Bug #81295 for Text-CSV_XS: escaped sep_char discarded when only item in unquoted field

Text::CSV_XS does not properly handled a escaped sep_char if it is the only item in a data field. i.e. if sep_char => ',' and escape_char => '\\' and data is 'testing,\,,fields' data will show up as 4 columns instead of 3. However, this is remedied if the escape_char is double used inside a column with only the sep_char in it i.e. if sep_char => ',' and escape_char => '\\' and data is 'testing,\\,,fields' data will show up properly as 3 columns. Example script below: #!/usr/bin/perl; use Text::CSV_XS; use Data::Dumper; my $csv = Text::CSV_XS->new({quote_char => undef, escape_char => '\\', sep_char => ','}); print "Properly escaped sep_char which should show up as 4 columns showing up as 5 columns:\n"; $csv->parse('hello,wo\\,rld,\\,,testing'); print Dumper($csv->fields())."\n"; $csv->parse('hello,wo\\,rld,\\\\,,testing'); print "Double escaped sep_char showing up correctly as 4 columns:\n"; print Dumper($csv->fields()); __END__ From testing it makes no difference what escape and sep chars are, the problem still exists.

I added a new attribute to make this possible without breaking backward compatibility: allow_unquoted_escape There is a backward compatability issue in that the escape character, when differing from the quotation character, cannot be on the first position of a field. e.g. with C<quote_char> equal to the default C<"> and C<escape_char> set to C<\>, this would be illegal: 1,\0,2 To overcome issues with backward compatibility, you can allow this by setting this attribute to 1.