Changeset 15839 for gsdl/trunk/bin/script/replace_srcdoc_with_html.pl

Timestamp:

2008-05-30T19:39:00+12:00 (16 years ago)

Author:

ak19

Message:

Corrected bug that failed on PDFs with funny characters. Instead of moving (copy & delete original) the html file created, it now reads the orig file's contents and ensures it is UTF8. Now replacing srcdocs with html in such cases succeeds, just as pdfs processed by PDFPlug.pm would.

File:

• gsdl/trunk/bin/script/replace_srcdoc_with_html.pl (modified) (1 diff)

gsdl/trunk/bin/script/replace_srcdoc_with_html.pl

@@ -187 +187 @@
     }
 
-    # Now we know we have no file name collisions
-    # use util to move it all to the import directory where the original file is located
-       #&util::mv($output_filename, &util::filename_cat($import_dir, $tmp_name.$ext)); # doesn't work
-       #&util::mv($output_filename, $import_dir); # doesn't work
-    &util::cp($output_filename, $import_dir);
+    # Now we know we have no file name collisions. We 'move' the html file by copying its
+    # contents over and ensuring that these contents are utf8. If we don't do this, PDFs
+    # replaced by html may fail, whereas those converted with PDFPlug with have succeeded.
+    open(FIN,"<$output_filename") or die "replace_srcdoc_with_html.pl: Unable to open $output_filename to ensure utf8...ERROR: $!\n";
+    my $html_contents;
+    {
+    local $/ = undef;        # Read entire file at once
+    $html_contents = <FIN>;  # Now file is read in as one single 'line'
+    unicode::ensure_utf8(\$html_contents); # turn any high bytes that aren't valid utf-8 into utf-8.
+    }
+    close(FIN); 
+
+    # write the utf8 contents to the new file and delete the original.
+    open(FOUT, ">$new_file") or die "replace_srcdoc_with_html.pl: Unable to open $new_file for writing out utf8 html...ERROR: $!\n";
+    print FOUT $html_contents;
+    close(FOUT);
     &util::rm($output_filename);
-    if(-e $assoc_folder) { # move any associated folders too
+    
+    # move any associated folders containing associated files too
+    if(-e $assoc_folder) { 
     #print STDERR "****Folder for associated files is $assoc_folder\n";
         #&util::mv($assoc_folder, $import_dir); # doesn't work for me