Changeset 1928 for trunk/gsdl/bin/script/gsConvert.pl

Timestamp:

2001-02-08T13:49:27+13:00 (23 years ago)

Author:

sjboddie

Message:

Added: pdftohtml.pl - Perl script that handles conversion of PDF documents into

HTML. Called by gsConvert.pl in sub pdf_to_html.

Modified: gsConvert.pl - Perl script that converts various formats (MSWord,

RTF, PDF, PS) into HTML when importing the collection.

File:

• trunk/gsdl/bin/script/gsConvert.pl (modified) (14 diffs)

trunk/gsdl/bin/script/gsConvert.pl

@@ -60 +60 @@
     my (@ARGV) = @_;
     my ($input_type,$output_type,$verbose,$timeout);
-
+    
     $timeout = 0;
     # read command-line arguments
@@ -76 +76 @@
     print_usage();
     }
+
     my $input_filename = $ARGV[0];
     if (!-r $input_filename) {
@@ -102 +103 @@
     } 
     elsif ($input_type eq "doc") {
-    print STDERR "I recognise this to be a Word document...\n"; # remove
     print &convertDOC($input_filename, $output_filestem, $output_type);
     print "\n";
@@ -149 +149 @@
     my $realtype = &find_docfile_type($input_filename);
 
-    print STDERR "The real type of this Word document is $realtype\n"; # remove
-
     if ($realtype eq "word6" || $realtype eq "word7" || $realtype eq "word8") {
-    print STDERR "I recognise this to be a word678 document...\n"; # remove
     return &convertWord678($input_filename, $output_filestem, $output_type);
     } elsif ($realtype eq "rtf") {
@@ -328 +325 @@
 
 sub doc_to_html {
-    print STDERR "/;-DG I am in doc_to_html...\n"; # remove
     ($input_filename, $output_filestem) = @_;
 
     my $wvWare = "";
     my $wv_conf = "";
+
     if ($ENV{'GSDLOS'} =~ /^windows$/i) {
     $wvWare = "$ENV{'GSDLHOME'}\\bin\\windows\\wvWare.exe";
@@ -343 +340 @@
     $wvWare = &util::filename_cat($wv_home, "bin", "wvWare");
     }
-    print STDERR "I am about to test if your file exists...\n";
     return 0 unless (-e "$wvWare");
+    
     $cmd = "";
     if ($timeout) {$cmd = "ulimit -t $timeout;";}
@@ -350 +347 @@
     $cmd .= " \"$input_filename\" > \"$output_filestem.html\" 2>\"$output_filestem.err\"";
 
-    print STDERR "$cmd\n"; #remove
-
     # execute the command
-    print STDERR system($cmd);
-    print STDERR "\n";
     if (system($cmd)>0)
     {
@@ -374 +367 @@
     }
     }
-    print STDERR "/;-DG I am leaving doc_to_html...\n";
+
     return 0;
 }
@@ -427 +420 @@
     ($dirname, $input_filename, $output_filestem) = @_;
 
-    # formulate the command
-    my $p_home = &util::filename_cat($ENV{'GSDLHOME'}, "packages", "unix", "pdftohtml");
-    my $pdftohtml = &util::filename_cat($p_home, "pdftohtml_0_20", "pdftohtml.bin");
-    return 0 unless (-e "$pdftohtml");
-
     $cmd = "";
     if ($timeout) {$cmd = "ulimit -t $timeout;";}
-    $cmd .= "$pdftohtml -noframes";
-    $cmd .= " \"$input_filename\" \"$output_filestem.html\"";
-    $cmd .= " >\"$output_filestem.out\" 2>\"$output_filestem.err\"";
+    $cmd .= "pdftohtml.pl -F ";
+    $cmd .= " \"$input_filename\" \"$output_filestem\"";
 
     if (system($cmd)>0)
     {
     print STDERR "Error executing $cmd: $!\n";
-    &util::rm("$output_filestem.html") if (-e "$output_filestem.html");
-    &util::rm("$output_filestem.out") if (-e "$output_filestem.out");
     return 0;
     }
@@ -466 +451 @@
 }
 
-
 # Convert a PDF file to text with the pdftotext command
 
@@ -502 +486 @@
     return 1;
 }
-
 
 # Convert a PostScript document to text with ps2ascii
@@ -560 +543 @@
 
 sub any_to_html {
-    print STDERR "/;-Dg I am in any_to_html!\n";
     ($input_filename, $output_filestem) = @_;
 
@@ -583 +565 @@
 
     &util::rm("$output_filestem.text") if (-e "$output_filestem.text");
-    print STDERR "/;-Dg I am getting out of  any_to_html!\n";
     return 1;
 }
@@ -625 +606 @@
     return 1;
 }
-
-
-