	my $section_tag	= $tag_list->{ 'SECTION' };
-
-	# <dd>, <col> are usually not the children but the
-	# desendents of <page> but I'm not sure about this
-	my $dd_tag		= $tag_list->{ 'DD' };
-	my $column_tag	= $tag_list->{ 'COL' };
-	my $frame_tag	= $tag_list->{ 'FRAME' };
-
-	# Check if there's any column or dd, what the heck is dd 
-	while (defined $child)
-	{
-		my $xpath = $child->path();
-
-		# if this child is <section>, then <column> and <dd> tag are grandchild of <page>
-		if ($xpath =~ m/\/$section_tag$/)
-		{
-			# Get the first grand child
-			my $grand_child = $child->first_child();
-
-			# Subloop
-			while (defined $grand_child)
-			{
-				my $grand_xpath = $grand_child->path();
-
-				# if this child is <column>
-				if ($grand_xpath =~ m/\/$column_tag$/)
-				{
-					my $column = new Omni::Omnicol();
-	
-					# Set raw content
-					$column->set_raw($grand_child->sprint());
-
-					# Update column list
-					push @tmp_objs, $column;
-
-					# Update content
-					$tmp_content = $tmp_content . $column->get_content() . "\n";
-				}
-				# if this child is <dd>
-				elsif ($grand_xpath =~ m/\/$dd_tag$/)
-				{
-					my $dd = new Omni::Omnidd();
-
-					# Set raw content
-					$dd->set_raw($child->sprint());
-
-					# Update column list
-					push @tmp_objs, $dd;
-
-					# Update content
-					$tmp_content = $tmp_content . $dd->get_content() . "\n";
-				}
-				# if this child is <frame>
-				elsif ($xpath =~ m/\/$frame_tag$/)
-				{
-					my $frame = new Omni::Omniframe();
-
-					# Set raw content
-					$frame->set_raw($child->sprint());
-
-					# Update column list
-					push @tmp_objs, $frame;
-
-					# Update content
-					$tmp_content = $tmp_content . $frame->get_content() . "\n";
-				}
-
-				# Little brother
-				if ($grand_child->is_last_child) 
-				{ 
-					last; 
-				}
-				else
-				{
-					$grand_child = $grand_child->next_sibling();
-				}
-			}
-		}
-		# if this child is <column>
-		elsif ($xpath =~ m/\/$column_tag$/)
-		{
-			my $column = new Omni::Omnicol();
-
-			# Set raw content
-			$column->set_raw($child->sprint());
-
-			# Update column list
-			push @tmp_objs, $column;
-
-			# Update content
-			$tmp_content = $tmp_content . $column->get_content() . "\n";
-		}
-		# if this child is <dd>
-		elsif ($xpath =~ m/\/$dd_tag$/)
-		{
-			my $dd = new Omni::Omnidd();
-
-			# Set raw content
-			$dd->set_raw($child->sprint());
-
-			# Update column list
-			push @tmp_objs, $dd;
-
-			# Update content
-			$tmp_content = $tmp_content . $dd->get_content() . "\n";
-		}
-		# if this child is <frame>
-		elsif ($xpath =~ m/\/$frame_tag$/)
-		{
-			my $frame = new Omni::Omniframe();
-
-			# Set raw content
-			$frame->set_raw($child->sprint());
-
-			# Update column list
-			push @tmp_objs, $frame;
-
-			# Update content
-			$tmp_content = $tmp_content . $frame->get_content() . "\n";
-		}
-
-		# Little brother
-		if ($child->is_last_child) 
-		{ 
-			last; 
-		}
-		else
-		{
-			$child = $child->next_sibling();
-		}
-	}
-}
-
-sub get_name
-{
-	my ($self) = @_;
-	return $self->{ '_self' };
-}
-
-sub get_objs_ref
-{
-	my ($self) = @_;
-	return $self->{ '_objs' };
-}
-
-sub get_content
-{
-	my ($self) = @_;
-	return $self->{ '_content' };
-}
-
-# Support functions
-sub GetNodeAttr 
-{
-	my ($node, $attr) = @_;
-	return ($node->att($attr) ? $node->att($attr) : "");
-}
-
-sub SetNodeAttr 
-{
-	my ($node, $attr, $value) = @_;
-	$node->set_att($attr, $value);
-}
-
-sub GetNodeText
-{
-	my ($node) = @_;
-	return $node->text;
-}
-
-sub SetNodeText
-{
-	my ($node, $value) = @_;
-	$node->set_text($value);
-}
-
-1;