Changeset 1692 for trunk/gsdl

Timestamp:

2000-11-22T16:40:38+13:00 (24 years ago)

Author:

jrm21

Message:

pdftohtml can't handle encrypted pdfs, but doesn't return an error code,
so check if it made the .html file - if not, return 0

File:

• trunk/gsdl/bin/script/gsConvert.pl (modified) (7 diffs)

trunk/gsdl/bin/script/gsConvert.pl

@@ -49 +49 @@
 sub print_usage
 {
-    print STDERR "Usage: $0 [-type doc|pdf] [-output html|text] filename\n";
+    print STDERR "Usage: $0 [options] filename\n";
+    print STDERR "Options are:\n\t-type\tdoc|pdf\n\t-output\thtml|text\n";
+    print STDERR "\t-timeout\t<max cpu seconds>\n";
     exit(1);
 }
@@ -57 +59 @@
 {
     my (@ARGV) = @_;
-    my ($input_type,$output_type,$verbose);
-
+    my ($input_type,$output_type,$verbose,$timeout);
+
+    $timeout = 0;
     # read command-line arguments
     if (!parsargv::parse(\@ARGV,
              'type/(doc|pdf)/', \$input_type,
              'output/(html|text)/', \$output_type,
+             'timeout/\d+/0',\$timeout,
              'verbose/\d+/0',   \$verbose))
     {
@@ -323 +327 @@
     my $wvWare = &util::filename_cat($wv_home, "bin", "wvWare");
     return 0 unless (-e "$wvWare");
-    $cmd = "ulimit -t 40;";
+    $cmd = "";
+    if ($timeout) {$cmd = "ulimit -t $timeout;";}
     $cmd .= "$wvWare --charset utf-8 --config $wv_conf";
     $cmd .= " \"$input_filename\" > \"$output_filestem.html\" 2>\"$output_filestem.err\"";
@@ -365 +370 @@
     $r_cmd = "rtf2html" unless (-e "$r_cmd");
     return 0 unless (-e "$r_cmd");
-    $cmd = "ulimit -t 40;";
+    $cmd = "";
+    if ($timeout) {$cmd = "ulimit -t $timeout;";}
     $cmd .= "$r_cmd";
     $cmd .= " \"$input_filename\" > \"$output_filestem.html\" 2>\"$output_filestem.err\"";
@@ -403 +409 @@
     return 0 unless (-e "$pdftohtml");
 
-    $cmd = "ulimit -t 40;";
+    $cmd = "";
+    if ($timeout) {$cmd = "ulimit -t $timeout;";}
     $cmd .= "$pdftohtml -noframes";
     $cmd .= " \"$input_filename\" \"$output_filestem.html\"";
@@ -416 +423 @@
     }
 
+    # make sure the converter made something
+    if (! -e "$output_filestem.html")
+    {
+    &util::rm("$output_filestem.out") if (-e "$output_filestem.out");
+    # print out the converters std err, if any
+    if (-e "$output_filestem.err") {
+        open (ERRLOG, "$output_filestem.err") || die "$!";
+        print STDERR "pdftohtml:\n";
+        while (<ERRLOG>) {
+        print STDERR "$_";
+        }
+        close ERRLOG;
+    }
+    return 0;
+    }
+
     &util::rm("$output_filestem.out") if (-e "$output_filestem.out");
     return 1;
@@ -434 +457 @@
     &util::rm("$output_filestem.text") if (-e "$output_filestem.text");
     &util::rm("$output_filestem.err") if (-e "$output_filestem.err");
+    return 0;
+    }
+
+    # make sure the converter made something
+    if (! -e "$output_filestem.html")
+    {
+    &util::rm("$output_filestem.out") if (-e "$output_filestem.out");
+    # print out the converters std err, if any
+    if (-e "$output_filestem.err") {
+        open (ERRLOG, "$output_filestem.err") || die "$!";
+        print STDERR "pdftotext:\n";
+        while (<ERRLOG>) {
+        print STDERR "$_";
+        }
+        close ERRLOG;
+    }
     return 0;
     }