Changeset 2241

Timestamp:

2001-04-01T21:19:25+12:00 (23 years ago)

Author:

sjboddie

Message:

Tidied up the ConvertToPlug stuff to get it working on Windows 95/98

Location:

trunk/gsdl

Files:

• bin/script/gsConvert.pl (modified) (9 diffs)
• bin/script/pdftohtml.pl (modified) (3 diffs)
• perllib/plugins/ConvertToPlug.pm (modified) (11 diffs)

trunk/gsdl/bin/script/gsConvert.pl

@@ -91 +91 @@
     # Deduce filenames
     my ($tailname,$dirname,$suffix) 
-    = File::Basename::fileparse($input_filename,'\..+');
-    my $output_filestem = &util::filename_cat($dirname,"$tailname");
+    = File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
+    my $output_filestem = &util::filename_cat($dirname, "$tailname");
 
     if ($input_type eq "")
     {
-    $input_type = substr($suffix,1,length($suffix)-1);
+    $input_type = lc (substr($suffix,1,length($suffix)-1));
     }
     
@@ -138 +138 @@
 
 
-# Document-type conversion fucntions
+# Document-type conversion functions
 #
 # The following functions attempt to convert documents from their
@@ -219 +219 @@
     # Convert to text
     if (!$output_type || ($output_type =~ /text/i)) {
-    $success = any_to_text($input_filename, $output_filestem);
+    $success = &any_to_text($input_filename, $output_filestem);
     if ($success) {
         return "text";
@@ -332 +332 @@
     my $wvWare = &util::filename_cat($ENV{'GSDLHOME'}, "bin", 
                      $ENV{'GSDLOS'}, "wvWare");
-    $wvWare .= ".exe" if ($ENV{'GSDLOS'} =~ /^windows$/i);
-    return 0 unless (-e "$wvWare");
+
+    # don't include path on windows (to avoid having to play about
+    # with quoting when GSDLHOME might contain spaces) but assume
+    # that the PATH is set up correctly
+    $wvWare = "wvWare" if ($ENV{'GSDLOS'} =~ /^windows$/i);
 
     my $wv_conf = &util::filename_cat($ENV{'GSDLHOME'}, "packages", 
                       "wv", "wvHtml.xml");
     
+    my $cmd = "";
+    if ($timeout) {$cmd = "ulimit -t $timeout;";}
+    $cmd .= "$wvWare --charset utf-8 --config \"$wv_conf\"";
+    $cmd .= " \"$input_filename\" > \"$output_filestem.html\"";
+    
+    # redirecting STDERR is a bad idea on windows 95/98
+    $cmd .= " 2> \"$output_filestem.err\""
+    if $ENV{'GSDLOS'} !~ /^windows$/i;
+
+    # execute the command
+    if (system($cmd)!=0)
+    {
+    print STDERR "Error executing wv converter: $!. Continuing...\n";
+    }
+
+    # Was the conversion successful?
+
+    if (-e "$output_filestem.html") {
+    open(TMP, "$output_filestem.html");
+    $line = <TMP>;
+    close(TMP);
+    if ($line && $line =~ /DOCTYPE HTML/) {
+        &util::rm("$output_filestem.err") if -e "$output_filestem.err";
+        return 1;
+    } else {
+        # An error of some sort occurred
+        &util::rm("$output_filestem.html");
+        &util::rm("$output_filestem.err") if -e "$output_filestem.err";
+    }
+    }
+
+    return 0;
+}
+
+
+# Attempt to convert an RTF document to html with rtftohtml
+#
+# rtf2html isn't distributed with Greenstone because it is not 
+# distributed under the GPL.  If you know of a better solution,
+# please let me know.
+
+sub rtf_to_html {
+    my ($input_filename, $output_filestem) = @_;
+
+    # we'll give up already if using Windows
+    return 0 if $ENV{'GSDLOS'} =~ /^windows$/i;
+
+    # formulate the command
+    my $r_cmd = &util::filename_cat($ENV{'GSDLHOME'}, "packages", "unix", 
+                    "rtf2html", "rtf2html", "rtf2html");
+    $r_cmd = "rtf2html" unless (-e "$r_cmd");
+    return 0 unless (-e "$r_cmd");
     $cmd = "";
     if ($timeout) {$cmd = "ulimit -t $timeout;";}
-    $cmd .= "$wvWare --charset utf-8 --config $wv_conf";
+    $cmd .= "$r_cmd";
     $cmd .= " \"$input_filename\" > \"$output_filestem.html\" 2>\"$output_filestem.err\"";
 
@@ -346 +401 @@
     if (system($cmd)!=0)
     {
-    print STDERR "Error executing wv converter: $!. Continuing...\n";
+    print STDERR "Error executing rtf converter: $!. Continuing...\n";
     }
 
@@ -363 +418 @@
     }
     }
-
     return 0;
 }
 
 
-# Attempt to convert an RTF document to html with rtftohtml
-#
-# rtf2html isn't distributed with Greenstone because it is not 
-# distributed under teh GPL.  If you know of a better solution,
-# please let me know.
-
-sub rtf_to_html {
-    ($input_filename, $output_filestem) = @_;
-
-    # formulate the command
-    my $r_cmd = &util::filename_cat($ENV{'GSDLHOME'}, "packages", "unix", 
-                    "rtf2html", "rtf2html", "rtf2html");
-    $r_cmd = "rtf2html" unless (-e "$r_cmd");
-    return 0 unless (-e "$r_cmd");
+# Convert a pdf file to html with the pdftohtml command
+
+sub pdf_to_html {
+    ($dirname, $input_filename, $output_filestem) = @_;
+
     $cmd = "";
     if ($timeout) {$cmd = "ulimit -t $timeout;";}
-    $cmd .= "$r_cmd";
-    $cmd .= " \"$input_filename\" > \"$output_filestem.html\" 2>\"$output_filestem.err\"";
-
-    # execute the command
-    if (system($cmd)!=0)
-    {
-    print STDERR "Error executing rtf converter: $!. Continuing...\n";
-    }
-
-    # Was the conversion successful?
-    if (-e "$output_filestem.html") {
-    open(TMP, "$output_filestem.html");
-    $line = <TMP>;
-    close(TMP);
-    if ($line && $line =~ /DOCTYPE HTML/) {
-        &util::rm("$output_filestem.err");
-        return 1;
-    } else {
-        # An error of some sort occurred
-        &util::rm("$output_filestem.html");
-        &util::rm("$output_filestem.err");
-    }
-    }
-    return 0;
-}
-
-
-# Convert a pdf file to html with the pdftohtml command
-
-sub pdf_to_html {
-    ($dirname, $input_filename, $output_filestem) = @_;
-
-    $cmd = "";
-    if ($timeout) {$cmd = "ulimit -t $timeout;";}
-    $cmd .= "pdftohtml.pl -F ";
+    $cmd .= "perl -S pdftohtml.pl -F ";
     $cmd .= " \"$input_filename\" \"$output_filestem\"";
     $!=0;
+
     if (system($cmd)!=0)
     {
@@ -491 +503 @@
 
 sub ps_to_text {
-    ($input_filename, $output_filestem) = @_;
-
-    my $cmd = "gs -q -dNODISPLAY -dNOBIND -dWRITESYSTEMDICT -dSIMPLE -c save ";
-    $cmd .= "-f ps2ascii.ps \"$input_filename\" -c quit > \"$output_filestem.text\"";
-    $cmd .= " 2> $output_filestem.err";
-    $!=0;
-    my $retcode=system($cmd);
-    $retcode = $? >> 8;  # see man perlfunc - system for this...
-    # if system returns -1 | 127 (couldn't start program), look at $! for message
-    my $error="";
-    if ($retcode!=0) {if ($!) {$error=$!;} else {$error="couldn't run.\n";}}
-    elsif (! -e "$output_filestem.text") {
-    $error="did not create output file.\n";
-    }
-    else 
-    { # make sure the interpreter didn't get an error. It is technically
-    # possible for the actual text to start with this, but....
-    open PSOUT, "$output_filestem.text";
-    if (<PSOUT> =~ /^Error: (.*)/) {
-        $error="interpreter error - \"$1\"";
-    }
-    close PSOUT;
-    }
+    my ($input_filename, $output_filestem) = @_;
+
+    my $error = "";
+
+    # if we're on windows we'll fall straight through without attempting
+    # to use gs
+    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
+    $error = "Windows does not support gs";
+
+    } else {
+    my $cmd = "gs -q -dNODISPLAY -dNOBIND -dWRITESYSTEMDICT -dSIMPLE -c save ";
+    $cmd .= "-f ps2ascii.ps \"$input_filename\" -c quit > \"$output_filestem.text\"";
+    $cmd .= " 2> $output_filestem.err";
+    $!=0;
+    my $retcode=system($cmd);
+    $retcode = $? >> 8;  # see man perlfunc - system for this...
+    # if system returns -1 | 127 (couldn't start program), look at $! for message
+
+    if ($retcode!=0) {if ($!) {$error=$!;} else {$error="couldn't run.\n";}}
+    elsif (! -e "$output_filestem.text") {
+        $error="did not create output file.\n";
+    }
+    else 
+    {   # make sure the interpreter didn't get an error. It is technically
+        # possible for the actual text to start with this, but....
+        open PSOUT, "$output_filestem.text";
+        if (<PSOUT> =~ /^Error: (.*)/) {
+        $error="interpreter error - \"$1\"";
+        }
+        close PSOUT;
+    }
+    }
+
     if ($error ne "")
     {
@@ -611 +633 @@
     open(HTML, ">$output_filestem.html");
 
-    print HTML '<html><head>
-<META HTTP-EQUIV="Content-Type" CONTENT="text/html">
-<META NAME="GENERATOR" CONTENT="Greenstone any_to_html">
-</head><body>';
-    print HTML "\n\n";
+    print HTML "<html><head>\n";
+    print HTML "<META HTTP-EQUIV=\"Content-Type\" CONTENT=\"text/html\">\n";
+    print HTML "<META NAME=\"GENERATOR\" CONTENT=\"Greenstone any_to_html\">\n";
+    print HTML "</head><body>\n\n";
 
     while (<TEXT>) { 
     print HTML "<p> ", $_;
-    
     }
     print HTML "\n</body></html>\n";
+
+    close HTML;
+    close TEXT;
 
     &util::rm("$output_filestem.text") if (-e "$output_filestem.text");
@@ -661 +684 @@
     }
     }
+
+    close OUT;
+    close IN;
+
     return 1;
 }

trunk/gsdl/bin/script/pdftohtml.pl

@@ -177 +177 @@
     print STDERR "pdftohtml.pl: $input_filename appears to have no ";
     print STDERR "textual data. Aborting.\n";
-    print STDERR "num: $unenc_stream_objects and $non_text_objects from $num_objects\n";
+    # print STDERR "num: $unenc_stream_objects and $non_text_objects from $num_objects\n";
     exit(1);
     }
 
     # formulate the command
-    my $pdftohtml = &util::filename_cat($ENV{'GSDLHOME'}, "bin", 
-                    $ENV{'GSDLOS'}, "pdftohtml.bin");
-    return 0 unless (-e "$pdftohtml");
-    
-    $cmd = "";
+    my $cmd = &util::filename_cat($ENV{'GSDLHOME'}, "bin", $ENV{'GSDLOS'}, "pdftohtml.bin");
+
+    # don't include path on windows (to avoid having to play about
+    # with quoting when GSDLHOME might contain spaces) but assume
+    # that the PATH is set up correctly - note also that on windows
+    # we use pdftohtml.exe not pdftohtml.bin
+    $cmd = "pdftohtml" if ($ENV{'GSDLOS'} !~ /^windows$/) {
+
     if ($timeout) {$cmd = "ulimit -t $timeout;";}
-    $cmd .= "$pdftohtml -noframes";
-    $cmd .= " \"$input_filename\" \"$output_filestem.html\"";
-    $cmd .= " >\"$output_filestem.out\" 2>\"$output_filestem.err\"";
-    
+    $cmd .= " -noframes \"$input_filename\" \"$output_filestem.html\"";
+    $cmd .= " > \"$output_filestem.out\"";
+
+    # attempting to redirect STDERR on windows 95/98 is a bad idea
+    $cmd .= " 2> \"$output_filestem.err\"" 
+    if $ENV{'GSDLOS'} !~ /^windows$/i;
+
     if (system($cmd)>0) {
     print STDERR "Error executing $cmd: $!\n";
@@ -201 +207 @@
     # Need to convert images from PPM format to PNG format
     my @images;
+
 
     open (IMAGES, "images.log");
@@ -212 +219 @@
     my $cmd = "";
     if ($ENV{'GSDLOS'} =~ /^windows/i) {
-        $cmd = &util::filename_cat($ENV{'GSDLHOME'}, "bin", "windows", "pnmtopng.exe");
-        $cmd .= " $image";
+        $cmd = "pnmtopng $image";
         if (system($cmd)!=0) {
         print STDERR "Error executing $cmd\n";

trunk/gsdl/perllib/plugins/ConvertToPlug.pm

@@ -94 +94 @@
     if ($class eq "ConvertToPlug") {$class = shift (@_);}
     my $self;
-# parsargv::parse might modify the list, so we do this by creating a copy
-# of the argument list.
+    # parsargv::parse might modify the list, so we do this by creating a copy
+    # of the argument list.
     my @arglist = @_;
-    my ($plugin_name,$generate_format, $kea_arg) = $class->parse_args(\@_);
+    my ($plugin_name, $generate_format, $kea_arg) = $class->parse_args(\@_);
+
+    if ($class eq "PDFPlug" && $generate_format eq "text" && 
+    $ENV{'GSDLOS'} =~ /^windows$/i) {
+    print STDERR "Windows does not support pdf to text. PDFs will be converted to HTML instead\n";
+    $generate_format = "html";
+    }
 
     if ($generate_format eq "text")
@@ -135 +141 @@
 sub tmp_area_convert_file {
     my $self = shift (@_);
-    my ($output_ext,$input_filename, $textref) = @_;
+    my ($output_ext, $input_filename, $textref) = @_;
 
     my $convert_to = $self->{'convert_to'};
@@ -142 +148 @@
     my $colname = &util::use_collection();
     my $tmp_dirname 
-    = &util::filename_cat($ENV{'GSDLHOME'},"collect",$colname,"tmp");
+    = &util::filename_cat($ENV{'GSDLHOME'}, "collect", $colname, "tmp");
     &util::mk_dir($tmp_dirname) if (!-e $tmp_dirname);
 
     # derive tmp filename from input filename
-    my ($tailname,$dirname,$suffix)
-    = File::Basename::fileparse($input_filename,'\.[^\.]+$');
+    my ($tailname, $dirname, $suffix)
+    = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
 
     # Remove any white space from filename -- no risk of name collision, and
@@ -153 +159 @@
     $tailname =~ s/\s+//g; 
 
-    my $tmp_filename = &util::filename_cat($tmp_dirname,"$tailname$suffix");
-
-    &util::soft_link($input_filename,$tmp_filename);
+    my $tmp_filename = &util::filename_cat($tmp_dirname, "$tailname$suffix");
+
+    &util::soft_link($input_filename, $tmp_filename);
 
     my $verbosity = $self->{'verbosity'};
-    if ($verbosity>0)
-    {
+    if ($verbosity > 0) {
     print STDERR "Converting $tailname$suffix to $convert_to format\n";
     }
@@ -166 +171 @@
     # making sure the converter gives us the appropriate output type
     my $output_type = lc($convert_to);
-    my $cmd = "gsConvert.pl -verbose $verbosity -output $output_type \"$tmp_filename\"";
+    my $cmd = "perl -S gsConvert.pl -verbose $verbosity -output $output_type \"$tmp_filename\"";
     $output_type = `$cmd`;
-
-    # Check STDERR here
 
     chomp $output_type;
@@ -175 +178 @@
     print STDERR "Could not convert $tailname$suffix to $convert_to format\n";
     return "";
-### exit 1;
     }
 
@@ -184 +186 @@
     $self->{'convert_to_ext'} = $output_type;
     my $output_filename = $tmp_filename;
+
     $output_filename =~ s/$suffix$/.$output_type/;
 
@@ -197 +200 @@
     my $colname = &util::use_collection();
     my $tmp_dirname 
-    = &util::filename_cat($ENV{'GSDLHOME'},"collect",$colname,"tmp");
+    = &util::filename_cat($ENV{'GSDLHOME'}, "collect", $colname, "tmp");
     &util::rm_r($tmp_dirname);
     &util::mk_dir($tmp_dirname);
@@ -229 +232 @@
 
     my $output_ext = $self->{'convert_to_ext'};
-    my $conv_filename = $self->tmp_area_convert_file($output_ext,$filename);
+    my $conv_filename = $self->tmp_area_convert_file($output_ext, $filename);
+
     if ("$conv_filename" eq "") {return 0;} # allows continue on errors
     if (! -e "$conv_filename") {return 0;} # allows continue on errors
     $self->{'conv_filename'} = $conv_filename;
 
-# Do encoding stuff
+    # Do encoding stuff
     my ($language, $encoding);
     if ($self->{'input_encoding'} eq "auto") {
@@ -253 +257 @@
     }
 
-    BasPlug::read_file($self,$conv_filename, $encoding, \$text);
+    &BasPlug::read_file($self, $conv_filename, $encoding, \$text);
     if (!length ($text)) {
         print $outhandle "$plugin_name: ERROR: $file contains no text\n" if $self->{'verbosity'};
@@ -301 +305 @@
     {
 
-    $ret_val = TEXTPlug::process($self,$textref,$pluginfo,
-                     $tmp_dirname,$tmp_tailname,
-                     $metadata,$doc_obj);
+    $ret_val = &TEXTPlug::process($self, $textref, $pluginfo,
+                      $tmp_dirname, $tmp_tailname,
+                      $metadata, $doc_obj);
     }
     else
     {
-    $ret_val = HTMLPlug::process($self,$textref,$pluginfo,
-                     $tmp_dirname,$tmp_tailname,
-                     $metadata,$doc_obj);
+    $ret_val = &HTMLPlug::process($self, $textref, $pluginfo,
+                      $tmp_dirname, $tmp_tailname,
+                      $metadata, $doc_obj);
     }
 
     # associate original file with doc object
     my $cursection = $doc_obj->get_top_section();
-    my $filename = &util::filename_cat($base_dir,$file);
+    my $filename = &util::filename_cat($base_dir, $file);
     $doc_obj->associate_file($filename, "doc.$doc_ext", undef, $cursection);