Changeset 22861 for main/trunk/greenstone2/perllib/plugins/PowerPointPlugin.pm

Timestamp:

2010-09-07T12:08:44+12:00 (14 years ago)

Author:

kjdon

Message:

now uses new AutoLoadConverters instead of AutoloadConverterScripting. This doesn't inherit from ConvertBinaryFile, so these plugins all inherit from that again. Now we can initialise the converters, fix up the modifications to the arguments, before parsing them when we do new ConvertBinaryFile. PowerPointPlugin incomplete and still needs lots of work done for processing the result on open office conversion

File:

• main/trunk/greenstone2/perllib/plugins/PowerPointPlugin.pm (modified) (10 diffs)

main/trunk/greenstone2/perllib/plugins/PowerPointPlugin.pm

@@ -3 +3 @@
 # PowerPointPlugin.pm -- plugin for importing Microsoft PowerPoint files.
 #  (basic version supports versions 95 and 97)
-#  (through OpenOffice extension, supports all contempoary formats)
+#  (through OpenOffice extension, supports all contemporary formats)
 #
 # A component of the Greenstone digital library software
@@ -35 +35 @@
 use gsprintf 'gsprintf';
 
-use AutoloadConverterScripting;
-
-@PowerPointPlugin::ISA = ('AutoloadConverterScripting');
-
+use AutoLoadConverters;
+use ConvertBinaryFile;
+
+sub BEGIN {
+    @PowerPointPlugin::ISA = ('ConvertBinaryFile', 'AutoLoadConverters');
+}
+
+my $openoffice_available = 0;
 
 my $windows_convert_to_list =
@@ -55 +59 @@
       ];
 
+my $openoffice_convert_to_list = 
+    [ { 'name' => "auto",
+    'desc' => "{ConvertBinaryFile.convert_to.auto}" },
+      { 'name' => "html",
+    'desc' => "{PowerPointPlugin.convert_to.oo_html}" },
+      { 'name' => "text",
+    'desc' => "{ConvertBinaryFile.convert_to.text}" },
+      { 'name' => "pagedimg",
+    'desc' => "{PowerPointPlugin.convert_to.pagedimg}" }
+      ];
+
 my $arguments = 
     [ { 'name' => "process_exp",
@@ -60 +75 @@
     'type' => "regexp",
     'reqd' => "no",
-    'deft' => "&get_default_process_exp(\$self)",  # delayed (see below)
+    'deft' => "&get_default_process_exp()",  # delayed (see below)
     }
       ];
@@ -77 +92 @@
       ];
 
+my $opt_office_args = 
+    [ { 'name' => "convert_to",
+    'desc' => "{ConvertBinaryFile.convert_to}",
+    'type' => "enum",
+    'reqd' => "yes",
+    'list' => $openoffice_convert_to_list, 
+    'deft' => "html" }
+      ];
+
 my $options = { 'name'     => "PowerPointPlugin",
         'desc'     => "{PowerPointPlugin.desc}",
@@ -89 +113 @@
     push(@$pluginlist, $class);
 
+    # this bit needs to happen later after the arguments array has been 
+    # finished - used for parsing the input args.
+    # push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
+    # this one needs to go in first, to get the print info in the right order
+    push(@{$hashArgOptLists->{"OptList"}},$options);
+
     if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
     push(@$arguments,@$opt_windows_args);
     }
 
-    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
-    push(@{$hashArgOptLists->{"OptList"}},$options);
-
-
-    my $self 
-    = new AutoloadConverterScripting("OpenOfficeConverter",
-                     $pluginlist, $inputargs, 
-                     $hashArgOptLists);
-
-
-    # plugin's process_exp can only be correctly determined once autoloading
-    # has taken place
-    my $plug_options = $self->{'option_list'}->[0];
-    my $plug_args = $plug_options->{'args'};
-    
-    foreach my $a (@$plug_args) {
-    # consider changing this to search for all values that are
-    # tagged as 'deft-delayed' = 1 ?!?
-
+    my $auto_converter_self = new AutoLoadConverters($pluginlist,$inputargs,$hashArgOptLists,["OpenOfficeConverter"],1);
+
+    if ($auto_converter_self->{'openoffice_available'}) {
+    push (@$arguments,@$opt_office_args);
+    $openoffice_available = 1;
+    }
+    # TODO need to do the case where they are both enabled!!! what will the convert to list be???
+
+    # evaluate the default for process_exp  - it needs to be delayed till here so we know if openoffice is available or not. But needs to be done before parsing the args.
+    foreach my $a (@$arguments) {
     if ($a->{'name'} eq "process_exp") {
         my $eval_expr = $a->{'deft'};
         $a->{'deft'} = eval "$eval_expr";
-
-        # Now see if process_exp needs updating
-        my $process_exp = $self->{'process_exp'};
-        if (!$self->{'info_only'} && ($process_exp eq $eval_expr)) {
-        # process_exp is only defined if not 'info_only'
-        #
-        # if it does exist and it equals the unevaluated $eval_expr
-        # then it was set to the default (rather than overriden by
-        # the collect.cfg file)
-
-        $self->{'process_exp'} = $a->{'deft'};
-        }
-    }
-    }
-
- 
+        last;
+    }
+    }
+
+    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
+
+    my $cbf_self = new ConvertBinaryFile($pluginlist, $inputargs, $hashArgOptLists);
+    my $self = BasePlugin::merge_inheritance($auto_converter_self, $cbf_self);
+
     if ($self->{'info_only'}) {
     # don't worry about any options etc
@@ -136 +150 @@
     }
 
+    $self = bless $self, $class;
     $self->{'filename_extension'} = "ppt";
     $self->{'file_type'} = "PPT";
@@ -150 +165 @@
    my $outhandle = $self->{'outhandle'};
 
-    # can't have windows_scripting and openoffice_scripting at the same time
-    if ($self->{'windows_scripting'} && $self->{'openoffice_scripting'}) {
-    print $outhandle "Warning: Cannot have -windows_scripting and -openoffice_scripting\n";
+    # can't have windows_scripting and openoffice_conversion at the same time
+    if ($self->{'windows_scripting'} && $self->{'openoffice_conversion'}) {
+    print $outhandle "Warning: Cannot have -windows_scripting and -openoffice_conversion\n";
     print $outhandle "         on at the same time.  Defaulting to -windows_scripting\n";
-    $self->{'openoffice_scripting'} = 0;
+    $self->{'openoffice_conversion'} = 0;
     }
     
@@ -193 +208 @@
     my $self = shift (@_);
 
-    if ($self->{'scripting_ext_working'}) {
+    if ($openoffice_available) {
     return q^(?i)\.(ppt|pptx|odp)$^;
     }
@@ -200 +215 @@
 }
 
+sub init {
+    my $self = shift (@_);
+
+    # ConvertBinaryFile init
+    $self->SUPER::init(@_);
+    $self->AutoLoadConverters::init();
+
+}
+
+sub begin {
+    my $self = shift (@_);
+
+    $self->AutoLoadConverters::begin();
+    $self->SUPER::begin(@_);
+
+}
+
+sub deinit {
+    my $self = shift (@_);
+    
+    $self->AutoLoadConverters::deinit();
+    $self->SUPER::deinit(@_);
+
+}
+
+# override AutoLoadConverters version, as we need to do more stuff once its converted if we are converting to item file
+sub tmp_area_convert_file {
+    my $self = shift (@_);
+    my ($output_ext, $input_filename, $textref) = @_;
+
+    if ($self->{'openoffice_conversion'}) {
+    if ($self->{'convert_to'} eq "pagedimg") {
+        $output_ext = "html"; # first convert to html
+    }
+    my ($result, $result_str, $new_filename) = $self->OpenOfficeConverter::convert($input_filename, $output_ext);
+    if ($result == 0) {
+        my $outhandle=$self->{'outhandle'};
+        print $outhandle "OpenOfficeConverter Conversion error\n";
+        print $outhandle $result_str;
+        return "";
+
+    }
+    #print STDERR "result = $result\n";
+    if ($self->{'convert_to'} eq "pagedimg") {
+        #my $item_filename = $self->generate_item_file($new_filename);
+        #return $item_filename;
+        return "/research/kjdon/home/gsdl/collect/openoffice/test.item";
+    }
+    return $new_filename;
+
+    }
+    else {
+    return $self->ConvertBinaryFile::tmp_area_convert_file(@_);
+    }
+    # get tmp filename
+}
+
+# override default read in some situations, as the conversion of ppt to html results in many files, and we want them all to be processed.
+sub read_XX {
+    my $self = shift (@_);  
+    my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
+
+    # can we process this file??
+    my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
+
+    return undef unless $self->can_process_this_file($filename_full_path);
+    
+    my ($process_status,$doc_obj) = $self->read_into_doc_obj(@_);
+    
+    if ((defined $process_status) && ($process_status == 1)) {
+    
+    # process the document
+    $processor->process($doc_obj);
+
+    $self->{'num_processed'} ++;
+    undef $doc_obj; 
+    }
+    # delete any temp files that we may have created
+    $self->clean_up_after_doc_obj_processing();
+
+
+    # if process_status == 1, then the file has been processed.
+    return $process_status;
+
+}
+
 1;