Changeset 1960

Timestamp:

2001-02-13T11:48:10+13:00 (23 years ago)

Author:

dg5

Message:

Modified pdftohtml.pl to reflect the change in location of pdftohtml.bin file

Location:

trunk/gsdl/bin/script

Files:

• gsConvert.pl (modified) (8 diffs)
• pdftohtml.pl (modified) (1 diff)

trunk/gsdl/bin/script/gsConvert.pl

@@ -60 +60 @@
     my (@ARGV) = @_;
     my ($input_type,$output_type,$verbose,$timeout);
-    
+
     $timeout = 0;
     # read command-line arguments
@@ -167 +167 @@
     # Attempt specialised conversion to HTML
     if (!$output_type || ($output_type =~ /html/i)) {
-    print STDERR "I am about to call doc_to_html...\n";
     $success = &doc_to_html($input_filename, $output_filestem);
     if ($success) {
@@ -280 +279 @@
     ($input_filename) = @_;
     
-    open(TMP, ">temp.txt");
-    binmode(TMP);
     open(CHK, "<$input_filename");
     binmode(CHK);
@@ -290 +287 @@
     
     $line = $_;
-    print TMP "$line\n\n";
+
     if ($first) {
         # check to see if this is an rtf file
@@ -494 +491 @@
     my $cmd = "ps2ascii \"$input_filename\" > \"$output_filestem.text\"";
     $cmd .= " 2> $output_filestem.err";
+    
     if (system($cmd)>0)
     {
@@ -499 +497 @@
     &util::rm("$output_filestem.text") if (-e "$output_filestem.text");
     &util::rm("$output_filestem.err") if (-e "$output_filestem.err");
-
-    # Fine then. We'll just do a lousy job by ourselves...
-    # Based on code nicked from:
-    # http://snark.ptc.spbu.ru/mail-archives/lout/brown/msg00003.html
-    # 
-    print STDERR "Attempting to strip text from postscript.\n";
-    my $errorcode=0;
-    open (IN, "$input_filename") 
-        ||  ($errorcode=1, warn "Couldn't read file: $!");
-    open (OUT, ">$output_filestem.text") 
-        ||  ($errorcode=1, warn "Couldn't write file: $!");
-    if ($errorcode) {print STDERR "errors\n";return 0;}
-
-    my $in_a_sentence=0;
-    while (<IN>) {
-        if (/^[^\(\)]+$/ && !$in_a_sentence) {next ;} # no brackets in line
-        # attempt to add whitespace between different lines...
-        s/F.?\(/\( /g; # this might break up some other words though...
-        ### remove all postscript control data
-        if (!$in_a_sentence) {
-        s/^[^\(\)]*?\(//;}  # rm start of line up to first open bracket
-        s/\\\(/\{/g;s/\\\)/\}/g ; # change quoted braces
-        s/\)([^\(\)])*?\(//g ;   # close bracket up to next open unquoted bracket
-        if (s/\)[^\(\)]*?$//g)  # last close bracket to end of line
-        {$in_a_sentence=0;chomp;}     
-        if (s/\\$//) # if line is a continuation
-        {$in_a_sentence=1;chomp;} 
-        s/^$//g ;             # remove empty lines
-        ### ligatures have special characters...
-        s/\\214/fi/g;
-        s/\\215/fl/g;
-        print OUT "$_";
-    }
-    close IN; close OUT;
-    }
+    return 0;
+    }
+
     &util::rm("$output_filestem.err") if (-e "$output_filestem.err");
     return 1;
@@ -574 +540 @@
     ($input_filename, $output_filestem) = @_;
 
-    #open(TEMP, ">temp.txt");
     open(IN, "<$input_filename");
     binmode(IN);
@@ -585 +550 @@
 
     # delete anything that isn't a printable character
-    #print TEMP $line;
     $line =~ s/[^\040-\176]+/\n/sg;
 

trunk/gsdl/bin/script/pdftohtml.pl

@@ -110 +110 @@
     else  {
     $p_home = &util::filename_cat($ENV{'GSDLHOME'}, "packages", "unix", "pdftohtml");
-    $pdftohtml = &util::filename_cat($p_home, "pdftohtml_0_22", "pdftohtml.bin");
+    $pdftohtml = &util::filename_cat($p_home, "bin", "pdftohtml.bin");
     }
     return 0 unless (-e "$pdftohtml");