source: gsdl/trunk/perllib/plugins/PagedImagePlugin.pm@ 15905

###########################################################################
#
# PagedImagePlugin.pm -- plugin for sets of images and OCR text that
#  make up a document
# A component of the Greenstone digital library software
# from the New Zealand Digital Library Project at the 
# University of Waikato, New Zealand.
#
# Copyright (C) 1999 New Zealand Digital Library Project
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
#
###########################################################################

# PagedImagePlugin
# processes sequences of images, with optional OCR text
#
# This plugin takes *.item files, which contain metadata and lists of image 
# files, and produces a document containing sections, one for each page.
# The files should be named something.item, then you can have more than one 
# book in a directory. You will need to create these files, one for each
# document/book.
#
#There are two formats for the item files: a plain text format, and an xml 
#format. You can use either format, and can have both formats in the same 
#collection if you like. If you use the plain format, you must not start the 
#file off with <PagedDocument>

#### PLAIN FORMAT
# The format of the xxx.item file is as follows:
# The first lines contain any metadata for the whole document
# <metadata-name>metadata-value
# eg.
# <Title>Snail farming
# <Date>19230102
# Then comes a list of pages, one page per line, each line has the format
#
# pagenum:imagefile:textfile:r
#
# page num and imagefile are required. pagenum is used for the Title 
# of the section, and in the display is shown as page <pagenum>. 
# imagefile is the image for the page. textfile is an optional text 
# file containing the OCR (or any) text for the page - this gets added 
# as the text for the section. r is optional, and signals that the image 
# should be rotated 180deg. Eg use this if the image has been made upside down.
# So an example item file looks like:
# <Title>Snail farming
# <Date>19960403
# 1:p1.gif:p1.txt:
# 2:p2.gif::
# 3:p3.gif:p3.txt:
# 3b:p3b.gif:p3b.txt:r
# The second page has no text, the fourth page is a back page, and 
# should be rotated.
# 

#### XML FORMAT
# The xml format looks like the following
#<PagedDocument>
#<Metadata name="Title">The Title of the entire document</Metadata>
#<Page pagenum="1" imgfile="xxx.jpg" txtfile="yyy.txt">
#<Metadata name="Title">The Title of this page</Metadata>
#</Page>
#... more pages
#</PagedDocument>
#PagedDocument contains a list of Pages, Metadata and PageGroups. Any metadata 
#that is not inside another tag will belong to the document.
#Each Page has a pagenum (not used at the moment), an imgfile and/or a txtfile.
#These are both optional - if neither is used, the section will have no content.
#Pages can also have metadata associated with them.
#PageGroups can be introduced at any point - they can contain Metadata and Pages and other PageGroups. They are used to introduce hierarchical structure into the document.
#For example
#<PagedDocument>
#<PageGroup>
#<Page>
#<Page>
#</PageGroup>
#<Page>
#</PagedDocument>
#would generate a structure like
#X
#--X
#  --X
#  --X
#--X
#PageGroup tags can also have imgfile/textfile metadata if you like - this way they get some content themselves.

#Currently the XML structure doesn't work very well with the paged document type, unless you use numerical Titles for each section.
#There is still a bit of work to do on this format:
#* enable other text file types, eg html, pdf etc
#* make the document paging work properly
#* add pagenum as Title unless a Title is present?

# All the supplemetary image amd text files should be in the same folder as 
# the .item file.
#
# To display the images instead of the document text, you can use [srcicon] 
# in the DocumentText format statement.
# For example, 
#
# format DocumentText "<center><table width=_pagewidth_><tr><td>[srcicon]</td></tr></table></center>"
#
# To have it create thumbnail size images, use the '-create_thumbnail' option.
# To have it create medium size images for display, use the '-create_screenview'
# option. As usual, running 
# 'perl -S pluginfo.pl PagedImagePlugin' will list all the options.

# If you want the resulting documents to be presented with a table of 
# contents, use '-documenttype hierarchy', otherwise they will have 
# next and previous arrows, and a goto page X box. 

# If you have used -create_screenview, you can also use [screenicon] in the format
# statement to display the smaller image.  Here is an example  that switches 
# between the two: 
#
# format DocumentText "<center><table width=_pagewidth_><tr><td>{If}{_cgiargp_ eq full,<a href='_httpdocument_&d=_cgiargd_&p=small'>Switch to small version.</a>,<a href='_httpdocument_&d=_cgiargd_&p=full'>Switch to fullsize version</a>}</td></tr><tr><td>{If}{_cgiargp_ eq full,<a href='_httpdocument_&d=_cgiargd_&p=small' title='Switch to small version'>[srcicon]</a>,<a href='_httpdocument_&d=_cgiargd_&p=full' title='Switch to fullsize version'>[screenicon]</a>}</td></tr></table></center>"
#
# Additional metadata can be added into the .item files, alternatively you can 
# use normal metadata.xml files, with the name of the xxx.item file as the 
# FileName (only for document level metadata).

package PagedImagePlugin;

use ReadXMLFile;
use ReadTextFile;
use ImageConverter;

use strict;
no strict 'refs'; # allow filehandles to be variables and viceversa

sub BEGIN {
    @PagedImagePlugin::ISA = ('ReadXMLFile', 'ReadTextFile', 'ImageConverter');
}

my $type_list =
    [ { 'name' => "paged",
        'desc' => "{PagedImagePlugin.documenttype.paged}" },
      { 'name' => "hierarchy",
        'desc' => "{PagedImagePlugin.documenttype.hierarchy}" } ];

my $arguments =
    [ { 'name' => "process_exp",
    'desc' => "{BasPlug.process_exp}",
    'type' => "string",
    'deft' => &get_default_process_exp(),
    'reqd' => "no" },
      { 'name' => "block_exp",
    'desc' => "{BasPlug.block_exp}",
    'type' => "string",
    'deft' => &get_default_block_exp(),
    'reqd' => "no" },
      { 'name' => "title_sub",
    'desc' => "{HTMLPlug.title_sub}",
    'type' => "string", 
    'deft' => "" },
      { 'name' => "headerpage",
    'desc' => "{PagedImagePlugin.headerpage}",
    'type' => "flag",
    'reqd' => "no" },
      { 'name' => "documenttype",
    'desc' => "{PagedImagePlugin.documenttype}",
    'type' => "enum",
    'list' => $type_list,
    'deft' => "paged",
    'reqd' => "no" } ];


my $options = { 'name'     => "PagedImagePlugin",
        'desc'     => "{PagedImagePlugin.desc}",
        'abstract' => "no",
        'inherits' => "yes",
        'args'     => $arguments };

sub new {
    my ($class) = shift (@_);
    my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
    push(@$pluginlist, $class);

    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
    push(@{$hashArgOptLists->{"OptList"}},$options);
    
    new ImageConverter($pluginlist, $inputargs, $hashArgOptLists);
    new ReadTextFile($pluginlist, $inputargs, $hashArgOptLists, 1);
    my $self = new ReadXMLFile($pluginlist, $inputargs, $hashArgOptLists);
    
    return bless $self, $class;
}


sub init {
    my $self = shift (@_);
    my ($verbosity, $outhandle, $failhandle) = @_;

    $self->SUPER::init(@_);
    $self->ImageConverter::init();
}

sub get_default_process_exp {
    my $self = shift (@_);

    return q^\.item$^;
}

sub get_doctype {
    my $self = shift(@_);
    
    return "PagedDocument";
}


# want to block everything except the .item ones 
# but instead we will block images and txt files
sub get_default_block_exp {
    my $self = shift (@_);
    
    return q^(?i)(\.jpe?g|\.gif|\.png|\.tif?f|\.te?xt|\.html?|~)$^
}

sub rotate_image {
    my $self = shift (@_);
    my ($filename_full_path) = @_;
    
    my ($this_filetype) = $filename_full_path =~ /\.([^\.]*)$/;
    my $result = $self->convert($filename_full_path, $this_filetype, "-rotate 180", "ROTATE");
    my ($new_filename) = ($result =~ /=>(.*\.$this_filetype)/);
    if (-e "$new_filename") {
    return $new_filename;
    }
    # somethings gone wrong
    return $filename_full_path;

}

sub process_image {
    my $self = shift(@_);
    my ($filename_full_path, $filename_no_path, $doc_obj, $section, $rotate) = @_;
    # do rotation
    if ($rotate eq "r") {
    # check the filenames
    return 0 if ($filename_no_path eq "" || !-f $filename_full_path);

    # we get a new temporary file which is rotated
    $filename_full_path = $self->rotate_image($filename_full_path);
    }
    
    # do generate images
    my $result = $self->generate_images($filename_full_path, $filename_no_path, $doc_obj, $section);
    #overwrite one set in ImageConverter
    $doc_obj->set_metadata_element ($section, "FileFormat", "PagedImage");
    return $result;
}

sub old_read_stuff_for_xml_version {
    my ($self, $filename, $file, $gli);

    # this bit same as ReadXMLFile read
#   $file =~ s/^[\/\\]+//; # $file often begins with / so we'll tidy it up
#   $self->{'file'} = $file;
#   $self->{'filename'} = $filename_full_path;
#   $self->{'processor'} = $processor;
#   $self->{'metadata'} = $metadata;
#
    eval {
        $@ = "";
        my $xslt = $self->{'xslt'};
        if (defined $xslt && ($xslt ne "")) {
        # perform xslt
        my $transformed_xml = $self->apply_xslt($xslt,$filename);
        
        # feed transformed file (now in memory as string) into XML parser
        $self->{'parser'}->parse($transformed_xml);
        ###$self->parse_string($transformed_xml);
        }
        else {
        $self->{'parser'}->parsefile($filename);
        #$self->parse_file($filename);
        }
    };
    
    
    
    if ($@) {
        
        # parsefile may either croak somewhere in XML::Parser (e.g. because
        # the document is not well formed) or die somewhere in XMLPlug or a
        # derived plugin (e.g. because we're attempting to process a
        # document whose DOCTYPE is not meant for this plugin). For the
        # first case we'll print a warning and continue, for the second
        # we'll just continue quietly
        
        print STDERR "**** XML Parse Error is: $@\n";
        
        my ($msg) = $@ =~ /Carp::croak\(\'(.*?)\'\)/;
        if (defined $msg) { 
        my $outhandle = $self->{'outhandle'};
        my $plugin_name = ref ($self);
        print $outhandle "$plugin_name failed to process $file ($msg)\n";
        }

        # reset ourself for the next document
        $self->{'section_level'}=0;
        print STDERR "<ProcessingError n='$file'>\n" if ($gli);
        return -1; # error during processing
    }

}


# The PagedImagePlugin read() function. This function does all the right things
# to make general options work for a given plugin. It calls the process()
# function which does all the work specific to a plugin (like the old
# read functions used to do). Most plugins should define their own 
# process() function and let this read() function keep control.
#
# PagedImagePlugin overrides read() because there is no need to read the actual 
# text of the file in, because the contents of the file is not text...
#
# Return number of files processed, undef if can't process
# Note that $base_dir might be "" and that $file might 
# include directories

sub read_into_doc_obj {
    my $self = shift (@_);
    my ($pluginfo, $base_dir, $file, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
    my $outhandle = $self->{'outhandle'};
    
    my ($filename_full_path, $filename_no_path) = $self->get_full_filenames($base_dir, $file);

    print $outhandle "PagedImagePlugin processing \"$filename_full_path\"\n"
    if $self->{'verbosity'} > 1;
    print STDERR "<Processing n='$file' p='PagedImagePlugin'>\n" if ($gli);
    
    
    # here we need to decide if we have an old text .item file, or a new xml 
    # .item file
    my $xml_version = $self->is_xml_item_file($filename_full_path);

    $self->tidy_item_file($filename_full_path);
    
    my $doc_obj;
    if ($xml_version) {
    # careful checking needed here!! are we using local xml handlers or super ones
    $self->ReadXMLFile::read($pluginfo, $base_dir, $file, $metadata, $processor, $maxdocs, $total_count, $gli);
    $doc_obj = $self->{'doc_obj'};
    } else {
    my ($dir);
    ($dir, $file) = $filename_full_path =~ /^(.*?)([^\/\\]*)$/;

    #process the .item file
    $doc_obj = $self->process_item($filename_full_path, $dir, $file, $processor);
    
    }

    my $section = $doc_obj->get_top_section();
        
    $doc_obj->add_utf8_metadata($section, "Plugin", "$self->{'plugin_type'}");
    $doc_obj->add_metadata($section, "FileFormat", "PagedImage");

    # include any metadata passed in from previous plugins 
    # note that this metadata is associated with the top level section
    $self->add_associated_files($doc_obj, $filename_full_path);
    $self->extra_metadata ($doc_obj, $section, $metadata);
    $self->auto_extract_metadata ($doc_obj);

    # if we haven't found any Title so far, assign one
    $self->title_fallback($doc_obj,$section,$filename_no_path);

    $self->add_OID($doc_obj);
    return (1,$doc_obj);
}

# for now, the test is if the first non-empty line is <PagedDocument>, then its xml
sub is_xml_item_file {
    my $self = shift(@_);
    my ($filename) = @_;

    my $xml_version = 0;
    open (ITEMFILE, $filename) || die "couldn't open $filename\n";
    
    my $line = "";
    my $num = 0;
    $line = <ITEMFILE>;
    while ($line !~ /\w/) {
    $line = <ITEMFILE>;
    }
    chomp $line;
    if ($line =~ /<PagedDocument/) {
    $xml_version = 1;
    }
    close ITEMFILE; 
    return $xml_version;
}

sub tidy_item_file {
    my $self = shift(@_);
    my ($filename) = @_;

    open (ITEMFILE, $filename) || die "couldn't open $filename\n";
    my $backup_filename = "backup.item";
    open (BACKUP,">$backup_filename")|| die "couldn't write to $backup_filename\n";
    my $line = "";
    $line = <ITEMFILE>;
    $line =~ s/^\xEF\xBB\xBF//; # strip BOM
    $line =~ s/\x0B+//ig;
    $line =~ s/&/&amp;/g;
    print BACKUP ($line);
    #Tidy up the item file some metadata title contains \vt-vertical tab
    while ($line = <ITEMFILE>) {
    $line =~ s/\x0B+//ig;
    $line =~ s/&/&amp;/g;
    print BACKUP ($line);
    }
    close ITEMFILE;
    close BACKUP;
    &File::Copy::copy ($backup_filename, $filename);
    &util::rm($backup_filename);

}
# de we need this? old read was the same as BasePlug read, not the same as ReadXMLfile read
sub read
{
    my $self = shift (@_);
    $self->BasePlugin::read(@_);
}

sub xml_start_tag {
    my $self = shift(@_);
    my ($expat, $element) = @_;
    $self->{'element'} = $element;
    
    my $doc_obj = $self->{'doc_obj'};
    if ($element eq "PagedDocument") {
    $self->{'current_section'} = $doc_obj->get_top_section();
    } elsif ($element eq "PageGroup" || $element eq "Page") {
    # create a new section as a child
    $self->{'current_section'} = $doc_obj->insert_section($doc_obj->get_end_child($self->{'current_section'}));
    $self->{'num_pages'}++;
    # assign pagenum as  what??
    my $pagenum = $_{'pagenum'}; #TODO!!
    if (defined $pagenum) {
        $doc_obj->set_utf8_metadata_element($self->{'current_section'}, 'PageNum', $pagenum);
    }
    my ($imgfile) = $_{'imgfile'};
    if (defined $imgfile) {
        $self->process_image($self->{'base_dir'}.$imgfile, $imgfile, $doc_obj, $self->{'current_section'});
    }
    my ($txtfile) = $_{'txtfile'};
    if (defined($txtfile)&& $txtfile ne "") {
        $self->process_text ($self->{'base_dir'}.$txtfile, $txtfile, $doc_obj, $self->{'current_section'});
    } else {
        $self->add_dummy_text($doc_obj, $self->{'current_section'});
    }
    } elsif ($element eq "Metadata") {
    $self->{'metadata_name'} = $_{'name'};
    }
}

sub xml_end_tag {
    my $self = shift(@_);
    my ($expat, $element) = @_;
    
    my $doc_obj = $self->{'doc_obj'};
    if ($element eq "Page" || $element eq "PageGroup") {
    # if Title hasn't been assigned, set PageNum as Title
    if (!defined $doc_obj->get_metadata_element ($self->{'current_section'}, "Title") && defined $doc_obj->get_metadata_element ($self->{'current_section'}, "PageNum" )) {
        $doc_obj->add_utf8_metadata ($self->{'current_section'}, "Title", $doc_obj->get_metadata_element ($self->{'current_section'}, "PageNum" ));
    }
    # move the current section back to the parent
    $self->{'current_section'} = $doc_obj->get_parent_section($self->{'current_section'});
    } elsif ($element eq "Metadata") {
    
    $doc_obj->add_utf8_metadata ($self->{'current_section'}, $self->{'metadata_name'}, $self->{'metadata_value'});
    $self->{'metadata_name'} = "";
    $self->{'metadata_value'} = "";

    }
    # otherwise we ignore the end tag
}


sub xml_text {
    my $self = shift(@_);
    my ($expat) = @_; 

    if ($self->{'element'} eq "Metadata" && $self->{'metadata_name'}) {
    $self->{'metadata_value'} .= $_;
    }
}

sub xml_doctype {
}

sub open_document {
    my $self = shift(@_);
    
    # create a new document
    $self->{'doc_obj'} = new doc ($self->{'filename'}, "indexed_doc");
    # TODO is file filenmae_no_path??
    $self->set_initial_doc_fields($self->{'doc_obj'}, $self->{'file'}, $self->{'processor'});

    my ($dir, $file) = $self->{'filename'} =~ /^(.*?)([^\/\\]*)$/;
    $self->{'base_dir'} = $dir;
    $self->{'num_pages'} = 0;

}

sub close_document {
    my $self = shift(@_);
    my $doc_obj = $self->{'doc_obj'};
    
    # add numpages metadata
    $doc_obj->set_utf8_metadata_element ($doc_obj->get_top_section(), 'NumPages', $self->{'num_pages'});

    
}


sub set_initial_doc_fields {
    my $self = shift(@_);
    my ($doc_obj, $filename_no_path, $processor) = @_;

    $doc_obj->set_OIDtype ($processor->{'OIDtype'}, $processor->{'OIDmetadata'});
    my $topsection = $doc_obj->get_top_section();

    if ($self->{'documenttype'} eq 'paged') {
    # set the gsdlthistype metadata to Paged - this ensures this document will
    # be treated as a Paged doc, even if Titles are not numeric
    $doc_obj->set_utf8_metadata_element ($topsection, "gsdlthistype", "Paged");
    } else {
    $doc_obj->set_utf8_metadata_element ($topsection, "gsdlthistype", "Hierarchy");
    }

    $self->set_Source_metadata($doc_obj, $filename_no_path);
   
    # if we want a header page, we need to add some text into the top section, otherwise this section will become invisible
    if ($self->{'headerpage'}) {
    $doc_obj->add_text($topsection, &gsprintf::lookup_string("{BasePlugin.dummy_text}"));
    }


}


sub process_item {
    my $self = shift (@_);
    my ($filename_full_path, $dir, $filename_no_path, $processor) = @_;

    my $doc_obj = new doc ($filename_full_path, "indexed_doc");
    $self->set_initial_doc_fields($doc_obj, $filename_no_path, $processor);
    my $topsection = $doc_obj->get_top_section();
    open (ITEMFILE, $filename_full_path) || die "couldn't open $filename_full_path\n";
    my $line = "";
    my $num = 0;
    while (defined ($line = <ITEMFILE>)) {
    next unless $line =~ /\w/;
    chomp $line;
    next if $line =~ /^#/; # ignore comment lines
    if ($line =~ /^<([^>]*)>\s*(.*?)\s*$/) {
        $doc_obj->set_utf8_metadata_element ($topsection, $1, $2);
        #$meta->{$1} = $2;
    } else {
        $num++;
        # line should be like page:imagefilename:textfilename:r - the r is optional -> means rotate the image 180 deg
        $line =~ s/^\s+//; #remove space at the front
        $line =~ s/\s+$//; #remove space at the end
        my ($pagenum, $imgname, $txtname, $rotate) = split /:/, $line;
        
        # create a new section for each image file
        my $cursection = $doc_obj->insert_section($doc_obj->get_end_child($topsection));
        # the page number becomes the Title
        $doc_obj->set_utf8_metadata_element($cursection, 'Title', $pagenum);
     
        # process the image for this page if there is one
        if (defined $imgname && $imgname ne "") {
        my $result1 = $self->process_image($dir.$imgname, $imgname, $doc_obj, $cursection, $rotate);
        
        if (!defined $result1)
        {
            print "PagedImagePlugin: couldn't process image \"$dir.$imgname\" for item \"$filename_full_path\"\n";
        }
        }
        # process the text file if one is there
        if (defined $txtname && $txtname ne "") {
        my $result2 = $self->process_text ($dir.$txtname, $txtname, $doc_obj, $cursection);
               
        if (!defined $result2) {
            print "PagedImagePlugin: couldn't process text file \"$dir.$txtname\" for item \"$filename_full_path\"\n";
            $self->add_dummy_text($doc_obj, $cursection);
        }
        } else {
        # otherwise add in some dummy text 
        $self->add_dummy_text($doc_obj, $cursection);
        }
    }
    }
    
    close ITEMFILE;

    # add numpages metadata
    $doc_obj->set_utf8_metadata_element ($topsection, 'NumPages', "$num");
    return $doc_obj;
}

sub process_text {
    my $self = shift (@_);
    my ($filename_full_path, $file, $doc_obj, $cursection) = @_;
    
    # check that the text file exists!!
    if (!-f $filename_full_path) {
    print "PagedImagePlugin: ERROR: File $filename_full_path does not exist, skipping\n";
    return 0;
    }

    # Do encoding stuff
    my ($language, $encoding) = $self->textcat_get_language_encoding ($filename_full_path);

    my $text="";
    &ReadTextFile::read_file($self, $filename_full_path, $encoding, $language, \$text);
    if (!length ($text)) {
    # It's a bit unusual but not out of the question to have no text, so just give a warning
        print "PagedImagePlugin: WARNING: $filename_full_path contains no text\n";
    }

    # we need to escape the escape character, or else mg will convert into
    # eg literal newlines, instead of leaving the text as '\n'
    $text =~ s/\\/\\\\/g; # macro language
    $text =~ s/_/\\_/g; # macro language


    if ($text =~ m/<html.*?>\s*<head.*?>.*<\/head>\s*<body.*?>(.*)<\/body>\s*<\/html>\s*$/s) {
    # looks like HTML input
    # no need to escape < and > or put in <pre> tags

    $text = $1;

    # insert preformat tags and add text to document object
    $doc_obj->add_utf8_text($cursection, "$text");
    }
    else {
    $text =~ s/</&lt;/g;
    $text =~ s/>/&gt;/g;

    # insert preformat tags and add text to document object
    $doc_obj->add_utf8_text($cursection, "<pre>\n$text\n</pre>");
    }

    
    return 1;
}

# do plugin specific processing of doc_obj
sub process_old {
    my $self = shift (@_);
    my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj) = @_;
    my $outhandle = $self->{'outhandle'};
    
    return 1;
}

sub clean_up_after_doc_obj_processing {
    my $self = shift(@_);
    
    $self->ImageConverter::clean_up_temporary_files();
}

1;