Converter Perl脚本修复

Question

I am trying to use a small Perl script in order to convert SQL INSERT statements from MySQL syntax into SQLite syntax. But the script has a bug and can't convert the string constants properly in some corner cases.

Here is the script:

#!/usr/bin/perl
while (<>){
    s/\\'/''/g;                # Use '' instead of \'
    s/\\"/"/g;                 # Use " instead of \"
    s/\\r\\n/\r\n/g;           # Convert escaped \r\n to literal
    s/\\\\/\\/g;               # Convert escaped \ to literal
    s/ auto_increment//g;      # Remove auto_increment
    s/^[UN]*?LOCK TABLES.*//g; # Remove locking statements
    print;
}

The problem is that in the case of the following MySQL string:

'It doesn\'t work :-\\'

It generates wrong SQLite string:

'It doesn''t work :-\''

instead of the proper:

'It doesn''t work :-\'

What is wrong in the script?

EDIT:

And how to fix it?

Answer 1

It's because you're processing \\' before \\\\ . You need to handle backslashes in the order they are encountered, so you need to do everything in one pass.

#!/usr/bin/perl

my %escapes = (
   n => "\n",
   r => "\r",
);

while (<>) {
    s{\\([\Wrn])}{ $escapes{$1} // ( $1 eq "'" ? "''" : $1 ) }eg;
    s/ auto_increment//g;
    next if /^(?:UN)?LOCK TABLES/;
    print;
}

Note the proper way of optionally matching UN .

Answer 2

The first substitution s/\\\\'/''/g; replace the last \\' by '' , so

'It doesn\'t work :-\\'
#                    ^^

becomes

'It doesn\'t work :-\''

I guess you want to replace \\' only when it is not followed by a non-word character \\W .

I'm using here only the 2 relevant substitutions for this case

my $str = <<'EOD';
insert into tbl values ('it doesn\'t work :-\\', 42, 33, 'and this doesn\'t work as well :-\\', 1024);
EOD
say "before: ",$str;
$str =~ s/\\'(?!\W)/''/g;
$str =~ s/\\\\/\\/g;
say "after: ",$str;

Output:

before: insert into tbl values ('it doesn\'t work :-\\', 42, 33, 'and this doesn\'t work as well :-\\', 1024);

after: insert into tbl values ('it doesn''t work :-\', 42, 33, 'and this doesn''t work as well :-\', 1024);